ABSTRACT DAILY ACTIVITY SEQUENCES AND TIME-SPACE CONSTRAINTS By John Dickson Stephens The principal aim of the research is to develop a methodology for investigating how individuals' decisions about their time-space behavior interrelate. What determines where, when, for how long, and in what sequence activities are performed? In attempting to understand more clearly the structuring of everyday behavior, the research adopts a methodology based on the time-space mechanics of constraints. The major research hypothesis is that the critical determinant of the structure of a person's day is the extent to which one feels con- strained relative to certain activities, times, and locations. The term structure is interpreted to mean not only the types and sequences of activities, but also their location and duration. In order to test the hypothesis it was necessary to isolate a sample population and to develop a survey instrument for obtaining the needed data. The sample was drawn from the faculty, staff, and student popu- lations of Michigan State University, located in East Lansing, Michigan. In order to study the time-space constraints operating on the formation of activity sequences, attention must be focused on the paths, or behavior, of the individual through time-space. The instrument chosen for collecting this type of information, the time-space budget, incor-} porates the sequence, linkage, timing, duration, and frequency of acti- vities, as well as the spatial and temporal coordinates of one's behavior. The time-space budget focuses on two related aspects: . people's overt behavior and the perceptions of their social and physical environment. This method of activity accounting is unique in that, as far as their activities for the sampled time span are concerned, indi- viduals are treated as totalities; their entire, unbroken sequences of activities in time-space are made available. A methodology was formulated which was built around the notion that subjective constraints are the critical determinants of one's behavior in time and space. The vehicle used to examine this hypothesis is that of simulation. Thus, an attempt was made to simulate the unknown variables--the timing, sequencing, and location of activities—-in terms of the parameters or known variables, the subjective constraints. As a first step toward the goal of developing a simulation model, an attempt was made to isolate that behavior which demonstrates recur- rent patterns in time-space. Hence, an initial hypothesis regarding the principle of consistency in human behavior was tested using the sample of time-space budgets. The claim was made that the concept of pattern could be applied to the set, or a subset, of time-space budgets and that this pattern is partly defined by the sequence in which activities are performed. Behavior patterns are also partly defined by activity dura- tions. The amount of time a person devotes to an activity follows a pattern over groups of individuals just as much as does the position it occupies in that person's sequence. Such simple patterns of activity are necessary conditions for the develOpment of a simulation model. In order to test this hypothesis, the time-space budget data for all_respondents in the university sample had to be compressed in a sys- tematic manner. Therefore, an algorithm was devised which groups indi- viduals on the basis of the amount of time which they spend on different activities and the order in which they are performed. Based upon the sequential behavior patterns of individuals, the algorithm decomposed the sample population into several groups each of which maximized within- group similarities and between—group differentials. As a result of the preliminary analysis of time-space budget data, it was concluded that patterned variations do exist between subgroups of the university popu- lation, where the concept of pattern is construed in terms of sequence and duration of daily activities. Can activities, their timing and location be associated with the pattern of time-space constraints throughout the day? This is the question that the simulation model addresses. In conjunction with this problem, it was hypothesized that time-space constraints and levels of activity commitment are the critical determinants of activity sequences in time-space. The first step in the computer simulation is to develop six cumula- tive probability distributions which are approximations of the activity structure for a given pattern group. A Monte Carlo procedure is then invoked to select activities and their spatial and temporal locations. The distributions are treated as discrete cumulative probability vectors, and uniform random numbers bounded by zero and one are drawn to deter- mine the activities, durations and locations. Thus, the time-space path of each individual's day is built up by assigning random numbers in accordance with the calculated probabilities, and by comparing num- bers drawn for every time period with these distributions. The structure of the model is necessarily tied to the conceptual framework of behavior and time-space constraints. Therefore, the simu- 1ation first establishes the major constraints in each person's day, and then, how one relates his behavior to these constraints. The findings of the modelling experiments demonstrate that the sub- jective constraints acting on the choice, timing, and location of acti- vities are modestly important in understanding the formation of activity sequences and the paths people follow through time-space. The statistical results of the model's output suggest that they are not as significant as they were hypothesized to be. Hence, it was concluded that for the sample population that subjective constraints alone are not the critical determinants which structure time-space paths at a daily scale. DAILY ACTIVITY SEQUENCES AND TIME-SPACE CONSTRAINTS By John Dickson Stephens A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Geography 1975 ACKNOWLEDGMENTS In the preparation of this dissertation, I have received assistance from many sources. Initially, I want to thank Barbara Stephens for the generous support that she gave throughout my doctoral program. I am especially indebted to my friend and advisor, Dr. Robert Thomas, whose continued assistance and encouragement have proved invaluable to the completion of the research. His critical analysis of the ideas, organization, and writing of this dissertation is greatly appreciated. Special mention must also be given to Dr. Robert Wittick who unselfishly gave of his time and skills toward the completion of this study. In addition, I was fortunate to have worked and exchanged ideas with my colleague, Brian Holly, who was engaged in related research. Special thanks are also due to Dr. Lawrence Sommers for making available to me the resources of the Department of Geography at Michigan State University. The University's Computer Institute for Social Science Research, under the directorship of Dr. Charles Wrigley, extended financial support during part of my doctoral program and, more importantly, provided a stimulating and congenial intel- lectual environment in which to work. ii TABLE OF LIST OF TABLES . . . . . . . . LIST OF FIGURES . . . . . . . . Chapter 1. INTRODUCTION . . . . . . TIME AND SOCIETY . . . Social Time and Space and the Freedom and Constraints . CONTENTS Individual TOWARD A TIME-SPACE FRAME OF REFERENCE . Interdependence of Time and Space . . . . Paths and Traces within the Time-Space Frame NATURE OF THE RESEARCH PROBLEM The Approach . . . . The Organization . . 2. TIME-BUDGETS OF HUMAN BEHAVIOR: OF SELECTED APPROACHES . THE TIME-BUDGET PERSPECTIVE . Deve10pment of Time-Budget Research . The Multinational Time-Budget Project . CURRENT THRUSTS IN URBAN AND REGIONAL ANALYSIS Time and Space . . Activity Conceptual System iii SYNTHESIS AND CRITIQUE Page vii ix 12 13 16 18 20 24 24 25 33 40 4O 43 Chapter 3. 4. Time-Use and Ecological Organization . . . . . Other Approaches to Time-Budget and Urban Activity ResearCh C O O O O O O O O O O O O I O O O O O O O TIME-SPACE PATHS OF HUMAN BEHAVIOR: CONCEPTUALIZATION AND IMPLEMENTATION OF THE RESEARCH STRATEGY . . . THE TIME-SPACE BUDGET PERSPECTIVE AND HUMAN GEOGRAPHY O O C O I O O O O O I O I O I O O O 0 Spatial Behavior, Spatial Structure and Location Theory 0 O C O O C O I I O O O O O O O O O 0 Choice- and Constraint-Oriented Approaches to Spatial Behavior . . . . . . . . . . . . . . . . CONCEPTUALIZATION OF HUMAN BEHAVIOR AND CONSTRAINTS . Time-Space Behavior: The Search for Assumptions Principle of Consistency and Recurrent Behavior Patterns . . . . . . . . . . . . . . . . . . . Priorities and Opportunities . . . . . . . . . The Interaction of Constraints: The Global View . Objective and Subjective Constraints on Time-Space Behavior . . . . . . . . . . . . . . . . . . . . . RESEARCH STRATEGY . . . . . . . . . . . . . . . . . . Methodology of the Time—Space Budget Diary . . . Survey Site . . . . . . . . . . . . . . . . . . Sampling Design . . . . . . . . . . . . . . Preliminary Survey . . . . . . . . . . Choice of Survey Format . . . . . . . . . . BEHAVIOR PATTERN RECOGNITION: EMPIRICAL ANALYSIS FOR MODEL DEVELOPMENT . . . . . . . . . . . . . . . . . SURVEY RESULTS . . . . . . . . . . . . . . . . . . DATA STRUCTURE . . . . . . . . . . . . . . Unit of Analysis . . . . . . . . . . . . . iv Page 56 77 88 88 91 97 104 104 107 109 110 113 117 117 122 127 131 139 147 147 149 149 Chapter Page Activity Modules . . . . . . . . . . . . . . . . . 153 ANALYTICAL PROCEDURES FOR THE RECOGNITION OF BEHAVIOR PATTERNS O I O O O O I O O O O I O O O O I C O O O O 1 5 3 Criteria for Pattern Recognition Analysis . . . . . 154 A Factor Analytic Model . . . . . . . . . . . . . . 155 Algorithmic Approach to Pattern Recognition Using a Non-Sequential Criterion . . . . . . . . . . . . 159 The Time-Space Map . . . . . . . . . . . . . . . . 163 Algorithmic Approach to Pattern Recognition Using a Sequential Criterion . . . . . . . . . . . . . . 168 Alternative Algorithmic Approach Using the L.A.W.S. MOdification O O O O O O O O I O O O I O O I O O O 177 RECOGNITION OF BEHAVIOR PATTERN GROUPS . . . . . . . 188 Termination Criteria for Group Formation . . . . . 188 Results of the Grouping Algorithm . . . . . . . . . 191 5. SIMULATION MODEL OF TIME-SPACE PATHS AND THE MECHANICS OF CONSTRAINTS O O O O O O O O O O O O O O O O O O O O 224 SOLUTION, MODEL-BUILDING AND SIMULATION . . . . . . . 225 DATA REQUIREMENTS . . . . . . . . . . . . . . . . . . 227 MODELLING OBJECTIVES, ASSUMPTIONS AND RULES . . . . . 230 Limiting Assumptions . . . . . . . . . . . . . . . 231 Procedural Rules. . . . . . . . . . . . . . . . . . 232 DESIGN OF THE SIMULATION MODEL AND EXPERIMENTS . . . 234 Model Design . . . . . . . . . . . . . . . . . . . 234 Verification . . . . . . . . . . . . . . . . . . . 243 Simulation Experiments . . . . . . . . . . . . . . 245 EVALUATION OF THE MODEL . . . . . . . . . . . . . . . 246 Chapter Correspondence Between Estimated and Observed Results . . . . . . . . . . . . . .'. . . . . Research Hypothesis . . . . . . . . . . . . . . 6. SUMMARY AND CONCLUSIONS . . . . . . . . . . . SUMMARY . . . . . . . . . . . . . . . . . . . . . CONCLUSIONS: RETROSPECT AND PROSPECT . . . . . . APPENDIX A: THE TIME-SPACE BUDGET DIARY . . . . . . . . . . APPENDIX B: THE SUPPLEMENTAL INTERVIEW SCHEDULE . . . . . . APPENDIX C: DATA CODEBOOK FOR INTERVIEW SCHEDULE AND TIME— SPACE BUDGET DIARY . . . . . . . . . . . . . . APPENDIX D: TWO-DIGIT ACTIVITY CODE: COLLAPSED FORM . . . APPENDIX E: SURVEY SAMPLE DESIGN AND RESPONSE RATES . APPENDIX F: THE COMPUTER SIMULATION PROGRAM . . . . . . . . APPENDIX G: SUMMARY OF SIGNIFICANCE TESTS . . . . . . . . . B I BLI OGWHY O O O O O O O O O O O O O O O O O O O O O O O 0 vi Page 247 267 272 272 280 288 293 300 317 321 323 352 357 10I 11. 12. 13. 14. 15. 16. 17. LIST OF TABLES TARGET SURVEY POPULATION . . . . . . . . . . . . . . . THE FOUR-PART SUBJECTIVE FIXITY QUESTION . . . . . . . RESPONDENTS AND NONRESPONDENTS BY UNIVERSITY AFFILIATION I I I I I I I I I I I I I I I I I I I I I I VECTOR OF INFORMATION DESCRIBING AN ACTIVITY MODULE . . SAMPLE SET OF OVERLAP RELATIONSHIPS . . . . . . . . . . REDUCED MATRICES FROM RECIPROCAL PAIRS ANALYSIS OF OVERLAP TOTALS IN TABLE 5 . . . . . . . . . . . . . . . THE RELATIONSHIPS (FROM TABLE 5) NOT ELIMINATED DURING RECIPROCAL PAIRS ANALYSIS . . . . . . . . . . . . . . . BACKGROUND VARIABLES CHOSEN FOR TESTS OF SIGNIFICANCE . GROUPING CYCLES ORDERED BY TERMINATION CRITERIA . . . . CHI-SQUARE STATISTICS FOR SIGNIFICANCE OF PATTERN GROUP MEMBERSHIP . . . . . . . . . . . . . . . . . . . TABLES OF OBSERVED AND EXPECTED FREQUENCIES FOR SELECTED VARIABLES OVER PATTERN GROUPS . . . . . . . . TOTAL AND OVERLAPPING GROUP MEMBERSHIP OF BEHAVIOR PATTERN GROUPS . . . . . . . . . . . . . . . . . . . . SUMMARY STATISTICS FOR BEHAVIOR PATTERN GROUPS . . . . . RATINGS OF ACTIVITY, TIME AND LOCATION FIXITY--GROUP I . RATINGS 0F ACTIVITY, TIME AND LOCATION FIXITY-- GROUP I I o o o o o o o o o o o o o o o o o o o o o o o 0 SUMMARY STATISTICS FOR CONSTRAINT AND COMMITMENT INDICES I I I I I I I I I I I I I I I I I I I I I I I I ACTUAL AND PREDICTED ACTIVITY SEQUENCES . . . . . . . . vii Page 130 141 148 148 179 181 183 190 192 193 195 198 198 213 217 219 251 Table 18. 19. 20. 21. CHI-SQUARE STATISTICS FOR PATTERN GROUPS . SIMULATED ACTIVITY ASSIGNMENTS BY LEVELS OF COWIMNT I I I I I I I I I I I I I I I I I SIMULATED ACTIVITY ASSIGNMENTS BY COMMITMENT AND FIXITY RATINGS--PATTERN GROUP I . . . . . . . SIMULATED ACTIVITY ASSIGNMENTS BY COMMITMENT AND FIXITY RATINGS—-PATTERN GROUP II . . . . . . . v MODEL PERFORMANCE BY CATEGORIES OF TIME-SPACE CONSTRAINT I I I I I I I I I I I I I I I I I THE TIME-SPACE BUDGET DIARY . . . . . . . . . TWO-DIGIT ACTIVITY CODE . . . . . . . . . . . SOCIAL CONTACTS CODES . . . . . . . . . . . . COLLAPSED CODING SYSTEM FOR SOCIAL CONTACTS . TWO-DIGIT ACTIVITY CODE: COLLAPSED FORM . . SURVEY SCHEDULE AND RESPONSE RATES . . . . . RESULTS OF SIGNIFICANCE TESTS FOR GROUP I . . RESULTS OF SIGNIFICANCE TESTS FOR GROUP II . viii Page 255 256 258 260 262 289 308 315 316 317 322 352 355 Figure 1. 10. 11. 12. 13. 14. 15. 16. 17. I8. 19. LIST OF FIGURES Page THE PATHS OF A GROUPS OF INDIVIDUALS IN A TIME- SPACE SYSTm I I I I I I I I I I I I I I I I I I I I I 15 TIME-BUDGET RESEARCH THROUGH THE MID-1960's . . . . . 27 TIME-BUDGET RESEARCH PERTAINING TO URBAN AND REGIONAL ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . 42 DEVELOPMENTS IN ACTIVITY SYSTEMS RESEARCH . . . . . . 54 AN INDIVIDUAL'S PATH IN TIME—SPACE . . . . . . . . . . 62 TIME-SPACE IN Two DIMENSIONS: THE TIME-SPACE PRISM (BUDGET SPACE) . . . . . . . . . . . . . . . . . . . . 62 TRANSPORT MODES AND THE DAILY PRISM . . . . . . . . . 64 FEASIBLE PATHS WITHIN THE TIME—SPACE PRISM . . . . . . 64 TIME-PATHS AND INTER-STATION MOVEMENTS . . . . . . . . 7o MUTUAL ADJUSTMENTS OF THE POPULATION AND ACTIVITY SYSTEMS AS SEEN FROM A DAILY PERSPECTIVE . . . . . . . 75 THREE DIMENSIONAL ARRAY OF ACTIVITY, TIME AND LOCAT I ON 0 o o o o o o o o o o o o o o o o o o o o o o 8 2 ILLUSTRATION OF CAPABILITY CONSTRAINTS . . . . . . . . 101 ILLUSTRATION OF COUPLING CONSTRAINTS . . . . . . . . . 101 ILLUSTRATION OF AUTHORITY CONSTRAINTS . . . . . . . . 101 FACETS OF THE BEHAVIORAL PROCESS . . . . . . . . . . . 111 SURVEY FORMATS TESTED IN PRELIMINARY SURVEY . . . . . 132 STRUCTURE OF TIME-SPACE DIARY INFORMATION . . . . . . 151 DIAGRAM OF ACTIVITY SEQUENCES . . . . . . . . . . . . 164 AGGREGATE TIME-SPACE MAP . . . . . . . . . . . . . . . 166 ix Figure 20. 21. 22. 23I 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. DISAGGREGATED TIME-SPACE MAP . . . . . . . . . . . UNIFORM SEQUENCING OF ACTIVITY VECTORS . . . . . . TYPAL ANALYSIS OF OVERLAP TOTALS IN TABLE 5 . RECIPROCAL PAIRS ANALYSIS OF OVERLAP TOTALS IN TABLE 5 I I I I I I I I I I I I I I I I I I I I I LAWS ANALYSIS OF OVERLAP TOTALS IN TABLE 5 . GROUP MEMBERSHIP CONFIGURATION OF DATA IN TABLE 12 TIME-SPACE MAP OF PATTERN GROUP I . TIME-SPACE MAP OF PATTERN GROUP II . . . . . . . . CONSTRAINT CONFIGURATION OF PATTERN GROUP I . . . . CONSTRAINT CONFIGURATION OF PATTERN GROUP II DISTRIBUTION OF ACTIVITY COMMITMENTS FOR PATTERN GROUP I I I I I I I I I I I I I I I I I I I I I I I DISTRIBUTION OF ACTIVITY COMMITMENTS FOR PATTERN GROUP I I I I I I I I I I I I I I I I I I I I I I I AGGREGATE TRIP LENGTH DISTRIBUTIONS FOR PATTERN GROUP S I I I I I I I I I I I I I I I I I I I I I I GENERALIZED PATTERN OF TIME-SPACE CONSTRAINTS . . DATA REQUIREMENTS OF THE SIMULATION MODEL . . . . FLOW CHART OF THE COMPUTER SIMULATION MODEL . . . SIMULATED TIME-SPACE MAP OF PATTERN GROUP I . SIMULATED TIME-SPACE MAP OF PATTERN GROUP II TIME-SPACE DIAGRAMS FOR PATTERN GROUP I . . . . . TIME-SPACE DIAGRAMS FOR PATTERN GROUP II . . SIMULATED TRIP LENGTH DISTRIBUTIONS FOR PATTERN GROUPS . . . . . . . . . . . . . . . . . . . . Page 167 172 179 181 183 198 201 202 205 206 208 209 210 222 229 235 248 249 263 265 266 CHAPTER 1 INTRODUCTION The industrialization of man's activities involved a most dramatic retiming of his whole way of life. One important change in the advancement of society from a simple to a complex state has been the growing need by man for accurate timepieces. Increasing specialization, more than anything else, demands this. Many years ago, Mullford observed that: The clock not the steam engine, is the key machine of the modern industrial age. For even phase of its development the clock is both the outstanding fact and the typical symbol of the machine ... the effect is pervasive and strict. It presides over the day from the hour of rising to the hour of rest ... when one thinks of time not as a sequence of experiences, but as a collection of hours, minutes, and seconds, the habits of adding time and saving time come into existence: it could be divided, it could be filled up, it could be expanded by the invention of labor-saving instru- mmts . .. the new medium of existence is abstract tin ... organic functions themselves were regu- lated by it, we ate not upon feeling hungry, but when the clock sanctioned it. As our societies have grown more complex, so have our cities and as the range of activities expands, the economic base of the city becomes more complicated and time disintegrates into smaller and smaller units.2 Some primitive societies have no word for time, for llais Mumford, Technics and Civilization (New York: Harcourt, Brace and Co., 1934), p. 14. 2Kevin Lynch, What Time is This' Place? (Caflaridge, Mass.: The M.I.T. Press, 1973), p. 64. 2 others the time it takes a pot of rice to boil is the basic unit of social time. The shortest unit of social time in the city, on the other hand, has been described as the period between a traffic light turning green and the sound of the car horn from behind.3 There can be little doubt that time is symbolic of the urban way of life,‘ and will become increasingly so. Cities are, above all, finely timed life spaces . TIME AND SOCIETY The individual in contemporary society is very aware of time. Time, as a pervasively awkward scarcity, is perhaps the main rival to mney in our society; in many situations, time is money. Time is not simply a quantity to be spent, saved, or squandered; we also have to "keep time," meet deadlines, and order our activities in particular sequences, simultaneously taking into account the amounts of time required to move from place to place. The property of time which distinguishes it from other commodities is that it comes in a unique, irreversible sequence as well as having an amount or duration. With the aid of clocks and calendars, the modern citizen organizes his daily life in accordance with deadlines over which he often has little or no control. Today formal time constraints are greater than ever before, and the resulting clock-watching obsession shows little sign of abating, despite continuing advances in time- and labor-saving technology. Indeed, these advances demand more 3W.E. Moore, Man, Time, and Society (New York: John Wiley 8 Sons, Inc., 1963), p.37. , 4Louis Wirth, "Urbanism as a Way of Life," American Sociological Review, 44 (1938), pp. 1-34. 3 accurate time-keeping. In a production-oriented society, the quest for scale economies, and the increased specialization and inter- dependencies have led to a greater spatial separation of some of the interdependent elements--in particular those which directly affect the ordinary citizen. 5 Social Time and Space and the Individual A.belief that has pervaded our thinking from the beginning of the industrial revolution through present post-industrial society is that ideas of "specialization" and "division of labor" ought to be applied without reservation to almost all spheres of human activity. As a result, a continually growing bureaucracy, a vast number of technical specialists, and powerful interest organizations attempt to regulate more and more narrowly defined sets of events, whether they be economic, social, or cultural activities. Consequently, individuals conducting their specialized roles have come to be treated in an increasingly piecemeal fashion. The growing division of labor and specialization, or increasing societal scale, have intensified the segregation of man's activities in time and space. This logic has resulted in bigger and bigger firms, buildings, institutions, and cities to contain them.6 5James Anderson, "Living in Urban Space Time," Architectural Basie, 41 (January, 1971), p. 41. . 6Melvin MS‘Webber, "The Urban Place and the Nonplace Urban Realm," Explorations into Urban Structure, eds. Melvin M. Webber, et a1. (Philadelphia: University of Pennsylvania Press, 1964), pp. 85- 86. As large metropolitan areas continue to grow even larger, they are simultaneously becoming the places at which the widest variety of . specialists offer the greatest variety of specialized services, thus further increasing their attractiveness to other specialists. This tremendous accumulation of specialized competence is on its way to creating an aggregate outcome which is much less than satisfactory. It is difficult to continue dividing up tasks systematically and still respect the facts that the human being is indivisible and that there are needs and wants of individuals which can never be satisfied with money transactions or professional mediation of one kind or another. It also tends to go unnoticed that things and processes in the real world do not only operate in the ways the experts think to be important. These events also interact in a variety of unexpected and unwanted configurations because of their coexistence in time and space. The limitations of time and space to accomdate things and events were intuitively clear in a less complicated, preindustrial society. In preindustrial society, inhabitants knew how their settle- ment functioned in terms of social relationships, working procedures, and land use. This was primarily because work and the conduct of everyday life was organized vertically.7 However, with the advent of industrialization and concomitant urbanization, society has under- gone a radical transformation which has been described as a transition 7Torsten Higerstrand and Sture Oberg, "Befolknings-fordel— ningen och dess f'érandringar," Urbaniseringen i SveriL: h: geografisk samhallsanalys. Bilagadel I till Balanserad regignal Ev‘ecklin , Statens offentligfiredningar, 1970: 14, bil. 1 (Stockholm: Esselte tryck, 1970), p. 2. from a vertically organized way of life to one with horizontal linkages.8 Industrialization outmoded the practice of self-sufficiency and brought an increased division of labor. With growing urbanization, work was subdivided into an increasing number of units which could only produce goods and services through collaboration. The complex interdependence of activities grew rapidly. The result was a society built on the principle of horizontal linkages which demand the constant mvement of material, people, and information among increasingly specialized work functions. The growing interdependencies show few signs of abating and are likely to be accentuated in the future. As things have been divided into smaller and smaller pieces of knowledge, responsibility, and action, the connections among events have become more and more invisible to today's policy makers and planners. Since so much is now unobservable, we have lost touch with the kind of direct information which is needed for intuitive under- standing and rational planning. Freedom and Constraints For some time now, the benefits of urbanization have been extended to an ever-growing preportion of the pOpulation. From among these benefits, the improvements in transportation and comunications media have enabled individuals to be relatively free of spatial constraints in establishing and maintaining their ties of association with others. However, this situation may have led us to believe that 8Ibid.; the ~aut'hvzrr's use of the term vertical linkages denotes independence or self-sufficiency while horizonal linkages signify interdependence. the loosening of local ties have given us an altogether free personal choice of events in which to engage. It is true that the assortment of possible continations, resulting in part from a shift of "place" cmmnunities to diverse commmities of "interest,"9 is greater than it ever was, but certain fundamental limitations still exist. Just as we can achieve only limited rationality in our day-to-day existence, so is only a limited freedom available when it comes to action. A vast nunber of situations in which the individual becomes involved are of a kind that he cannot escape or counter. Circumstances like availability of employment, housing, education, health care, recreation, and transportation are among environmental resources out- side inediate control by the average citizen. As Harvey10 has recently pointed out,» the manner in which urban . areas are structured and the ways in which urban activities are organized have important consequences for the distribution of real income among the urban population. Real income is construed not only as money income in a narrow sense, but also as access to environmental resources and Opportunities of all kinds. Given this point of view, one can conclude that an urban population, given its particular location and configuration, sends it inhabitants into sequences of events which, by their distributions 9Melvin M. Webber, , "Order in Diversity: Comunity Without PrOpinquity," Cities and Space: The Future Use of Urban Land, ed. Lowdon Wingo, Jr. (Baltimore: The Johns Hopkins University Press, 1963), p. 29. 10David W. Harvey, Social Justice and "the City (Baltimore: The Johns HOpkins University Press, 19737, pp. 53-54. For a similar view, see: R.E. Pahl, Whose City? And Other Essays on Sociolggy and Planning (London: Longman Group Ltd., 1970), pp. 215-225. 7 and characteristics, describe the performance of the system as a whole. An innovative effort to estimate performance in a comprehensive way has been suggested by Parka.” His emphasis is on measures Of urban time relevant to planning considerations. Through timing of cities he suggests that we will be able to develop "an urban time-use classification which will serve a more flexible and sensitive purpose than land use classification schemes, if only because the utility derived from spending time has much to do with our satisfaction rating "12 The transactions included in his for the quality of urban life. aggregate time-allocation accounts do not exactly measure what is under discussion here. From a cross-sectional perspective, they cannot show how roles become divided in the process and how events come to affect individuals in a longitudinal perspective. For the moment no effort seems tobe more inportant than to find a just distribution of benefits and sacrifices between all individuals. If so, we should give greater attention to individual behavior as an unbroken sequence of actions. A step in this direction is taken more clearly in the surveys collecting so-called "social indicators" of the quality of life.13 The growing interest in 11Donald Parkes, "Timing the City: A Theme for Urban Environ- mental Planning," Royal Australian Planning_Institute Journal, 11 (October, 1973), pp. 130-135. 13Richard L. Meier, "Human Time Allocation: A Basis for Social Accounts," igurnal of the American Institute of Planners, 25 (January, 1959). PP. 27-33. 8 time-budgets as a research tool is an indication of a further move toward a less aggregated and abstract form of observation.“ These trends in research are clear achievements. Empirical approaches of this kind, however, entail those disadvantages which follow measuransnt without theory. It is true that by observing the behavior of peeple, one learns something about their living conditions. But this information does not clearly distinguish what are wants and needs from what are various degrees of necessity. In the field of human geography, in particular, the problem with advances in this area is perhaps not so much that we have failed to understand man's needs and wants as that our geography is too incomplete to be able to catch the conditions which circumscribe man's actions. Behavior, in and of itself, does not fully reveal the underlying pattern of constraints which shapes the situations in which action occurs. And this also means that no good clues areprovided for how to reshape the living conditions, if that is the goal. I I Instead, we must look to latent structure and latent processes to find the clues. Purposeful changes in the distribution of risks and opportunities among individuals requires an understanding of how constraints interact and how choice potential is affected by changes to one or more of those which can be influenced at all. If it is 1"See, for exnple: Richard L. Meier, A Commnications Theog Of‘Urban‘GrWth (Calflaridge, Mass.: The M.I.T. Press, 1962), pp. 45- 59; and, William Michelson, "Time Budgets in Environmental Research: Some Introductory Considerations," Environmental Designjes earch: Volume No, Smosia and Workshops-~Fourth International EDRA Confer- e ed. Wolfgang F.E. Preiser (Stroudsburg, Pa.: Dowden, Hutchin- enc , son 8 Ross, Inc., 1973), pp. 262-267. 9 assumed that people survive in "niches" of possible actions, then the shapes and volumes of these niches are more fundamental objects of researdh than actual behavior at a particular point in time.15 Actual behavior at a given moment is only a subset from the universe of permitted events. The pattern of niches as formed by constraints in Operation describes something different--a map of potmtial events. The notion of potentials contains a.more general measure of perfor- mance.16 The question of how to go about finding the map remains. TOWARD A.TIME-SPACE FRAME OF REFERENCE In order to get at this issue, consider first geography's dominant instrument of recording, the map. We require that observa- tions be rendered graphically on maps. This habit is perhaps more risky today than it once was, because we have started to take the map much more seriously. Earlier the map was used primarily as a means of general orientation and as an aid to discussion. Our private field experience has helped us to extract much more synthetic information intuitively than was directly displayed symbol by symbol. wa the symbols are the direct basis for serious and precise measure- ment. To a considerable extent the recent quantitative analysis in geography represents an in-depth study of the patterns of points, 15Torsten Hagerstrand, "The Impact of Social Organization and Environment upon the Time-use of Individuals and Households," Plan: Tidskrift fUr planering av landsbygd och tutorter, 26 (1972), p. 27. 16T'orsten Hugerstrand, "What About People in Regional Science?" ngers of the Regional Science Association, 24 (1970), p. 11. 10 lines, areas, and surfaces depicted on maps of some sort or defined by co-ordinates in a two- or three-dimensional space. The real significance of this development has been the strengthing of our feelings regarding the value of the geometrical outlook. The danger lies in its influence on how'we come to view the relation between map symbols and realdworld phenomena.17 Quite possibly, the map's limitations have forced us to consider only certain classes of spatial phenomena for investigation while neglecting others. This is to suggest that perhaps, we have been so interested in the distributional arrangements of things that we have tended to overlook the space-consuming properties of phenomena and the consequences for their ordering which these properties imply. The map is a poor instrument for depicting packing problems,18 except in the simpler cases (e.g., land use). The map has to be supplemented with some more abstract spatial notation before we can deveIOp an ' understanding of these processes. 17Robert D. Sack, "The Spatial Separatist Theme in Geography," Economic Geogggpgz, 50 (January, 1974), p. l. 18The term packing refers not only to the space-consuming pro- perties of objects, but also to the process whereby Objects compete for space. In discussing the issue of packing, Hagerstrand states that "as soon as one object has found a location, the space it occupies is not available for a host of other 'weaker' objects and the probabi- lity field of their location has changed. Of course, objects are some- times more or less closed and elastic relative to each other, and this makes the packing process very complex. Whatever the complexities are, however, packing adds a meaning to interaction which is different from what we usually have in mind. Many objects interact just because they have come to be located adjacent to each other and for no other reason. This very kind of interaction can have great significance both for the structure of an area and for the sequence of events occurring there." See: Torsten Hagerstrand, "The Domain of Human Geography," Directions in Geogggpgz, ed. R.J. Charley (London: Methuen & Co., 1973). p. 71. 11 This observation leads to another feature of the traditional geographical format, namely, a noticeable emphasis is laid upon the thin spatial cross-sectional view of the flow of terrestrial events. Indeed, the widespread dissatisfaction with existing geographic theories may be due to a preoccupation with spatial patterns and a neglect of small-scale generating processes over time.19 Of course, various efforts have been made to include a time-perspective of some sort, and historical geography in its most ambitious form tries to reconstruct the geography of selected dates in the past. Change, then, is visualized as a strip of film where the sequence of frames provides some idea of long term trends. Those concerned with process approaches to the analysis of spatial behavior often proceed in the same fashion, only the number of time intervals is considerably reduced.20 That is, researchers commonly map some index of change, calculated between two points in time. Mbdels for making spatial extrapolation into future time have been developed21 but, on the whole, researchers have not yet succeeded in handling events as located and 19Gunnar Olsson, "Inference Problems in Locational Analysis," Behavioral Problems in Geography: A Symposium, Studies in Geography No. 17, eds. Kevin R. Cox and Reginald G. Golledge (EvanSton, 111.: Department of Geography, Northwestern University, 1969), p. 21. 20For a more detailed review of process approaches to the study of spatial behavior, see: Reginald G. Golledge, Process Approaches to the Study of Human Spgtial Behavior, Department of Geography, Discussion Paper No. 16 (Columbus, Ohio: The Ohio State University Press, 1970). 21Michael Chisholm, Allan E. Frey, and Peter Haggett, eds., Rggional Forecasting, Proceedings of the Twenty-second Symposium of the Colston Research Society (London: Butterworth 8 Co. Ltd., 1971). l! 12 and connected within a compact time-space block. Again, to overcome this problem.requires a new type of notation. Interdependence of Time and Space To obtain a proper perspective on the quality of life in modern cities, it helps to consider time as well as space; better still, consider them simultaneously, thinking of peoples' activities and their environments in terms of locations in time-space. Time and space are.among man's most fundamental resources, and they are also among the most difficult to study due to their scarcity characteristics. Not the least of the problems is their unique pattern of interdependence and interaction. Time and space combine to form a four-dimensional "volume" which continuously surrounds us, and these dimensions also provide a co-ordinate reference system which we use to organize our own activities, and which can be used to describe the environments and the activities of others.22 Time and space are interrelated dimen- sions of human behavior because individuals cannot consume one without the other.23 In a sense, movement in either case must be continuous, since one must experience every point in time sequentially, and one must pass through a series of points in space in order of their conti- guity. Alternatively, the continuous function of time is more ZZHngrstrand, "The Domain of Human Geography,‘ op. cit., p. 77. 2.3In a sense space and time are independent of eadh other because while we continually consume both, the allocatable time is fixed for everyone, but the space available varies. Time, therefore, as distinguished from.space, comes in a unique, irreversible sequence having an amount or duration. Movement in time can only be in one direction, whereas movement in space may proceed in any of an infinite number of directions. ‘1 In: 13 demanding of the individual than the more discrete function of space. Although an individual is required to experience every point on the time-scale, he need only be at some point in the physical and social environment that provides him.with the minimum requisites for survival. Given either interpretation, it is hardly reasonable to assume that locations in space can effectively be separated from the flow of time. Paths and Traces within the Time-space Frame A promising way to overcome the problems discussed above has been suggested by Hagerstrand.24 He recommends translating all necessary concepts, old or new, into a strictly "physical" language. In this procedure, it is essential that the unity of time and space be fully respected. The use of a single spatial-temporal system to identify the particulars of human discourse is fundamental. With regard to this matter, Sack comments: The single physical time-space system makes it possible to identify and individuate events that are separate from any single observer and to communicate, discuss, or make public, observations that are not immediately apparent to the senses by everyone. Clearly, these functions cannot be provided by a plurality of physical spaces, for that would lead to confusion of identification and perhaps ultimately to solipsism. Only one system of physical space and time can provide the function of identification and individuation of events.25 2('Hligerstrand, "The Domain of Human Geography,’ op. cit., pp 0 77-790 25Robert D. Sack, "A Concept of Physical Space in Geography," Geographical Analysis, 5 (January, 1972), pp. 25-26. The use of a system of physical space and time is also treated by Sack in: Sack, "The Spatial Separatist Theme in Geography," op. cit., p. 17. 14 A physical time-space system is illustrated in Figure 1. For purposes of illustration, three-dimensional space has been collapsed into a two-dimensional map, leaving the third dimension to represent time. This procedure makes it possible to incorporate time and space into one unified geometrical time-space picture with full continuity in the time direction. This also means that form and process would not seem to be so essentially different as they appear today. Process takes shape as four-dimensional form. To man, time and space are not only dimensions for viewing and analyzing the location of events, they are also, in a very real sense, scarCe resources. This makes the time-space outlook fundamental. It is in this context that the "packing problem" reaches its full weight and it is here that the importance of not omitting any portion of man's time and surrounding space becomes critiCal. In reality one must always keep both dimensions in mind, for to a certain extent they are interchangeable. FUrthermore, in the conduct of human affairs, time often makes itself felt as a more demanding dimension than do spatial donstraints. The acting individual, therefore, describes an unbroken path in time-space (Figure 1). It is easy to imagine how the total popula- tion of interacting individuals within the block form a network of unbroken time-space paths, but with the fundamental prOperty that identity is retained. Similarly, the time-space traces of elements of known human importance in the environment, such as other organisms, tools, buildings, materials, can be depicted. The scenario makes it imediately clear that compatability among individuals, buildings, 15 mud SPACE FIGURE 1 THE PATHS OF A GROUP OF INDIVIDUALS IN I A TIME-SPACE SYSTEM* *The numbers 1, 2, 3, and 4 indicate different kinds of activity bundles which are locations in time-space where people come together in order to exchange information, transact business, or perform a common activity. 1 is connected with some fixed installation; 2 and 3 are meet- ings with different durations. All three types of bundles require move- ments, before and after, in order to be formed. 4 represents a telephone call which does not require movement but, nevertheless, ties two individuals together into one activity. (After Hagerstrand, "A Socio-environmental Web Model," Op. cit., p. 24). 16 tools, materials, and signals requires both (temporal) synchronization and (spatial) synchorization. Seen as a whole, the system exhibits a skeleton of meeting times and meeting places, much of which is frequently built up in advance, in order to fit activities to the timetables of other people, facilities, and institutions. This is a complex situation even in a space-free timetable sense. Even if members of a population do nothing more than exchange messages by telecomnication media (so that transportation is practi- cally instantaneous), the indivisibility of the human being is a severe constraint on what can occur. As soon as a comunicating group has come into being, the duration of its activity inevitably creates waiting-times among those who want to come into contact with one or more members of the group. The conflicts that arise will become a more obvious difficulty as we move toward a society in which the processing of information develops into the main activity.26 NATURE OF THE RESEARCH PROBLEM When trying to disentangle the seeming chaos of paths and traces inside the time-space block, several modes of investigation come to mind.27 In attempting to understand more clearly the structuring of 2 6Webber, "Order in Diversity," op. cit., pp. 42-44. 27For various approaches, see: James Anderson, "Space-time Budgets and Activity Studies in Urban Geography and Flaming," Environment and Planning, 3 (1971), pp. 353-368; Ian G. Cullen, "Space, Time, and the Disruption of Behavior in Cities," Environment and Planning, 4 (1972), pp. 459-470; and, Alan G. Wilson, "Some Recent Deve10pments in Micro-economic Approaches to lbdelling Household Behavior, with Special Reference to Spatio-temporal Organization," Papers in Urban and Regignal Analysis, Alan G. Wilson (London: Pion Ltd., 1972), pp. 216-236. These and other alternative approaches will be covered in more detail in the following chapter. 17 everyday behavior, this research adOpts a methodology that can be described as the time-space mechanics of constraints.2 Conceptually, the behavior, or activities, of an individual can be viewed as a series of discrete episodes occurring in a sequence through some specified period of time. These episodes have meaning at different temporal scales. Mbst relate to a day's activities, such as work, housekeeping, or eating, but some have meaning in the weekly time scale, such as visiting with relatives, in the yearly time scale, such as taking a holiday, or in a genera- tional time scale, such as migration.29 The activity sequence is seen to be a continuous stream of events where days flow into weeks, weeks into months, seasons into years, and years into a lifetime. FUrther, the events in this sequence are seen to possess an essential order in the sense that certain types of events occur in fairly predictable cycles or routines and take place in a space of fairly predictable locus. Activities, therefore, occur in a time-space continuum: there are temporal regularities inherent in Spatial regularities, and temporal rhythms obviously vary over space. 28Hngrstrand, "What About People in Regional Science?" Op. cit., p. 11; see also: Torsten Hngrstrand. "A Socio-environmental Web Mbdel," Studier i planeringsmetodik, ed. GOsta A. Eriksson, IMemorandum.fr3n ekonomisk-geografiska institutionen, nr. 9 (Abo [Turku], Finland: Handelshbgskolan vid Abo akademi [Abo Swedish university School of Economics], 1969), pp. 19-28. 29Torsten Hngrstrand, "On the Definition of Migration," VHestUntutkimuksen vuosikirja (Yearbook of POpulation Research in ‘Finland, 11 (1969), p. 65. 18 This research addresses itself to the general question: How do an individual's decisions about his time-space behavior inter- relate? In striving to answer such a question, practical as well as theoretical censiderations must be taken into account. At a very basic level, the strategy of this research is based upon a simple belief: if one is to improve the lot of the individual living and working in the city, than one essential element is a study of the behavior of that individual, seen as an unbroken sequence of actions. The study of this behavior, through time and space, in its normal metrOpolitan context cannot be avoided if one's intention is to shed some light upon potentially dysfunctional relationships between the two. No alternative approach will reveal the ways in which the context structures, constrains, or even frustrates the behavior which it is intended to facilitate. The_Approach Two general approaches to the study of time-space behavior can be identified. One orientation, clearly the more common, focuses on choices of activity that are manifest in behavior. The second approach, and the one adapted here, focuses on the constraints which circumscribe man's actions. Since choices can only be realized within the context of constraints, the latter approach is more general and more likely to lead to a discovery of how the decisions regarding one's time-space behavior interrelate. Two broad groups of constraints are prOposed. Objective constraints constitute the first group and are those which are imposed by the environment. It is this group that tends to shape the "niches" 19 of possible action Open to an individual. The pattern of niches formed by objective constraints describes a map of potential events. These potential events can be more narrowly defined by a second group, the subjective constraints, which are more or less self- imposed. Such constraints occur as a result of the individual's physical and social environment. In most contexts, a given environ- mental situation is subjectively perceived by different people as constraining choice to varying degrees. Although the peculiarities of the environmental context are important, an individual's reaction to that situation is crucial. Any situation to which an individual reacts has at least three dimensions: a temporal position, a location in space, and a potential activities set. These dimensions interact to produce a feeling of constraint upon the individual. The basic proposition of this research is that the critical determinant of the structure of a person's day, given environmental forces, is the extent one feels constrained relative to certain activities, times, and locations. These points at which one feels most constrained and around which one's day tends to be organized will be used as the basis for modelling experiments. The principal aim of the present inquiry is to develop a methodo- logy for investigating just how an individual's decisions about his time-space behavior interrelate. The approach adopted is based on the time-space mechanics of constraints. The adequacy of the methodology ‘wiJJ.be judged against its ability to replicate observed behavior. For the purposes of this research, it is assumed that one goal of geography, and social science in general, is to construct mechanisms sufficient to produce the observed results. These mechanisms take the form of 20 constraints which determine the structuring of paths of behavior in time-space. As noted earlier, an alternative approach is based on choice-mechanisms which manifest themselves in actual behavior. But, by observing the actual behavior of members of avapulation in a given period of time, this approach risks becoming lost in a descrip- tion of how aggregate behavior deveIOps as a sum total of individual behavior, without revealing the really critical determinants of behavior in time-space. What we "see" may be only a very small portion of the phenomena and, thus, behavior does not fully unveil the underlying pattern of constraints which shapes the situations in which action takes place. We must go well beyond a description of overt behavior to answer the question posed above. We must look for latent structure and latent processes in human behavior, or the hidden mechanics of constraints, to find the clues. Therefore, it seems more promising to try to define, using simulation procedures,30 the time- space mechanics of constraints which determine how paths are built-up, coordinated, or constrained. The Organization The primary objective of this inquiry is to report on a research and modelling strategy that attempts to reveal the processes or rules ‘which govern the integration of the individual's time-space behavior. The remainder of this inquiry is devoted to five main tOpics: (a) Timeebudgets of Human Behavior: A Synthesis and Critique of Selected Approaches. 3oHagerstrand, "What About PeOple in Regional Science?" op. cit. 21 In this section the development of the time-budget approach to the study of behavior is reviewed. The purpose is, on the one hand, to comment on the applicability of time-budgets other than the one which is developed and used here and, on the other hand, to reformulate the conceptual foundation within which this research tool can be considered. (b) Time-space Paths of Human Behavior: Conceptualization and Implementation of the Research Strategy. Building on the synthesis and critique of time-budget use in urban and regional analysis, this section formulates a conceptualiza- tion and research prOpositions designed to investigate the structuring of daily behavior through the time-space mechanics of constraints. In order to substantiate or illuminate the research prOpositions, a research strategy is established and involves the design and implementation of a survey. (c) Empirical Analysis for Model Deve10pment. . The third section involves the application of several analytical techniques to the survey data. Many modes of analysis apprOpriate for time-space budget data exist and.are briefly reviewed. However, only those methods which are intimately related to the ultimate goal of developing a simulation model will be treated here. Employing data on the daily activities of members of Michigan State University, East Lansing, Michigan, these analytical procedures will be employed in the evaluation of specific hypotheses about time—space behavior. (d) Simulation Model of Time-Space Paths and the Mechanics of Socio-environmental Constraints. F‘ 1‘ § 22 Given the results of the preceding section, a simulation model is formulated for determining the time-space paths of individuals in an urban environment. Through the use of simulation procedures, it is recognized that the full complexity of individual paths in time- space cannot be reproduced. Rather, it is intended as a deductive device for partial verification and as a mechanism for controlled observation of constraints interacting in the context of daily behavior. The findings of the simulation modelling experiments will -then be evaluated in relation to the major research propositions. (e) Summary and Conclusions: Retrospect and Prospect. This section is divided into two parts: (1) a summary of the methodology and the main conclusions about the validity of the model- ling strategy that attempts to reveal the processes which govern the integration of the individual's time-space behavior and (2) a dis- cussion of possible lines of further inquiry. i In this introductory chapter, the intention has been to explicate the motivation and purpose of the following study. Detailed arguments for the rationale of the viewpoint have been omitted. To have in- cluded these would have involved some reasonably abstruse arguments from the philosophy of science--for instance, the relationshipof description, explanation, and prediction,31 the implications of 3¥May Brodbeck, "Explanation, Prediction, and Imperfect Know- ledge," Readings in the PhilosOphy of the Social Sciences, ed. May Brodbeck (Toronto: The Macmillan Co., 1968), pp. 363-397; Abraham IKaplan, The Conduct of Inquiry (Scranton, Pa.: Chandler Publishing 60’, 1964), pp. 327-3690 23 determinism and probability,32 and the individual in scientific inquiry and the reductionist viewpoint.33 As this study is not the proper place for these comments, the reader is referred to the more detailed treatments in the bibliography. The purpose of the present inquiry is simply to develOp a conceptual framework for the study of the integration of an individual's time-space behavior; the philosOphic issues serve only as a background against which to view these efforts. 32Gustav Bergmann, "Imperfect Knowledge," Readings in the Philosophy of the Social Sciences, ed. May Brodbeck (Toronto: The Macmillan Co., 1968), pp. 413-436. 33M1.B. Turner, Philosophy and_the Science of Behavior (New York: Appleton-Century-Crofts, 1967), pp. 301-374; May Brodbeck, "MethodolOgical Individualisms: Definition and Reduction," Readings in the Philosophy of the Social Sciences, ed. May Brodbeck (Toronto: The Macmillan Co., 1968), pp. 280-303. CHAPTER 2 TIME-BUDGETS OF HUMAN BEHAVIOR: SYNTHESIS AND CRITIQUE OF SELECTED APPROACHES Many interesting patterns of social life are associated with the temporal distribution of human activities, with regularities in their timing, duration, frequency, and sequential order. Certain techniques of data collection based on direct observation, inter- viewing, and the examination of records permit the establishment of. itemized and measured accounts of how people spend their time within the bounds of a working day, week-end, a seven-day week, or any other time period. Investigations of this particular aspectJof social life based on the quantitative analysis of such accounts, have produced a lineage of research commonly referred to as "time-budget studies." The purpose of this chapter is to briefly review the deve10p- ment of the time-budget approach to the study of human behavior. The objective is, on the one hand, to comment on the applicability of time- budgets other than the one which is developed and used in this inquiry and on the other hand, to review selected uses of this research todl in urban and regional analysis. THE TIME-BUDGET PERSPECTIVE The term "time-budget" signifies an accounting scheme describing the allocation of time to activities during a given period--how many hours and minutes people spend daily on chores and past-times such as doing work, shopping, eating meals, socializing, leisure-time pursuits, 24 25 and sleeping. The time-budget investigation is somewhat similar to the procedure by which the allocation of funds for different purposes in financial budgets is analyzed. As far as personal or household budgets are concerned, the similarity will even extend to many specific types of expenditure, because a great number of everyday activities involve not only the expenditure of time but also money.1 At this point, however, the resemblance comes to an end. Time can only be spent, not "earned" therefore, time-budgets have no income side. Due to the fixed limits of time, everybody disposes equally of the 24 hours of the day. The fund of time which is being allocated to various activities (24 hours in daily time-budgets) serves simply as a frame of reference for setting out the temporal proportions of peOple's engagement in the whole gamut of their daily activities. Thus, it is not time itself, either as a physical or as a subjectively perceived entity, but rather the use people make of their time which is the real subject of time- budget studies. Development of Time-Budget Research The time-budget approach was first developed in social surveys reporting on the living conditions of the working class. The long working hours characteristic of early industrial development and the struggle which organized labor led from its very beginning for the shortening of the working day, make it fully understandable that the 1Historically, the time-budget concept is apparently derived from the practice of maintaining a set of household income and expendi- ture accounts which dates to the nineteenth century. It was a simple step to move from money accounting to time accounting with the movement to the hourly wage as the predominant kind of income workers received. 26 preportions of work and leisure in the daily life of laborers became a matter of considerable public concern in all countries where industrialization was in progress.2 The famous "3 x 8" password of labor movements around the turn of the century, claiming eight hours of sleep, eight hours of work, and eight hours of recreation as a rightful daily schedule of all laborers, expressed a social demand in the form of a concise time-budget. Just about then, chronometric time- and-motion analyses were introduced into industrial practice by the pioneers of "scientific management".3 This also meant time-budgeting of a sort, by setting up precise accounts of the amounts of paid time spend by workers on all kinds of "necessary" or "wasteful" activities during their work in the factory. The bulk of time-budget studies published before world War II originated in Great Britain, the Soviet Union, and the United States (see Figure 2).4 In general, these earlier studies focused on the following topics: 1. the share that such broad categories of activity like paid work, housework, personal care, family tasks, sleep and recreation have in the daily, weekly, or yearly time-budget of the pOpulation; 2. characteristic time expenditures of certain social groups or strata (e.g., industrial workers, farm homemakers, college students, unemployed men) on more or less specified types of everyday activities; 2Alexander Szalai, "Trends in Comparative Time-Budget Research," The American Behavioral Scientist, 9 (May, 1966), p. 3. albido’ pp. 3". 4Ibid., pp. 5-7. vb-‘ \‘(I\ \\\.,.Wxst\\.\ 27 ToomfiJ—Hx 8.5. 58% 53v”? akgbmtg .N gr.» Ihuwaomn. Iomdmmwm hmogmamgh ._863... Agugh Unssuu 4383.53 A.u.m.m.uv zuafisgm a figv «a Rum: 2H 3 Hana v.35“: Tiny No.3 8 kg Annamuud 9.1.4 3cm :1: 73.“: an»: e: gnaw 32. 8 55.8: 1‘5: Emu.— 4 . Ammunahm SEA! 2 2.: A.u.uv §r~hu;)ect," op. cit. [Appendix] 38 A common proposal has been the longitudinal analysis of data on time use as a means of gauging social change.” Szalai compares the time-budgets of workers in Moscow in 1924 and 1959.30 Robinson attempts to ascertain changes in the use of time in the United States, using the Lundberg data on Westchester County and the Sorokin-Berger data on Boston, both collected in the 1930's, a Ward-Mutual Broad- casting Company national survey from 1954, and the Jackson, Michigan and national sample data obtained in 1965-66 as part of the Multi- national Project. All of the conclusions are only suggestive, however, as there are serious problems with the populations surveyed, the sampling procedures, and the non-comparability of the various activity- classification schemes . 31 Szalai has developed a measure of the relative "compressibility" of different activities when the amount of available time is reduced. The concept is analogous to elasticity in economics. For example, workers with less free time because of a longer journey-to-work will devote considerably less time to some leisure pursuit, but virtually the same amount of time sleeping as other workers. Thus, the g 29Richard L. Meier, "Human Time Allocation: A Basis for SOC-131 Accounts," Journal of the American Institute of Planners, 25 January, 1959), p. 27. 30Szalai, "Trends in Comparative Time-Budget Research," op. Cite, p. 40 J 31John P. Robinson, '..social Change as Measured by Time-Budgets," —°u\rnal of Leisure Research, 1 (Winter, 1969), pp. 75-77, t1 hu.\\.a\..d :0. \u (7.. saw ......\..u. 39 leisure-time activity is more compressible than- sleeping.32 The Hungarian microcensus includes this type of analysis.33 Converse employs a slightly more sophisticated approach in comparing the use of time among nations. A Euclidean measure of "distance" between time-budget profiles, developed by Szalai, gives measures of intergroup differences. Guttman-Lingo smallest—space analysis is then used to determine the primary dimensions of difference among nations. The results are fascinating, since, in the two dimensions, the nations cluster geographically. Unfortunately, Converse does not suggest the value of this approach.“ Soviet researchers are focusing attention not on individual time-budgets but on the aggregate time balance of a region. For economic planning, they argue the primary consideration is the total amount of time spent by everyone on a given activity. Plans can then be devised for the most efficient use of time.35 The Bulgarians take this approach for the forecasting of time use in the future.36 _ 32 A Alexander Szalai, "Differential Work and Leisure Time-Budgets 88 a Basis for Inter-Cultural Comparisons," New Hungarian Quarterly, 5 (Winter, 1964), pp. 105-119. 33Hungarian Central Statistical Office, op. cit. U 3"Phillip E. Converse, "Country Differences in Time Use," The Wee. of Time: Daily Activities of Urban and Suburban POpulations in ‘1 elve Countries, ed. Alexander Szalai (The Hague: Mouton & Co., 972 3, pp. 145-177. £0 35V.D. Patrushev, "Aggregate Time-balances and Their Meaning Ur: Socio-economic Planning," The Use of Time: Daily Activities of \8 an and Suburban Pogulations in Nelve Countries, ed. Alexander zalai (The Hagu : Mouton & Co., 1972), pp. 429-440. 36 Zahari Staikov, "Time Budgets and Technological Progress," fie Use of Time: Dail Activities of Urban and Suburban Po ulations 197:"elve Countries, ed. Alexander Szalai The Hague: Mouton & Co., ), ppe 461.482e 0! 3‘ nun. \Ir ‘a as q ‘ e I. a\~s‘m\ lit h‘i» 40 Virtually all of the analyses of time-budget data consider either the freguency in which peOple engage in different activities,‘ or the duration of those activities. No one has. successfully solved the very difficult methodological problems involved in looking at other aspects of time use such as the sequencing of activities. Another problem which underlies much of the analytical work is the seeming lack Of direction in much of the research. Time seems important, so time- budget surveys are conducted and attempts are made to analyze the data. Since so many of the researchers have little notion of exactly what they want to find out, the analyses often appear diffuse and Purposeless . CURRENT THRUSTS IN URBAN AND REGIONAL ANALYSIS The preceding review of the time-budget concept serves as an introduction. Out of these roots there has arisen an interest on. the Part of urban and regional researchers in the use of time-budget analysis in both theoretical and planning applications.37 33‘1“ and Space The primary difference between the time-budget studies discussed to this point and the work to be discussed in geographic and planning researchcenters on the fact that a spatial dimension is explicitly \ 37For reviews of time-budget analysis in both theoretical and planning applications, see: Gutenschwager, 0p. cit., pp. 381-386; hevor MacMurray, "Aspects of Time and the Study of Activity Patterns," Wing‘geview, 45 (April, 1971), pp. 195-209; and, F. Stuart chapin, Jr., The Use of Time-Budgets in the Study of Urban Living tattems," Research Previgws (The University of North Carolina), 13 Ii<>vember, 1966), pp. 1-7. 41 included in planning and spatial analyses while, heretofore, the spatial concerns have been submerged. Hence, the planning researcher sees the activity choice occurring within the dimensions of time and space, and carries his time-budget survey beyond a concern for the when, what, and with whom to also include the "where" of activity choice. During the period immediately following the. Multinational Survey, there occurred a noticeable increase in the use of time-budgets and their logical extension, the "time-space budget."38 The period can be characterized by an ever-increasing saphistication in conceptual aPproa’ches to time-budget analysis. A preoccupation with accounts of human time allocation as ends in themselves has passed, to be replaced by a search for theory of human behavior in time and space. Human behavior, though highly variable, does display some marked temporal regularities because of physiological, physical, and social cC'anstraints. Activities occur in a time-space contimimg: there are temporal regularities inherent in spatial regularities, and temporal rhythms obviously vary over space. Although this premise has been advanced by a number of researchers over the past decade, their c<>xzceptua1 apparatus used to investigate human time-space behavior <1-‘-"~~‘Efers considerably. The ensuing discussion attempts to highlight the various conceptual approaches evolving over the past decade in North An‘erica and Western Europe (Figure 3). \ 1] 38James Anderson, "Space-Time Budgets and Activity Studies in I'flJan Geography and Planning," EnVironment and Plannigg, 3 (1971) , DP - 353-354. . .- zuhL'I‘! iuilll A 4| 1 ! {MU‘tIflz‘ 1.5. tflvz -. I\\ .1.» \ 42 .m 559E » .1 + 822.3. 2328» an: mus—m 82 HQ .088 82 was flat as: 3 mane-<3 n.5- mquoEuu .3 A338.“ .3 duo: £35m“: =53 _ Seamus. 5qu amen-33: . 338m?m55 and: :2 .5 .5333 22.-631 wflmmnfluflahfla _ 95.. 3 35.32: 5:... 52335623 . . H: 1338 exam-mos. A 58 78.38 3.83522 .255 _ _ .3 19.383355 m E: oz; ._ 5 e A.e.m.m.3 >335: 352...: 33... .223 63.228.- A a: H n 3 13:3 mam—Emu: 555 .3 muse-FE: _ 0&9 e e e I _ A a 8 .2 E 3.38 5.523 .2 .3 .32: 353...: ...-:32- 525 32.8 5.38 833 :2 5 £55.26 GEE: can»: 745:: :35 85am 28:35:80 Guava—z: Canaan: 55:: 45838”. 32 mm: axes 31.4.4: mag: A333 33.—.235. 3353 35:9- 5:55 2 _ 1.3.3 5.3.0 no 33552 Supp-nus.— 33338 :55 23» 3:63 52: .3 38:.- 303555.88 33.3 «was. 060‘ 22.3 25.53 i 32.3 zcmhmuk sag- $3 . ma: 4'- 54 x «J 5.. , 4 a '1 IL I." 43 Activity Conceptual System Apart from some of the Russian examples,39 the longest tradition in the use of the time-budget approach in urban planning is found in the work of Chapin, his students and research associates at the University of North Carolina."0 Drawing primarily on ideas from within planning, Chapin first proposed that the city be viewed in terms of systems of activities. For more than a decade, Chapin has been ' intimately concerned with the planning and policy implications of the relationships between urban space, time, and activity. Until then, Planners and geographers alike had not sought to record the activity dynamic of how the environment is used, and the activity component "of human time allocation... is missing from the available urban accounting system. "Planners have jumped directly into land use Studies, essentially studying the effects of activity systems rather than seeking to define and understand activities themselves as producers 42 Of land-use patterns." The behavioral antecedents of land use are not adequately represented by the usual constants and proxies of mapping t echniques . \ 39$zalai, The Use of Time, op. cit., pp. 838, 861-867. 40For some of the earliest statements by Chapin, see: F. Stuart CI'lapin, Jr., ed., Urban Land Use Plannipg, 2d ed. (Urbana, 111.: iversity of Illinois, 1965), pp. 221-253; and, F. Stuart Chapin, Jr. and Henry C. Hightower, "Household Activity Patterns and Land Use," J\°urnal of the American Institute of Planners, 31 (August, 1965), pp. 222-231. “Richard L. Meier, A Communications Theory of Urban Growth (Cambridge, Mass." The M.I.T. Press, 1962), p. 303. 42Chapin, 0p. cit., p. 221. 44 Origins of the Approach. Mitchell and Rapkin were among the first researchers to suggest the relationship between activities and the spatial structure of the city."3 Their work undoubtedly has had considerable impact upon urban transportation planning. Although their concept of activities is narrowly restricted to travel behavior, their statement of the role of activities in the city anticipates the formulation offered by Chapin nearly ten years later.“ Another crucial influence in the development of the activity aImproach was Meier's prOposal to use time allocation as a measure of Well-being of urban residents. The amount and variety of free-time aCtivities available to pe0ple in the city could provide a form of 8(kcial accounting. Meier was also among the first to suggest the t1Otion of a time-space budget, realizing that the allocation of time is closely linked to the use of space."6 Urban Activity Systems. Chapin synthesizes these two approaches into a theoretical framework of urban activity systems. An urban mOdifier is used to indicate the metrOpolitan community as a convenient study locale. In a systems framework, the metrOpolitan community is \ 43Robert B. Mitchell and Chester Rapkin, Urban Traffic: A @tion of Land Use (New York: Columbia University Press, 1954). "AChapin, op. cit., pp. 221-253. “Meier, "Human Time Allocation," op. cit., pp. 27-30; Meier's in‘pact on the purpose and directions of Chapin's research can be found (it: 3 F. Stuart Chapin, Jr., "Activity Analysis or the Human Use of ' 33:15am Space," Town and Country Plannipg, 38 (July/August, 1970), pp. 45-348. “Meier, A Communications Theory of Urban Growth, Op. cit. , PP- 48-59. H‘ fi.‘ lie , I'- 45 conceived as a series of interrelated activity systems fashioned around the pursuits of various entities—persons, firms, voluntary organizations, churches, governments, and other institutions. These entity categories are seen, in turn, to consist of subcategories within which individuals act in certain similar, purposive ways and, at the same time, respond to certain more or less comon constraints. Thus, the person category is seen to consist of some set of subcategories (e.g., various socioeconomic classes, stages in the life-cycle, etc.), each consisting of persons which as a class are seen to follow similar lifeways .47 In the activity system rationale, time allocation is seen as a IIleana of studying how human activity affects and is affected by the environment and how they jointly influence the structure and functioning Of the metropolitan community. It conceives of time as a resource allocated by individuals in the pursuit of their everyday affairs. ‘Through measures of time allocation and the use of a systems framework, Possibilities are foreseen of analyzing interrelationships between activities and population and those of metropolitan area assemblages of firms, governments, various organizations, and Other institutional agen- And clies. In contemporary society, daily life is organized around time. the time schedules of various institutions (e.g., work places, shape, 8el'vice establishments, and public agencies) affect the time schedules \ A 47F“ Stuart Chap-m, Jrn "Activity Systems and Urban Structure: Working Schema," Journal of the American Institute of Planners, 34 (January, 1968), p. 12. 46 of individuals and of one another. Moreover, the environment man has deve10ped for himself, especially the facilities developed to accommodate his personal and his institutions' daily needs, also affect time schedules. For example, time spent in co-uting or traveling to activity centers is a familiar measure for establishing how land use Because of as an environmental variable affects activity patterns. the complexity of dealing with time allocations of all entities in all Of these pursuits, a general systems approach is thought to be useful in the study of interrelations involved and the dynamics of these 8It'stems. Since "general systems theory seeks to classify systems by the way their components are organized (interrelated) and to derive 'laws' of typical patterns of behavior for the different classes of systems 8ingled out by the taxonomy,"48 it offers a useful framework in which to conceive of the organization within activity systems and the dynamic relationships between activity systems. Under this rationale, the behavior of a person, firm, or other entity is broken down into discrete .'episodes" which have meaning to individuals as they go about their particular affairs. When episodes are found to serve similar purposes, they can be grouped at a more generic level into particular classes of a"citivities as defined by some pre-established activity classification 83’8 tem. Thus, in the case of person entities, a variety of episodes car‘- be grouped into classes such as work-related activities, social—i 121133 activities, homemaking activities, or activities involving \ B 48Anatol Rapaport's Forward to Modern Systems Research for the ce\‘1£urioral Scientist, ed. Walter Buckley (Chicago: Aldine Publishing ° ‘ 1968). xvii. , 47 recreation and relaxation. It follows, then, that an activity system consists of a sequence of activities of some classified content and some specified order. More particularly, when an individual within an entity category-ma person, a firm, a voluntary organization or some other entity—exhibits regularities in the content and ordering of its activities in time and space, this sequence is referred to as an activity system, a firm's activity system, and so onf‘9 The preceding, then, is a brief overview of the conceptual frame- work. Chapin's conceptualization differs from previous theoretical Orientations in that it sees the city as composed of activities occur- ring not only over space, but also through time. This time-space frame of reference eventually led urban researchers to ask questions never before posed. Urban activity systems encompass a wide range of possibilities. The work at the University of North Carolina focuses on "activities" as measurable components of living patterns in the urban scene, while t:llne allocation serves as one of several dimensions, that include the type, number, and variety of activity choices, their sequence, and their spatial locus. Special attention is given to activity routines of urban residents functioning as individuals. But, since many person- activities are merged with institutional activities, the total system 18 of concern as well. Increasingly more detailed research has been \ . “Phillip G. Hamer, Jr., and F. Stuart Chapin, Jr., Human Time ££catiom A Case Study of Washington, D.C. (Chapel Hill, N.C.: etiter for Urban and Regional Studies, University of North Carolina, 19 72) a PP- 13-14. 48 conducted in an attempt to understand specific aspects of the whole problem. The ultimate goal, however, is to reverse the process and combine these results into a single theory of urban activity systems. The lineage of activity systems research is represented in Figure 4. The first step taken to narrow the research from a global activity systems formulation was to restrict attention to one segment of the urban pOpulation-namely, heads of households and spouses of heads. This was done because of their important decision-making role in the household. The first small-scale surveys undertaken in Durham, North Carolina,had respondents complete a time-budget listing all activities performed on the preceding day, including the timing, location and duration of activities. The temporal (duration and frequency) and Spatial dimensions of respondent's activities were analyzed. Respondents were also asked to "play" a trading-stamp game which was designed to elicit leisure time activity preferences.50 Drawing on the trading-stamp ritual of modern-day retailing, the game is a device 4 to simulate the way in which an individual might budget the use of his time for leisure. The results of the game were then compared with actual activity choices.51 Patterns of activities at various temporal scales were discussed theoretically, with special consideration given to the effects of changes in daily and weekly activities on the important \ 50F. Stuart Chapin, Jr. and Henry C. Hightower, "Household getiVity Patterns and Land Use," Journal of the American Institute of lamers, 31 (August, 1965), p. 225. II SJ'Ibid.; and, P. Stuart Chapin, Jr. and Henry C. Hightower, flsehold Activity Systems: A Pilot Investigation (Chapel Hill, N.C.: elitter for Urban and Regional Studies, University of North Carolina, 1966L 49 life—cycle activity of moving behavior.52 The important step, however, was the use of choice theory to explain the patterns of activities of urban residents. Following his early surveys, Chapin outlined a working schema for the development of a conceptual framework for urban spatial structure using activity analysis and activity systems. Essentially, Chapin sees activity decisions as arising out of an evolutionary process involving motivation-choice-activity.53 He suggests that motivation is derived from two sets of needs, fundamental and supplemental. Fimdamental needs are involved with shelter, clothing, food, and so on, i.e., choices to minimize feelings of discomfort or deprivation. Supplemntal needs for reason of achievement and status are requisite to a "full sense of well being" and require choices to maximize satisfaction. He suggests that each of these needs is satisfied by, or is sought for, in different roles and arenas, some— times simultaneously and sometimes separately. 52Chapin, "Activity Systems and Urban Structure," 0p. cit., PP- 16—17. Relating to the topic of life-cycle mobility, a current Proj ect by Michelson applies and tests many of the ideas and techniques of activity study, in analyzing the activity constraints that prompt residential relocation and the time-budget changes that occur, over . as a result. See: William Michelson, "Discretionary and Non— Activity and Social Contact in Residential giscretionary Aspects of eleetion," The Form of Cities in Central Canada: Selected Papers, 3:8' L.S. BOurne, R.D. MacKinnon, and J.W. Simona (Toronto: Univer- Lity of Toronto Press, 1973), .pp. 180-198; and, Brian P. Holly, "Urban fe Styles and Environment: The Effects of Location and Residence t1 Behavior in a Time-Space Framewor " (unpublished doctor's disserta- ona Michigan State University, 1974). 53Chapin, "Activity System and Urban Structure," op. cit., P. 50 In a discussion of the choice component, Chapin specifies the activity as output, with motivation and the values it represents forming the inputs to a final decision. "The context is the social system consisting of the environment and all other human activity relevant to the situation in hand." He suggests that in making choices of activities and in budgeting his time, the individual attempts to find an optimal combination based on his needs for "security, achievement, status, and other needs essential to his sense of well-being."55 The final output is, of course, his set of activities in various time scales: daily, weekly, annually, and 56 throughout his entire life cycle. In examining the activity patterns that result for the motivation-choice-activity sequence, it is apparent that all three elements of the sequence must be influenced by the alternatives available, which quite obviously are numerous. One clear distinction My be dram between those activities that occur in and out of the home. Chapin, through the use of time-budgets, conducted an \_ 541bid., p. 15. 551bid., p. 15. 56Activit ies are clearly cyclical. Some are daily-~the trip Others may be weekly-- to work, eating, some sort of recreation. Still others may be monthly, grocery shapping, trips to the bank. anfinal, or at even longer periods (e.g., the decision to save). A , general discussion of cyclical activity is given by Chapin, Urban % Use Planning, op. cit., pp. 221-253. 51 investigation on this division of the various activities which compose each subset.57 In brief, the patterns observed resulting from the motivation-choice-activity sequence of urban residents are fundamntal factors in many decisions that affect the physical structure of the urban area. At the sane time, the components of the physical structure--the spaces adapted to various activities--exert an influence on the choices made. The situation is strongly analogous to the economic concepts of supply and danand and their interaction with price and production. Next came the national surveys of activities in 1966 and 1969. In using these survey data, their interest focused on household time allocation, but largely ignored the spatial dimension. Thus, the research was quite similar to the time-budget studies reviewed earlier in this chapter. Efforts during this research phase were directed toward identifying factors significantly associated with the occurrence of various classes of activity in a weekday's itinerary. An additional result of this research was the construction of a typology of life styles based‘upon the activity patterns of groups distinguished by various social and economic variables.58 Further research was conducted in Washington, D.C., including 8:1 activities survey of the entire metropolitan area, as well as a \ 57Chapin and Hightower, Household Activity Systems, 0p. cit. 58F. Stuart (hapin and Richard K. Brail, "Human Activity SYBtem in the Metropolitan United States," Environment and Behavior, (December, 1969), pp. 107-130; and, F. Stuart Chapin, Jr. and now H. Logan, "Patterns of Time and Space Use," The Quality of £1:me Environment, ed. Harvey S. Perloff (Baltimore: The Johns Pkins University Press, 1969), pp. 305-332. 52 somewhat more intense survey in a low-income black neighborhood.S9 During this stage of the research, the focus narrowed to activities of a discretionary nature. Time-allocation to discretionary, or "free—time" activities was considered a most important indicator of differences between groups. Attempts were made to account for discretionary activity choice in terms of both opportunity and propensity to engage in the activity. The latter included both preconditioning factors (i.e., background social variables that constrain choice such as sex, income, and occupational status) and predisposing variables (i.e., variables that stimulate choice, such as felt needs for security, social status, etc.). This represents the first detailed application of choice theory to activity selec- tion. 61 Further research is underway on the spatial patterning of activities which involves a combination of temporal and spatial allocation of activities into a more comwehens ive theory of human aetivity systems. This phase is seen to be particularly important, for until now, it is unlikely that the technique can be adapted for Practical use by urban planners and policy makers. The conception \ 59Hammer and Chapin, op. cit.; and, F. Stuart Chapin, Jr., Edgar W. Butler, and Fred C. Patten, Blackways in the Inner City (Urbana, 111.: The University of Illinois Press, 1973). 60F. Stuart Chapin, Jr., "Free Time Activities and the guality of Urban Life," Journal of the American Institute of laliners, 37 (November, 1971), p. 411. (1 61See: Ibid., pp. 411-417; Hammer and Chap“, °P- C1"; ha13111, et al., Blaclnvsys in the Inner City; and, Richard K. Brail F. Stuart Chapin, Jr., "Activity Patterns of Urban Residents," and %ment and Behavior, 5 (June, 1973), pp. 163-190. 53 of the city as a system of activities is still quite new. The utility of the approach for understanding urban phenomena, despite its popularity, remains ‘to be proven. Still, an orientation encompassing both the tanporal and spatial dimensions of human behavior is appeal- ing and is well worth pursuing. . In summary, the activity systems research of Chapin and others :18 composed of two phases. The first, descriptive phase has been briefly sketched out above and summarized in Figure 4. In the descriptive phase, the concern has been with activity analysis which involves identifying the kinds of activity a population of individuals in the metropolitan comunity engages in over some defined period of time, classifying them, and then ordering the persons into groups on the basis of similarities in the occurrence or sequences of activities. The aim of this ordering process is to minimize within- group variation and to show variation between groups at significance levels sufficient to enable the investigator to draw inferences COncerning the activity patterns of these groups in the population. The activities found to cluster under these arch types are treated as Person systems and generalized to the appropriate subtotals in the Populat ion. 62 The second and more recent phase has focused on the explan- ation and simulation of human activity systems. ‘ Just as Chapin has provided the impetus in the descriptive phase of urban activity \ . 62F. Stuart (hapin, Jr., "Activity Systems as a Source of InPuts for Land Use Models," Urban Development Models, Special Report “0° 97, ed. George C. Hmens (Washington, D.C.: Highway Research Board, 1968, pp. 81-83. S4 SCOPE OF THE RESEARCH EFFORT URBAN ACTIVITY SYSTEMS ' \ HOUSEHOLD SECURITY SYSTEMS ' asses/enve- HOUSEHOLD TIME PHRASE ALLOCATION DISCRETIONARY TIME ALLOCATION k INTERDEPENDENCE BETWEEN TIME AND SPACE USE | EXPLANATORY PHRASE SIMULATION OF HUMAN ACTIVITY SYSTEMS . I DEVELOPMENTS IN ACTIVITY SYSTEMS RESEARCH FIGURE 4. 55 analysis, so too has the work by Henmens been the seminal force with regard to the modeling of human activity systems. In the develop- ment of a capability for simulating the behavior of one or more urban activity system, the concern shifts to acquiring an understand- ing of not only what patterns this arch type person's activities take, but why these patterns evolve. The concern is with the interdepen- dencies of the temporal and spatial aspects of the flow of events—why activity patterns develop, what structures them, and how sequences in one system relate to those in another system. In short, in the simulation phase, the aim is to reproduce the regularities in the ordering of activities observed in the descriptive phase in such a way as to approach the sequencing, timing, and spatial distribution that occurs in reality.“ In conclusion, the studies by the North Carolina group have made considerable progress in quantifying the amounts of time which different groups spend on various sorts of activity and many interest- his ideas have been generated—notably the trading stamp concept of a(2'::l.vity choice. But despite the group's use of time-budget diary techniques to collect information, there has been very little in the way of developing either ideas or data relating explicitly to the \ 63George C. H.ens, The Structure of Urban Activity Linkages ‘ Center for Urban and Regional Studies, University (Chapel Hill, N.C.: North Carolina, 1966); and, George C. Hemens, "Analysis and 0f Sinhalation of Urban Activity Patterns," Socio-Economic Planning Sciences, 4 (1970), pp. 53-66. 6['Brail provides a comprehensive discussion of modeling Richard K. appl’oaches to the simulation of activity systems; see: Strategy for Model Design" fire-11, "Activity Systems Investigations: nuPublished doctor's dissertation, University of North Carolina, 1969) . 56 question of interdependence between activity choice decisions over time. The task of pattern recognition has not yet taken the form of identifying sequences of activities which can be viewed as activity modules or of trying to understand the processes which govern the integration of the individual's time-space behavior. Time Use a_nd Ecolofim Organizat iog Theoretical work on time use is surprisingly limited, given its considerable history. An interesting set of theoretical considerations relating to time-use andecological organization have been presented by a group of Swedish geographers at the Royal University of Lund. The following is but a brief account of some of their studies that have a close connection with the work presented here. Origingfof the Approach. In a number of research projects recently emanating from Swedish geographers, the organizational aspects of society and of its substructures (e.g., manufacturing and industrial firms, political administrations, etc.) are the focus of interest.” When forming concepts and models, researchers have \ 65Some recent exanples of this research thrust (in English) include: Gunner Tornqvist, Contact Systems and Regional Development, I"find Studies in Geography, Series B, Human Geography, No. 35 (Lund: ~W.K. Gleerup, 1970); Allan R. Fred and Gunnar Tornqvist, System o\f‘cities and Information Flows, Lund Studies in Geography, Series 3. Human Geography, No. 38 (Lund: C.W.K. Gleerup, 1973), Olof aI‘laeryd, Interdependence in Urban Systems, Meddelanden fran G'ote- borgs universitets geografiska institutioner, Ser. B, Nr. 1 (Gate- I3031‘s: Kulturgeografiska institutionen, Goteborgs universitet, 3:968); and a review of other studies, Claes-Frederik Claeson, SYatemic preaches in Present Swedish Social Geography," Svensk %afisk rsbok, 44 (1968), pp. 140-150. ‘— 57 found inspiration and guidance in organization theory, system theory, and like modes of thinking. Not long ago, for example, location of industry was viewed mainly in terms of the transportation costs of goods. Today's researchers seem to find it more profitable to look into the links of comunication, both internal and external, between functional units of firms and organizations. These links require nearness or allow spatial independence in quite clearcut ways which seem to give a more profound understanding of locational behavior than earlier approaches, as well as better clues as to how influence and control might be exerted.66 H'égerstrand suggests that not only firms and organizations may be studied in terms of linkages and systems, but also the very local aspects of the modes of daily life may be analyzed in a related manner. He writes further that: One of the questions asked when the discussion of future regional and urban policy and planning started was how exactly did the physical and social environment--for example as reflected in city size--affect the life of individuals and families? In order to find out, we need a picture of how the daily activities of people are canalized through time-space. flerstrand's "Time-Geograflry" lbdel of Society. Throughout Haserstrand's writings on migration, there is a common underlying theme linking social comunication networks and "the changes in time \ 66W'drneryd, op. cit., pp. 19-22. U 67Torsten Hfigerstrand, "Methods and New Techniques in Current lrban Research and Planning," Plan: Tidskrift f‘dr planering av - Wow tutorter, 22 (1968), p. 10. 58 and space jointly experienced by the individual and society as a whole. "68 In recent years Hagerstrand has experienced a growing preoccupation with the fate of the individual in an increasingly complicated environment. This contination of circumstances has led him and his research associates to undertake an ambitious project to devise a "time-geography model of society" (en tidsgeografisk samhallsmodell) for the purposes of guiding urban and regional planning and locational policies in general.69 The basic problem of the future as seen by H'ageratand, is how society should be organized and how settlement patterns ought to be 68See, for example, some of H'égerstrand's more recent state- ments on the tepic of migration: Torsten H'a'gerstrand, "Geographical Measurenents of Migration: Swedish Data," Les dé lacements humains, ed. J. Sutter (Monaco: Entretiens de Monaco en Sciences Humains, 1962), pp. 61-83; and, Torsten H'a'gerstrand, "On the Definition of Migration," V‘a‘estBntutkimuksen vuosikirja [Yearbook of Population Research in Finland], 11 (1969), pp. 64—72. 69Some published reports of the project include: Torsten Hfigerstrand, "A Socio-environmental Web Model," Studier i planeringsmetodik, ed. G.A. Eriksson, Memorandum er; ekonomisk- geografiska ins itutionen, N . 9 (Abo [Turku] , Finland: Handel- shbgskolan vid 0 akademi [ bo medish University School of Econanics], 1969), pp. 19-28; Torsten Hagerstrand, "What About People in Regional Science?" Papersg The Regiogjg Science Associgion, 24 (1970), pp. 7-21; Torsten Hagerstrand, "Tidsanv'andning och omgivings- struktur," Urbaniseringen i Sverige: En geografisk samhallsanalys. Bilagadel I till Balanserad regional utveckling, Statens offentliga utredningar, 1970: 14, bil. 4 (Stockholm: Esselte tryck, 1970); Torsten Hagerstrand, "Frihet och tvdng 1 Stockholm och Ruskele. Nigra observationer av individ och familj i skilda svenska omgivnin- gar," Forskning och samh'a'llsutveckling (Stockholm: AB Allm'a’nna Forlaget, 1970), pp. 66-77; and, Torsten Hagerstrand, "Tutortsgrupper son: region-samhallen: Tillgdngen till fbrvarvsarbete och tjnnster utanfbr de starre staderna," Regioner att leva i, Expertgruppen for Regional Utredningsverksamhet (Stockholm: AB Allmanna Forlaget, 1972), pp. 141-173. 59 structured so as to ensure a "livable" day-to—day existence for the individual. Or, given the time restrictions on human movement and the fact that every economic and noneconomic activity is space- consuming, how ought the system of human activities to be organized spatially so as to provide substance to that portion of each individual's environment which lies outside the realm of income acquisition.7O In order to assault the details of his broadly defined problem, a.Swedish research project on "time-budgets and ecological organiza- tion" was initiated during the mid-1960's by Hagerstrand and other Swedish geographers.71 The purpose of the project was to discover in what way urban environments affect and constrain the daily activities of individuals, to see how activity schedules can be accsmmsdated in a "fixed" spatial-temporal system. The project is characterized by a fresh approach to location problems, using notions derived from organization and systems thinking, and new modes of classification.721 In Hfigerstrand's time-budget study of household 7oTorsten Hagerstrand, "The Impact of Social Organization and Environment upon the Time-use of Individuals and Households," Plan: Tidskrift. fUrgplaneringgav landsbygd och tutorter, 26 (1972), pp. 24- 25. 71Torsten Hagerstrand, Arbetsplan r8rande,projektet-- lgidsanvflndning och miljU (Lund: Institutionen for kulturgeografi Ocrh ekonomisk geografi vid Lunds universitet, 1967). 72Swedish geographic research is not necessarily technique- orniented, but in their pragmatism, the Swedes have in fact developed nevi techniques and ideas with both theoretical and utilitarian value. For more comprehensive reviews of concepts and methods coming from Shmsdish human geography, see: Allan R. Pred, "Urbanization, Domestic Plazuning Problems, and Swedish Geographic Research," Progress in Esglgraphy: International Reviews of Current Research, Vol. 5, eds. Christopher Board et al. (New York: St. Martin's Press, 1973), pp. 1‘76; James Anderson and John Goddard, Some Current Approaches. to 60 activity patterns, individuals are classified not in conventional socio-economic terms but in terms directly related to how these activities are constrained, and to the constraints they impose on other members of their household. Relevant components of the "fixed" spatial-temporal system include given locations (houses, workplaces, shOps) communications networks (roads, telephone, public transport routes), and a variety of institutional timetables (work hours, Opening and closing times of service establishments, transport timetables). These inter- related components of the urban system are influenced by activity patterns, but at the same time, they also restrict the activities of individuals and of households. They result in spatial and temporal regularities in human behavior. In addition, human behavior displays inherent "pattern" due to other factors of which the physiological (e.g., food and sleep requirements) are perhaps the most important. Physiological needs, that provide basic motivation for many activities, are themselves time-consuming and have a direct influence on peOple's time-use and movements. But, while such factors are taken into account, as they were in the research of Chapin, the Swedish project focuses on relationships between daily behavior and thus spatial and temporal variations in the environment.7 ¥ Eggnen Geography in Sweden, Graduate Discussion Paper No. 33 (London: Ixnrdon School of Economics and Political Science, 1969); and, Claeson,‘ op. cit. 73Two of the earliest pieces of documentation on the project include: (1) a comprehensive annotated bibliography of relevant tinB-‘budget studies--T. Carlstein and Solveig Martensson, Bibliografi Egggggde tidsanvandniggloch ecologisk organization, Urbaniserings- Processen, rapport nr. 3 (Lund: Institutionen fbr kulturgeografi och ekonomnick geografi vid Lunds universitet, 1967); and (2) a description, 61 Time:geographic Conceptualization. In his time-geography model, Hagerstrand views every individual as being surrounded by an "environ- mental structure" (omgivingsstruktur), or a pattern of activity or resource opportunities (conceived very generally to include, for example, employment, services, social and informational contacts, leisure-time pursuits, and a variety of resourCe goods) that are necessary to satisfy needs and wants and which are distributed unevenly in time and space. The environmental structure is relative to the individual; its composition depends on one's information and economic resources and psychological make-up. The time-space movements of any individual can be depicted graphically by compressing geographical space into a two-dimensional surface and representing the scale and direction of time along a vertical axis (Figure 5). Seen from this perspective, movement is transformed into geometrical form, which permits us to view an individual as a unique object, whose time- geography (defined by time and space coordinates) proceeds along a path. While earning a living and satisfying one's informational, social, and recreational needs and wants, an individual in time-space describes such a path, starting at a point of birth and terminating at a point 0f death. Depending on the perspective desired, the individual path nary be strictly defined either as a "daily-path" or a "life-path" tflrrough the use of time and space coordinates.74 k bassed on survey work of the daily time—use and movements of household menmbers in a number of selected households--T. Carlstein, Bo Lenntorp, and Solveig Mdrtensson, Individers dygsbanor i nggra hushfillstyper, Urbémniseringsprocessen, rapport nr. 17 (Lund: Institutionen fbr kulturgeografi och ekonomisk geografi vid Lunds universitet, 1968). 7“Hagerstrand, "Tidsanvandning och omgivingsstruktur," 0p. cit., pp ' 14-15 0 TIME 62 m = MOVEMENTS BETWEEN ESTZYTKDBES ~‘§~£ZL~“ J S, AND 82 = STATIONS v”?;:”‘ SPACE FIGURE 5. AN INDIVIDUAL'S PATH IN TIME-SPACE I O 1 I STATIONS I ' I I I I I ' : I : J I _ I I s. o o b c 82 SPACE HOME BASE FIGURE 6. ‘ TIME-SPACE IN TWO DIMENSIONS: THE TIME-SPACE PRISM. The prism, or "budget-space," represents the time- space frame within which a person can move. 63 While engaging in production, consumption, and social activi- ties, the individual stops at physically permanent "stations" (e.g., homes, workplaces, and recreation locations), or locations where move- ment is not observed over time. Individuals "spend time" at particular. stations, send messages, and move between stations to satisfy various needs and perform specific functions. Hagerstrand's station concept is quite flexible in terms of both its time and space scales. For instance, a city of residence, viewed as a station from the "life-path perspective," can be dissolved into a complex of stations when viewed from a "daily-path perspective."75 The environment with which the individual path comes in contact can be viewed at two scales. First, the "daily-living environment" is the geographical space that a person can reach within a single day and still return to his home base. Theoretically, it has a fixed definable outer boundary (as seen in Figures 6 and 7), owing to the capability of available means of transport to carry the individual back and forth within a fixed time-interval. Naturally, the space within the boundary of the daily-living environment (or daily-prism) is never fully exploited in every direction, but there remains the potential for choice of stations, such as workplace, shopping, recreation, and social contacts . The second scale exposes those places in which the individual nmfght choose to reside during longer periods of his lifetime. The 'restxictions imposed on an individual's life—path are not quite as 75Ibid., p. 15. I2:OO | I 9100 64 Illumumm 1’ ' ' 'illlllli'llf‘ I I k l "W!!!“ l"! l!!l!!!!|!|ll!!m"Hum”; I I ' | 6:00 : II I I n| | 3:00 : ' I I I I .‘ '- “’ ' III II ' E [2:00 I Illu. .IIl'I I F d 9100 _ 6100 1 3:00 *‘I .4 REGION 3 REGION C I REGION D REGION E SPACE STATION A FIGURE 7. TRANSPORT MODES AND THE DAILY PRISM DAILY PRISM ILLUSTRATING THE RANGE ERG! REGION A GIVEN CERTAIN SPEEDS OF I” IN MINATION WITH TIHE RESTRICTIONS. WITHIN REGIONS C AND D, THE MAXIMUM POSSIBLE TIME-STAY IS 8 HOURS. WITHIN REGIONS I AND E. THE HAXIHUH TIME-STAY IS 3 HOURS. OUTSIDE REGIONS B AND E TIME-STAY IS NOT SUFFICIENT FOR A CONTACT. IF TRAVEL TIMES ARE ASYHHETRIC (E.C., ONE ”DE 0? TRANSPORT IN ONE DIRECTION AND A DIFFERENT MODE IN THE OTHER) 'II'IIS HAY MAKE IT POSSIBLE TO EXTEND TIRE-STAYS OR TO REACH OTHER REGIWS. I ACTIVITIES I-4 ARE OF FIXED DURATION AND LOCATION DJ E ... HOME BASE SPACE FIGURE 8. FEASIBLE PATHS WITHIN THE TIME-SPACE PRISM non HM BASE rut cumczs LII! BETWEEN 1 AND 3 ON THE on: HAND, on 2 on m: omen HAND. ALTHOUGH 3 atoms AFTER 2 runs. nu: mnnveumc Tm: Is Too snout T0 PERMIT MOVEMENT BETHEEN non: STATIONS. 1. CANNOT BE cuosm smca us run POINT LIES BEYOND THE TIME-SPACE PRISM. 65 clear as those imposed on the daily-path environment and, consequently, the outer boundary of the life-path region is not as easily determin— able.76 Needs and desires of the individual can be considered as variables that make themselves felt, more or less, in a regular rhythmical way as time passes.77 The satisfaction of an individual's needs and wants usually requires movement from one station to another. However, the individual's set of potentially possible actions is severely restricted by the presence of other people, the complexity of private and public decisions, and social and behavioral norms. The individual's freedom to move from station to station is also limited by a number of more obvious physical and physiological restrictions. Thus, time-space paths become captivated in a network of constraints of which they can never free themselves. Constraints can become 76Ibid., pp. 16-17. Although Hagerstrand suggests that the life-perspective environment has no "ineluctable outer boundary," Jacobsson has documented that such a region for the majority of the present day Swedish pOpulation does not usually extend for more than some ten kilometers in any direction. Within such a range occurs most 0f the movement. See: A. Jacobsson, Omflyttningen i Sverige 1950- 12§0z Komparativa studier av migrationsfalt,»flyttningaavstfind och mObilitet, Meddelanden £E§h Lunds universitets geografiska institu- tion, Avhandlingar 59 (Lund: Lund universitets geografiska institu- tion, 1969), pp. 78-83. 77W’estelius has attempted to model the cyclical nature of needs and desires and their satisfaction in the context of urban travel behavior; see: Orvar Westelius, "The Individual Pattern of Travel Within an Urban Area-—An Interaction Between the Need of Contact with Different Activities and the Structure of the Location Pattern," P‘lam Tidskrift for planering av landsbygd och tutorter, 22 (1968), PP- ‘92-100; and, Orvar westelius, The Individual's Pattern of Travel in imn Urban Area, Document D-2 (Stockholm: Statens institut for bysgnadsforskning, 1972) . 66 imposed by society and interact against the will of the individual.78 Normally, the individual has limited means for influencing these varied restrictions, most of which fall into one of three general categories: 1. Capability constraints, being those physiological needs which regulate behavior, on the one hand, and distance restraints or areas within reach in the time available, on the other; 2. Couplinglconstraints, being those paths within the capability constraints, the timing and synchronization of activities, and the bundles of converging paths facilitated by telecommunications; and, 3. Authority constraints, being time-space entities under the control of particular individuals for groups (e.g., patterns of ownership and jurisdiction); the hierarchy of domains and the resulting limits on accessibility imposed by those in control.79 The individual is thus seen by Hagerstrand to Operate within this complex of constraints which together constitute a "highly institution- alized power— and activity-system."80 In summary, Hagerstrand envisages the individual's day to be IJDdertaken in a two-dimensional time-space framework, in which ¥ 78Hagerstrand, "What About PeOple in Regional Science?" op. cit., p. 12. 79Ibid., pp. 12-14. 80Ibid., p. 12. 67 a person's path through the day is structured by the fact that at certain times he has to be at particular places and that in order to meet these commitments his area of potential movement in the intervening periods is limited to varying extents depending on the type of transport available to him. His choice of activities in these periods, between those activities which were time and space fixed, was limited to those which could be undertaken within the time-space prism defining the feasible region of activity. The main features implicit in this model of daily behavior are, therefore, the notion that certain activities are fixed in both time and space, the division of the day into two types of period-- fixed and unfixed, and the delimitation of feasible prisms within which the entire day is confined, but more specifically, within which unfixed activities may be performed. Thus, the very essence of the model is the explicit demonstration of the vital interdepen- dence of time-space decisions. The decision of where to shop at 3:00 p.m. is no longer taken in the context of a purely theoretical and cross-sectional "action space" surrounding the individual's home- base, but is taken in terms of a highly specific time-space prism anchored between the individual's location at that time and his next forecast commitment (Figure 8). Applications of the model. In the past, time-use studies have been mainly concerned.with average or "typical" time allocations to \narious activity categories for various pOpulatiOn classes (e.g., the Nhaltinational Comparative Time-Budget Research Project). Where the time dimension has been considered in spatial analysis, it has been . 68 given only partial treatment, being used as a measure of distance or disutility, usually without explicit reference to overall time- budgeting.81 In contrast, the Lund project attempts an integrated analysis of time use, with time spent overcomdng physical distance as just one general category in the time-budget, but a category which constrains other time-uses. And, unlike many of the sociological time-use studies, the project does not attempt to typify populations or areas by time-use. It is mainly concerned with the sequence, timing, and location of activities (i.e., organizational aspects of time-use) in particular environments. These co—ordination aspects exist in greatest degree in large cities.82 In the first stage of the "time-use and ecological organiza- tion" research project, surveys on time-budgeting were completed in Lund and Stockholm, following an exploratory study in a small village 83 The daily-paths of individuals from different household in Skfine. types were recorded on time-budgets. These diaries showed the organization in time-space of their activities, both individually and as households. The samples were small and no attempt was made to 81James Anderson. Time-Budgets and Human Geography: Notes and References, Graduate Discussion Paper No. 36 (London: London School of Economics and Political Science, January, 1970), p. 11. 82Meier, A Communications Theory of Urban Growth, op. cit., Pp 0 48-54 0 83 Carlstein et al., 0p. cit. 69 obtain samples which were statistically representative, of either areas or households. Obtained in the Lund and Stockholm questionnaire interview were data on household composition, facilities, modes of transporta- tion, and the frequency of various activities which do not follow a daily cycle.’ For individual members there were questions regarding the frequency and location of recreation activities, school activities, and work-related activities. 0n the following day, members were required to record on time-budget diaries their activities, the respective locations visited, mode of transportation, and also telephone contacts. The published results of these surveys used, for tabular and graphic presentation, a five-category classification of household members: (1) full-time worker, (2) part-time worker, (3) students, (4) children under seven requiring care, and (5) home-centered individuals (e.g., housewives and elderly, but not those whose work- place is in the home). This classification is related to the duration and frequency of activities outside the home. Combinations of these individual "types" gave a range of household compositions.8 The time-budgets of the members of thirteen selected households there represented diagrammatically. A sample two-dimensional diagram :is presented in Figure 9, where the vertical axis defines the 24-hour 84Ibid., p. 27. Given the quality of official Swedish census data, these surveys did provide realistic sets of activity schedules envied: could be assigned to households in areas not surveyed. ' BSIbid. 7O E .u with uuuuuuuuuuuuuuuuuu , - - - . . e . n u m “ mm 53 utegq u _. _ u _ .AISR . fl, . u. PWE . 2.03 . . . EHM 4 _ . _ TOM . u. u .- m uh . l2? l()'- ' _ _ III... 24:21- A - .n 12¢- ” n. _ 4 =05 - . 2.30.5 5:30 .5233: 4830. >34! it ’ DWELLING it‘s... W TI In— :35: UOIKO . b l2.+- LIJ E ... J\ RECREATION AND SHOPPING STATIONS WORK PLACES SCHOOLS PERSONAL SERVICES LEISURE AND FIGURE 9. TIME-PATHS AND ImR-smnon uovmum'rs 71 period, and the horizontal axis is divided into seven station categories. Activities are defined in terms of stations visited; each station has three divisions that indicate different stations in that category. In Figure 9 an individual's time—use (or time- path) is shown by a continuous line, vertical segments represent time-periods spent at particular stations; slaping segments represent travelling-time between stations. Since it is difficult to adequately represent the spatial dimension in this type of diagram, space is only treated implicitly. Thus, the distance between stations is a component of travel-time and not a measure of spatial separation. The authors provide interpretations to such diagrams according to background information on household members, their available means of transport, telephone calls,mealtimes, and home location (in relation to workplaces, shops, and service locations). Using examples from these selected time- budgets, various types of activities (e.g., physiological, contractual, recreation, etc.) and some facets of household and family organization are discussed. By portraying graphically the different activities I of family members with specific reference to spatial settings and time, it is possible to observe the functioning of the family as a system.86 86This type of graphical analysis is quite versatile, and llichelson, for example, suggests that one can learn from it at what age and under what environmental conditions children become independent of their parents for specific purposes. "Since the accomodation of cfliildren is an extremely sensitive aspect of current housing programs,‘ this would appear to be a promising analytical tool to provide infor- nation for fruitful design." See: William Michelson, Selected Asgects of Environmental Research in Scandinavia, Research Paper No. 26 (Toronto: Centre for Urban and Community Studies, University of Toronto, March, 1970), pp. 8-9. 72 In more recent reports that relate to Hagerstrand's time- geography model, there has been more explicit treatment of spatial\ aspects of time-use-station locations, means of transport, travel- times at different times of day, institutional timetables, and variability in service standard between stations in the same category.87 Lenntorp has developed a computer model to determine the alternative station-to-station movements and daily paths of individuals in a given environment.88 In conjunction with this, information on the locational pattern and opening hours of eight types of stations in nine differently characterized urban environments has been gathered and restrictions on contact possibilities have been charted.89 Using a time-geographic perspective, Sweden's regional variations in the accessibility to dental care, eye clinics, apothecaries, libraries, and other services have been studied and interpreted.90 Hagerstrand himself has embellished his model in order 87Hngrstrand, "Tidsanvlndning och omgivingsstruktur,‘ op. cit.; Solveig Martensson, Tidsgeografisk beskrivning av stationsstruktur, Urbaniseringsprocessen, rapport nr. 39 (Lund: Institutionen for kulturgeografi och ekonomisk geografi vid Lunds universitet, 1970); Bo Lenntorp, Tidsgeografiska synpunkter p8 upplgggnigg;av transport- analyser-Sammanfattning av nagra foredrag, Forskagruppen i kultur- geografisk process- och systemanalys (Lund: Institutionen far kul- turgeografi och ekonomisk geografi vid Lunds universitet, 1973). 8830 Lenntorp, PESASP--en modell far berakning av alternativa banor, Urbaniseringsprocessen, rapport nr. 38 (Lund: Institutionen :fbr kulturgeografi och ekonomisk geografi vid Lunds universitet, 1970). 89Martensson, Op. cit. 90Hagerstrand, "Tidsanvandning och omgivingsstruktur," Dp. cit. 73 to explore, empirically and theoretically, the frequent local incom- patability of existing activities in urban environments, and to evaluate the individual's possibilities for making use of his environ- ment.91 He chose to summarize the ways in which the daily time-uses of a population constitute a system in which activity bundles are dependent upon one another. This necessitates very complex models, which Hagerstrand argues, must be Operationalized through computer simulation.92 It is not possible to go into detail showing how the simulation proceeds analytically. However, the basic scheme is as follows: A Step 1. Assume a "population system” and a related "activity system” in timetable terms. The population system is comprised Of all individuals in an area and their biological and social relations I to one another. The daily time-income is the 24 hours x the total number of inhabitants. Since needs, wants, and obligations vary according to age, one can imagine the individuals in question, who are tied together in households, as being depicted by daily lines arranged from the youngest to the oldest (Figure 10a). The activity system, on the other hand, consists of all those activities performed at'a given location, regardless of their necessity or value. These k 91Hagerstrand, "Tatortsgrupper som regionsamhfillen,‘ op. cit. See also: Torsten Hagerstrand, "En rattvis stadsstruktur," Plan: ftidskxift; f8r_planeringflav landsbygd och tutorter, 24 (1970), pp. 112-119. 92Hagerstrand, "Methods and New Techniques in Current Urban Research and Planning," 0p. cit., p. 11; and, Hagerstrand, "The Ihqmact of Social Organization and Environment," 0p. cit., p. 29. 74 time-consuming activities are seen as a whole, detached from the local population (Figure 10b). Step 2. Sieve the population and activity systems through an environment. Unlike the situation represented in Figure 10b, all individuals do not have identical time and activity budgets. Rather, the activity system.is divided into separate "time packages." Each of these "seeks a bearer in the population." 0r, appositely, each individual picks his way through a series of activity bundles in the offered system. The matching procedure of individuals and activities must occur within the framework of two unavoidable physical restrictions. First, once an individual is matched to and participates in a specified activity bundle at a given time and place, he cannot be somewhere else doing something else at the same time. Secondly, once a "delegatable" activity bundle is fully staffed, it becomes closed to others.93 Hence, two budget problems exist: how to divide the individual's allocatable time among activities; and - how to delegate activities among the total population? Actual match- ing is only crudely suggested in Figure 10c. Due to the indivisibi- lity of the individual, and the fact that movement from one activity bundle to another is time-consuming, the individual cannot participate :tn temporally successive bundles that are separated by excessive 93Figure 8b represents the "activity system" as being comprised oi? two major types: (1) Non-delegatable activities-those for which every individual is responsible and cannot be delegated to someone else (e.g., sleep, personal care, and eating); and (2) Delegatable activitieswthose which can be delegated either from one household member to another or between individuals of other large groups. 75 (3M) 3W”. (31“) 3WIl N. 0. e~n o v N. w. ON .323 .vcfiumpwwmm snows. gwfimmmfie .453 < 29$ 53 we mzmpmwm war/Eu... 92 28258.. E ..8 252.539.. 53.5: 20.59520 5.254 .3 5.34 8 ON a DJOIuwDOI no mmugm! v h 5.03 g l v Sumac... I T T n we: BE .83; I Fllllll N mtg ..(Smmwa _ emu-....” mm.._..>..PU< AN. .OH MESH... mwdzdxm m2.» aux... Am. A. 80.. no 5252 mmaoz 02.2wmo manor 02.283 oavos MJ'BM P-‘I ouhs‘ofiwo mm 240 BEER? owhdzofiuo mm #02240 xuvs H some 1306 H P—‘I ‘2 O N v N 38018000! A. m..<:o.>.oz. ...0 18232 Owed no ON n I IN. 1 Im. » H _ _ cm 0401330: 5:388 3 ('SN) 3Wli ("5‘”) 3W|l 76 time-distances. From. a daily perspective, it is the relationship of transport-time to the general time-budget situation which, at a given location, delimits the population and activities that can be matched.“ Step 3. Register the outcome as it is distributed among individuals in the pOpulation. The assigned activity schedules of individuals can then be matched to a "fixed" spatial-temporal environ- ment. A simple graphic example of this is portrayed in Figure 10d. Hagerstrand's time-accessibility evidence for three urban areas demonstrates that problems of matching home, work-place, and service consumption are especially acute in those instances where the individual resides in a small town and Opportunities are concentrated to a nearby larger central place and those cases where the individual d<=Des not have access to a car.95 Only through repeated runs will the model be able to show the ways in which activity schedules are accommodated within the system, In addition, &111:! the constraints which the system imposes on behavior. a~3l-I:ernatives in the system may be specified (e.g., change the locations or operating-hours of shops or workplaces) and their effects on activity patterns simlated. This has obvious relevance to planning-- the probable effects of various proposed changes could be simulated Q16 choices made between alternative proposals. \ 94liligerstrand, "THtortsgrupper som regionsamhallen," 0p. cit. , pp - 94-109; and, for a more detailed review of step 2 in English, see: Pred, op. cit., pp. 49-50. 95Hagerstrand, "Tfltortsgrupper som regionsamhallen," 0p. cit. 77 A considerable part of the work is basic research dealing with modelling as such. A second component covers empirical testing of the patterns of constraints and Opportunities inherent in typical present-day urban environments. Two time scales have been chosen. One part of the work deals with daily and weekly activities. The other part takes a long-term perspective. Questions are asked about how life chances and environ- ments are related. In neither case are broad statistical surveys of actual behavior essential. Emphasis is laid upon the working of constraints. And, these are in various ways sensed with the aid of ideal-typical sequences of actions which are confronted with actual or model environments. Questions are also asked about how these Various environments perform as providers of employment, training, 8Get‘vices, recreation, social communication, and free-time. Hagerstrand's time-geographical analysis should have consider- ab 1e appeal to scholars in other countries confronted with acute PrOblems that arise from rapid urbanization. This should be the case because it provides a means of considering the plight of the in~<'1:l.vidual while simultaneously attacking large-scale local or regional plat'ining problems within a systems framework. Wt Approaches to Time-budget and Urban Activity Research Following the lead of Meier, Chapin, and Hagerstrand, there has been an increased use of time-budgets in urban spatial research, particularly by planners, architects, and geographers in Great Britain (Figure 3). Much of the work in Britain is of a practical as opposed 78 to a theoretical bent.96 The two areas of interest include activity schedules of individuals and households, and institutional and work- place linkages. One focus is on overall time and space patterns of individuals or groups, plus particular activities, such as leisure, The other focus personal contacts and communication, 7 and travel.98 is on personal activities, especially face-to-face information exchanges that often involve travel, with further attention on working hours and on spatial links and comunications underlying functional organization of particular establishments.” Although the type of data obtained, the population and time-period. the classification of activities and locations, and the precision of location in time and space varies considerably with study purpose, the same time-budget has Proven adequate for the different types of time-space behavior study. 96James Anderson, Time-Budgets and Human Geograjm: Notes and Leferences, Graduate Discussion Paper No. 36 (London: London School °f Economics and Political Science, January, 1970), p. 7. 97See: Ian G. Cullen and Vida Godson, Networks of Urban The Structure of Activity Patterns, Final mities Volume II: I. Ort (London: Joint Unit for Planning Research, University College, ondon, 1971). 9 d1 98The London Traffic Survey in 1964 and 1966 used a 24-hour Saar)? to analyze aggregate volumes over time and space; London Traffic fly“ 1964 and 1966, 2 vols. (London: Greater London Council, a 1966). S 9gsee, for example, Vida Nichols, An Institution in Metropolis, 13! Paper NS 12 (London: Joint Unit for Planning Research, Univ- era l Ne: it)? College, London, 1969); and, Ian G. Cullen and Vida Godson, Nita of Urban Activities Volume 1: Internal and External Links es Urban University, Interim Report (London: Joint Unit for Plan- Wu“ 8 Research, University College, London. 1971). 79 A group of architects affiliated with the Centre for Land Use and Built Form Studies of Cambridge University have also contributed to the growing body of literature on activity patterns and the spatial organization of urban areas. The influence of the built environment on the way in which individuals use those structures has long been at the heart of architectural studies.100 Similarly, the relationship between the spatial organization of an urban area and the expected activity patterns of groups of individuals has been implicit in much of urban design. However, explicit relationships between the layout of a development area or even an entire new town and the patterns of interaction within them have seldom been demons- trated, principally because urban design, like its parent subject, architecture, has traditionally been an art rather than an "exact aczltence." A number of leaps of faith have therefore been required in postulating a relationship between, for instance, urban density and levels of interaction. The work at the Centre, some of which is described in the essays of Urban Space and Structures, 1 suggests that such leaps of fa1th may no longer be necessary. The work demonstrates how mathe- matical models applied to large bodies of data can be used to describe \ B looTerence Lee, "The Effect of the Built Environment on Human e1lavior," International Journal of Environmental Studies, 1 (1971), pp - 307-314. ' loI'Leslie Martin and Lionel March, eds. , Urban Space and Cambridge University Press, 1972). In parti- 8 \ct buctures (Cambridge: Lbular, see the essays contained in "Part 2: Activities, Space, and. Qanon," pp. 109-157. ' 80 both the relationship between building structures, space and time available, and potential use at the micro-scale. With the aid of such models, these researchers believe that the consequences of alternative building designs and urban spatial structures may be more readily evaluated in the planning stages. On the basis of empirical evidence derived from individual "time-space budgets" gathered in a number of British universities, these researchers have demonstrated a clear connection between these budgets and student numbers, timetabling, the arrangement of build- ings, and their location in the urban area.102 Data from the diaries are used as a starting point for the modelling of activities in time and space. A model has been developed whose purpose is to predict the distribution of individuals in different activities and locations during the course of a typical day, depending on the effective restrictions imposed by the spatial distribution of buildings and sites, and by administrative and social constraints on the timing of ac t ivit ies . 103 ¥ J'OzJanet Tomlinson et al. , "A Model of Students' Daily ACfiiVitY Patterns," Environment and Planning, 5 (1973), pp. 231-266. 103In the first of their series of working papers, Bullock disc”asses the theoretical basis for an approach to the simulation of a‘~'3':'-V:l.ties; see: Nicholas Bullock, An Approach to the Simulation of Muss: A University Example, Land Use and Built Form Studies, Working Paper No. 21 (Canbridge: School of Architecture, Cambridge University, August, 1970). A more recent report on the deve10pment °f a model of daily activity patterns may be found in: Nicholas gun-Ock et al., Development of an Activities Model, Land Use and nil-t Form Studies, Working Paper No. 41 (Cambridge: School of . Archi- tecture, Cambridge University, April, 1971), pp. 53-76. 81 The assumptions made about the behavior of pe0p1e in aggregate for the purposes of the model are simple. First, it is assumed that for a given group of peOple, over some repeated period (a day or a week), the proportion of time spent in various activities will remain the same, although the sequence of activities and their locations would, of course, differ. This overall division of time between activities, the time-budget, is expected to vary for different groups of perle. As an input to the model, the time-budget and the hypothe- sis of its stability for similar pOpulations under dissimilar physical conditions, is of considerable importance. Second, it is assumed that the behavior of people is subject to a number of limiting restrictions which determine either the times or the locations of activities, or both.1°" With his simplified view of behavior the problem of modelling day—to—day activities may be seen as distributing the population to activities in time and space in such a way that, first, the proportion of time spent in different activities by the pOpulation as a whole is Bullet to the time-budgets, and second, that the restrictions for different activities on the availability of times or locations (or bOth) are observed. More formally, the problem may be seen as distributing numbers of people in a three—dimensional space whose dimemaions are activities A, times of day T, and locations L (Figure 11a). Not all combinations of A, T, and L are available (Figure 11(1)- \ o loaTomlinson, et al., "A Model of Daily Activity Patterns," p0 Cit.’ pp. 24-29. 82 \V‘ " a. A three—dimensional array of cells representing activity, time, and location. 3 )- L' 2 '6 < TIME (7) 3 The total population in " E cells in each activity/ 5 location plane must obey 2 the population constraint. ‘ \V o“ 11.5 (T) " c. The total population in ' cells in each location/ time plane must obey the I time-budget constraint for IllllIlllll that ECU-Vi”- Some- activity/location ’ combinations are not available, nor are some time/activity combinations . FIGURE 11, THREE-DIMENSIONAL ARRAY OF TIME, ACTIVITY, AND LOCATION 83 Both time and location are treated as discrete entities, and thus, a value in a particular cell could represent, for example, the number of pe0ple watching television between 6:00 and 7:00 p.m. at a particular address. 105 Throughout the day the number of people in the AL plane must remain constant (Figure llb) . Equally, the amount of time spent in particular activity at all times in all locations (i.e., the sum of the values on the LT plane) must be equal to the number of hours per day for the whole population (Figure llc) determined from the time-budget. Apart from the limiting restrictions of the timing of acti- vities and the constraint of the time-budget, two modelling assump- tions are made. People are allocated to activities and locations, first, subject to the availability of locations for that activity at that time and, second, to the constraint that the total amount of travel time in the time-budget for the whole pOpulation is maintained. Thelmodel uses an entropy-maximizing method to derive the most PrObable distribution of the population in time and space, subject '10 the population and time-budget constraints and the restrictions on the availability of time and space for activities.106 The resulting \ 1°5Ib1d., pp. 32-35. 106 For a review of the entrapy-maximizing technique as applied t° “than and regional analysis, see: A.G. Wilson, Entronv in Urban Wonal Modellipg (London: Pion Limited, .1970). Wilson has also PrOpo Bed a similar model employing the entrapy technique to the spatial-temporal organization of household behavior; see: A.G. Wilson, Bi“ Recent Developments in Micro-economic Approaches to Modelling Behold Behavior, with Special Reference to Spatio-temporal Organi- tation " (Paper 11) , Papers in Urban and Regional Analysis (London: “on Limited, 1972), pp. 216-236. 84 distribution gives the number of peOple engaged in each activity, during each time period, and at each location over the day. Thus, no attempt is made to identify or follow the sequence of activity and locational choices made by individuals; only the overall distri- bution of people in any cell or the flow of peOple between cells from one time period to another is modelled.107 Much of the work at the Centre over the past ten years has been with the application of these ideas first at the scale of an individual institution (e.g., a university) set in its urban context, and second to the entire urban area. However, their approach is not without its disadvantages. One drawback is that once individual tine-budgets are input to the model, they lose their individual identity.108 Furthermore, when working from a totally disaggregated data base to the more aggregated form required by the entropy-maximiz- ing technique, both time (i.e., in terms of activity sequencing) and location come to be treated as discrete entities. Another disadvant- 882 is that their approach deliberately neglects perceptual processes 33 8 Vital mechanism relating individual behavior to the built en"1'-li‘onment. In spite of these caveats, their work reflects numerous \ 107Because this resulting distribution is statistically the no": probable, it is precisely that distribution which assumes least knowledge about activities of individual decision-makers. 10813116 reader will note some similarity between this modelling approach and that prOposed by Hagerstrand (above). Unlike Hager- strmd's approach, however, the modelling effort described here makes no attempt to preserve an individual's identity, seen as a continuous Sequence of actions in time-space. 85 insights into the operation of the urban system. In this respect, the philosOphy behind much of their work lies in the mainstream of thought that has linked architecture with urban planning by stressing the relationships between activities and the structures that exist at all scales. A team of British planners, led by Cullen, is taking on an activities approach to individual behavior. Their framework emphasizes the role of activities as linkages between institutions and the patterning of sets of activities. Research is being conducted in the institutional setting of a university where time-budgets are obtained f ran the students and faculty.109 The hOpe is that such a micro-view Will enable the researchers to decipher the linkages and constraints which serve to determine the sequences of activities.1m. To accomplish this goal, they employ some of the more advanced analyses of time- budget data. 11]” Cullen's approach draws on Hagerstrand's conceptualization, in which he views the greater part of individual behavior as structured \ 109Cullen and Godson, Networks of Urban Activities Volume I, 0p- cit.; Cullen and Godson, Networks of Urban Activities Volume II, 0" - cit. 110Ian G. Cullen and Vida Nichols, "A Micro-Analytical Approach :0 the Understanding of Metropolitan Growth" (paper read at the e‘fenth World Congress of Sociology, Varna, Bulgaria, 1970). 8 111See: Ian G. Cullen Vida Godson, and Sandra Major, "The tI‘lleture of Activity Patterns," Patterns and Processes in Urban and Pkwu Systems, ed. A.G. Wilson, London Papers in Regional Science' "- 3 (London: Pion Ltd., 1972), pp. 281-296. 86 about a whole range of environmental constraints. However, Cullen considers Hagerstrand's approach "over-deterministic" in that it allows for "no variation in the perception of constraints, but rather treats each as an unambiguously objective fact."11:z Thus, the work to date of the group at the Joint Unit for Planning Research of the University College, London has been based upon a theoretical framework which emphasizes those elements in a person's day which, from his point of view, lend it coherence, or give it the shape that he feels it possesses. (In place of Hagerstrand's "fixed-unfixed dichotomy,"113 they have characterized the individual's day as an integrated function of a much more elaborate range of flexibility, defined by one's degree of commitment to each activity and by the time and space constraints to which one is subject. These additional items, which reflect their extensions of Hagerstrand's time-geography model, permitted the group to test the hypothesis that an individual's day is structured at‘Ound certain key events, and that this structure derives from the Way in which individuals perceive constraints in their environment. To date, the information which these researchers have collected is I"~1l‘ely descriptive of the constraints people experience, and thus, cannot directly inform planning decisions, which is their ultimate obj ective. ‘ Indirectly, however, they were able to contribute to the \ 1 112Ian G. Cullen, "Space, Time, and the Disruption of Behavior Cities," Environment and Planning, 4 (1972), p. 465. cit 113Hagerstrand, "Tidsanvandning och omgivingsstruktur," Dp. 114Cullen et al., op. cit., p. 284. 87 problem of locating large institutions by virtue of the fact that they treated a relatively homogeneous group—-the members of one institu- tion. 115 This review of current research thrusts in urban and regional analysis relating to time-budget approaches is by no means compre-' hensive of the literature in this field. Rather, these selected projects exemplify some possible uses of a time-budget approach. The studies chosen for review are particularly important for human geography in that they stress the spatial dimension and treat'it and the temporal dimension in an integrated manner. Hence, they are seen as precedents in the literature that relate to the general question confronted by this research: How do'individuala' decisions about their time-space behavior interrelate? 115Cunen, op. cit., pp. 460-461. CHAPTER3 TIME-SPACE PATHS OF HUMAN BHiAVIOR: (DNCEPTUALIZATION IMPLmENTATION OF THE RESEARCH STRATEGY Building upon the preceding synthesis of time-budget research, the objectives of this chapter are twofold. The first intent is to assess the potential of a time-space budget perspective for theory development in human geography. Two possible approaches toward this end are identified and critically evaluated. The second objective is to develop a conceptualization and to formulate research prOpositions designed to investigate the structuring of daily behavior through the tine-space mechanics of constraints. In order to substantiate or illuminate the research propositions, a researchstrategy is established and involves the design and implementation of a survey. THE TIME-SPACE BUDGET PERSPECTIVE AND HUMAN GEOGRAPHY The term time-budget signifies an accounting scheme to describe the allocation of time to activities during a given period. A time- b'-ldget's logical extension, the time-space budget, is ostensibly a behavioral approach to geographic and planning research: it incorpor- ates the sequence, linkage, timing, duration, and frequency of act ivities, as well as the spatial and temporal coordinates of one's behavior. Time-space budgets focus on two related aspects: people's u eil':t behavior, and their perceptions of their physical and social 88 89 environment.1 The time-space budget, therefore, can record spatial behavior (moving and stationary),2 and it can be indirectly related to environmental perceptions (via questionnaire data) as these interact with overt activities.3 Although the device has potential for a variety of research applications, the previous chapter demonstrates that its use in urban analysis predominates. It is particularly well-suited to metropolitan areas where distance is often considered temporally and where many activities are precisely scheduled by the clock. Perhaps the primary attribute of time-space budgets is that they record behavior patterns which are not directly observable due to their spatial and temporal extent. Furthermore, this method of h 7 1James Anderson, "Space-time Budgets and Activity Studies in Urban Geography and Planning," Environment and Planning, 3 (1971), PP . 353-368 . 2The use of a related device, the "travel diary", by Marble and Nystuen has also contributed to the growing interest in relation- 8hips between time and space uses. The purpose of a travel diary, Owever, is to record movements from place to place throughout some 8l>ecified period of time. It does not attempt to record the totality of behavior in time-space for the selected period (including station- ary periods at particular locations or stations), and thus, the 8equential properties of activity modules cannot be accomodated. . tivities at "fixed locations or stations are not described though 8(”lie of them may be inferred from trip destination and purpose. For eJuamples of travel diaries and their use, see: Duare F. Marble, "A eoretical Exploration of Individual Travel Behavior," Quantitative Mphy, Part I: Economic and Cultural Topics, Studies in Geography, N0 - l3 (Evanston, 111.: Department of Geography, Northwestern Univ- era ity, 1967), pp. 33-53; and, John D. Nystuen, "A Theory and Simula- . Intra-urban Travel," Quantitative Geoflphy, Part I: Economic and matural'hpics, Studies in Geography, No. 13 (Evanston, 111.: Depart- ent of Geography, Northwestern University, 1967), pp. 54-83. of 3Ian G. Cullen, Vida Godson, and Sandra Major, "The Structure 8 Activity Patterns," Patterns and Processes in Urban and Regional tans, ed. A.G. Wilson, London Papers in Regional Science No. 3 London: Pion Limited, 1972), pp. 287-291 90 activity accounting is unique in that, as far as their activities for the sampled time span are concerned individuals are treated as totalities, their behavioral integrity is preserved, and the entire sequence or pattern of activities can be analyzed. Although isolated parts, particular activities such as commuting to work, shopping, and leisure pursuits, may be extracted for analysis, they can be considered in the context of the respondents' overall time and space uses throughout the recorded period of time. Hence, they provide an overall behavior-context within which particular activities may be viewed realistically. This does not mean that the activity sequences Or patterns are fully explained as intended and understood by the Persons who create them, but at least the totality is there for eRaininatrion. Nystuen identifies distance, direction, and connection as three fundamental spatial concepts." Through the use of time-space blltlgets, distance can be expressed temporally, which is frequently the way urban residents perceive distance, and connection can be given I“Ore comprehensive treatment than is customary in geographic research. w1th regard to the concept of connection, Anderson points out that in the recent trend toward studying cities as contact structures, the euphasis has shifted from distance to contacts and from Euclidean 89°th to graph theory and topology. At the same time, the discon- tihuous and anistropic properties of urban space (e.g., as measured in terms of time, costs, or perceptions) have received increased \ C "John D. Nystuen, "Identification of Some Fundamental Spatial oncepts," Papers of the Michigan Acadg of Science, Arts, and t. M. 48 (1963), pp. 373-384. 91 attention, accompanied by a corresponding. dissatisfaction with the asstmptions of classical location theory.S Whether the use of time- space budgets can directly contribute to improving location theory or underline its shortcomings remains to be seen. Spatial Behavior, Spatial Structure and Location Theory In revealing the inadequacies of location theory, behavioral research in geography has demonstrated that it is difficult to improve its postulates, or to match its formal elegance. This has been particularly evident in research that treats the interact ion between spatial behavior and spatial structure.6 Although the spatial structure of activities in an urban area will reflect both current and past patterns of behavior, explanations of spatial structure based On such patterns of behavior often seem to be tautological since it would appear to be just as reasonable to explain behavior as a func- tion of structure as to explain structure as a function of behavior. 'Ihe relationship is clearly one of mutual dependence; that is, changes in spatial structure elicit changes in spatial behavior and vice versa. In this regard, the use of time-space budgets entails certain disadvantages that should not be overlooked. They sometimes tempt 1:‘easoning and explanation into a "circle of causality" formed by \ 5Anderson, op. cit., p. 355. 0 6Gerard Rushton, "Behavioral Correlates of Urban Spatial St ructure," Economic Gem, 47 (January, 1971), pp. 49-58. 92 spatial behavior+-~+spatial structure.7 A behavior pattern which contributes to spatial structure is the way individuals make choices between alternatives distributed over an area. Central place theory is only one of many areas of human geography where assmnptions about individuals' behavior patterns are incorporated in explanations of spatial structure. Indeed, Curry's work has focused on the problem of developing theory from postulates which do not inherently contain the deduced facts which interest us.8 Curry argues that since it is possible to deduce the distribution of central places from an accurate description of spatial behavior patterns of people in an area, no insight is gained from studies which explain spatial struc- ture in terms of behavior patterns that occur within it. He points out that the description of spatial behavior is no more a process type explanation of a central place pattern than is the description of the pattern itself.9 Thus, he concludes that: "A term such as christaller's 'range of a good' suffers from this conditionality of definition."1° \ 7Reginald G. Golledge, Process Approaches to the Study of Human \Behavior, Department of Geography, Discussion Paper No. 16 (Columbus, oh10: The Ohio State University, 1970), p. 2 8Leslie Curry, "Central Places in the Random Spatial Economy," J‘:’\‘lrnal of Regional Science (Supplement, 1967), p. 218. 9Curry attempts to replace behavioral assumptions with mathe- :Qt 1cal (Poisson) process descriptions of spatial activity. This t rategy is attractive if one's purpose is to construct aggregate zodels incorporating human behavior. Howaver, the lack of explicit avioral assumptions reduces its reliability for inferring disag- Ere sate results. . 1'OCurry, loc. cit . 93 Observed behavior is partly determined by the structure of the environment in which it occurs, and Rushton argues that parameters of behavior in a particular environment (e.g., empirically derived distance decay functions) are therefore "not admissable as a behavioral postulhte in any theory."11 What, then, are the critical characteris- tics which a spatial behavior postulate must have to be admissable in a viable theory? Curry states, in reference to central place theory, that: A postulate on spatial behavior should not directly describe the behavior occurring within a central place system, since it is obvious that the system can then be directly derived without providing any insight. The behavior postulate must allow a cen— tral place system to be erected on it in a suffi- ciently indirect manner that a measure of initial surprise is occasioned by the results, and this pos- tulate must still describe behavior afterthe system has been derived.12 ThuB, a behavioral postulate should incorporate the rules of spatial (tho ice which underlie behavior patterns, irrespective of the particular environment in which the behavior has been observed. Christal-ler's I)"81:lrlate that consumers patronize the nearest place offering a - Irefill-tired itan is, according to Curry's criterion, a logically admissable 13 me, although consumers frequently disobey it. If mre realistic 1miles of spatial choice are to be discovered through behavior studies, \ S 1'J'Gerard Rushton, "Analysis of Spatial Behavior by Revealed 5591: gee Preference, " Annals of the Association of American Geographers, (June, 1969), p.192. 12 Curry, op. cit., p. 219. Ran 13N.A.V. Clark, "Consumer Travel Patterns and the Concept of lgsge." Annals of the Association of American Geographers, 58 (June, ) . p. 396. . 94 and studies employing time-space budgets specifically, we must avoid the circle of causality. An attempt to overcome this problem is reflected in the work by Rushton concerning the locational preferences underlying a pOpula- tion's spatial behavior.“ In his approach, Rushton assumes that an individual's spatial behavior is affected by his "preference function." Thus, by aggregating the preference functions of individuals, a pattern of behavior can be generated for any distribution of spatial opport- unities. In proposing the concept of "revealed space preference," he notes that" in econanic consumption theory the spatial distribution of Show is not considered a significant variable. However, it is significant when the choice is between different conmodities. Although his "spatial preference structures" may have more generality than 8I’at:ial systems, he concludes that they are not‘independent of the Particular system in which they are derived. Furthermore, his method is descriptive in the sense that no attempt is made to explain why one opportunity is favored over another in the recovered preference func- tion. This approach might eventually provide useful postulates, but thus far it only indicates how elusive general and independent "rules" really are. As yet there is little evidence to suggest that the use of t-‘-'-lne-space budgets would be any more successful in finding these rules. \ ‘ . S 1"Rushton, op. cit.; and, Gerard Rushton, "Temporal Changes in Page Preference Structures," Proceedings of the Association of \ LilleErican Geographers, l (1960), pp. 129-132. 95 The notion of trade-offs between time allocations, and between time and space preferences, is conceptually appealing, and the analysis of behavior patterns in a wide variety of environments might produce interesting results. Wolpert insists that an understanding of spatial behavior involves sorting out the regularities or constants in time- space patterns and distinguishing these from the variables.15 Since there is still so little spatially-oriented comparative time-use research, it is hardly realistic to believe that the constants could be useful as a basis for a deductive location theory; but the variables night prove to be more enlightening. Accessibility, both in a temporal and spatial sense, is a key VaIl':':l.ab1e in determining both the extent to which behavior is shaped by the spatial environment, and the way in which'individuals evaluate a"-?-t-:I.Varities and locations.16 However, such an abstract variable is itself a very complex set of variables. Their "objective" values (e- 8., time and money) vary widely with different populations, transport modes, and their implications also differ considerably depending on a wide range of factors (e.g., age, income, occupation). The relevant set of important accessibility-Opportunities varies both with life-cycle stage and with life-style. And although a probable \ L11 ”Julian Wolpert, "Behavioral Aspects of the Decision to ppgrate." Papers of the Regional Science Association, 15 (1965), J 16M.A. Stegman, "Accessibility Models and Residential Location," %1 of the American Institute of Planners, 35 (January, 1969) . 96 set of opportunities may be inferred from time-budget data,17 it may be misleading since some opportunities which are felt to be relevant by the respondent may not be revealed in his activity pattern because they are too inaccessible. Chapin and others have discussed a general framework for study- ing urban living patterns in which time-use is seen through motiva— tion—-+choice-—--vactivity set of relationships. They contend that spatial behavior can be conceptualized as the outcome of choices which reflect people's motivations and values. But such choices are realized within the constraints set by personal circumstances (i.e., socioeconomic requisites), accessibilities, and the environment. merefore, the limits of free choice differ greatly on a wide range of factors (e.g., age, income, mobility, life-cycle stage, etc.) These trends in research are clear achievements. By observing the behavior of members of a pOpulation, one learns something about their living conditions. But this information does not clearly differentiate what are wants and needs from what are various degrees of necessity. For the purposes here, behavior (seen as a manifestation of choices) does not fully reveal the underlying pattern of ‘ constraints \ 17F. Stuart Chapin, Jr., "Activity Systems and Urban Structure: A Working Schana," Journal of the American Institute of Planners, 34 (January, 1968), p. 16. , 18Chapin, ibid.; F. Stuart Chapin, Jr. and Thomas H. Logan, .Patterns of Time and Space Use," The Opality of the Urban Environment, . Harvey S. Perloff (Baltimore: The Johns Hopkins University Press, ed 1969), pp. 305-332; and, F. Stuart Chapin, Jr., "Activity Systems as 3 Source of Inputs for Land Use Models," Urban Developpent Models, pecial Report No. 97, ed. George C. Hemens (Washington, D.C.: High- "Qy Research Board, 1968), pp. 77-96. 97 which shapes action and the situations in which it occurs. This also means that clues are not provided for how to reshape the living conditions, if that is the goal. In order to find the clues, we must look to latent structure and latent processes. Purposeful changes in the distribution of Opportunities and risks among individuals necessitates an understanding of how constraints interact and how choice potentials are affected by changes to one or more of those constraints. Choice- and Constraint-oriented,Approaches to Spatial Behavior Constraints are implicit in choice, but, depending on the relative emphasis given to "positive" and "negative determinants" of behavior,19 a distinction.may be made between "choice-" and "constraint-oriented" approaches to spatial behavior. If psychological motivations of spatial behavior (i.e., the rules of spatial choice) are sought, then observed behavior must be seen in terms of choices 20 ‘Hmwever, the danger exists between alternative courses of action. that important constraints may either be underestimated or go unnoticed. Observed behavior can be conceptualized as reflecting the constraints of the environment (i.e., "objective" constraints) and personal circumstances (i.e., the "subjective" ones). If highly constrained situations are not recognized as such, observed behavior 19Torsten Hagerstrand, "What About PeOple in Regional Science?" _§apers of the Regional Science Association, 24 (1970), p. 11. 20Rushton, "Analysis of Spatial Behavior by Revealed Space Preference," op. cit., p. 392. 98 may be misinterpreted as what peeple "choose" to do, rather than what they are "forced" to do. With the prevailing emphasis on choice and positive determinants in general, situations where actors do not have effective choice may be neglected.21 That is to say, in the absence of effective choice, individuals may be asked to decide between hypothetical alternatives which in reality they have little chance of achieving.22 A less common tack focuses on the outer limits within which behavior can take shape, emphasizing negative determinants of behavior.’ Such an approach is exemplified in the constraint-orientation adopted 23 by Pahl. Pahl advocates studying "the pattern of spatial and social constraints which operates differentially in given localities" and " He argues that ‘which fundamentally "affects people's life chances. there are fundamental spatial and social constraints on access to urban resources and facilities. Spatial constraints are generally expressed in time/cost distance while social constraints reflect the distribution of power in society. The latter are illustrated by 21This is exemplified in the activity systems and time-budget research where activities are simplified into obligatory as opposed to discretionary activities. See: Richard K. Rail and F. Stuart Chapin. Jr., "Activity Patterns of Urban Residents," Environment and Behavior, 5 (June, 1973), pp. 163-190. 22One example is the use of the trading-stamp game of choice theory to elicit leisurertime activity preferences; see: F. Stuart Chapin, Jr. and Henry C. Hightower, "Household Activity Patterns and Land Use," Journal of the American Institute of Planners, 31 (August, 1965), pp. 23R.E. Pahl, "Urban Social Theory and Research" (Chapter 13), Whose City? And Other jEssays on Sociology and Plannin (London: Longman Group Ltd., 1970), pp. 209-225. 99 bureaucratic rules and procedures, and the actions Of what Pahl calls "social gatekeepers" (e.g., local government officials and policy-makers, landlords, employers). The gatekeepers are those who distribute and control urban resources and, thus, regulate the quality and accessibility of Opportunities such as educational facilities, the housing market, and the job market. Conflicts of interest in this socio-spatial system.are inevitable, and the greater the scarcity of valued Opportunities, the greater the conflict. Pahl sees populations limited in their access to scarce urban resources and facilities as dependent variables, while those controlling access, the gatekeepers or managers of the system, are the independent variables. He notes that the current emphasis on diversity of choice in physical planning implies that the access to facilities is an independent variable, in contrast to being dependent on the allocation by gatekeepers. This suggests that there are ideological as well as methodological differences underlying the variation between choice and constraint orientations (e.g., differing attitudes to "free market" mechanism).24 The interrelationships between time and space uses have perhaps been most clearly worked out by Hagerstrand and his 2[‘Ibid” pp. 215-216. A similar attitude has been echoed by Harvey in his conceptualization of the city; see: David w. Harvey, Social Justice and the City (Baltimore: The Johns HOpkins Univer- sity Press, 1973, pp. 91-95. 100 colleagues.25 In the physicalist tradition,26 their approach similarly focuses on the outer limits within which behavior can occur and emphasizes negative determinants of behavior rather than positive factors such as the attitudes, motives, prefer- ences, and choices which contribute to structuring activities in a timeLspace framework. The emphasis on positive factors is more common in social science, and the more common response to time-space data sets has been to abstract certain characteristics from them (e.g., time-spending characteristics, in aggregate, Of population groups and areas).27 In many cases, geographers and other social scientists treat a pOpulation as a mass of objects, almost freely interchangeable and divisible; i.e., we often impute upon any given' individual of the population the modal characteristics Of that pOpulation. Also, it is common practice to segment the mass into such aggregates as shOppers, migrants, and age and occupational 25See, for example: Torsten Hngerstrand, "Tidsanvandning och omgivingsstruktur," Urbaniseringen i Sverigfi: En geografisk samhallsanalys. Bilagadel I till Balanserad regional utvecklin , Statens Offentliga utredningar, 1970:14, bil. 4 (Stockholm: Esselte tryck, 1970); TOrsten Hagerstrand, "A Socio-environmental Heb'MOdel, "_Stndigz_i_plangzingametgdik, ed. GBsta A. Eriksson, Memorandum fran ekonomiskrgeografiska institutionen, Nr. 9 (Abo [Turku], 31mm: Handelshbgskolan vid Abo akademi [Abo Svedish University School of Economics], 1969), pp. 19-28; and, T. Carl- stein, BO Lenntorp, and Solveig‘Mfirtensson, Individers dyggbanor i nigra hushfillstyper, Urbaniseringsprocessen, rapport nr. 17 (Lund: Institutionen for kulturgeografi och ekonomidk geografi vid Lunds universitet, 1968). 26Hagerstrand, "What About People in Regional Science?" 0p. cit., p. 11. 27Anderson reviews examples of these research trends and discusses their limitations; see: Anderson, Op. cit., p. 356. TIME TIME TIME (I I cameo I l2-~ CD I 101 FIGURE 12. * 'm ILLUSTRATION 0? - CAPABILITY CONSTRAINTS nu: DAILY nISN's MAXIMUM DIMENSIONS m AN INDIVIDUAL m0 ms 10 SPEND I‘IMF. I'd. , ' AT A wounuca. A PERSON um: LOU 7 " savanna CAPABILITY (a) HAS A MORE watts saunas or HORKPLACE CHOICE 'lllAN A mason (b) Hl'l'l-l A NION CAPABILITY. a b ' o—DlSTANCEfi FIGURE 13. , ILLUSTRATION OF .5; 43:!" comm: CONSTRAINTS \“ THE COUPLING or INDIVIDUAL urns (5.0.. gem STUDENTS AND INSTIUCTORS) IN AN ACIII°Ifl BUNDLE Inc., scuouu. scuoon. ACTIVITIES ‘ ARE ASSUHED 1'0 NAI'I; rssnsmamrn 1m;- 1 "V ‘ \' TABLES AND occuII AT A FIXED LOCATION III mlfll (‘Shll SPACE. SPACE nouns 14'. ILLUSTRATION OF AUTHORITY CONSTRAINTS HIBRARCNY OF DOMAINS AND ITS INPACT ON INDIVIDUAL PATHS. ”I REPRESENIS A SUPERIOR DOHAIN (S.G., A HUNlCIPALITY) AND D; AND D3 ARE SUNONDINATN DOMAINS. ACTIVITIES WITHIN D) AND ”3 CAN TO A CERTAIN EXTENT BE NECULATED NY 0 . THIS MIGHT OCCUR IF SHRVICN HAS [5- TRICTED T0 PERSONS RhSIDfiNT IN 0 (b-¢ BUT NOT A AND I). 102 groups. Through segmentation, each aggregate is analyzed in isolation from the others. It was this particular problem that prompted ngerstrand to write: "...one risks becoming lost in a description of how aggregate behavior develops as a sum total of actual individual behavior, without arriving at essential clues toward an understanding Of how the system works as a whole." In order to discover these clues, he focuses on what may be termed the "time-space mechanics of constraints."28 Even if constraints are formulated as general and abstract rules of behavior, they can be provided with a "physical" shape in terms of location in space, areal extension, and duration in time. An individual's time-space path is constrainedby a set Of physio- logical and physical factors that arise in part from the social and/or private network of decisions and actions surrounding the individual. Hagerstrand suggests that these constraints are composed Of three interacting groups: capability constraints (Figure 12), coupling constraints (Figure 13), and authority constraints (Figure 11.).29 The individual is thus seen by HAgerstrand to operate within this complex of constraints which together constitute a "highly institutionalized power- and activity-system." Viewed from a time-space 2BIHngrstrand, "What About People in Regional Science?" loc. cit. 29Hngrstrand, "What About PeOple in Regional Science?" Op. cit., pp. 12-18; and, Hagerstrand, "A Socio-environmental Web Model," Op. (cit., pp. 20—25. These groups of constraints are also described in (:hapter 2. 103 perspective, two recognizable, but diverse, systems of interaction are seen. One is primarily a time-oriented warp Of individual paths which constitutes the population of an area and its capability constraints. The second is a set of constraints that domains impose (including within them coupling constraints) to which the individual may or may not have access according to his needs and wants.30 Social scientists know very little of the interaction of constraints, expecially as viewed from the daily—path perspective. Hagerstrand suggests that at this stage of research into the interaction Of constraints, a simulation approach would be the most appropriate method of analysis, until more general mathematical techniques. become available. Reasonably solvent simulations may aid in survey- ing whole systems and help to reduce the trial and error component of empirical applications.“ mgerstrand's conceptualization of constraints is instructive. Based upon his recommendation for a simulation methodoloy, it seems worthwhile to define the time-space mechanics of constraints which rule how paths are channeled, diverted, or even routinized. It is believed that such an approach will shed light on how decisions about one's time-space behavior interrelate. Some authors believe that such 3(’Hleigerstrand, "What About People in Regional Science? " op. cit., pp. 16-18. In a more recent paper Hagerstrand discusses the political aspects of entrance conditions attached to domains; see: Torsten Hagerstrand, "The Domain Of Human Geography," Directions in Geography, ed. R.J. Chorley (London: Methuen and Co., Ltd., 1973), pps - 60 3J‘Torsten Hngrstrand, "The Impact of Social Organization and Environment upon the Time-use Of Individuals and Households," P_1____an: Tidskrift fBrjlanerinLav landsbyLoch tutorter, 26 (1972),—.- pp. 29.300 104 a negative determinants approach to social science may indeed be the safest and most productive.32 In using this methodology, the simula- tion will attempt to construct each individual's day separately, generating'the activities he performed and the sequence he adapted from generalized rules. The rules take the form of constraints or probability distributions which are estimated from a time-space budget data base. A technique for grouping constraints in time-space terms is formulated in order to collapse their considerable variety into a tractable mmber. wNCEPTUALIZATION OF HUMAN BEHAVIOR AND CONSTRAINTS The renainder of this chapter is devoted to a conceptualization of human behavior and constraints and the development of a methodo- logy, including a time—space budget diary and associated survey. In accordance with the research strategy, the following chapters will develop analytical procedures to compress the survey data into a more couprehensible form (Chapter 4) necessary for modelling, via simula- tion techniques, the time-space mechanics of constraints (Chapter 5). Time-space Behavior: The Search for Assumptions Although the emergence and formal discussion of the "behavioral "33 approach is rather recent, the employment of behavioral assumptions ‘ 32Hagerstrand, "What About People in Regional Science?" Op. cit., P. 11. ' 33For an overview Of behavioral approaches in geography, the reader is referred to: J .M. Doherty, Deve10pments in Behavioral Geo- graphy, Discussion Paper NO. 35 (London: Department of Geography, London School of Economics and Political Science, November, 1969); and, Reginnld G. Golledge, Lawrence A. Brown and Frank Williamson, "Behav- ioral Approaches in Geography: An Overview," The Australian Geom- ahg, 12 (1972), pp. 59-79. 105 is not at all new. Harvey has noted that human geographers have almst always made assumptions about behavior—however, these assumptions have cOIImIonly been implicit in the analysis, rather than explicit in theoretical statements.“ The most notable exception has Of course been the case of the "Economic Man" assumption. This normative behavioral postulate has been used extensively by geogra- phers in both implicit and explicit fashions--to the point where much of location theory bears a close relationship to classical economics. Thus Olsson and Gale have noted that most spatial theories rely on the same behavior assumptions as the theory Of the firm.35 The well-known behavioral assumptions embodied in the Economic Han concept are reflected in the knowledge and goals attributed to the individual. Thus, it assumed that behavior is based on knowledge which is omniscient and perfect, and which precludes uncertainty from predictions. Behavior is oriented towards goals which Optimize profits or utility. The extension Of these goals into a spatial context results in the producer attempting to increase the size of his market area, while the consumer attempts to reduce purchasing costs by minimizing the expenses and effort associated with distance.36 Similarly, the theory of route choice behavior-~for whatever purpose-- 3"David W. Harvey, "Behavioral Postulates and the Construction of Theory in Human Geography," Geographia Polonica, 18 (1970), p. 27. 35Gunnar Olsson and Stephen Gale, "Spatial Theory and Human Behavior: A Study of Anarchistic Vector Spaces," _apers of the Regional Science Association, 21 (1968), p. 230. 36Loc. cit. ' 106 is characterized by decisions which minimize distance, travel cost, or travel effort. The fact that the above behavioral assumptions are unrealistic and inaccurate is well known. Economists, for example, have long been aware that producers rarely attain optimum profit levels. Non-norma- tive economics, therefore, recognizes that business decisiondmakers possess imperfect knowledge and problemrsolving ability, and that their goals under certain circumstances may be non-Optimizing. The failure Of real man to correspond with the actions Of the firm have promoted various reformulations of the behavior assumptions. Simon, for example, has prOposed the principle Of "bounded rationality"--which emphasizes the limits to problem-solving capacities and the avail- ability Of knowledge.37 Geographers have widely criticized the Economic Man assumption in a spatial context. Particularly, spatial analysts have recognized the need for adjusting the distance Optimizing function. Thus, linear distance has now been largely replaced in geographic models by "functional" measures such as accessibility or travel time. The use 37In so doing, Simon relates the notion Of "satisficing behavior"; see: Herbert Simon, Models Of Man (New York: John Wiley & Sons, Inc., 1957). WOlpert recently reintroduced the concept into geography in writing of farmers in Central Sweden: "The ' concept Of the spatial satisficer appears more descriptively accurate of the behaviOral pattern Of the sample pOpulation than the norma- tive concept of Economic Man. The individual is adaptatively or intendedly rational rather than omnisciently rational." The satisficing notion assumes that in the absence of perfect knowledge, 'behavior is directed towards an alternative which is satisfactory, but not necessarily Optimal. See: Julian Wolpert, "The Decision IProcess in a Spatial Context," Annals Of the Association Of American (Seographers, 54 (December, 1964), pp. 537-558. 107 of surrogates in these adjustments, however, has not accounted for observed discrepancies between the normative—economic behavioral model and real-world locational and trip activity.38 The widespread recognition of thesediscrepancies has prompted many geographers to seek reformulations or alternatives to the Economic Man concept. Principle of Consistency and Recurrent Behavior Patterns As an alternative, this research is committed to the funda- mental principle of consistency in human behavior. The notion of consistency is not to be confused with assumptions of the rationality of behavior (i.e., Economic Man). The purposeful element in individual behavior varies in degree between individuals and among the different activities undertaken by an individual and, thus, some behavior seems to be virtually instinctive while other behavior is highly calculated. All behavior is purposeful to some degree. Whether spatial behavior is construed to be rational or not, depends on our depth of understanding the values, goals, and purposes towards which that behavior is directed. Moreover, what is rationality to one Observer may not correspond to the interpretation of another observer; the assessment Of rationality is a dubious process which varies with cultural, ideological, and personal biases. In the conceptualization presented here, the notion of rationality is avoided in favor of the principle of consistency. In ¥ 38Ibid.; Marble, Op. cit.; Clark, op. cit.; and W.A.V. Clark and Gerard Rushton, "Models of Intra-urban Consumer Behavior and Their Implications for Central Place Theory," Economic Gegraphy, 46 (July, 1970), pp. 486-487. 108 so doing, an attanpt is made to isolate that behavior which demonstrates recurrent patterns in time-space.39 And, although behavior is not seen as consistently rational and well-informed in the classical econOmics sense, it is seen as containing highly organized episodes which give structure and pattern to the whole stream of behavior in time and space. Given this interpretation of consistency, the claim is made that the concept of pattern can be meaningfully applied to any set, or sub-group, of time-space budgets and that this pattern is partly defined by the sequence in which activities are performed. This hypothesis will be validated if it can be demonstrated that, as individuals move through their days, the probabilities of their engaging in such activities as working, eating meals, relaxing, and sleeping vary significantly from one time period to another. The physiologically-determined activities tend to be those of a more routine nature. Although discretion may sometimes be exercised as to where the activity occurs, these activities occur in a day's sequence at about the same times. Working, attending classes, shOpping, and homemaking—the culturally defined extensions of sustaining activities-- tend to fall into a person's daily sequence at approximately the same time. The idea that activities differ significatnly from one time 39.Jiri Kolaja, Social System and Time and Space: An Intro- duction to the Theory of Recurrent Behavior (Pittsburgh: Duquesne University Press, 1969), pp. 48-51. 109 period to another is at least guaranteed by the fact that, historic- ally, broad margins of choice over these matters has been institu— tionally removed.40 A similar assumption incorporated in the general idea of consistency is that the amount of time a person devotes to an activity fOllows a pattern over sub-groups of individuals just as much as does the position it occupies in that person's sequence. This is to say that the physiologically determdned activities and their culturally defined extensions tend to have the same durations each day just as they tend to fall into a person's daily sequence with similar timing and frequency. Such simple patterns of activity must exist if simula- tion is to be a meaningful exercise. Priorities and Opportunities As was the case in Hagerstand's time-geography model, the individual in this conceptualization is seen to Operate within a framework which is fundamentally structured by physical patterns and needs. This environmental structure is institutionalized to a considerable degree by the availability of services and by the norms, expectations, and habits of the individual. Within the environmental structure, or budget-space, the individual selects from.a set of Opportunities that consists of possible activities. The order of selection is made after priorities Ihawe been assigned to the various activities in the Opportunities set ¥ "OHngrstrand, "The Impact of Social Organization and Environ- nmt," Op. Cite, pp. 24-250 110 in accordance with their attributes. Several factors contribute to an individual's assessment and ranking of priorities. Perhaps the first cOnsideration is the importance Of the activity to the individual. A.second consideration might be the presence of participants and their characteristics (e.g., their relationship to the decision—maker, their frequency of contact with the decision~maker, distance to be travelled or time spent travelling, etc.). Preferences will also produce a tendency for an individual to choose one kind of activity having a certain set of qualities over another activity having a different set; hence, preferences will enter into a person's ranking of alternatives. , The Interaction of Constraints: The Global View Priorities, however, are only realized within the context of constraints. And whereas priorities are self-generated, most con- straints are imposed externally (Figure 15). Daily activities are seen to possess a certain temporal rhythm defined by the manner in which physiological and culturally defined constraints impinge on human activity. Along with the temporal aspect of a person's activities, there is also the spatial dimension.which has to do with the locations of his activities. Although some of the same con~ straints that regulate the timing are also involved, activity location is strongly influenced by environmental cOnstraints. The physiological, cultural, and environmental constraints Operate differently on the activity sequence in different contexts. Fbr example, the physiological needs of an individual (i.e., sleep and.austenance) constrain other activities in the daily activity cytzle. They are activities in themselves, but they also serve as 111 *L OBJECTIVE ENVIRONMENT ]\ ENVIRONMENTAL AND SOCIAL PERMITTED j 1 BEHAVIOR 1 FIGURE 15 . FACETS OF THE BEHAVIORAL PROCESS. V KRCEPTUAL HOCESS Jk ,..f L._ BEHAVDRALlE _:;.$ :5 E ENVIRONMENT ‘ a ** \ I I I I \ / . ACCEPTED \ \ / REJEOTED/‘\ ‘\ \ \ \ \ \ x W mm INFORMATION I MM ”GIO- SWIG STATUS V IMAGE PERCEPTION OF f ENVIRONNENT AND OPPORTUNITIES 5 MOTIVATION - g; PRIORITIES g _ E O O 2 @STRAINTS tWSTRA'NED PHYSIOLOGICAL, V BEHAVIORAL PROCESS 112 constraints by affecting the timing of other activities. Similarly, the culturally defined extensions of these requirements (e.g., working, attending school, shopping, and homemaking) fall among the key activities in a person's day, but they also have an affect on other activities and, therefore, also serve as constraints. Indeed, all of these activities are considered to be so fundamental to an activity sequence, forming what may be termed a daily routine of an individual, that they structure the remainder of the activity sequence affecting not only the timing, but also the character of other activities in the sequence. Environmental constraints are no less complex in the way they impinge on the flow of a person's activities. Individual activities may be directly influenced by the physical setting-the natural surroundings and the spatial configuration of the built environment- as well as the technical means with which to negotiate a path through the environment. Environmental constraints affect not only the spatial distribution, but also the timing of activity. The routines of urban residents must conform to the schedules of various institu- tional entities. In short, firms, organizations of various kinds, governments, and other institutional entities serve to constrain a person's activity simp1y=by the schedules they set. For example, a person's place of employment maintains certain regular working hours, and stores are Open only at certain times. Therefore, a person's routine is shaped by others, and the rest of his activity sequence is determined for him, to a considerable extent, by these constraints. 113 Objective and Subjective Constraints on Time-space Behavior The next task is to refine, or simplify, this global view of physiological, cultural, and environmental constraints into a con— ceptualization more suitable for empirical work and eventual modelling. At the same time, I wish to incorporate and expand upon Hngrstrand's simple fixed/unfixed dichotomy of constraints to a much more elaborate range of flexibility as Cullen and others have proposed.41 The simplification of the global view of constraints, along with the expansion of Hngrstrand's conceptualization, is reflected in the prOposition that people are exposed to objective and subjective constraints. Objective Constraints. The objective constraints on their behavior include those which are imposed by the environment. The notion of objective constraints is similar to Hflgerstrand's idea of capability constraints. FOr example, a person at any given location may only move a certain distance away from that original station in a certain amount of time. The distance of this movement is deter- mined by existing modes Of transport and time available to him. Objective constraints tend to shape the potential "budget-space" available to an individual (Figure 12). To the extent that these constraints operate equally on groups of individuals, they frequently fix patterns of behavior, alluded to earlier in reference to the principle of consistency. Many of the movement constraints, however, [Operate quite independently on each individual (i.e., according to 41Cullen et al., Op. cit. 114 one's capabilities, technical and Otherwise, for movement through time-space). Despite their greater variation, or unfixed quality, they are essential in the development of a simulation model. Subjective Constraints. Even more important than objective constraints is the considerable range of subjective ones which, to some degree, are self-imposed. Such constraints probably occur as a result of certain features of the individual's physical and social environment, but it is unlikely that there will be anything more than a contingent relationship between the Objective facts and the sub- jective state associated with these."2 In most contexts, a given environmental situation is subjectively perceived by different peOple as constraining choice to varying degrees. Although the peculiar- ities of the environmental context are important, an individual's response to that situation is critical. Any situation to which an individual reacts has at least three dimensions: a temporal position (e.g., in a day or sequence), a location in space, and a potential activities sat. These dimensions interact to produce a feeling of constraint upon the individual. One's evaluation of the way in which these dimensions inter- relate is complex. But as a result of the priorities and constraints felt by an individual, any activity has a subjective fixity rating associated with it. This subjective fixity rating is prOposed in place of Hngrstrand's fixed/unfixed dichotomy, since it permits a ¥ AzIan G. Cullen and Vida Nichols, "A Micro-Analytical Approach to the Understanding of MatrOpolitan Growth" (paper read at the Seventh WOrld Congress of Sociology, Varna, Bulgaria, 1970). 115 more elaborate range Of flexibility defined by the degree of comitment to the activity and the extent to which it is constrained in time and space. These two dimensions of activity f1exibility--commitment and time and space fixity-—are highly correlated. Cullen, however, advises that such dimensions are theoretically independent aspects of the problem and may be treated separately.43 The degree of commitment affects the flexibility of activity choice. Cullen has distinguished four categories of commitment.44 l. Arranged activities. Such activities involve interaction with Others, where time and place of the activity have usually been specified and, therefore, fixed. 2. Routine activities. These activities, which frequently recur at regular intervals, attain the status of virtually immoveable points in a person's day. Such routine activities may Often have more stability and thus a greater degree of commitment, than is strictly necessary. 3. Planned activities. These include activities which an individual plans to perform at sometime in the future. Since planned activities are not usually coordinated with other individuals or groups, their degree of flexibility is generally greater. 4. Unexpected activities. Activities of this kind are commonly those which arise, or "just happen", while the individual is performing some other activity. Such activities as chance meetings, accidents, or 43Cu11en, et al., op. cit. aalbido 116 some leisurely pursuits have no long-term planning perspective in the pursuit of one's everyday affairs, and are of the "spur of the moment" variety. The remaining component of the subjective fixity rating is the degree to which activities are fixed with respect to particular times of day or to particular locations, or both. An individual schedules a variable proportion of his day according to these ratings of fixity, in order to facilitate synchronization and synchorization of activities and movements in time and space. .Although much of the scheduling process is undoubtedly routine, the ordering of more complicated, unfamiliar, or crowded combinations of activities is objectively calculated. Activities to which the individual is strongly committed and which are space- and time-fixedor merely time-fixed, tend to act as points around which the ordering of activities, accord- ing to their flexibility, are scheduled. Research Proposition. The central proposition of this research is that it is meaningful to view the constraints to which peOple feel subjected as relative to either a particular time of day or a certain activity, or both. It follows that for any particular time of day, it seems reasonable to ask a person.whether or not he felt "tied down" either with respect to his current activity or his location. And for any given activity, it also seems reasonable to ask the individual whether or not he felt its timing or location to be similarly fixed. It would appear to be acceptable to fix on times of day at which one has to be at a particular location or at which one has to perform certain activities and to fix upon activities that must be undertaken 117 at certain times or at specificlocations. Alternatively, it would not appear acceptable to fix on locations since it seems unlikely that many peOple perceive many places to be either specific to a certain activity or a particular time Of day, or both.45 The basic hypothesis is, therefore, that the critical determinant of the structure of a person's day, given environmental forces (i.e., Objective constraints), is the extent to which one feels constrained relative to certain activities, times, and locations. These fixes, or points, about which one's day tends to be organized will act as the crux of the simulation experiment. RESEARCH STRATEGY In order to substantiate or illuminate the preceding pro- positions, a survey was undertaken. The remainder of this chapter reports on a research design for the collection of pertinent data and includes: (1) a consideration of some methodological problems of time-space budgets, (2) the selection of a survey site, (3) the design of a sampling procedure, (4) the development of survey site, and (5) the administration of the survey. methodology of the Time-space Budget Diary FOr any given group of peOple, a full set of time-space budget diaries provides a very comprehensive data base from'which to approach an understanding of the problems that this research confronts. An individual's time-space budget can give spatial as 45Cullen and Nichols, op. cit. 118 well as temporal coordinates of one's behavior in a given period. In most time-budget accounts, however, spatial coordinates have not been obtained or, if they have been obtained, they have not been used in the analysis. Where time spending has been considered explicitly in a spatial context, it usually has been given only partial treatment. Fbr example, it has been used as a surrogate for distance to shape or workplaces, without reference to overall time- budgeting and spatial organization.46 Problems of Data and Analysis. Considerable difficulties confront the use of a time-space budget approach and in combining temporal and spatial analysis. Time-space budget data are expensive to collect and, for this reason, many studies are based on a short period of time such as a day or a week. Perhaps more as a matter of faith than of fact, this short time period is assumed to be typical. Activities which follow a longer time cycle may be neglected (e.g., seasonal ones),l'7 thus altering the factual information. It is not easy to check for accuracy, or to check that recorded activities are in general typical of recurring behavior patterns. Therefore, it is frequently necessary to Obtain information, via questionnaires, on 46James Anderson, Time-bpdgets and Human Gquraphy: Notes and References, Graduate Discussion Paper NO. 36 (London: Department of Geography, London School of Economics and Political Science, 1970), p. 3. As an example, see: Chapin and Logan, Op. cit. 47MEad is one of the few researchers to have concentrated on time-budgets and seasonal variations; see: W.Ro Head, "The Seasonal Round: A Study of Adjustment on Finland's Pioneer Fringe," Tijdschrift voor Economische en Socials Geografie, 49 (July, 1958), pp. 157-162. 119 activities which occur irregularly or follow a longer cycle than the activity record period, although seasonal variability presents difficulties. Also, the willingness of respondents to record activities declines fairly rapidly, whether or not payment is involved. Anderson has found that about three days seems a critical 48 limit for relatively comprehensive diary records. However, a shorter activity record period can facilitate comprehensive treatment of important daily and weekly rhythms. Lipggistic A_spects of Time-space Budgets. Additional problems of the use of time-space budget data can perhaps be clarified by noting a distinction between substancelanguage and coordinate language."9 The semantic rules of a language must specify the manner of designating and identifying the objects in, its domain of discourse, a process referred to as "individuation." The designation of Objects commonly uses prOper names, which is the mode for substance and thing languages, or positional coordinates, which characterize coordinate or time-space languages. The former, an aspatial language, describes characteristics or properties of individuals (e.g., incomes, occupa— tions, life styles, types of activity, etc.), while their locations or positions are described in terms of a time-space language. The latter, a locational coordinate system of four dimensions (conventionally written as x, y, z, t), is associated with the so-called "physicalists R 48Anderson, "Space-time Budgets and Activity Studies in Urban Geography and Planning," Op. cit., p. 357. 49David W. Harvey, Egplanation in Geography (New York: St. Martin's Press, 1969), pp. 191-229. 120 who advocated the unificatiOn of science through the language of physics.50 Wilson,51 followed by Dacey,52 suggests that "individuation" requires the use of time-space coordinates. Individuation of an individual may result either from attributes that the individual manifests or from the position one occupies. The individuals in the domain of most languages are commnly identified by a name or by presence of non-positional attributes and the time-space regions occupied by each individual are seldom explicitly taken into account. For this reason, the semantic rules of designation and individuation generally ignore spatial-temporal attritubes of indi- viduals. And if individuals are defined only in substance terms (i.e., non-positional attributes), they lose part of their essence. Harvey points out that this is particularly essential to geographic study in that the ordering of geographic information "amounts to the difficult logical problem of working with two different language soThe physicalists generally contend that the physical, cultural, and biological things considered by empirical science are largely time-space regions of the four dimensional time-space continuum: each thing occupies a definite region of space at a definite instant of time and occupies a temporal series of spatial regions during the whole history of its existence. Thus, all empirical science may be unified through the language of physics. See: Oscar Neurath, "Foundations of the Social Sciences," Inter- gational EncyclOpejdfiie of Unified Science, Vol. 2 (Chicago: Univ- ersity of Chicago Press, 1944), pp. 1-51. 5J‘N.L. Wilson, "Space, Time, and Individuals," Journal Of Philosophy, 52 (1955), pp. 589-598. 52Michael F. Dacey, "Linguistic Aspects of Maps and Geographic Information," Ontario Geography, 5 (1970), p. 78. 121 systens in the same context";53 a problem magnified by the fact that geographers usually deal with both discrete and continuous data. The problem of using both languages lies at the heart of time-space budget analysis. The substance, or non-positional, language is itself multi-dimensional (e.g., different types of respondents and activities, attitudes, non-locational cont raints) , and adding the positional, or time-space, language further complicates the problem. Hagerstrand's use of computer simulation models, where activity schedules are sieved through a time-space environment in order to evaluate the constraints imposed by the environment, is perhaps the most complete answer to these linguistic problems.54 Data Requirements. All surveys which involve the collection of time-space budget diaries may be used to collect a set of activity modules which together may be taken as a definition of that individ- ual's behavior over a fixed period of time. Each module contains information which has between two and four dimensions of variation. The most basic dimensions are those of activity, time, and location. Thus, each activity performed must have some descriptive title, a 53David W. Harvey, Explanggon in Geography (New York: St. Martin's Press, 1969), pp. 216-217; and for additional coments on the problem of individuation, see: David W. Harvey, Social Justice and The Cit , Op. cit., pp. 38—40. 5"Torsten Hagerstrand, 'Tatortsgrupper som regionsathllen: Tillgdngen til fbrviirvsarbete och tjanster utanfor de storre stHderna," _R_e_gioner att leva i, Expertgruppen fbr Regional Utredningsverksamhet (Stockholm: AB Allmanna Forlaget, 1972), pp. 141-173. 122 rather precise location in time, including start and end times, and a spatial location. A final dimension that is sometimes included 13 relational; some information is collected about other individuals, if any, in whose company the event occurred. Furthermore, each of these four dimensions lend themselves to differing degrees of elaboration and integration. FOr example, if a respondent is watch- ing television while eating dinner, a secondary activity might be recorded. And, if appropriate, an activity may be described in terms of the degree to which it was anticipated or unexpected, and further subjectively ranked according to the constraining effect of the major dimensions of time, space, and activity. Survey Site With this overview of the data requirements in mind, the first stage in the research design was the selection Of a survey site and sampling procedures. The sample of respondents was drawn from the faculty, staff, and students of Michigan State University, located in East Lansing, Michigan. The university, a territorial enclave‘within the Lansing, Michigan metropolitan area, maintains a population of some 35,403 students, 3,463 faculty, and 4,703 staff.55 Owing to its sheer size, in terms of population and areal extent, such an institu- tion has often been referred to as "megaversity" or "multiversity." To many of its residents, the university is Often viewed as a 55These data are for the Fall Term.of the 1973-74 academic year, and represent only full-time faculty, staff, and students; source: Department of Information Services, This is‘Michiggn State Universi_y: 1974 Facts Book (East Lansing, Michigan: Department of Information Services, Michigan State University, 1974). 123 well-defined and self-contained entity. That is, within the confines of the institution, many of its members can satisfy almost all of their needs and wants in the conduct of thEir everyday affairs. These range from such physiologically defined needs as sleep and sustenance, to culturally defined extensions such as working, shOpping, and home- making, and to a variety of leisure pursuits. The spatial configuration of the university is in some ways analogous to that form commonly associated with the city, i.e., seen as a city in microcosm. It is characterized by a core of administra- tive and information-processing units that have important links and interdependencies necessary for personal communication and face-to- face contact. This core is ringed by classroom buildings and Office structures in which the majority of business is transacted in the process of education and research. Also located in this zone is the majority of commercial establishments such as stores and restaurants. This zone is, in turn, engulfed by living quarters for the academic staff and students, ranging from high density dormitories to lower density apartment complexes as one moves toward the periphery. In terms of aggregate population shifts from.core to periphery, and vice versa, during the course of a day, the diurnal movements of.the univ- ersity's pOpulation are not unlike those of a city's inhabitants. Also, the variety of transport modes available within this institu- tional complex rivals that common to the city. There is, however, the danger of carrying the city/university analogy to extremes. .Although the university may be thought of as an independent or semi-independent entity, it is in reality, quite dependent on its surroundings. This is borne out by the fact that, 124 just as the linkages among the members of the university are numerous and complex, so too are the linkages it maintains with the surround- ing environment. The university is seen, therefore, as just one of a variety of subsystems of the larger urban system. Subsystems of the urban system can be distinguished in several ways. They can be described as spatially defined areas of the city; alternatively they can be delimited as operational units, for example, offices, organizations, industrial plants, or educational institutions. In each case, a more detailed knowledge of how these subsystems operate will assist in understanding the apparent complexity of the larger urban scene.56 In striving toward this understanding, geographers and planners are giving increased attention to subsystems, such as institutions and organizations and their internal and external linkages over space, rather than from sampling randomly from the larger urban population or from.socio-economic strata.57 In moving from the urban scale to that 56Nicholas Bullock, Peter Dickens, and Philip Steadman, "The MOdelling of Day to Day Activities," Urban Space and Structures, Cambridge Urban and Architectual Studies 1, eds. Leslie Martin and Lionel March (Cambridge: Cambridge University Press, 1972), p. 127. Equally of course, these parts or subsystems--such as, for example, the particular institution here chosen for investigation, the unive ersity--are deserving of study in their own right for the special geographic, planning, and architectural problems they present at their particular scale, besides as components Of the larger urban problem. 57Some examples include: Vida Nichols, An Institution in Metropolis, Seminar Paper No. 12 (London: Joint Unit for Planning Research, University College, London, 1969); Britta Ohlsson, Inter- institutionella kontaktflbden, Urbaniserins-processen, rapport nr. 8 CLund: Institutionen fUr kulturgeografi och ekonomisk geografi vid Lunds universitet, 1967); John Goddard, Office Linkages in Central London: Volume II (London: London School Of Economics and Political Science, 1971); Gunnar E. Tbrnqvist, Contact Systems and Regional Development, Lund Studies in Geography, Series B, Human Geography, No. 35 (Lund: C.W.K. Gleerup, 1970); and, Olof WHrneryd, Interde- dependence in Urban Systems, Meddelanden fran institutionen 125 of the institution, a much.wider and more complex range of activities ‘must be examined than just those distinctions that might be made for an urban model. FOr the university, the great variety of types of activity in which many of its members engage, and the relative free- dom which they have to organize their time, make this perhaps an especially complicated case to consider, when compared with the more regular routine of the factory or Office worker, or housewife. In many respects, the sample of the pOpulation affiliated with a university might be expected to be a highly abnormal one, since university personnel generally are thought to have a degree of free- dom.governing their working hours far in excess of the working popu- lation. The general expectation would be that the greater the degree of freedom within this group, the greater the variation between individuals in the ways in which they structure their days. And this variation would be expected to be much larger than within other homogeneous groups of working peOple.58 However, for these and other reasons, this pOpulation is particularly interesting and promising for study. The complex range of activities Open to members of the univer- sity institution leads one to expect a greater variety of subjective constraints in accommodating the greater range of activities in their kulturgeografiska, nr. 1 (Gateborg: Kulturgeografiska institutionen vids theborgs universitet, 1968). 58This problem, however, does not appear too serious since the research methodology is amenable to replication. Underlying this point is the conviction that a research event should be viewed as a single step in the context of a larger, continuing process. 126 daily routines. For example, the university institution depends, to a great extent, on personal and face-to-face contacts in the conduct of its everyday affairs. The commitments and time-space fixities assoiiated with such negotiations produces a great variety of subjec— tive constraints Operating on daily behavior. In this respect, the university institution is seen as a limiting case of a variety of tertiary and quaternary institutions.59 Therefore, a.model of the time-space mechanics of constraints Operating on the daily activities of university personnel should have relevance to other subsystems or institutions in the larger metropolitan system, where the modelling may indeed be a more simple one.60 The emergence of clearly defined patterns of behavior among members of this group, despite their loosely structured budget-space, provides interesting information concerning people's habit of structuring their day in a standardized form regardless of the lack of direct constraints. In relation to this matter, Cullen and others note that: It must also be remembered that only about half the population is employed in any sort of regular job and that there are many housewives, old peOple, and children who have even fewer hard and fast commit- ments imposed upon them from outside than these supposedly unfettered students and dons. In some respects, university people, although abnormal, are perhaps quite middle-of-the-road in terms of the degree of freedom which they have to structure their own time.61 59Cullen and Nichols, Op. cit., p. 5; Nichols, op. cit., p. 14. 6oBullock et al., Op. cit., p. 130. 61Cullen et al., op. cit., p. 11. 127 Sampling;pesigg As the first stage of the sampling design, one institution, the IMichigan State University, was selected as a base for investigation. Although the rationale for its selection as one subsystem of the larger metropolitan system was discussed above, a few additional comments are necessary. In order to adequately study the time-space mechanics of constraints operating on the formation of activity sequences, atten- tion.mnst be focused on the paths, or behavior of the individual through time-space. The instrument chosen for collecting this type of information is the time-space budget diary. The collection of diary information from a sample of people randomly selected from the entire metropolitan complex, or even from.a number of its various subsystems (e.g., business or service organizations and institutions) would not be feasible in that responses would be spread so thinly among them as to make it virtually impossible to discern sequences or patterns of behavior. At this stage of research into the structuring of daily activities, more is to be gained through investigating the members of a single organization or subsystem in order to establish the pattern-of linkages associated with one unit. Once the first step in the sampling design is completed, the selection of the university institution, the procedure of sampling from among its members remains. When the problem.of sampling elements from any universe arises, the issues of representativeness and reliability come to mind. Typically, these issues are important in the methodology of cross-sectional studies dealing with the aggregation 128 and prediction of individual behavior.62 In this research project, however, the focus is on the structuring effects that the interaction or mechanics of constraints have on daily behavior. It was in regard to these issues that Hagerstrand pointed out that, "a large sample survey of actual behavior is not essential since emphasis is laid on the working of constraints rather than the aggregation of their "63 In a behavior patterns, the latter demanding a larger sample. previous section of this chapter a distinction was made between choice- and constraint-oriented approaches to time-space behavior. Had the former methodology been adapted, the necessity of a large sample would have been a significant issue. With the issue of sample size relaxed, a survey population of some 200 individuals, or members of the university population was sought.64 The obvious sampling frame is the total university population, or, the list of all individuals who are affiliated with the institu- tion. However, many who could be considered a part of this group seemed inapprOpriate to the purposes of the study. As a result, it 62R.A. Carr-Hill and K.I. Macdonald, "Problems in the Analysis of Life Histories," Stochastic Processes in Sociology, ed. R.E.A. Mapes, The Sociological Review Monograph 19, ed. Paul Halmos (Keele, England: University of Keele, 1973), p. 58. 6311agerstrand, "The Impact of the Environment and Social Organization," op. cit., p. ; Carr-Hill and Macdonald agree that studies with similar objectives have too often tried "to deploy the methodology of cross-sectional studies, using large samples when a few subjects would suffice for the discovery of a developmental sequence." See: Carr-Hill and‘Macdonald, loc. cit. 64Realizing the likelihood of non-response, the sample size was set at 200 (i.e., a sampling ratio of 0.5 to the survey popula- tion), so that the number of respondents, or, successfully inter- viewed persons would amount to at least 150. 129 was necessary to define the survey population in a somewhat more restricted fashion. The final definition included three general groups which are defined by university affiliation: (l) full-time students registered for the Fall Term, 1973, on the East Lansing campus of the university, (2) full-time faculty, including those in instructional and research progr-s, and administrative officers employed on the main campus during the Fall Term, 1973, and (3) full-time clerical-technical and labor staff employed during the Fall Term, 1973, on the East Lansing campus of the university. The complete lists of individual's names and addresses were obtained for the Fall Term, 1973, and were stratified by type of affiliation as described previously.65 Precautions were taken to ensure that no one individual's name appeared in more than one of the lists, so as to guarantee that all members of the survey pOpulation had an equal chance of being selected in the sample. Once the individuals' names had been arranged by affiliation, a systematic sample was selected across-the entire list. The sample size for the study was set at 200 observation units. To achieve this sample, the sampling procedure employed a ll200 sampling fraction, or a sampling ratio of 0.5. Thus, in using the systematic sampling method, every 200th individual in the list was chosen for inclusion in the sample. To guard against any 65The data sources for the three groups included: (1) a roster of all full-time students enrolled during the Fall Term, 1973, on the main campus was obtained from the registrar's office; (2) a roster of all full-time faculty and administrative officers was obtained from the office of the executive vice president; and (3) a list of all clerical-technical and labor staff was received from the personnel office, Michigan State University, East Lansing, Michigan. 130 possible bias, the first element was chosen at random. The random start was made by generating a random number within the range of the sampling interval. The totals for the sample, stratified by type of university affiliation, are given in Table 1. TABLE 1. TARGET SURVEY POPULATION TOTAL PROPORTION TARGET SURVEY AFFILIATION POPULATION OF TOTAL POPULATION STUDENTS: (full-time only) 35,403 81.2 162 EACULTY: 3,463 8.0 16 Instructional Programs [2,229] Research Grants & Others [ 557] Administrative- Professional [ 657] STAFF: 4,703 10.8 22 Clerical-Technical [2,128] Labor [2,575] TOTAL: 43,569 100.0 200 The next step in the sampling design was to assign each individual in the sample to a weekday for survey. In other words, the sample'was further stratified by weekday of participation. In order to counteract 131 any possible bias in the stratification by weekday, the list Of individuals in the sanple was first randomized. After the reordering of elements, via random number techniques, the list was divided into quintiles; the first 40 individuals being assigned to a MOnday survey date, the second quintile of individuals being assigned to a Tuesday survey date, and sO on through the fifth quintile which was assigned a Friday survey date.66 Preliminary Survey Before administering time-space budget diaries and supplemental questionnaires to the selected sample, several considerations made a preliminary survey advisable. Perhaps most important was the need to test several different methods Of data collection pertaining to individuals' activities in time and space. In a preliminary survey conducted at Michigan State University, four survey formats were tested and compared. (See survey options l-4 in Figure 16.) Each of the four Options included a time-space budget diary and a supplemental questionnaire. The survey Options differed according to two criteria: (1) the extent to which the survey relied upon interview techniques and (2) the degree to which the diary format was structured. The four alternatives were designed and implemented in such a way as to determine 66The survey was conducted over a continuous five-week period during the months Of October and November, 1973. In the case Of the first quintile of 40 individuals, for example, an equal nunber Of eight individuals were assigned to each Of the Mondays falling in the survey period. The same practice was followed for the remaining quintiles and their respective weekday Of Observation (Appendix E). 132 O S m m m 3 3 n- ’55 m Ea ... (D U) >* :5 .5 >* .. . .. . :2: 5: O O {E -o'o -o 'o r4 '0 'o ,4 O O >a O O c: O O I: 'U'U 5§ ‘U '0 O O o o o O A 0) OJ Q U U H U U '5 :4 :4 a: O c: Q a c: O n. O. a: H O o O G) O G- t: A C t: O m (D U a F‘l U Q] g 0) U) Q U) (0 ca -H °H O O o. O O V c: U DO H “H O H “H :1 -H c: u v g u c: > >~ s-a -a In O -H .4 > c: > G a U i O) 'H LIJ g 0H O H U a u a u c) E.A< +4 -a O <3 -a O E! ,\ " E3 Efl‘d £3 £4 ‘< INTERVIEW “‘ H N m 338‘ 3. 3 9 ‘ INTERVIEW SITUATION 1) Diary completed in OPTION retrospect; 3 2) Self—administered questionnaire. ([6) PARTIAL INTERVIEW SITUATION 1) Diary kept during OPTION OPTION 24-hour period; 4 5 2) Administered interview schedule. (I7) ‘ NONINTERVIEW SITUATION 1) Diary kept during OPTION op'non 24-hour period; I 2 2) Self-administered ' questionnaire . (I 9) ( l9) NONINTERVIEW SFTUAHWOWI FIGURE 16 SURVEY FORMATS TESTED IN PRELIMINARY SURVEY *In the preliminary survey, a total Of 20 individuals were asked on four occasions to complete diaries (Options 1 through 4). The value within parentheses indicates the number of respondents completing this survey format. **Format Option chosen for the Michigan State University Survey. 133 'which method, or combination Of methods, would produce the most reliable results, keeping in mind the time constraints of the survey as a whole, and the respondents in particular. In selecting individuals for pretests, it was neither possible nor advisable to look for anything like a representative sample. Repre- sentativeness is no criterion in a pretest in which the Objective is rather to test the degree to which variations in procedure affect the reporting Of results. Given such an Objective, the respondents need to be chosen in such a way as to represent a maximum variation with respect to whatever characteristics are likely to influence reactions to a particular data collection technique.67 Also, the decision to use cOmbinations Of survey formats for describing the behavior of one individual presupposed that a stronger motivation to cooperate was needed than is necessary in a normal survey. The latter consideration resulted in an overrepresentation Of persons in the student and faculty groups Of the survey population. Accessibility was the main considera- tion in selecting cases for joint use Of Observational techniques, and to some degree the investigator was obliged to rely on networks of acquaintanceship. Survey Formats. The first method (Option 1), a noninterview, was designed tO collect the desired survey information by mail. This method of self-reporting involved the respondent filling in a structured diary in which time units for the 24-hour period were precoded by fifteen mdnute intervals and classes Of activity were predefined. For each 67Earl R. Babbie, Survey Research Methods (Belmont, CA: Wads— worth Publishing Company, Inc., 1973), pp. 206-207. 134 activity at a given point in time, the respondent was asked to indicate its location, if anyone else was involved, and if the respondent was doing anything else at that time. In addition to the diary, the respondent was asked to complete a self-administered questionnaire that would provide the needed background information about the individual being surveyed. [For a similar questionnaire, see Appendix B.] [The second method (Option 2) also relied on self-reporting in a noninterview situation. However, this alternative differed from the first in that the respondent was asked to complete an unstructured, or "Open," diary. Here, the respondents were to describe in their own words their activities over the 24-hour period, giving their times and locations as exactly as possible. FOr each activity, the respondents were also asked to report any secondary activity and to indicate if any- one else was involved. In addition, the completion Of a questionnaire, similar tO that given in Appendix B, was required. Unlike the previously mentioned alternatives, the third method relied on "yesterday interviews" in which the respondent was asked to provide an interviewer with an account of activities for the preceding day. The diary format used by the interviewer was partially precoded, in that only time units were defined by fifteen minute intervals. Other- wise, the remaining categories Of response were Open. These included: a description of the activity, its location, if anyone else was involved, and if the respondent was performing a secondary activity. The use of interview techniques permitted the interviewer to pose more detailed questions about activities, especially secondary ones, to be reported in the diary, and tO administer an interview schedule. 135 The fourth and final survey format tested had the respondent keep a diary during the assigned day Of Observation (Figure 16). The diary format was partially structured, in that only time intervals of fifteen minutes duration were precoded, leaving the activities and their loca- tions to be described in the respondents own words. The respondent was also instructed to record secondary activities where appropriate and other persons in whose company the activities were performed. On the following day, an interviewer administered an interview schedule and reviewed the diary with the respondent, probing for clarification in recorded activities, if necessary, and asking the commitment and fixity questions for all activities (Appendix B). With all Options, questions relating tO one's degree Of commitment to activities and their subjective fixity ratings were posed. In cases where the diary information was obtained via interviews, the questions were asked by the interviewer, whereas in noninterview situations, the questions were printed in the diary with space allocated for responses for each activity reported. In attempting to discern an individual‘s commitment to a given activity, the respondent was asked whether or not the activity had been planned. -For information concerning the spatial and temporal fixity, the following questions were asked for each activity reported: (1) Could you have done this (activity) elsewhere? and (2) Could you have done this (activity) at another time? It was thought that the use Of precoded diaries in non-interview situations (Option 1) would allow for a greater number of survey respondents than noncoded diaries, but with the risk that this information might be too sketchy or too innacurate to be Of any use in modelling individuals' activities in detail. A diary in which both the activities 136 and the time intervals are precoded would be simple to fill in and would thus guarantee a minimum level of detail. But by precoding, much of the detailed information would be forfeited. First, the range Of activities would be greatly reduced, although space was provided for respondents to add further activities which did not fit 'within those categories listed. Secondly, by precoding the time units in intervals of fifteen or twenty minutes, it would become necessary either to disre- gard trips or activities Of short durations or to round them Off. Thus, much Of this type of information would necessarily be inaccurate, particularly for trips around campus or in town, in which cases the majority are less than fifteen minutes. An additional disadvantage to the use Of this kind Of format is the ease with which fictitious activities might be entered into the diary. The noncoded, or Open, diary (Option 2) used in a noninterview situation would not only make it possible to collect several times as many diaries as in the interview situation (Options 3 and 4), but it would also preserve this information in a more detailed form than the precoded diary. The coding Of activities could be undertaken after the survey and could thus take into account the actual range Of actiVities described. FUrthermore, the difficulty Of choosing a fixed time unit would be overcome automatically. An additional advantage Of this diary is that it would become very tedious to enter fictitious activities. However, an argument against this is that the quality Of the diary might deteriorate over the 24-hour period. It was thought that the interviews (Options 3 and 4) would provide both reliable and detailed information of an individual's activities for a 24-hour period. Some advantages claimed for the interview were the 137 possibility, with an experienced interviewer, of detecting if the inter- viewee were making up activities, and of prompting the respondent should he find it difficult to recall his activities or should he have difficulty interpreting the commitment and fixity questions. Results of the Preliminary Survey. The results of the preliminary survey confirmed expectations that data on time and space use are very sensitive to even relatively minor variations in the procedures of data collection. This sensitivity effects both the reliability and the validity of the data. There is no obvious and absolute measure by which to judge the results of any of the techniques used during the pretests as "true"; the only basis for judgements are the variations among the techniques that record the behavior of the same person. However, if a particular technique consistently produces more complete and detailed records, then the technique that produced more complete records may reasonably be treated as "better." The greatest difference between the various combinations of formats was Observed with respect to activities that were incidental to other activities. Consequently, completeness Of recording secondary activities was used as one means of judging the quality of data collection. Another noticeable difference among the various formats was the level of detail ‘in recording the spatial locations of activity and movement between activity locations. The permutations of combinations allowed then a rank- ing of different techniques with respect to their completeness in record- ing behavior, and with regard to their suitability in reproducing various types Of behavior. This procedure is analogous to the reasoning in using paired comparisons as a scaling procedure--each recording technique or sur- vey format serves as a yardstick for every other technique. 138 No one technique with which the investigator is aware will result in "perfect" data. But among the techniques tested during the preliminary survey, a rather definite sequence in the reliability and validity of data collection procedures could be observed. The results of the survey showed that the partially precoded diary used in an interview situation yielded the best results (Option 4 in Figure 16). Although it did not produce as many diaries as some other methods in a comparable amount of time, it did provide the best level of detail. Respondents were willing to complete these diaries in considerable detail and there was no con- sistent reduction in the number of entries toward the end of the day. The diaries completed in retrospect, the yesterday interviews with aided recall (option 3), consistently did not produce sufficient detail as those kept during the 24-hour period. It should be noted that the open diary method (option 2) provided an adequate level of detail. Considering that the open diary method was used in a noninterview situation, suggests that the method has even greater potential in an interview situation for providing greater detail. The effects of an interview versus a noninterview situation were also noted. A comparison of the account of a given day in an interview with that from a noninterview revealed that much additional detail can be recovered from the use of diaries in interview situations. In the case of the total interview situation, the yesterday interview (Option 3), the interviewer could aid recall, pose questions of clarification where appropriate, or probe for further detail. The interviews were particularly helpful for respondents and interviewers alike in clarifying ambiguities in the commitment and fixity questions. 139 The noncoded and partially precoded diaries were considerably more detailed than the precoded diaries in the number of entries and the range of activities described for both primary and secondary activities. Although provision was made for respondents to enter extra activities in the precoded diary, only the most obvious additions were made and these diaries were too unspecific about the type of activity. unexpectedly, a number of respondents remarked that it appeared easier to fill in the noncoded diaries than the precoded ones, largely because with the latter it was difficult to find a suitable description for an activity in terms of the small number of categories provided and also because of the difficulties of rounding the duration of their activities to fit within the time period provided. Choice of Survey FOrmat 0f the four survey options tested, the second and fourth methods (Figure 16) yielded the best results. The primary advantage of the second method was its use of an unstructured, or open time-space budget in a noninterview situation. The only recognized weakness of the fourth method was that the diary format was partially precoded--time units being defined by fifteen minute intervals. The format chosen for the final survey was, therefore, a combination of the second and fourth methods. [See Option 5 in Figure 16.] Final Time-space Budget Diary. The fifth survey method involved the respondent keeping a diary during the assigned day Of observation. Since the diary format was unstructured, the respondents were asked to describe their activities over the 24-hour period, giving their times and locations as exactly as possible. In addition to this information, the 140 respondents were asked to further indicate for each activity the other persons present and whether or not a secondary activity occurred at the same time. Thus, all of the core information of the time-space budget diary would be recorded by the respondent without an interviewer present. Since the remaining information of the time-space budget--the data on commitment to and fixity of activities-are particularly important to this research, they were obtained by interviewers on the day following the completion of the diary. The pretests were especially helpful in revealing the inadequacies of the questions relating to commitment and subjective fixity ratings of activities. The pretests demonstrated that the response categories to the commitment question, planned and unplanned, were too limited. The decision was made, therefore, to expand the categories of response to include a broader range of flexibility, as discussed in an earlier part of this chapter. In attempting to discern one's degree of commitment to a given activity, the question--"TO what extent was the activity planned?"-dwas changed to include four categories of response: (1) arranged with others, (2) routine, (3) planned independently, and (4) unexpected. Although some of the categories were thought to be self- evident, the interviewers were instructed to define each of the possible responses for the purposes of standardization. Interviewers were to advise respondents that routine activities were to be construed as those activities which almost invariably occur with a particular set of circum- stances, even though the set of circumstances might only recur at intervals of a week or more. Hence, the degree of regularity of routine activities remained flexible. The degree to which an activity was planned, as in response categories 1 and 3, was similarly left Open. If 141 the respondent claimed to have planned to do a given activity at some time during the day, it was considered as planned. The format of the subjective fixity questions was also altered to better match the conceptualization of the constraining effects that the dimensions of time, space, and activity have on the structuring of daily behavior. Only two of these dimensions were treated during the pretests. Following Cullen and Nichols' suggestion,68 a four-part fixity question ‘was included in the final time-space budget. Each of the four questions, designed to reveal subjective constraints, seeks simple yes/no responses (Table 2). TABLE 2. THE FOURPPARI SUBJECTIVE FIXITY QUESTION subjective Fixity Question Type of Constraint (1) Could you have done anything else at Activity choice this time? fixity (2) Could you have done this (activity) Temporal fixity of at any other time? activity (3) Could you have done this (activity) Spatial fixity of elsewhere? activity choice (4) Could you have been elsewhere at Spatial fixity of this time? activity location The first question is phrased in such a way as to discover whether or not the individual had any choice of activity Open to him. The second 68Cullen and Nichols, Op. cit., p. 12. 142 question inquires about the timing of the chosen activity and is phrased in such a way that the respondent is forced to isolate the timing from the choice of activity.69 The third and fourth questions seek informa- tion on the spatial fixity of a given activity. The third part is the direct spatial equivalent of the second in that it is similarly dependent on the choice of activity. The subject of the fourth question is the spatial fixity of the activity location. The fourth question is similar to the first in that the first focused on the choice of activity avail- able to the individual during a given unit of time, while the fourth focuses on the same unit of time but on the individual's potential choice of location in it. Since the importance of this may not be immediately evident, consider the following example: A sequence of events may effectively tie a person to a particular place for short periods simply because there is not sufficient time to go elsewhere during the intervening period (e.g., successive meetings in the same office building). A need arises for a standard of comparison for the detailed fixity question. Prior to answering the four-part fixity question (Table 2), the respondent was asked to review the day's activities as he had recorded them. While doing so, he was asked to isolate those activities or episodes which, in his opinion, were important in the sense of having had to be done at a fixed time or location and thus having been points about which he felt his day had to be organized. Once these activities had been isolated, the respondent was asked to further subjectively rank 69Ibid. In order not to confuse the timing of an activity with the choice of a particular activity, Cullen and Nichols recommend two inde- pendent questions rather than the more direct question: Did you have to do this activity at this time? 143 the activities in order of their importance. This simple ranking, then, was to act as a measure against which to assess the detailed fixity question. As a result of the pretests, further decisions were made regard- ing the format of the final survey. For example, the 24-hour weekday was retained as the unit of observation. Such a limited time span might have grave restrictions on research whose purpose is the quantitative analysis of aggregate behavior patterns, and where a typical day of observation is a significant issue.70 However, the 24-hour observa- tion period was not considered a restriction on this research. The objective focuses on the understanding of constraints, both objective and subjective ones, that affect the structuring of daily behavior, regardless of the typicality of the day of observation. However, it was observed during the pretests that a great many activities of con- siderable importance for the interviewees such as participating in sports, going to the theater or church, or meeting regularly with friends at a bar were quite infrequent during any one day. In order to have some idea of the relative frequency of such activities, in the event that such data might prove useful at some unforeseen stage of the research, data relating to these usually infrequent activities were collected in the interview schedule (Appendix B). An additional problem in recording use of time derives from the fact that many peOple during a large portion of the day do more than one thing at a time. To catch this aspect of time use, a distinction was 7oErwin K. Scheuch, "The Timeébudget Interview," The Use of Time, ed. Alexander Szalai (The Hague: Mbuton & Co., 1972), p. 75. 144 made between primary and secondary activities. A primary activity was defined as any act that was determined by a person's location, and/or his interaction partner plus his commitment response. A secondary activity was defined as an activity performed concurrently with primary activities. By and large, a primary activity represents the rough organization of a day, as a consequence of the institutionalization of society and the combination of duties that result from a person‘s configuration of roles; and, in general, a secondary activity represents the preferences of an individual. Although a number of respondents managed to do three things simultaneously such as eating, watching television, and conversing with Others present, it was decided for practical reasons not to record more than two activities during any one time period. While the term activity seems to be a notion which is clear enough, it became evident during the preliminary survey that a more formalized definition had to be adOpted for recording and coding purposes. Activity, therefore, was defined as any behavior where any of the following conditions remained unchanged: the common sense term used by the respon- dent either for primary or secondary, the location; and the interacting partner. Or to define activity more positively: as soon as any one of the four dimensions characterizing behavior changed, this would mark the beginning of a new activity. Interview Procedure. The standard interview procedure used during the final survey was as follows: 1. During the day t-3 (i.e., three days preceding the one during which the respondent kept records) an interviewer telephoned the respondent, briefly explaining the intent of the survey and asking for the respondent's OOOperation. 2. During the day t-Z 145 an interviewer visited the respondent's university address and‘ familiarized the respondent with the self-recording procedure. 3. The respondent would fill out his diary for the day t (i.e., the assigned day of Observation), either during the day or during the evening of this day. 4. During the day t+l (i.e., the day following the one for which the respondent's diary was kept) the interviewer would return and conduct an interview based upon the diary of the respondent and an interview schedule. The interview procedure in step 4 was as follows: 1. The inter- viewer carefully reviewed all entries that the respondent had recorded in the time-space diary, checking to see that: a) the reporting of activities had been accomplished at an acceptable level of detail; b) no portions of the respondent's diary remained unreported; c) activity locations were clearly indicated; d) the times spent travelling to and from activities were recorded as separate activities e) modes of travel were specified; and f) activities and associated information were reported sequentially. If the level of reporting was too general, the interviewer prObed for more details about the activities. 2. After any probing for clarification, the interviewer asked the respondent to go through the diary and for each activity to provide an answer to the commitment question: To what extent was the activity planned? 3. The » interviewer then instructed the respondent to review the day's activities as recorded. While doing so, the respondent was asked to pinpoint those activities or episodes which were important in the sense that they had to be done at a fixed time or location. Once these activities were selected, the respondent was asked to rank them in their order of importance. 4. Then, the respondent was asked the four-part fixity question for all 146 recorded activities except sleep and travel. 5. Finally, the respondent was to answer the questions in the interview schedule (Appendix B), designed to collect certain socioeconomic background information about the respondent. CHAPTER 4 BEHAVIOR PATTERN RECOGNITION: EMPIRICAL ANALYSIS FOR MODEL DEVELOPMENT The purpose of this chapter is to evaluate the preliminary hypothe- sis regarding the principle of consistency in human behavior. The claim was made earlier that the concept of pattern can be applied not only to the sequencing of activities, but also to their durations. The recogni- tion of pattern and consistency in the recorded behavior of the sample population is essential prior to the develOpment of a simulation model. Before testing the consistency hypothesis, the survey results are briefly summarized and the structure and coding of the resulting data are discussed. Several analytic approaches are then reviewed in order to determine how best to reduce the vast battery of survey data into a more manageable form for analysis and eventually simulation. SURVEY RESULTS The university survey generated 138 24-hour time-space budget diaries. Out of a total survey pOpulation of 200, this represents a response rate of 69.5 percent (Table 3 and Appendix E). Each of the diaries obtained in the survey was composed of an average of 35 activity modules. The description of any activity module entails a vector of 15 elements of information (Table 4). Although the survey is a modest one, it has yielded a final data set of about 80,000 elements. 147 148 TABLE 3 RESPONDENTS AND NONRESPONDENTS BY UNIVERSITY AFFILIATION University ' Survey . Response Affiliation POpulation Respondents Nonrespondents Rate Students 162 105 57 64.8 Faculty 16 14 2 87.5 Staff 22 20 2 90.9 Total 200 138 62 69.5 TABLE 4 VECTOR OF INFORMATION DESCRIBING AN ACTIVITY MODULE* Element ' Description 1 Starting time of activity 2 Ending time of activity 3 X coordinate of activity location 4 Y coordinate of activity location 5 Two-digit primary activity code 6 Two-digit secondary activity code 7 One-digit activity location code 8 Two-digit social contact code 9 One-digit code for travel mode 10 Ordinal ranking of activity importance 11 Degree of commitment code 12 Subjective fixity code--(l) activity choice fixity 13 Subjective fixity code--(2) temporal fixity 14 Subjective fixity code-(3) spatial fixity 15 Subjective fixity code--(4) spatial fixity *For more details on the organization and coding of activity modules, see Appendix C. 149 DATA STRUCTURE In order to comprehend such a Vast amount of information, the data must be represented in a compressed form. Regardless of the analytical strategy chosen to accomplish this compression, the use of a digital computer will be essential for anything more than the simplest exercise. The issue of how the data set is to be made more manageable and how comprehension may be achieved through collapsing the data set is the subject Of the ensuing discussion. The data set generated by the survey presents a variety of problems for analysis. The research prOpositions outlined earlier provide a comprehensive framework within which to perform analysis. They suggest a variety of questions, each of which might illuminate one or more facets of behavior which are considered important to this research. The problems that derive both from.the complicated nature of the questions and from the structure Of the data from'which the answers are sought require that a decision be made as to the appropriate unit of study and an appropriate analytical procedure. Unit of Analysis In most survey research, the unit of analysis is not a problem since a sample of individuals is asked a battery of questions and responses are converted into a vector of information. The vector is of fixed length and the individual remains the unit of analysis throughout the study. However, this is not necessarily the case when information is taken from a time-space budget diary and associated questionnaire. There is only one vector of information of a fixed length. It contains socio-economic information about the respondent as 150 well as some background information relevant to the interpretation of the diary.1 But there is also the diary information which composes a vector of variable length. Actually, it can be regarded as a.matrix of information, with the length of one dimension fixed, while the other dimension varies in length (Figure 17). The size of the first dimension is determined by the number of indiceschosen to describe each activity module. The second dimension, however, varies with the actual number of activity modules or episodes performed. Each element of the array may be of cardinal, ordinal, or nominal form depending on the aspect of the activity module being described. Inevitably, one is confronted with a choice in any analysis of behavior patterns as to which unit of analysis is most meaningful. In many instances, the research questions determine the unit of analysis, but this is not always the case. Consider, for example, the case where interest is focused on time allocations to various types of activity. FOr some activities which are performed by a few peOple, it may be best only to investigate those data describing the duration of activity episodes, i.e., an average amount of time that some aggregate of people devote each day to a particular activity. However, for other activities which tend to be undertaken by the majority of a pOpulation, especially activities spread throughout the day (e.g., work, eating, personal hygiene), the most apprOpriate unit of study is the individual. Thus, attention would not be focused on the data describing the duration of activities, but rather on that describing the amount of time that the individual allocates to activities of that type. In many cases the moat 1See codebook for supplemental interview schedule and time-space budget diary in Appendix C. 1- j- 151 ]. - ' 'l 2 [I 3 1 III] II, ' 11 ° Hi [I] n JIIIII l,2,...,n; where n = the total number of individuals in the sample; l,2,...,m; where m - the total number of activity modules, and where m varies from individual to individual. —__I l I l ——‘ 'h__\ I l J i e ' 3 I l l l I l — ——— _ — — —I—— ——— _ — ——_ ——— —_—‘ —__u —__ _—fi L4. I III Hll Ulllujufi thne-—- l,2,...,n; where n - 15, the total number of dimensions describing activity module j; l,2,...,m; where m - the total number of activity modules for individual i; the time an individual awakens at the beginning of the day; the time at the end of the day at which the individual retires. The units along dimension j are of unequal length to repre- sent the varying durations of activities. FIGURE 17. STRUCTURE OF TIME-SPACE DIARY INFORMATION 152 meaningful unit of analysis may not be as obvious. For example, when the question involves comparing one measure of an activity with another --its location with the number of participants involved--the focus of interest is the episode that has these characteristics. But, in order to convert such ordinal or nominal data into a common cardinal measure for comparative purposes, one is forced to convert it into accounts of time spent in certain locations or with certain size groups by each member of the sample. That is, one has to transform the problem into one with the individual as the unit of study. This phapg of the research is concerned with the quantitative analysis of large numbers of sequences of temporally ordered Observations as an observational unit. This will involve the analysis of a sequence of observations on an individual. The emphasis on the quantitative analysis of large numbers of sequences is not to deny the importance and validity of biographic or interpretive accounts of the lives of individuals. Indeed, it has been noted that these have been geographic- ally more rewarding.2 But on the continuum between biography and aggregate statistics is a "twilight zone" where common patterns of behavior among individuals can be discerned while preserving their individual identity.3 If the problem of pattern recognition is to be adequately treated, it is this level of data representation that must be explored; a point that will become clearer in the subsequent discussion of analytical procedures. 2Torsten Hagerstrand, "The Domain of Human Geography," Directions in Geography, ed. R.J. Chorley (London: Methuen & Co., 1973), p. 75. 3Torsten Hagerstrand, "What About People in Regional Science?" Papers of the Regional Science Association, 24 (1970), p. 8. 153 Activity MOdules A11 surveys which collect time-space budget diaries generate a set of activity modules which together-define the individual's behavior over a fixed period of time. Each module varies according to several dimensions (Table 4). The most basic of these dimensions include activity, time, and location. Thus, the approach adopted in this inquiry conceives of the individual's day as an integrated n-dimensional unit. As a final stage in the comprehension of this unit, an attempt will be made to simulate the behavior of such an individual. Prior to the development of a simulation model, however, this chapter considers an intermediate stage of comprehension. Although attention must focus on the core of the inquiry, the respondents' behavior patterns, any simulation most proceed from an adequate understanding of the various factors involved. And since these factors and their integration are quite complex in the conception of time-space behavior, the techniques employed to achieve such understanding must be correspondingly compre- hensive. ANALYTICAL PROCEDURES FOR THE RECOGNITION OF BEHAVIOR PATTERNS At the outset, it is mandatory to decide how to reduce the data set to a more comprehensible form. Univariate statistics and cross- tabulations are considered unsatisfactory since they are not sufficiently broad to cope simultaneously with even two or three of the closely related demensions involved in an individual's behavior pattern. Although these statistics can provide the analyst with a wealth of background information, they facilitate the comparison Of no more than three variables 154 simultaneously. And,,when used repeatedly to enable many such compari- sons, this approach would result in some 1,200 tables and several thousands of summary indices even for this modest data set. The draw- ing of conclusions and generalizations from a full description of the data at this basic level would clearly be difficult given the unwieldy number of tables and summary statistics. This would also involve one in the dubious process of inferring individual behavior patterns from aggregate statistics. Even if one only considers portions of the data set, the tables and statistics at this level would require that they be treated individually since they defy rigorous consolidation. Criteria for Pattern Recgggition Analysis The nature of the research problem defines the criteria to be employed in the selection or construction of an adequate analytical technique. First, a single analytic approach is considered insufficient, since the problem of describing and correlating a set of bhhavior patterns has two distinct levels. Initially, there are the data describing the set of activities which constitute each person's day. This set produces the series of activity modules which fully define what can be termed the time-space behavior pattern of the indiVidual. In this case, only the time-space budget information is needed to satisfy the problem of pattern recognition. At a second but distinct level, are the data describing background characteristics of the respondents. Indeed, these background data are so different in form and substance from the diary data that it is difficult to imagine that an analytical procedure exists or could be formulated to treat both simultaneously. While attempting a solution to the problem of pattern recognition, the background 155 information relative to each respondent, may be ignored. In the event that patterns can be identified, it may be possible to relate subsequently some index of the pattern for each individual such as membership in some behavior pattern group to one or more of the variables taken from the respondents' background data files. Given the interest in pattern differentiation, a major problem is the vast amount of activity information relative to each individual in the sample. Some method must therefore be found by which all or part of these data may be condensed into a smaller, yet readily comprehensible, body of information. One should eventually be able to isolate a small set of fairly distinct behavior patterns, describe them in terms of the indices and variables which constitute them, and group individuals on a basis of the proximity of their behavior patterns to the generalized ones already described. A Factor Analytic Approach The use of a factor analytic technique might well be an ideal approach. Factor analysis would permit one to isolate the major dimensions of variation within a large data set defined by a consider- able number of variables. In other words, factor analysis uncovers the independent sources of data variation. Because interdependencies may exist among the data, the technique would allow one to determine whether the same amount of variation in the data can be represented equally well by dimensions smaller in number than the columns necessary to tabulate the data. The variables which make up an individual's activity pattern can be converted to cardinal form by treating each in terms of units of time. 156 Suppose one were to hypothesize that an individual's behavior pattern was constituted by the activities he performed, the degrees to which they were planned in advance and were treated as constraints, and the extent to which they involved other people. Then the cardinal measure in each of these sets of variables is the amount of time the person devoted (D) during the day to each variable category. Therefore, the total number of variables would be divided into three broad groups, and the activity pattern of the i-th individual, P1, could be represented in the following vector form: P1 - f(D1,‘D2,...,D‘j,Dj+1,Dj+2,...,Dk,Dk+1,Dk+2,...,Dm) (4.1) where m is the total number of input variables, j is the total pertain- ing to the content or nature of activity, k-j is the total relating to degree of commitment or anticipation, and m—k is the total pertaining to numbers of participants. One should be able, then, to adapt a very detailed breakdown of behavior patterns (as in 4.1) as input to a factor analysis. The technique should produceta highly'consolidated description which accounts for a high proportion of the variance within the data after the generation of a fairly small number of factors. The dimensions disclosed by a factor analysis can be interpreted as measures of the amount of ordered or patterned variation in the data. The degree to which such regularity or interdependency exists can be gauged by the number and strength Of the dimensions. Perhaps, some or all of the most important dimensions might be readily interpretable in terms of generalized activity patterns. Fer example, a pattern might result that would be typified by high factor loadings on on-campus activities such as attendance at classes and 157 seminars, on variables representing a high level of commitment and constraint, and on variables describing large group activities. Such a dimension would then represent a day that is tightly ordered. By inspecting the factor scores, one would be able to identify the individuals that share this type of behavior pattern. Alternatively, another dimen- sion might be characterized by low factor loadings on the above groups of variables, with higher loadings on independent study, socializing, or leisure, on smaller group size indices, and on the commitment and con- straint items representing a much more loosely structured day. Unfortunately, several problems confront the use of a factor analytic technique for the recognition of behavior patterns. Not the least of these is the fact that such a technique is based in parametric statistics. The analysis proceeds from a product-moment correlation matrix which is computed under the assumptions that all variables, commonly in cardinal form, are logically independent and have an under- lying mmltinormal or near multinormal frequency distribution. Given a careful conceptualization of the research problem, the first assumption is relatively simple to ensure. The second assumption, that each variable is normally distributed about its respective mean, presents a more serious problem. Consider, for example, those variables which relate to the nature of the activities performed. The cardinal representation for such var- iables is the amount of time the i-th individual devotes to the j-th activity during the 24-hour period. In this research, activities have been. coded according to a 98-way classification (Appendix C). Although the average number of activities performed per respondent per day in the univ- ersity sample was found to be 35, the mean number of different activities performed was only 16. SO even if the general assumption is made that 158 types of activity are randomly distributed throughout the sample days, the prObability that any one individual will perform any given activity at least once is only 0.163. Thus, the probability of a zero score occurring in any cell of the matrix, where columns are defined by the 98dway classification, is almost six times that of a non-zero score. It becomes apparent, therefore, that most of the activity variables of this form are heavily skewed to the right of the mean and are seriously distorted by the very large number of zero duration entries. One possible solution to this problem would be to collapse the 98dway activity Classification into, for example, the 37dway breakdown developed by Szalai and others in the Multinational Time-Budget Project.4 If adopted, this would reduce the accuracy level to less than 38percent of that attainable with the larger classification. The problem becomes much more complex that this, however. The claim was made earlier that the concept of pattern can be applied not only to the sequence in which activities are performed, but also to the duration of activities within the sequence. The assumption that acti- vity types are not randomly distributed among individuals is critical to this hypothesis of activity pattern. Even at the increased level of activity aggregation, the 37-way classification, it was found that the activity variables remained heavily skewed to the right. This finding reinforces the belief that the zero values which produce skewness are fundamental in defining the patterns. Hence, the significant issue is that if one begins to search “Alexander Szalai, ed., The Use of Time: Daily Activities of Urban and suburban Populations in Twelve Countries (The Hague: Mouton 8 CO., 1972), pp. 564-5660 159 for behavior patterns among individuals, one implicitly makes the assumption that peOple behave differently but do so in a non-random, predictable way. And if this is so, then some variables with a large number of zero entities will undoubtedly occur, and such variables may, in the end, be the most critical of all in defining generalized patterns. It appears just as important to know what activities are not performed by an individual as those that are. Thus, any attempt to satisfy the multinormal distribution of factor analysis would risk the loss of any regularity or pattern contained within the original data matrix. To search for only variables with a large number of zero values would be dangerous for several reasons. First, it would be quite prOblematic to conclude a priori whether an activity type involved a large number of zero entries simply because it was that type of activity that no one performed very often or whether the zero and non-zero entries were systematically related to a characteristic of the particular sample or the environmental setting that was important. A second problem stems from the fact that there is a continuum of activities ranging from a high to a low number of zero entries. Thus, it would inevitably be difficult to decide what is to represent a high number. Alggrithmic Approach to Pattern Recognition Using a Non-Sequential Criterion ‘ A The problems of using any parametric multivariate statistical technique such as factor analysis seem too great to provide a satisfactory solution. It may be possible, however, to use it in conjunction with an algorithmic approach to the problem of pattern recognition. Thus, an algorithm.might be formulated to partition the original data matrix so 160 as to yield a set of paired matrices. A group of individuals' activity patterns, as defined by selected variables, would correspond to each pair. The variables would be maximally separated into two groups on the basis of performance or non-performance of activity. The variables included in the first of each pair would be those which either all, or at least a sufficient number, performed to guarantee a more or less normal distribution. And those in the second group would include the variables that were heavily skewed owing to the preponderance of zero values. Thus, as a first step toward the comprehension of behavior patterns, groups may be described in terms of the variables which differentiate them on the binary criterion of whether or not the activity is performed by a majority of the members of the group or not. The algorithm devised for describing distinct behavior patterns using a nonsequential criterion is described below. 1. Begin with one vector of information unique to each individual within the sample, as in (4.1). 2. Generate the basic data matrix, D13, where i - 1,2, ... ,n (total number of individuals) and j - 1,2, ... ,m (total number of variables). The ij-th element is some function of the amount of time the i-th individual devotes in total over a 24-hour period to all activities that come within the scope of the definition of the j-th variable. The definition in this case relates to the nature ofactivity, the participants involved, and the levels of commitment and constraint. 3. Create a binary matrix, , as a summary of the basic data B11 matrix, by substituting each non-zero element of Dij with a one. 4. Compute a symmetric matrix, 81k, where i - 1,2, ... ,n and k - i+l,i+2, ... ,n-l (with the main diagonal entries deleted). The 161 ikrth element can vary between zero and m, and is the total number of variables which have non-zero values for both individuals 1 and k: m 1" -151 ”ii I '1: " ”In: (M) S 5. Find the largest value of S and assign individuals or groups ik i and k to a new group p. 6. Delete row i and k from B11, replacing them with a new composite row p in which each non-zero element represents corresponding non-zero elements in both old rows 1 and k: In - a (4.3) 3 pi 1O Pi "' ”13 where j - 1,2, ... ,m. 7. Replace the i-th and k-th rows and columns of 811‘ with a new composite row and column p. The p-th element can vary between zero and m. 2 Bpj’ and is the total number of variables which have non-zero values 1'1 for both the newly created group p and any existing individual or group i: Spi ' 1:1 BPJ ' 3p: ' 311 (“‘4’ where i - 1,2, ... ,n'-l; i f p; n' - n-1; and n - n'. 8. Test against criterion for termination of grouping procedure.5 If this has not been reached, return to step 5. 5The determination of this criterion should be something less than arbitrary. The limiting case would be the presence of no non-zero elements in matrix B. However, this might carry the algorithm to the point where 162 Once the grouping procedure has been terminated, the basic data matrix, D11, can be partitioned horizontally so that each consecutive set of rows relate to a homogeneous group of individuals created during steps five through eight above. Within each partition the columns could be rearranged in such a way that the total number of non-zero elements for each variable increase from right to left. The ranking of variables that results for each group would then form one possible description of the behavior pattern of that group. FOr some of the larger groups generated by the algorithm the partitioning of the matrix might then be used as the basis for a factor analysis. The use of a factor analytic technique in this case would probably increase the sOphistication of the description of the behavior patterns by emphasizing within-group variance. Thus, the set of variables for which non-zero values predominate would constitute the data base for a factor analysis. The factors produced and the scores for each individual might provide a basis for further differentiation and possibly the establishment of subgroups. The preceding discussion demonstrates how one possible algorithmic approach, in conjunction with the multivariate technique of factor ambiguous groupings would result. This would undoubtedly be the case since the presence of just one variable with initially positive values for all records would mean that the criterion would never be reached and individuals would be continually grouped until only group remained. A more promising solution would be to set some "mini-max" value of the 8 matrix as a criterion. This would mean that, if the value was set at 5, then for every group there would have to be at least five variables that had non-zero scores relative to each individual in that group. Such a criterion as this would probably mean that some residual groups would be very small, and a few individuals might not be grouped at all. How- ever, this would only occur to the extent that these groups or individuals had significantly unusual behavior patterns. 163 analysis, can be used for the recognition of behavior patterns. How- ever, the way in which the variables were structured in this approach did ignore the sequential ordering of the individual's activities. Although each variable was expressed in time units, time was treated only as a summary measure of quantity rather than an index of temporal distribution. There appears to be no obvious way in which the variables might be structured so as to retain their important sequential prOperty. The Time-SpaceiMep One technique which does maintain the sequential ordering of events is the time-space map. This is similar in form to Hagerstrand's activity diagram.(Figure 18),6 in that it maps behavior onto a two- dimensional surface, where the vertical dimension represents movement through time and the horizontal dimension provides some definition of the activity performed. The maps are designed to rillustrate in detail the manner in which a relatively small homogeneous group, such as a household, interacts over a period of time. The mapping technique is less valuable for a relatively large sample population such as the one treated here. A more productive use of the maps is to treat larger groups in a more generalized way. Thus, if a uniform symbol is used to represent one or more persons and if these symbols are plotted relative to the appropriate activity at each chosen time interval, a picture evolves of how people tend to distribute themselves throughout the day. The illustration in Figure 19 provides a crude summary of how T. Carlstein, Bo Lenntorp, and Solveig Martensson, Individers gyggbanor i negra hashallstyper, Urbaniseringsprocessen, rapport nr. 17 (Lund: Institutionen for kulturgeografi och ekonomisk geografi vid Lunds universitet, 1968), pp. 27-30. Time—o- 24 1 22- use 164 ...-..i Qg’” r'-------' ‘9' ' ------‘ I I -.1 ..--------‘J ...-..I‘ a- r- 0-0-.~.-.-.-B------------------------ P. -0-0-‘4 WORK-RELATED , I r._m_4 : bo-o_o-o—o-;1 I. I .! I .l '0 :l .I '0 'l o I X o (n °=- § 5 a «f: 3&5 E g 2 9- m .— :0- .J a: Q 0' Z < 0‘ - 9 -’ (I) m ‘5‘ '35 d ‘3 ’- 52- m I- Q .12 8 u 2 Z 01 U) - o C ‘ 5 o 0- " J < Activity—o- FIGURE 18. DIAGRAM OF ACTIVITY SEQUENCES PASSIVE LEISURE ‘0‘ I. a 165 part of such a map would appear. If each asterisk represents, for example, graduate students in the university, the map (Figure 19) would provide a picture of how such a group behaves as a whole. In general, the most noticeable weakness of this approach is that it does not maintain the spatial-temporal continuity of the individual. In other words, what any individual is doing between 9:30 and 9:45 is impossible to relate to what that same individual was doing between 9:00 and 9:15. The type of representation is somewhat analogous to a chronological record of a series of events using still photographs rather than a film where the continuity of events is preserved. However, one advantage of the technique is that it is a relatively simple way of generating ideas about the way in which people behave, albeit in a less rigorous fashion. At the very least, it might aid in the selection of backgroundvariables against which less constrained approaches to pattern recognition might be tested. This is to say that although the mapping technique requires that individuals be grouped a priori, as in the case of graduate students in Figure 19, it does permit simultaneous treatment of respondents' activity patterns and background data. If a group defined by one or more background variables is fairly small, 35 or less, then it is relatively easy to maintain the continuity and individual identity of each member by simply using an array of alphameric symbols such as A,B, ... ,Z,l,2, ... ,9, instead of the uniform asterisk (Figure 20). Although the map produced using the new symbolization conveys more information, it does pose difficulties in 166 nou- (On—- roun— (¢.o can: (can hunt tone (Out cou~ enum— non—u (out. noao none r CI n h 5‘ u once _ a — _ ~ . . — u _ _ mm-O-D-N-bb-—.-h——---bo——-.-.——b _—---o-..—.----.-— FI- b C. CI. '06. .000 00.00 I. bCFW—Ud 2.. y s.- so. 050 0000* 0.0!. _ assoc— so. so— QOOI— r OCC- COCO.— cocci. (a... _ sas— I- O! 0— as m 66 a. a. a» o. oo— s a. s so. as. m e . - _ p o _ m o a _ _ _ _ o. w a c _ _ . _ . _ . . _ . . _ u w . — a. s— es— s— C _ 00* _ as. w .m m a—Cus ” ompmc _ v¢<8 .0. ” saucu .h- _ _ _ mwh I s ~- ...'- ———-—~..- . - ~. ---..-. '.-.' --'.‘-—. '--~. .U...’ 0-0.0 (III. ...I C U I. C II 60...! ...C. IIJI ... O. ..a Jaw}... _ _ addiw a ms at. ‘0 .Ct II O;_a arm .o .-....IIIIIDIIOIOSIISSS m26H8<0HhHummflo 5BH>HBO< _----------- _ .66 '1 3.3:... A I -------'---.-----'---'- .60 .10. b... DID! 'CIO I‘- 3. 0' It I. as (.9! this 6' in“ maVu“. .m- . a . _ _ _ _ poo _ or as I. .0. so. CCU... .¢.ll no. 10000 ..I'. Osl.lo recess. 00:.- CI... II.I¢ Ob... Lorll_ l.¢-os_ .0. I. SrII lrla .I-ID J—II out. Ciro. Ills- sass (III-o. OCOIPOI IOOOOuoo CICCOVE. ...-a- mi... Ills IIOI case. '0. O —'.--.'-'-' sans ._. n.».- 3 H.I & 167 D D... ”..---.- no.0 (one cc“. _ no”; toss 00.0 A (Cut _ anus vou- (unu— co... hung. — noun _ _ s... A teak _ 60.0 : n catnc ...........L s 3 .....u v»& “I. 6 u Fee .a .- I u 4“ his I? “oh-J Us CC mean—ma no. x uuwuocuu MI‘ ...cwz vans .c. _ mdz moBH>HHU< ho ozHozgowm :xomHZD 172 1 4'25... a e a a A h A a a a a IFS— E— _~_d _ :_~_ Q o. _ o_-_m.,H ._om TL al.—El __ Trev—ewe TL mm TL? _n_on T—om ___- Tom 8 t. _ on a me 8 o. 8 on \ 23:23 53:3. access—com 3 220; 22. met. to m was or» 8 e. ore. oi me u .. oi. he. m o TIL use use am e use New 0. n «no mus 8 m J 36 cos on _ m. e. e n N _ 2322 53:04 173 uniformity of behavior patterns occurring between midnight and about 7:00 a.m., during which period most peOple are sleeping. When first tested, the outcome of the algorithmic procedure was one large group composed of 96 percent of the respondents in the sample, with the remaining four percent of respondents constituting one-member groups. Since the majority of individuals shared a common activity between about midnight and 8:00 a.m., this pattern so dominated the grouping that it was impossible to break out any patterns based upon daytime activities. For this reason, the decision was made to begin each sequence at 6:00 a.m. and continuing through midnight, the end of the recorded period. Therefore, all data prior to 6:00 s.m. were disregarded. The second problem relates to the index of content for each position in the sequence. It is by no means obvious how much detail is feasible in the activity description. As already noted, the 98-cate- gory activity classification developed by the Multinational Time-Budget Project9 was adopted because presumably it is the most‘W1d31Y recognized scheme and described activities at a fine level of detail.10 However, the sample population treated here is considerably more homogeneous than the populations surveyed in the Multinational Project. Given this greater degree of homogeneity, the activities which were coded using the 98~way classification were considered too detailed for analysis. In fact, as many as nine activity types were not even performed by 9Szalai, loc. cit. 1oAlthough clearly beyond the scOpe of this inquiry, the adaption of such a well-known classification system would provide the necessary standardization to allow for comparisons with other research using a similar system. 174 individuals in the sample population. If the 98~way scheme is adopted, the probability for any individual that one activity occurs in one's day is 16/98 or 0.163, given a random distribution of different activities between persons, with a mean ofl6. However, once the day is divided into time periods, the probability of any one activity occurring in any one time period is inversely prOportional to the number of time periods in total and directly prOportional to the mean number of those periods occupied by that one activity. If the mean duration of activities is constant, then the probability p of activity x occurring at time t is: put) - (g) - (g) - (3),) (4.6) where a is the mean number of different activities performed, b is the size of the classification, y is the size of the time interval, and z is the length of time. Thus, in this case the probability is 1/98 or 0.0102; and for any given activity and time period, one can only expect about 1 out of 138 peOple to be doing this at the same time. Alternatively, if the 98-way classification is condensed into a 65-category scheme, such as that given in Appendix C, little detail *would be forfeited. In fact, when using the collapsed activity classifi- cation, the mean number Of activities performed was about 33, only two less than with the larger categorization. And the mean number of different activities performed was 15, only one less than the total ‘using the larger classificatory scheme. Thus, by using the formula (4.6), the probability of activity x occurring at time t is 1/65 or ().0155. Given the slight increase in the probability, therefore, one 175 can now expect about two out of the 138 respondents to be doing the same thing at the same time. Very little detail is lost by collapsing the classification of activities to 65 categories. Moreover, the reduction should make the prediction of activities more manageable in the simulation phase. Even though the probability of a number of individuals performing the same activity at the same time of the day is very small, the likelihood of the null hypothesis (i.e., a random distribution of activities) being rejected remains high. For example, at 12:30 p.m. the probability of finding peOple eating lunch is very high, and at 6:00 a.m. the prObability of finding people asleep is, likewise, high. Although the theoretical probabilities of establishing patterns may be very low owing to the potentially wide variety of possible activities, the actual prObabilities are fairly high despite this variety, since people do behave in a predictable manner. It remains to be seen whether the practical variety of activities generated by the 65-way classification tends to obscure or clarify behavior patterns. The Algorithm. Given a matrix of sequenced durations, each indexed according to the activity performed, Si (4.5), the matrix can be partitioned horizontally to groupindividuals with similar patterns. The algorithm designed to accomplish this task focuses on those points in an individual's sequence, or vector, where the activity performed is the same as that performed by another individual. Consequently, it is possible to sum this total "overlap time" over individuals. The total overlap time is the linking index to be used as a basis for grouping; i.e., the greater the overlap time the greater the similarity 176 in behavior patterns. The steps in the algorithmic approach are briefly outlined below. 1. Generate the basic data matrix, [Agj], where i-1,2, ... ,n (the total number of individuals); j-1,2, ... ,m (the total number of sub- divisions of the day); and, t-l,2, ... ,k (the total number of activity categories). Each row vector of information unique to each respondent in the sample is given in (4.5). Each cell entry in the matrix repre- sents a sequenced activity duration. If the main activity during a given time interval was "attending class," then the cell entry is its numeric code (33) indexed to its duration (15 minutes), or 33.15. 2. Create a symmetric overlap totals matrix, [Opq], where p-l,2, ... ,n and q-1,2, ... ,n (the total number of individuals). Delete the main diagonal entries so that for the p-th individual there is a single figure relative to each other individual q representing the total amount of "activity overlap time": k c t t v Opq ' t2. [Min(Apj ,qu) I t‘t ] (407) 3. Select the maximum value of O and collapse the appropriate Pq individuals or groups p and q into a new group g. 4. Construct an amalgamated time series vector, A21, to represent the activity pattern of individuals p and q by selecting only those elements of each individual vector (4.5) which overlap: t _ t t' ,, . Ag-1 [Min(Apj,qu) I t t ] (4.8) t t t' . A81 - [0(ApJ,qu) I t 4 t ] (4.9) where t-1,2, ... ,k. 177 5. Delete rows p and q from Aij and rows and columns p and q f 0. “qu 6. Replace rows and columns p and q of 0pq with a new row and column g, so that relative to each existing individual or group there is a figure for activity overlap time for a new group g: k t t' o , t - _ O8p - 2:1 [Min(Agj,qu) I t-t 1, n n 1. (4.10) 7. Test against criterion for termination of grouping procedure. If this has not been reached, return to step 3. Computational Problems. When inspecting the overlap totals matrix generated by (4.7), several elements of the matrix were found to obtain identical or tied values. The problem of tied values pre- sents considerable difficulty, especially in step 3. If the maximum, value of O is obtained by more than one element in the matrix, then, Pq given the computational procedure of the algorithm, the first of the tied values is selected as the maximum value. This method of treating tied values is unsatisfactory, since it might seriously misrepresent the predominant structure of interrelationships in the overlap matrix. Alternative Algorithmic Approach Usingythe LAWS Modification In an attempt to solve this problem, the algorithm was modified to include an alternative approach to grouping based upon the Largest [Average'flithin-group Similarity (L.A.W.S.).11 The modification would 11Leighton A. Price, Hierarchical ClusteripgyBased on a Criterion of Lapgest AvereggyWithin-cluster Similarity, Research Report (East Lansing, MI: Computer Institute for Social Science Research, Michigan State University, March, 1969). 178 allow for all the relationships between pairs of elements to be pro- cessed in order of decreasing similarity. Thus, the procedure of processing interrelationships by rank order was chosen in preference to the reciprocal pairs approach described above. This was done because it may be demonstrated that reciprocal methods, which pro- ceed by combining elements and forming estimates of the relationship of the combination of other elements, are merely approximations of a rank order approach. Reciprocal Pairs and Typal Analysis. The purpose of the follow- ing discussion is to demonstrate that a reciprocal pairs approach deviates from an otherwise comparable method which processes rela- tionships by order of magnitude. It deviates to the extent that procedures for estimating the similarity between a combination of elements and other elements (or groups) have the effect of distorting and misrepresenting the nature of the original relationships. This demonstration will be made using McQuitty's typal analysis12 and the reciprocal pairs method, of which the algorithm above is but one example. Table 5a presents a 5x5 matrix of overlap totals and Table 5b shows a rank ordering of the off-diagonal relationships.13 This small data set will be used for comparison in lieu of the larger survey data set. In performing a typal analysis, the group consisting of elements 1 and 2 is formed first since {1,2} is the highest ranked pair in lzL.L. McQuitty, "Typal Analysis," Educational and ngcholggical Measurement, 21 (1961), pp. 677-696. 13The tables are after Price, op. cit. 179 TABLE 5 SAMPLE SET OF OVERLAP RELATIONSHIPS values LII-#UJNH IO l2 a. Matrix Form 1 2 3 4 5 - 19_ 2 4 .19 - 3 5 9 - 8 6 3 - 7 4 5 7 — FIGURE 22 b. Rank-order Form Pair Value 1,2 10 2,3 9 3,4 8 4,5 7 3,5 6 2,5 5 1,5 4 2,4 3 1,4 2 l 3 l TYPAL ANALYSIS OF OVERLAP TOTALS IN TABLE 5 (I ,3,42, (h,2) o /\ ’5).\/.(3 4 5) (3, 4) / EhzoMs \\ 180 Table 5b. Because the second ranked pair overlaps with group {1,2}, the second group formed is {1,2,3}. Other groups are formed in the same manner in arriving at the final structure (Figure 22). The reciprocal pairs analysis is performed as follows. Given that a reciprocal pair is defined as two elements having a higher relationship to one another than to any other elements, the first .reciprocal. pair in Table 5a is {1,2}. Once the reciprocal pair is identified, the first and second rows and columns are replaced by a single row and column, as in (4.10) of the preceding algorithm. Within the new row and column are estimates of the relationships of the other elements (or groups) to the first group. As conveyed by (4.10), these estimates are based on the assumption that the relation- ship Of an element (or group) to a newly formed group cannot be greater than the smaller relationship with the two components of the new group. This has been termed the classification assumption.14 The process is repeated until pairs of rows and columns can no longer be replaced. The matrices resulting at each step are given in Table 6 while the resulting structure is shown in Figure 23.’ The structures in Figures 22 and 23 obviously do not match. The classification assumption has the effect of eliminating a number of relationships before they would be encountered during typal analysis. In other words, reciprocal pairs analysis is equivalent to a typal analysis which ignores some values instead of processing all of them in order of magnitude. 14M'cQuitty, Op. cit. 181 TABLE 6 REDUCED MATRICES FROM RECIPROCAL PAIRS ANALYSIS OF OVERLAP TOTALS IN TABLE 5 1,2 3 4 5 1,2 3,4 5 1,2 3,4,5 1,2 - 1 4 1,2 - 1 1,2 - .1 1 - §_ 6 3,4 1 - 9_ 3,4,5 .1 - 2 .g - 7 5 g. - ' 4 6 7 - FIGURE 23 RECIPROCAL PAIRS ANALYSIS OF OVERLAP TOTALS IN TABLE 5 ' d . “12130435, 0 2 - (1.2.3.4) / a. '2‘ O ,,,... / “ “7<\. . 5. I 2 3 4 3 Elements comi typai tical thes. is m obta anal a fi in a pair and rel: and MC typa Only ‘10 m "be: Subs: Order in: pa iDEVit 181mm 182 Given in Table 7 are the values which were not eliminated by combining rows and cOlumns in the reciprocal pairs analysis. When a typal analysis is performed with these values, the outcome is iden- tical to the results of the reciprocal pairs analysis. Therefore, these results clearly demonstrate that the reciprocal pairs method is merely an approximation of the solution which would have been obtained had all relationships been processed in order of magnitude. The discrepancies between typal analysis and a reciprocal pairs analysis can be further clarified by a substitution procedure. As a first step, a list is created in which all pairwise relationships in a matrix are ranked. Then following each step in a reciprocal pairs analysis, the estimate of the relationship between the new group and another element (or group) is substituted in place of the pairwise relationships between each of the elements included in the new group and the other element (or all elements of the other group). This procedure is repeated for each new estimated value. The results of a typal analysis and a reciprocal pairs analysis will be identical if and only if, at each step in a reciprocal pairs analysis, the substitutions do not alter the original pairwise ordering of relationships prior to ‘when they would be encountered in a typal analysis. As long as the substitutions merely yield tied values where ties already existed, the ordering may be regarded as unaltered. The problem is that any method which condenses a matrix by replac- ing pairs of rows and columns with a single row and column will almost :hnevitably misrepresent the relationships in the original matrix. It .is impossible to force such a collapse without a considerable loss in Cycles 183 TABLE 7 THE RELATIONSHIPS (IN TABLE 5) NOT ELIMINATED DURING RECIPROCAL PAIRS ANALYSIS Pair Value 1,2 10 3,4 8 3,5 6 1,3 1 5 1 (l,2,3, 4 ,5) 64 , \(z,34,a 7... \K/éfl. (4, 5) 4 3-1‘,(3 /4] 9 - 0(2, 3) [0... (l,/ 2) o .\;/\ O 5 Elements FIGURE 24 LAWS ANALYSIS OF THE OVERLAP TOTALS IN TABLE 5 184 information. Thus, replacement procedures will often have the effect of eliminating relationships before they would be encountered if processed in order of rank. moreover, the larger groups generated by a replacement approach may not adequately describe predominant group- ings in a matrix.15 The LAWS Modification. The LAWS method incorporated in the algorithm focuses upon within-group characteristics. Typal analysis also has a similar focus, but with the latter method, a larger group is automatically formed whenever a pair is encountered that overlaps with a group already accepted. Therefore, grouping decisions do not consider any between-group calculations. In the LAWS approach, this consideration of within—group charactertistics is‘represented in cal- culations of average within-group similarity which are computed from the original matrix values throughout the analysis. Furthermore, these indices of within-group similarity are also used in making decisions regarding the acceptance or rejection of some possible new groups. The decision rule is, basically, that among a set of tenta- tive groups, only those having the largest average within-group simi- larity should be accepted. Hence, larger groups are not formed necessarily at the first point of overlap (Figure 24). Most reciprocal pairs and typal analysis methods tend to yield groups which might have been quite different had the relative magnitude 15This was clearly the case when the overlap totals matrix was partitioned by the algorithm, prior to the incorporation of the LAWS method. Moreover, this will almost always be the case for grouping algorithms which arrive at a one group termination in N-l cycles. 185 of a few rather similar pairwise interrelationships been ordered differently. There appears to be two reasons for this sensitivity. First, the matrix collapsing procedures used in reciprocal pairs methods.may, in effect, cause relationships to be eliminated. Second, all reciprocal pairs and typal analysis methods impose grouping restrictions which may misrepresent the nature of the original rela- tionships in a matrix. The primary restriction is, therefore, that once a set of elements has been accepted as a group, subsequent groups including any of these elements must include all of them. In the LAWS approach, the first form of sensitivity is eliminated by processing all element interrelationships by order of magnitude. The second restriction is treated by introducing a decision rule whereby tentative groups are accepted if and only if the average within-group similarity is greater than the value for the pair in- dicating that these groups should be considered. The LAWS method is based on the idea that estimates of average within-group similarity provide excellent indices upon which to base grouping decisions. The formulation of such a decision criterion is necessaryin.that the Operational characteristics of the method do not automatically lead to the formation of larger groups, as is the case with all reciprocal pairs and typal analysis methods.16 16This is due to the fact that the LAWS method may yield over- lapping groups. Partial overlap may result from the method's ties solution or from application of the criterion of largest within-group similarity, which will sometimes reject a larger group in favor of a pair of elements. Moreover, pairwise relationships are processed in order of magnitude, and once overlapping groups have been accepted subsequent pairs may overlap with more than one group or link more than one pair of groups (Figure 24). The frequency and nature of the over- laps which may result depends upon the complexity of the original rela- tionships. Therefore, the number of cycles needed to arrive at a one- group solution is determined by the relationships within the data and not by the grouping algorithm. 186 The values of the relationships between all pairs of elements are processed in order of decreasing similarity. All the pairs of elements having a given index value are compared with groups that may already have been accepted and which are not completely included in some larger group. For each pair associated with a particular index value, a record is made of all non-included groups in which one, or both, of the elements of the pair can be found. At this point, all pairs in the set are processed. On the basis of the over- lap information for a given pair of elements, some new group possibi- lities are formed. Whenever more than one possibility results, some decisions must be made before groups may be accepted. These decisions derive from the general rule of accepting groups that have the largest average within-group similarity. After all pairs in a set have been processed, the procedure is repeated for the next set, and so on, until all the data have been examined. The decision to build upon only the nonéincluded groups can be regarded as an attempt to satisfy the ob- jectives of reflecting the predominant structure of interrelationships while keeping the number of groups as small as is reasonable.17 For each pair in the ordered list, four decision situations may be encountered. The situations and their rules are as follows: 1. Situation: Neither element of the pair appears in any of the groups that are in the list of final groups. Rule: If so, add the pair to the list of groups which have been accepted thus far. 2. Situation: One element of a pair is included in one or more of the existing groups and the other included in none. Rule: This 17Price, Op. cit., p. 18. 187 being the case, tentatively form larger groups by expanding overlap groups to include the non-overlapping element of the pair. Then, determine whether any of these groups have greater average within-group similarity than the pair. If so, accept the largest ones among those having greatest average within—group similarity. Otherwise, add only the pair to the list of groups already accepted (rule 1). 3. Situation: The pair of elements links previously accepted groups (i.e., one element in some group and the other is another group). Rule: If so, tentatively form all groups which result from combining the indicated pairs of groups. Then determine which groups are the largest ones among those having greatest average within-group similarity and add them to the list of groups already accepted. 4. Situation: Both elements of the pair already appear in some group(s). Rule: If this is the case, proceed to the next pair in the ordered relationships. Two additional decision rules not directly associated with the processing of pairs in the ordered list are as follows: 5. Situation: The average within-group similarity associated with an accepted group reflects lower average similarity than a group in which it is included. Rule: If this occurs, which is seldomly the case, then eliminate the included group from the set of accepted groups. 6. Situation: The element pairs associated with a block of tied values include all possible pairs for some set(s) of elements. Rule: If so, identify all sets which are not subsets of others. Then, add each of these to the list of accepted groups, provided that none of its 188 elements are included in the groups accepted prior to processing the block of pairs.18 The present method should do a better job in producing groups with higher within-group similarity than the algorithm first prOposed. As a result, group differentiation, or pattern recognition should be greatly"facilitated. RECOGNITION OF BEHAVIOR PATTERN GROUPS The algorithm, incorporating the LAWS modification, was then used to partition the matrix of sequenced activity durations (Figure 21). A major consideration in the analysis was how to arrive at a meaningful termination to the grouping procedure. Clearly, the termination criterion should be something less than arbitrary. Termination Criteria for Group Formation Chi-square (x2) was considered a suitable technique for estab- lishing a realistic termination to the grouping procedure. Thus, as pattern groups were formed, they were tested for significance against the set of background variables (Appendix B) using the chi- square statistic. The use of chi-square served two purposes. First, it provided a test for the significance of pattern groups at various stages of group formation. And secondly, it provided a method whereby one could assess the importance of background variables in describing Specific attributes of behavior pattern groups. Chi-square is considered the most useful univariate test of group membership since much of the data obtained through the interview schedule (Appendix B) is nominal or at best ordinal in nature, which 18Price, op. cit., pp. 20—23. 189 precludes the use of parametric tests. The technique tests the null hypothesis that k independent samples have been drawn from the same population or from k identical pOpulations.19 When pattern group membership is tabulated against any background variable, the problem arises of determining whether the distributions associated with each group differ significantly from those that might have been expected had each group been drawn randomly from the population. The chi—square statistic is calculated as follows: 2 2 n m (Oij - Eij) x = 2: 2 EU (4.11) i=1 j=1 where i a 1,2, ... ,n (the total number of categories generated by the background variable), j = 1,2, ... ,m (the total number of pattern groups generated by the previously applied algorithm). Then, is the 011 number of people in the variable category 1 and group j. The degrees of freedom are given by the rule: d.f. = (n-l)(m-l) (4.12) The higher the value of chi-square derived for any variable, the greater is the probability that a given variable is of significance in defining the activity pattern groups. Ten background variables were chosen for tests of significance against pattern group membership (Table 8). Prior to the chi-square tests, however, all cycles of grouping analysis were arrayed in order of the amount of loss in the average within-group similarity index between v.7 *7 v 19Hubert M. Blalock, Jr., Social Statistics (New York: McGraw- Hill Book Co., 1960), pp. 214-219. 190 one cycle and the next.20 Within the first quartile of the ordered list, the cycles were further ordered according to a second criterion. Cycles were arranged in descending order according to the number of individuals TABLE 8 . BACKGROUND VARIABLES CHOSEN FOR TESTS OF SIGNIFICANCE Variable Categories No. Name Variable Description of Response 1 AGE Age 5 2 SEX Sex 2 3 CIVIL 'Civil Status 4 4 UAFFIL University affiliation 7 5 YAFFIL Years associated with the institution 4 6 OEMPLY Employed elsewhere 2 7 INCOME Annual income 5 8 HOUSE Housing type 7 9 TRAVLM Common mode of travel 6 10 LOC Residential location 5 20Before reordering the cycles according to this criterion, the final 42 cycles (out of a total of 286) were eliminated since only one group existed. It was during these final cycles that the remaining two-member residual groups were added one by one until only one group of all elements remained. 191 who had membership in a group whose size was greater than two members.21 The results of the rankings are given in Table 9 for the first quartile of the reordered list.22 In order to determine which one cycle of those given in Table 9 is to be chosen as the termination of the grouping procedure, chi-square tests were performed. Ten cross-classification tables were prepared for each of the nine cycles where the rows of the tables were defined by the number of response categories in a given background variable (Table 8) and the columns of the tables were defined by the number of groups formed during the cycle. The calculated chi-square statistics for each of the cycles, over all variables, are given in Table 10. Results of the Grouping Procedure: Behavior Pattern Groups' The highest calculated values of x2 are most consistently obtained by the second ranked cycle in Table 10. At the 99 percent confidence level, four variables were found to be significant: civil status (CIVIL), years of affiliation with the institution (YAFFIL), secondary employment (OEMPLY), and location of residence (LOC). The variables of age (AGE), sex (SEX), type of affiliation with the institution (UAFFIL), annual 21The method for seeking a realistic termination to the grouping procedure was developed through consultation with the originator of the LAWS method. Person communication, Leighton A. Price, Computer Institute for Social Science Research, Michigan State University, August 24, 1974. 22The resulting quartile originally included eleven cycles. How- ever, the last six of the eleven cycles do not appear in Table 9 since they had to be deleted. The cycles were ignored due to restrictions on tabular cell frequencies for calculating x2. Given the larger number of groups formed during these cycles, the dimensions of the tables increased substantially with a similar increase in the number of zero cell frequencies. For restrictions on zero cell frequencies, see: Blalock, op. cit., pp. 220-221. 192 .moooEoE oz» zaco wcwcflmucoo omosu ohm mmaopw Hmaowmomea .osp cozy hopmopm mawnmeooEms new: masosm som hacoa .huwamaflEwm asosmncwnpwz mmMpm><+ moa Hm mmm.m m:m.mba mm mm and m me an mmm.m mma.maa am m mma m Hm mm omw.m mmo.mm mm o HAN : om m: Hmm.m omm.mo :m m :mm m m: a: mh>.: m:o.mm Hoa m Hmm m a: on moo.: Hmm.o: 20H m omm a mawnmpmoEmz exoz w mao>o «amasosw @5090 mash comzpmm mgzopo canvas «masoso .02 .oz Hmsowmwm mz< ampoe . . . o .o oaaauasz mcfl>mm +m3< aw mmoq + cowpmasaom Hopoe mo 02 macho zoom m z masses>aecH Hamnm>o m.n 8:313:00 mm.o any um unmowmwmwwmsa .Hm>ma moameumcoo mm.o one am pamomcuewama om.mm Hm.m: Ho.mm «amm.b: «a:m.mm ma.mw NN.Hw abo.H: «amm.am mm.mm OH O mh.Hm No.3m no.0: ma.dm #ma.ma HH.mm No.5m «mo.mm «5:.oa wm.:m m m mm.om hm.mm b©.mm «amm.hm flamm.ba .«1m.mm «rm.©m «amd.am «om.aa mm.ma m : m:.mw mm.bm Nb.mm #«om.mm «amm.ma mo.ma mm.HN «amm.mm «mo.m hm.ha m m seam.om «mm.mm ha.wa «mm.mm ##Hm.HH «abm.ba «hm.:m ##mw.:m «ma.© ewo.ma m N 55.0 «abm.mm amm.ma mm.m «smm.ma «mm.m H:.m «mm.aa m©.m «am.m N H 004 24> AHmMHU me mo< masonw wwwmo lose Ame Ame and Ass Ame Ase Ame Awe Ass to .02 smxcmm mmHmmmmmzmz mDomO zmmeeNo Nev mmbomu ZMNBHo mmam amaumnmm mom mMHUzmadmmm nmaoumxm 92¢ nu>mmmmo mo mMAmNNN NNV ANN vmacfiuaoolt.HH NAm- .I_ om—ua .aaru .N. .xnuu ”amazon:— Ioouuom Iran I I ----Mmmmmmw-_ I.--‘----"-= --- -.--:-. -u ------:-I-- I--. "— --. ------ p u - IIIII.I¢ I” III.IJ In IIIIIO III— II IIu II I II— ONIrN. I II— I: I IIuIIm -r_ I .."... ' IIrIII_ I IIIIINI_ a 'IfIII_ IIIIIIII. IIIIIIIII IIIIIIIIII IIIIIIIcaII I OOIIIIIIII IIIJIII _ IIIINOI I— III I I III_ III III III b..‘9 -5. I I-Ir_ II I IIIIbIc I III IIIII I IvaoI . I III p I IIIIII I IIII.I I“ IIIIII'I II IIIIIIII II” IIIIIIIII — _ IICIIIII .— IIII III— oao N IIII “ II. I I: ; I, O I I I II I II IIII I II IIIIII II IIIIIII_ _ IIII'I I III II I II I III“ 0 II I III I_ III II III .- I III IIII. . III p a v I _ a. _ III— OIIIII— _ IIIIIIu III-leI— _ IIIIIIII- I IIIIIIINW I Ic-IIII-III I IIIaaIIIIIII— I IIIIIIIIINOI— I III-IIIIIIII— 9..“II91v .N:rvaea — .r. u I. CC CC CC «any eN_2c_ _ .n. ---- ------U- I_ II .IQI (III I.sII In —--"---"--- Ft. ta! N .. N .N. 3_------------_ monsduHmHmmddo MHHDHHU< Iv-aw_ I IVIIIrI.II II III-II II I I I (OI IIIr-I_ INIIIirl_ IsIIIIINII I!I?OII-II IIIIIIIIIII IIIIIII.II DIIONIoDI IIIII:a0_ OIODIIII nIVI¢I~II_ III-.II—II I I.IIIIII.II I IIIIIINIIII IIH IIIIIIrII It. IOIQIIII IL .III.’ IL III 7 Ll II a. u_ Icufi .III II C. 0.. IrII (III C ..I I'III .P.' I... I..— C. l.‘ _ a» .0— rcs~ I- not Ir I.II IIII INII IrIrII IIIIIII.II IIIIvInII IIIIIIOIII IrIsIIIIerI_ IIIsIIIIIvII IIIIIIII-II _ 'II---’---II-I— N {an} _ .—- :fiur 99N~_ unu- Nonn— .9u0 us.- 21 P! 1: J3 203 activities such as leisure, mass media, and personal needs. Following approximately 6:00 p.m. the pattern of activities for Group 1 is not as clear, however. Throughout the late afternoon and evening the only noticeable patterns that evolve relate to personal needs (mainly eating meals) and education-related activities. The map for Group 2 (Figure 27) differs from the first in that work-related activities predominate throughout much of the day. Educa- tional activities also attain a high frequency, especially during the daytime and early evening hours. Presumably, then, the similarities in time use among members of each group for daytime activities provided the greatest amount of overlap time, and this considerable degree of overlap was the basis for the grouping decisions. The time-space maps for the third group (Figures 19 and 20) exhibit a totally different configuration. One appreciable difference is the reduction in time devoted to educational activities. Although the smallest group, it is perhaps the most homogeneous in terms of the sequence and frequency of activities. The variety of activities throughout the entire time period is considerably less than for the other groups. The daytime hours are dominated by work-related activities, while work, mass media, and leisure constitute the greatest frequencies during the evening. The recognition of these distinct behavior pattern groups satisfies the criteria for simulation that were established by the research strategy. If the null hypothesis-that the population exhibited random behavior-had been accepted, then simulation would have been a trivial exercise. However, having identified behavior pattern groups, the 204 simulation of individuals' time-space paths would appear to be a worth- while exercise. Before considering the simulation model, additional insights into the structuring of activity sequences for both groups may be gained from a number of graphic and tabular summaries. These include: (1) the configuration of subjective constraints throughout the observation period (Figures 28 and 29), (2) the distribution of individuals' com- mitment to activities during the day (Figures 30 and 31), (3) the pattern of objective constraints as measured by aggregate travel times, and (4) tables summarizing the degree to which the most highly constrained episodes, or pegs, fix the choice, timing, and location of activities that precede and follow them (Tables 14 and 15). For the first pattern group (Figure 28), the choices of activity appear to be most constrained during the late morning and mid-afternoon with a lesser peak occurring during the early evening. Compared to Figure 28, the time-space map for the first group (Figure 26) indicates that these constraints are realized within the context of education- related activities. Constraints to the timing of activities follow a similar trend to that of activity choice with a more noticeable period of fixity occurring during the morning hours. The first category of spatial fixity (Figure 28) deviates considerably from the preceding distributions in that the early morning and late evening hours repre- sent the greatest degrees of constraint relative to one's location. This pattern can be partially understood by the fact that home-centered (personal) activities, which generally occur at these times, cannot be performed elsewhere, thereby being fixed at one's residence. The second distribution of spatial fixity reflects responses to the question-- 205 H mbomu zanaadm ha ZOHH55 45.2.3 it >h.x.& 02.8: >P—XK >h.>.._.04 I“. I: IO. to Us. to In I? In 37”]. 10. 19 1h 206 HH macaw gait mo ZOHBSBUHHIOU gang . m~ gum a. >h.x.m 45.9; a: rth 45...!” TEX: 08.8; rtxi >P.>.h0< an. I: re. ya no .L. to rm TI 3"”. 10 I: IO. Ya In 323-4. 0.. 8 ’8‘} 4r 3.4L 1r 3! 8‘ 207 could you have been elsewhere at this time? Comparatively, this cate- gory of fixity appears to have the least constraining effect. Despite the consistently smaller proportion of "yes" responses throughout the day, this distribution of spatial fixity corresponds to the peaks of activity choice and timing fixities (Figure 28). Additional information may be obtained by examining the pattern of degrees of commitment to various activities (Figure 30). Coincidental with the periods of maximum activity, time, and location fixities are activities that are routine in nature or involve coordination with others. Once again, these activities are primarily educational (Figure 26). Activities which are either planned independently or occur unex- pectedly reach their maximum proportions during the late afternoon and evening. According to the time-space map (Figure 26), the types of activity associated with these forms of commitment are education, mass media, and leisure. The graph of aggregate travel times (Figure 32) displays the aver- age amount of time that group I members devote to travel during a given lS-minute interval of the day. Sizeable shifts in location can be observed during the morning and late afternoon. However, the overall pattern of aggregate travel times suggests that this group is highly mobile over short distances during the working day and early evening.26 26For group I the average distance separating an individual's residence from his workplace (i.e., location of work, classes, seminars, etc.) is 0.943 miles. The modes of travel common to the majority of group I members include walking and bicycling. Therefore, the aggregate travel times displayed in Figure 32 measure the effects of objective constraints (i.e., modes of travel) in overcoming distance. 6800 7300 8300 9200 10:00 11:00 12800 1800 2:00 3:00 6800 5:00 6800 7300 8200 10:00 11:00 12:30 I 208 LEVtLS 0F COHHITHENT (l) ARKANGEO (2) ftANNED (3) (I) leH OTHERS lNOEPcNOENILV ROUTINE UNtXPECYED ....ICI....I. IIIIIIIIIIIII I IIIIIIIIIIII . ......I.l... . ....I....... .. IIIIIIIII-I ..I. ....I... ... ... . I... .. .. ...... . ... ... ....... .... .. ......I ... .. ...... . ..... .. ...... ..... . ....... .... I. ...l.. I .... .. ...... . ..... .. ...... ..... 3 ....... ..... .. ..... ...... I ....I . ..... I. ....II .... .. ...... ... ... ....II. I ... ........ . .. I. ....I... . .. .I .....I... . .. . ....I... .. ... .. .... . . ... ... IIIII-I I IIIII III-III ....I I...‘... It... . ......l . .... ...... ..... . ....... ..... .. ...... ...... . ...... . .... .. ...... .. ..... I ....I.. ..... . ...... ..... ... ..... . I... I. ....I . as: ... IIIII .. ... I.... .... .. .. ..... . . ... .. .... ...t. .. ... ... ...... I III: I. ....... ... I.- I...I.I .. ... ...t... . .. ... ....C. .. . ... ...... ... .. ... ...... II .. I... ..... . can ... ...... I I. ... ....l.. .- ... .. ....... . .... ... .... ... ... .... . ... .... ... Icons I .... III-I III. .. .... ...... . I .... ........ . ....t ....... . I II. ....I... I I. .. ......II. I. ......III. I . I Instinct: .. . I ..IIIIII. . .. ........... I t .....I... .. ......l... I 'IIIIIIIIIII o . FIVE PERSJNS FIGURE 30. DISTRIBUTION 0? ACTIVITY COMMITMENTS FOR PATTERN GROUP I EOUP I 6800 7:00 0800 9:00 10:00 11:00 12:0V 1800 2:00 JIOL #800 5800 6800 7100 8100 9300 lutJL 11800 12:00 209 lEVELS CF COMMITMENT GROJP 2 (ll ARflANGED (2) VLANNEO (3) (b! HIYH OTHERS INDEPENDtNILV ROUTINE UNEXPEC'ED 6800 I ....III.... 0136 I IIIIII-IIII I IIIIIIIIIII I IIIIIIIIIII 7200 ' IIIIIIIIIII 1,00 I I IIIIIIIIII I. II IIIIIIIII II I IIIIIIII I 3300 II II ......II 3:30 III ... .IIIIII .I III. IIIIIII I III IIIIIII II 9:30 II III IIIIIII I Otto I. III IIIIIII I III III III I III II ....IIII 10:00 IIII III III II 10:00 I... II IIIIII I III. I. ..IIII I II... I IIIIII [[300 .... II III... I 11:00 IIII II IIIIIII IIII IIII ....II III IIII II... II [2:00 II IIII IIIII I. 12:09 II IIIII IIII I IIIII IIIII I. III IIII IIIII I 1300 II IIII IIIIII I IIOU III. II IIIIII I I. IIIIIII IIIII I IIIIIII 2:.0 IIII I. IIIIII I 2150 IIIII I IIII I IIII IIIIIIII IIII I IIIIIII I 3:;9 IIII I IIIIII II 3300 III II IIIIIII I III II IIIIIII I IIII II IIIIIII 4340 III III IIIIII I #130 II IIII IIIII I III III III-I II IIII III IIIII I 5:00 III III ...II II 5:90 I. III. .IIIII I II III IIIIIII I III II .III.. I 5:90 III I ..IIII I. osgo I IIIIIII III IIII II III... I III III IIIIII I 1:30 I IIII IIIII II 11(0 II IIII IIIIII I III III ..I II. . III IIIIIIII 8:00 III ....II II 3,00 IIIII I IIIIII IIII I IIIIIII I .I II IIIIIII .I 9330 II II. IIIIII I. 9:03 II I... III... I I III-I III... I II IIII II... II [0200 I IIII IIIII II [0300 III IIIIII IIII I III IIIIIII I II IIIIIIII II 11:00 . II ......I. . 1130c III IIIIIIIII I II IIIIIIIIIII II .......I..I 12:90 12:00 I - THO PERSONS FIGURE 31. DISTRIBUTION 0!" ACTIVITY COMMITMENTS FOR PATTERN GROUP 11 210 O l I Q /’ 9." ... N ’r" CL CL I .. 8 8 " a: a: I, (D 0 S 1 Z Z ’J a: f — 5 Lu < - 8 t: :2 \ m \ _> 3‘. 3‘. <3" - ‘W~ i , ’ | ” —4 I I , O .. .4 Q I, ‘0 \\’ \~~ .4 ‘5 I ” «- ‘3 I __ ,, _ 8. \ L-~_ - a”, 2’ I .. g _ \ _ ,8 ) N / _ llllj :1 1 J n L_, L E3 ‘2 Q '0 "0 (samugw u!) EINIi WEAVELL BLVSEIHOSV TIME -—F FIGURE 32. AGGREGATE TRIP LENGTH DISTRIBUTIONS FOR PATTERN GROUPS 211 Since full-time students represent the majority of group membership, it is not surprising that moderately high and fluctuating rates of mobility result. The pattern of geographical mobility reflects a group in constant transit among a concentrated set of class, study, work, and home locations. Generally, it can be observed that periods of con— siderable mobility occur during times of least constraint (Figure 26). The general picture that results from comparisons among Figures 26, 28, 30, and 32 is that educational pursuits dominate the activity pat- tern of group I. These activities reach their maximum frequency during the morning but peak again in the afternoon and late evening. The times at which the majority of individuals are moving from place to place occur between these peaks. The maximum frequencies of education- related activities correspond to the time intervals which are most con- strained with respect to the choice, timing, and location of activities. These are also the times at which group members undertake activities to which they are routinely committed or involve planned interaction with others. The graphic information pertaining to the first group suggests that one or more educational-related activities represent the pegs about which most of their days are structured and which influence the constraints governing the choice, timing, and location of preceding and subsequent activities. In order to evaluate this assertion, the types of activity which respondents declared as pegs are summarized in Table 14. The table presents information regarding the effects of constraints upon the structuring of activity sequences; information which cannot be inferred by visual inspection of Figure 28. 212 The data contained in Table 143 describe the degree to which the choices of activity, preceding and following a peg, are fixed. For example, of those who declared a work-related activity as a peg, 80 percent indicated that their choices of activity during the 15 minutes preceding work were fixed by it. Similarly, 60 percent indicated that for as long as 30 minutes prior to an important work activity, activity choices were constrained. In locking across the rows of the table, it becomes apparent that activity choices are fixed for the greatest amount of time before and after important educational activities. Similarly, Table 14b summarizes the structuring effect that a particularly important activity has on the timing of preceding and subsequent events. The entries in the table indicate that activities leading to a peg generally cannot be performed at any other time. Conversely, activities which follow a peg are not as time-specific as those which precede it. Activities occurring before an all-important education—related event experience the greatest time fixity for the longest period of time. The data contained in Table 14c pertain to the degree to which activities were fixed at particular locations. Events that occur prior to activity pegs tend to be more place—specific than subsequent ones. Both work- and education-related activity pegs are preceded by activi- ties that are more noticeably fixed with respect to particular activity locations. Finally, the second category of spatial fixity describes the degree to which individuals are tied to locations at given points in time. For the time periods preceding a peg, the relative magnitudes of the data are moderate in comparison with the other tables. And of the time 213 TABLE 14. RATINGS OF ACTIVITY, TIME, AND LOCATION FIXITY—-GROUP I 4----- PRECEDE -----’ 4----- FOLLOW ----- > (in minutes) (in minutes) 0 Ch Q Ch Q Q '0‘ Q 0‘ 0 Category 0‘ 0 st N H N 3 H N c m ox of I I I I I 0,2 I I I I I Peg 3 g .9, 2 ° 0 :3, ° :2 .9. Q 8 z 2 Work 20 20 6O 6O 80 (5) 40 20 20 0 20 Personal 25 50 50 63 75 (8) 50 25 13 O 0 Education 25 42 65 83 92 (48) 71 58 29 40 35 a. Activity Choice Fixity: Could you have done anything else at this time? WOrk 20 6O 6O 80 100 (5) 80 6O 20 20 40 Personal 25 25 50 63 75 (8) I 40 25 25 13 25 Education 40 54 88 65 83 (48) 69 63 44 25 27 b. Timing Fixity: Could you have performed this activity at any other time? WOrk 60 6O 60 60 100 (5) 80 20 4O 20 40 Personal 38 50 25 50 75 (8) 63 25 25 38 25 Education 44 58 69 65 83 (48) 67 38 42 42 13 C. Spatial Fixity: Could you have performed this activity elsewhere? WOrk Personal Education 0 25 17 4O 40 60 13 38 25 33 29 60 4O 50 73 (5) (8) (48) 4O 38 38 20 0 0 13 13 38 40 23 19 20 13 21 d. Spatial Fixity: Could you have been elsewhere at this time? 214 intervals following a peg, the proportion of individuals affected is quite low. Again, activity pegs that are education-related stand out as having the most profound effect on the spatial fixity of acti- vities that precede them. 'In turning attention to the second pattern group, Figure 29 suggests that maximum constraint over activity choice is confined to working hours (i.e., 8:00 to 11:30 a.m. and 1:00 to 4:30 p.m.). The time-space map for group II (Figure 27) shows that primarily work-related acti- vities coincide with these time intervals. Activities that cannot be performed at any other time (Figure 29b) occur most frequently during the hours of work and, to a lesser extent, during the early evening. The categories of spatial fixity generally conform to the preceding distributions but differ in one important way. The relative propor- tion of group members experiencing constraint with respect to location is much greater. Individuals appear to regard work-related activities as specific to particular locations more so than to particular times of day. This suggests that of all activities undertaken during working hours, the majority cannot be performed elsewhere, and the individuals find it impossible to be elsewhere at these times. Unlike the distri- butions for pattern group I (Figure 28), the shapes of the constraint histograms for the second group tend to conform to one another. The greatest frequencies over all categories of fixity are recorded during the late morning and afternoon. On the other hand, periods of lowest frequency transpire at midday and during the early evening. Overall, the group is quite homogeneous with reference not only to activity sequences (Table 13), but also to temporal patterns of constraint. 215 Corresponding to the periods of maximum activity, time, and loca- tion fixity are activities that are overwhelmingly routine in nature (Figure31).27 Again, these activities are commonly work-related and educational. Activities which are either arranged with others or planned independently occur most frequently during late afternoon and evening. The types of activity associated with these forms of commit- ment include education, work, and mass media (Figure 27). The distribution of aggregate travel (Figure 32) reveals that the mobility of members of the second group is restricted to certain clearly defined periods of the day. Major changes in location take place in early morning, midday, and late afternoon, while intervening time periods obtain only modest rates of movement from place to place. This pattern is understandable since at these times individuals are noticeably tied to particular locations (Figures 29c and 29d).28 Membership in group II represents a mixture of university faculty, staff, and graduate students. For these individuals, employment and educational pursuits dominate the activity patterns. work-related activities reach maximum frequency throughout the morning and early afternoon. Educational activities, on the other hand, peak during 27In this case, there may have been a tendency on the part of the respondents to consider activities such as meetings, work sessions, classes, seminars as routine simply because they recur frequently and on a regular basis. In so doing, the label (routine) may have masked activities planned well in advance and arranged with others. 28The average residence-to-workplace distance for group II is 1.284 miles. Modes of travel common to the majority of members include private automobile and walking. The aggregate travel times shown in Figure 32 represent the degree to which objective constraints, or avail- able means of transport, limit movement over space. 216 late afternoon and early evening. The maximum frequencies of work- related activities correspond to the time intervals which record the maximum amount of constraint. Degree of commitment at these times is decidedly routine. The types of activity which members of group II reported as pegs are given in Table 15. work- and education-related events constitute the majority of these pegs. The entries in Table.15a show that one's choice of activity is more highly constrained prior to an important activity than following it. However, the data pertaining to fixity over the timing of activities (Table 15b) suggest that activities preceding a peg are as time-specific as those following it. Many activities performed before and after an important educational peg are those which cannot be undertaken at any other time. Table 15c describes the extent to which activities are fixed at particular locations. Events occurring before and after activity pegs are noticeably place-specific. In fact, the span of time over which this category of spatial fixity is recorded is greater than all other types of fixity. All-important work activities, however, appear to have the greatest impact on the locations where preceding and subsequent activities take place. The second category of spatial fixity identifies the extent to which individuals are tied to specific locations. A sequence of events may effectively tie a person to a particular place for short periods simply because there is not time to go elsewhere in the period between, for example, two classes in the same place or a number of interdependent work obligations at the same location. It can be observed in Table 15d that a large number of individuals have limited choice of location for 217 TABLE 15. RATINGS 0F ACTIVITY, TIME, AND LOCATION FIXITY--GROUP II 4.--” PRECEDE -‘-----> <----- FOLLOW -----r (in minutes) (in minutes) :3 as ~¢ as .3 m ~¢ as a- 0‘ :3 Category 0‘ In Q N H “a 3 H N Q ln 05 of I I I I I .‘3 I I I I I Peg 8 51‘ 8 ‘3 ° 2 £3 ° ‘2 8 3 8 work 31 31 46 46 69 (13) 62 31 15 31 23 Personal 25 50 50 50' 100 (4) 75 25 50 50 0 Education 13 38 50 75 75 (8) 50 38 13 0 13 a. Activity Choice Fixity: Could you have done anything else at this time? work 13 54 39 69 92 (13) 62 39 39 31 23 Personal 0 25 50 100 100 (4) ‘75 300 25 50 25 Education 38 25 50 50 88 (8) 50 25 ‘0 0 13 b. Timing Fixity: Could you have performed this activity at any other time? Wbrk 39 62 77 85 92 (13) 69 62 46 31 54 Personal 25 0 75 50 100 (4) 100 100 50 50 25 Education 25 25 38 75 88 (8) 75 38 38 13 25 c. Spatial Fixity: Could you have performed this activity elsewhere? work Personal Education 46 0 25 23 25 50 54 75 50 ‘ 69 .77. 7s 75 63 63 (13) (4) (8) 54 75 63 31 50 38 46 50 25 31 75 13 39 50 13 d. Spatial Fixity: Could you have been elsewhere at this time? 218 activities. Potential choice of location is most restricted before and after personal activities and especially prior to important work activities. In comparing the data for both categories of spatial fixity, it would appear that the first (Table 15c) has the more con- straining effect and for a greater duration both before and after activity pegs. This trend can be partially understood by realizing that the location of activity at a given time is dependent upon acti- vity choice. Thus, activity pegs are considered to have a direct influence on the location of activities which precede and follow them. Tables 14 and15 provide considerable information about levels of. fixity over limited portions of the day. But for the entire day, is the mean time spent on activities regarded as totally fixed in both time and space greater or less than that regarded as totally free? The results for both pattern groups indicate that the amount of time which is totally free exceeds the amount of time being totally fixed. Table 16a shows the mean amounts of time over which the four categories of fixity prevail. The greatest average amount of time which is fixed corresponds to the first category of spatial fixity. Therefore, acti- vities throughout the day are, on the average, slightly more place- specific than time-specific.29 Deviations about the means of timing and spatial fixity (i.e., the first category) are comparatively small suggesting, among other things, that there is a concensus in the sample's interpretation of these subjective constraints. 29For group 1, approximately ten percent more time is spent doing things where location is a more critical factor than timing of activities. Alternatively, 18 percent more time is devoted to space fixed activities than to time fixed ones for members of group II. 219 ¢~.¢om mm.aom~m no.aaa oo.Hm~ «a.mommm -.¢m~ vouooaxosp oo.mmH om.mnam~ co.~ce H~.mHH oe.mhama sm.mom oawusom ao.mm~ es.ean~m ma.maa mo.ama ma.oaama om.mom sausoeamaaeaa emacmam m~.¢om oo.m~o~ma oN.amm «H.Hmd o~.-m~m cc.wwm mumnuo saws owmmmuu< w mm m um HzMZHszoo .n HH gnome cuouumm H noose snouumm mm.maa no.am~¢a mm.ma¢ m~.num «a.maanofl «H.~¢~ Auv coaumoog em.oo ma.nmmm o~.wmm ae.wca ma.aomm~ nw.mmc adv mowuwuoq Hn.~w Hm.ocmo «a.mmc om.mHH Ha.fiuu< m mm m um mmHHHme .m HH macaw auouumm H nsouo cumuumm .oH MAQ USING THE DACT ARRAY, WHICH LIVES TIIIL I’KUIIAIIIJ UI’RAIIIWS III ACTIVITY CHOICE PROCEDURE FOR ASSIGNING ACTIVITY PROCEDURE T" ACTIVITY. ESTI‘IATE THE DL'IIATIIIN “F TMI MJIVIIY DRAWN I,‘.' ‘IIII ESTI‘IATE THE PRECEDINC STI‘II’. DURATION OF A FIX + IIIII'ZS ACTIVIIY IN'RATION «WIRIM' R'IIII A 51 RSI.QI'I.N‘I ALII' III ' 0 IF SI), CHOOSI' ANIITHHI ACTIVITY. ACTIVITY LOCATION II‘ NUT, CONTINUE IIY DPAIa'INI. I'an-I "II”. All" ARRAY nu runmmp PROCEIII'PF. TH LOCATION (IF THE ”X. DETFD‘IINE IHI. LOCATION (IF A FIX II A PEI} PRICCIILIJLS A IIX IINI) ‘a‘III IIII H IIIHII LOCATIONS ALI (JIMI‘AI IIIII. m: NnI. II NI)‘I,IIRANCII In Sham?” IIIAI IZAIH'IAII‘. IIII TIMI. AVAILABLE FUR IRA‘.'II . IIIIIITS PEI: P°IICFIIIE FIX INCPF‘IENT T I “E INTERVAL IF A PEI; MILLINA'S A I'I.\ IINII HIII IIIIII HI’ NUT THEIR IIIIIATIIINS ARI (.‘I'WAIIIIII. IF NUT. BRANCH TH SILVINI INA] 'AII'TA'I TIIIZ ‘II"I' AK‘AIIAMI .HIII IPA” II. IIIiIZPI‘IITNT I.\' 'IHI. HIST A SI'A'IIAI «mm mm. rm ,m- I , .=.:.-.u.-. .\a nu. TIVI; Irnwvm. TYI’I Axn IJN‘AIIHN 1w: ,ugnz-In- «mun.» (a). ASSIUN MIIIVITY .\.\'II I-;«.VI.'4I.':1 m 3‘33”“ I‘III‘ArI'I‘; CODES T“ ‘ II‘II RHWII I I." .\I‘1I"l1'v ‘mnru; I.'.'II.P'.\I III'IIJIVAI I FIGURE 35--CONTINUED ADVANCE TIME TO FIRST UNSCHEDULED 213i! INTERVAL ALL .* IME INTERVALS SET TIME ASSIGNED TO ZERO I SET DIRECTION FLAG INCREWENT TIME TO FIRST VNSCHEDVLED INTERVAL. II ALL INTERVALS ARE ASSIGNED BRANCH TU SEGMENT THAT OUTPUTS VECTOR 0F SEQUENCLD ACTIVITY DURATIONS. SET DIRECTION FLAG FOR FORUARD OR BACKWARD ITERATION. SUM EARLIEST UNSCHEDULED BLOCK OF TIME DEPENDING UPON DIRECTION, IIND EARLIEST UNSCHLDCLED HLULF HI TIME. I SUM EARLIEST UNSCHEDULED BLOCK OF TIME ADVANCE TIME TO LAST INTERVAL OF UNSCHEDULED BLOCK II ‘I INCREMENT TIME INTERVAL FROM FIRST COMMITTED DECREMENT TIME INTERVAL FOLLOWING TIME INTERVAL SCHEDULE 0 ARE LOCATIONS THE SAME? PRECEDING TIME INTERVAL SCHEDULED IF LOCATIONS OF ACTIVITY DIFFER, BRANCH Tu TRAVEL TIME SLUWLNT. FIND DURATION PROBABILITY OF FOLLOWING ACTIVITY FIND DURATION PROBABILITY OF PRECEDING ACTIVITY LOCATION FIXED: PROCEDURE TO FIND MOST PROBABLE ACTIVITY AT FIXED TIME 8 LOCATION .9 l’ IF ACTIVITY DURATION IS LESS IMAM TIME INTERVAL, FIND MOST PROBAHIL ACTIVITY FOP BALANCE OF TIWI. 4. ASSIGN ACTIVITY AND LOCATION OF FOLLOHING INTERVAL ASSIGN ACTIVITY TYPE B DCRATION TO TIME INTERVAL IF DURATION excexus TIME INTIRVAI. ASSICN ACTIVITY CHARACNTERISIIIS TO PRECEDINO 0R POLLOVINC INTERVAL. ASSIGN ACTIVITY AND LOCATION OF PRECEDING INTERVAL l J FIGURE 35--CONTINUED ADVANCE TINE TO FIRST UNSCHEDULED 238 INTERVAL ALL .* IME INTERVALS SET TIME ASSIGNED T0 ZERO ? SET DIRECTION FLAG INCREWENI TIME TO FIRST UNSCHEULLEO INTERVAL. IT ALL INTERVALS ARE ASSIGNED BRANCH Tu SEGMENT THAT OUTPUTS VECTOR 0F SEQUENCE" ACTIVITY DURATIONS. SET DIRECTION FLAG FOR FORHARD OR BACKWARD ITERATION. SUM EARLIEST UNSCHEDULED BLOCK OF TIME DEPENDING UPON DIRECTION, IIND EARLIEST UNSCHLDULED bLuIV OI TIME. I SUM EARLIEST UNSCHEDULED BLOCK OF TIME ADVANCE TIME'H) LAST INTERVAL OF UNSCHEDULED BLOCK J7 {I INCREMENT TIME FRO?! FIRST COMMITTED INTERVAL DECREMENT TIME INTERVAL FOLLOWING TIME INTERVAL SCHEDULE , ARE LOCATIONS THE SAME? S PRECEDING TIME INTERVAL SCHEDULED 7 ARE LOCATIONS THE SAME? IF LOCATIONS OF ACTIVITY DIVFEK, BRANCH TO TRAVEL TIME SEGMENT. FIND DURATION PROBABILITY OF FOLLOHING ACTIVITY FIND DURATION PROBABILITY 0F PRECEDING ACTIVITY LOCATION FIXED: PROCEDURE TO FIND MOST PROBABLE ACTIVITY AT FIXED TIME 8 LOGAT ION + 1* IF ACTIVITY DURATION IS LESS THAN TIME INTERVAL, FIND MOST PROHAHIL ACTIVITY FOP BALANCE OF TIMI. ... ASSIGN ACTIVITY AND LOCATION OF FOLLOII INC INTERVAL ASSIGN ACTIVITY TYPE 6 DURATION TO TIME INTERVAL [ IF DURATION EXCELDS Tm: INTHIVAI . ASSIGN ACTIVITY CHARACHTERISIIIS T0 PRECEDINI; OR FOLLOWING INTERVAL. ASSICNIACTIVITY AND LOCATION or PRECEDING INTERVAL J FIGURE 35--CONTINUED 239 ITI' HIT "III“I! w'e .il IIIITI I Hunts “" \(VI IVI IY CHOIU: LSIVL ALISK ARRAY. IIVII WISI IWIIIIAIOJJ M I HI I'" LINK AITII I W III PATIO“ PWKI’IH uI. IO E'STI‘AIP DI PATI"\ III A('1I\IT\ LINK ACTIVITY LOCATION: PROCEDURE In Itmmmuuunmx OI‘ ACTIVITY DOES ACTIVITY PRECEDE LINK? DECRE"E.\'T I IME INTERVAL INCREMENI I I‘W INTERVAL ""15 + SPATIAL + AI’TIVIIY 'OI.LO'.' MlSFI'I.‘ L LINN: O ,m. ‘ .- - ISIIREMENT “ “ ' II‘II‘. INTERVAL + DECRE‘IENI ‘I'I‘IE INTERVAL ASSN}; \I I I I I”: .\'.I' I"‘.\II"A I I'I'- ‘I \' II I "I' " I" II FIGURE 35--CONTINUED HHS SM)" )0 II' ] AI II'. LHI'III. IIII ‘ III HI. '.‘|, I‘l'IMIV. I- nn Haw-MU MILHA.‘~IS'IS. :vI IIR‘II‘IS III RATI||\, .\.\II IIIAI I‘IIIIIHI .II. -'I' II\. "“.\‘l 1|" lcl «It - ‘IIH A?! n \ .II x n 240 [ ADVANCE TIME LAST INDIVIDUAL SIMULATED FIND ACTIVITY THAT CONSUHES MAXINUV AMOUNT OF TIME DURING INTERVAL FIND LOCATION OF PRINARY ACTIVITY nacoao ACTIVITY. TYPE. DURATION. AND LOCATION cooss AI)\':\.\'(.I. II‘°I*. BY UNIV. INTIPVAI. 4. HAVE ALI. TI‘IL INTIIR‘I'ALS FOR AN INDHIDI'AI BEEN ASSIGNED? SEQ‘IENT TO OI’TPI‘T "I’ICTOP- (II SI QI'ENLH) A(.’TI'.'I IY DURATIONS FOR INDIVIDIAL (RISNCWI HI PATTERN GROUP 0. RECORD TYPE, LENGTH, AND OF TRAVEL ORIGIN/DESTINATION I DETERMINE TIME AVAILABLE FOR TRAVEL IF TIME 0 MIN TRAVEL TIME SEGMENT - I/O PROCEDURE T0 ESTIMATE TRAVEL TYPE AND TIME BETUEEN ACTIVITY LOCATIONS. (9* PRULMII R}, In ISTIHA’II’. IRA.” II'II BLTRIES AIIIVIIV LHLAIIUSS DETERMINE TIME OF DAY ASSICN TRAVEL [‘3 TYPE AND TIME TO ACTMOD IXIFS TENTATIVL .CIIVIIY PRELLDE THAVIL ASSIflN TRAVEL TYPE AND IIWE TO ACPIOI) FIGURE 35--CONTINUED 241 the respondent's importance ranking and his degree of commitment. Once a peg has been established, uniform random numbers are drawn in order to determine the activity, duration, and location to be asso- ciated with it. During the next step of the simulation a second pass is made through the data. Those activities having the highest subjective fixity ratings are then isolated.7 Consistent with the conceptual framework, these fixed points in the day, the fixes, are assigned activities. For each combination of commitment and constraint indices, there is associated with it a cumulative probability distribution defining the frequency of activity occurrence. A uniform random num- ber is then drawn to determine the type of activity to be assigned to a given fix. In a similar fashion, random numbers are generated to determine the probable activity duration and location for each fix. Once an activity, its duration and location have been tentatively assigned to a given fix, the program then determines whether or not another fix or peg immediately precedes or follows it. This being the case, it is then necessary to determine whether the two successive activities have compatible locations. If a spatial misfit does not arise, the activity, duration, and location assignments are retained. In the event of different locations for successive activities, the program checks to see if there is any intervening time available for 7For instance, if the responses to the four subjective fixity questions were no for a given activity, then that activity would be assigned the highest degree of fixity or constraint. Activities for which three of the responses were no would be next in order of fixity, and so on. 242 travel between the two locations. If not, the tentative activity assign- ments for the fix in question are rejected and the procedure for selecting an activity, duration, and location is reactivated. If, on the other hand, some time is available for travel, a travel time is estimated by sampling a discrete probability distribution of trip. lengths while taking into consideration the travel modes (objective constraints) available to the individual. If the trip length is less than the time available for travel, then the travel time and mode are retained; otherwise, they are rejected along with the tentative acti- vity, duration, and location of the fix preceding the trip. In simulating around the major fixes in a person's day, it became necessary to employ look-ahead and look-back mechanisms.8 Beginning with the first fix in an individual's timetable the program looks back in time in order to determine the amount of time available for activity. If a fix occurs at time t then the program iterates backward to interval t-l. An activity j is already associated with interval t. Therefore, it is possible to enter the table of activity linkage coef- ficients (Figure 34) until reaching the j-th column vector. This vector represents a cumulative probability distribution of linkages to all of the 1 activities. A uniform random number is then drawn to determine the activity that is tentatively to be assigned to the pre- ceding time interval, ti-l. In a similar fashion, random numbers are 8The look-ahead and look-back procedures were chosen in favor of the introduction of a Markovian property to the model. The Markovian property would unfortunately require that the performance of an activity at sequence t+l or t-l, as the case may be, be independent of the acti- vity performed at time t. The data recorded in the original time—space budgets indicate that activities preceding and following a highly con- strained activity are not independent of one another (Tables 14 and 15). 243 generated in order to determine the probable activity duration and location of activity 1 while still looking back in time. If intervals t and t-l are compatible in terms of activity and location, i.e., ti-l=t1, then the data describing the type, duration, and location of activity 1 are assigned to interval t—l. If two successive activities differ in type, are incompatible with respect to their locations, and no time is available for travel between them, then the solution to this instance of a spatial misfit is identical to that described earlier for successive fixes. If, on the other hand, a spatial mis- fit does not arise or the activities occurring at t-l and t allow for travel between their respective locations, then the data pertaining to type, duration, and location of activity are assigned to interval t-l. The simulation then looks ahead to time interval t+l and performs the same operations. Following the look-back and look-ahead operations around the first fix or fixes, the program then advances through the day until reaching the next fix. At that time, the same backward and forward iterations are performed until a solution to the assignment of activity, duration, and location is reached. This procedure is repeated for as many passes through the day as are necessary to build up a con- tinuous, unbroken sequence of events that describe the simulated time- space behavior of an individual. Verification The third stage of model development was that of verification. The procedure involved comparing the model's responses with those which would be anticipated if indeed the model's structure were programmed as intended. 244 The third group resulting from the pattern recognition analysis consisted of only eight members (Figure 25 and Table 12). This rela- tively small data set was employed during the verification phase. The data pertaining to such a small number of individuals made it possible to hand tabulate many segments of the computer simulation model. The following paragraphs describe some problems of model design revealed during verification. It is central to the simulation procedure that what peOple feel to be the major constraints upon their day should be established first and the remainder of the day built up around these. But the actual manner in which they are ascribed a substantive label is highly prob- lematic. This is because the way in which an episode may be regarded as constraining differs from occasion to occasion. At some times a constraint is felt to be relative to a particular activity--a person could not have done this activity at any other place or time. And yet other times, it is felt to be totally independent of the activity per- formed--he could not have been elsewhere at that time. To deal with both sorts of constraint in the same way seems somewhat unreasonable. Further, even when the fixes have been established, there remains the problem of simulating around them. This involves specifying the manner in which they affect both preceding and subsequent activities. For instance, it is reasonable to expect that a fix which involves being at a certain location at a particular time will operate in a different way from one which just involves doing something at a certain time, independent of location. It was realized during the verification phase of model development that the various combinations of constraint had to be treated differ- 245 ently. The decision was made, therefore, to include a three-dimen- sional data array that would define the probability of activity occur- rence according to the types of constraint that described a particular fix during the course of a day (see FIXT array in Figure 34). A second decision made during the verification phase was that the array containing the trip length distributions (TDIST in Figure 34) should be composed of the entire sample. If the TDIST array contains the trip length distribution relative to only the members of one group, then trip lengths are too limited by time of day. Simulation Experiments Deriving numerical values of the endogeneous variables from a single simulation run constitutes an experiment. The parameters, which are varied from one run to another, are the factors of the exper- iment; each factor can be assigned several values or levels. A single simulation run, or experiment, involves a given assignment of values to all the parameters manipulated in the investigation. The steps of the simulation as described above and depicted in Figure 35 were repeated as many times as there were members of a behavior pattern group. Before the completion of each experiment, i.e., simulated activity sequence, a final check was made to ensure that all time intervals in the day had been assigned the necessary information to describe an activity module.9 Each module in a sequence contains the data elements 1, 2, 5, and 7 (Table 4), and for those episodes that 9The possibility of an unassigned activity module exists. If after five attempts a spatial misfit is unresolved, all data elements of the activity module in question are coded as being void of activity. This precautionary measure was taken in order to avoid excessive use of computer time. 246 were fixes or pegs the activity module contains the additional data elements 10-15 (Table 14). The data modules describing the simulated activity sequences were then transformed into vectors of sequenced activity durations (Figure 21). The vectors were computed in the same way as that described in Chapter 4. For each of the 72 lS-minute time intervals, the type of activity which was performed for the greatest length of time was entered. weighting was maintained by using the cardinal duration (between 1 and 15 minutes) as the base information of the vector and it was indexed both according to its sequential position and the nature of the acti- vity performed. If EIj is the estimated amount of time the i-th individual devotes to main activity j during time period t, then the individual's simulated activity sequence, Pi’ can be represented by the vector: P . (Et t t Et 5‘ (5.1) 1 11’ E12’ "' ' Eij’ "' ’ im-l’ in) where: i - l,2,...,n (the total number of individuals); j = l,2,...,m (the total number of subdivisions of the day); and t = l,2,...,k (the total number of activity categories).10 EVALUATION OF THE MODEL The simulated activity sequences may be aggregated by behavior pattern groups and summarized in the form of time—space maps (Figures 10Each element in the vector represents a sequenced activity duration. If the main activity code during a given time interval is 47, then the element is represented by that numeric code indexed to its duration (e.g., 10 minutes), or 47.10. 247 36 and 37). When comparing the observed time-space maps (Figures 26 and 27) with the simulated ones, it becomes difficult to accurately determine the degree of correspondence between them. For instance, it is obvious, when comparing the observed (Figure 26) and simulated (Figure 36) maps for pattern group I, that considerable discrepancies occur following 6:00 p.m. Similarly, observed differences can be noted between the actual (Figure 27) and simulated (Figure 37) maps of the second pattern group, especially during the mid-afternoon and evening hours. For example, during the mid-afternoon there is a noticeable under-estimation of education-related activities, while during the evening hours a serious over-estimation of educational activities occurs. It is difficult to measure, by visual comparison alone, the depar- tures of the simulated from the observed patterns. Moreover, it is impossible to discern individual differences between the actual and the predicted time-space paths. A more reliable measure of corres- pondence must be sought. Correspondence Between Estimated and Observed Results Once the activity sequences for members of a pattern group have been simulated (5.1), it is necessary to compare these results against the observed sequences. The tests for evaluating the ability of the chosen methodology to replicate the observed time-space paths must satisfy two criteria. First the tests should measure at the micro- scale the degree of correspondence between the actual and predicted time-space paths of individuals. Second, and more importantly, the tests should measure the goodness-of-fit between the actual and 248 00-. 00.» H .595 guy—Eh ho a flub-D-bud 60a- ocuu‘ 70.3— )9“. 30.. 00.- 00.n- so... acac- 60.. _ 00-h 30.0 fl uln- 249 HH game 5E5— hO a Bdmmug ENS—gm . on gun— nlOn-we as» e e I I II on _ oo-- no"~‘ e0 0.. e eeeeee ee ee eee ee eeee ee e eee ee eee e_ e eee noun. e ee e e eeee e eee son-u e eee eeee ee eeeee e ee eee eeee Hun" e ee eeee eee ...o- ee ee eee e eeeee 00.0— ee e eeeeee e eee eeeee ee e eeeee e eeeee e e eeeeee 00.0 e e eeeeeee e eee so». ee eeeeee ee eee ee eee eeeeee ee e . eee eeeeee e ee 09.. eeee eeeee e eee 00.. ee eeeeeee e e ee e e e eeeeee e eee ee eee e e ee eeee on.» e ee e ee ee e eeee aauh e eee e ee eeeee e e eeee e eee-e ee ee e ee e eeeeee e 00.0 e e e e eeee e ee ee no.0 O O O .I ... CI ... e e eeee eee e eee ee ee ee ee e eee oo-o e ee eee ee eeeee 14.0 e e e e e eeeeeeee e e eee e e eeeeee ee eee e eeeeeee 00.: ee ee e e eeeeeeew $0.0 e eee e e e eeeeee eeee e e e eeeeee ee eeeeeeeeee 00.0 e ee eeeeeeeeee ooun e eee e eeeeeeee ee e e eeeeeeeee e eeeeeeeeeeee afinu ee eee eeeeeee oo- e eee e eeeeeeee ee e eeeeeeeeee e ee eeeeeeeeeee 00.. e e eeeeeeeeeee oou‘ e e O ele e eeeee "e N H ee e eeeeee e eee ee eeee oouua e ee e eeeeeeeee oa-a e e eeeeeeeeeee e ee e eeeeeeeee e ee ee eeeeeeee 90.-- e e ee eeeeeeeee no.4- e e eeeeeeeeeee "u en eeeeeeeeee eeeeeeeeee ocean e e eeeeeeeeeoe 03.0- e e eeeeeeeeeee H H .0... COO... eeeeeeeeeee 00.. ee ee 0 eeeeeeeu sea. e ee ee e eeeeeee ee eeee e eeeeee 8.. 3 he." . . 3.8 e ee e e e e e eeeeeee e eeu eeu acne e e eeeee eeel eee .... .u u ...... .. .. e eeeeeee ee e e ee eeeeee eee u 60-h eee uneeeeee ee ee e eeeeel. 00.0 e eeeeeeeeee, e ee n 00.0 u 04 c-OUI we! .npue oJJOU 85- ewsea an: e 08 ado!“ wade J-xu 180' : Cdm— g.” as. ~0 ”man ”0 .m- -mwao A a“. v 03-» ~ .20 n!O—u¢u_&—nn<4u >b—>-—u¢ 250 the predicted activity sequences of behavior pattern groups. By satisfying these criteria, it is possible toassess the effective- ness of the overall methodology in simulating the time-space paths of individuals based upon the mechanics of constraints. A test which satisfies the criteria is the Kolmogorov-Smirnov (K28) test.11 The technique is based on the principle that one expects the cumulated frequency distributions of two samples to be similar if they are random samples drawn from the same population. The test takes advantage of the situation in which there can be some logical ordering of intervals so that the profiles of the frequency distribu- tions can be compared. In.this application, the logical ordering of intervals is c1ear--the 36 sequential time intervals throughout the observation period (Table 17).12 The two profiles are represented by: (1) the actual distribution, where each entry in the list is a fre- quency (in minutes) for the major activity performed during a given time interval and (2) the predicted distribution, where each entry represents a frequency (in minutes) for the same activity recorded in the actual distribution. If a predicted activity is different from the actual, or observed, activity during the same time interval, then 11The method of construction and relative advantage of this statistic may be indicated most clearly for the data presegted in Table 17, where the minimum frequency restriction of the X test made that test inappropriate. For additional information pertaining to the KrS test, see: Hubert M. Blalock, Social Statistics (New York: McGraw-Hill Book Co., 1960), pp. 203-206. 121a order to assess the goodness-of—fit between the actual and predicted time series, the 72 lS-minute time intervals were con- densed into a sequence of 36 30-minute intervals. Thus, interval 1 represents the time period 6:00-6:29 a.m., interval 2--the time period 6:30-6:59 a.m., through interval 36--the time period 11:30-11:59 p.m. 251 TABLE 17. ACTUAL AND PREDICTED ACTIVITY SEQUENCES Actual Predicted Time Interval Frequency Proportion Cum. Frequency Proportion Cum. Difference l 30 .0363 .036 30 .051 .051 .015 2 30 .036 .072 30 .051 .102 .030 3 30 .036 .108 30 .051 .153 .045 4 22 .027 .135 30 .051 .204 .069 5 12 .015 .150 0 .000 .204 .054 6 10 .012 .162 24 .041 .245 .083 7 30 .036 .198 17 .029 .274 .076 8 24 .029 .227 0 .000 .274 .047 9 30 .036 .263 12 .021 .295 .032 10 30 .036 .299 30 .051 .346 .047 11 12 .015 .314 5 .009 .355 .041 12 18 .022 .336 20 .035 .390 .054 13 20 .024 .360 0 .000 .390 .030 14 16 .019 .379 0 .000 .390 .011 15 22 .027 .406 0 .000 .390 .016 16 30 .036 .442 15 .026 .416 .026 17 25 .030 .478 4 .007 .423 .055 18 10 .012 .490 23 .040 .463 .027 19 14 .017 .507 0 .000 .463 .044 20 25 .030 .537 0 .000 .463 .074 21 30 .036 .573 17 .029 .492 .081 22 15 .018 .591 30 .051 .543 .048 23 23 .028 .619 8 .014 .557 .062 24 30 .036 .655 12 .021 .578 .077 25 30 .036 .691 0 .000 .578 .113* 26 21 .025 .716 22 .038 .676 .040 27 14 .017 .733 30 .051 .667 .066 28 30 .036 ’ .769 30 .051 .718 .051 29 30 .036 .805 30 .051 .769 .036 30 15 .018 .823 30 .051 .820 .003 31 14 .017 .840 0 .000 .820 .020 32 21 .025 .865 0 .000 .820 .045 33 30 .036 .901 30 .051 .871 .030 34 27 .033 .937 11 .019 .890 .047 35 30 .036 .973 30 .051 .941 .032 36 30 .036 1.000 30 .051 1.000 .000 NA- 830 NP- 580 D - .113 xz- 17.433 252 the actual duration is recorded while the predicted duration is denoted by a zero frequency (Table 17).13 Once both sets of data are ordered in an internally logical sequence (Table 17), each frequency is then expressed as a proportion of N, which in the case of the actual (A) distribution, NA = 830, and in the case of the predicted (P) distribution, NP - 580.14 The pro- portions are then accumulated. Finally, the absolute values of the differences between the accumulated prOportions for each row are calculated. The largest of these is designated D and, in the example of the first member of pattern group I, this is 0.113 (interval 25 in Table 17). When the samples, NA and NP’ are large and unequal in size, a significance test which pools the two sample sizes must be employed. This test makes use of the fact that the sampling distribufion of D I can be transformed into the chi-square (X2) distribution as follows: N N 2_A_Ii_ (5.2) NA+ NP X I 4 D where the degrees of freedom associated with chi-square are two.15 13As an example, see interval 15 in Table 17. The observed activity which occupied the majority of time (22 minutes) during this interval is "eating." However, the predicted activity for the same time interval was "study consultation" for a duration of 18 minutes. Since the activity of eating was not predicted to occur for any dura- tion during the time interval, a zero frequency registers in the predicted time interval. This means, therefore, that no amount of activity overlap, or correspondence, exists between the observed and the predicted. 14As would be expected, the sizes of NA and NP vary among individuals as well as between pattern groups. 15Blalock, op. cit., p. 205. 253 In the example for one member of pattern group I (Table 17), the maximum difference (D) value is 0.113. The chi-square which results from substituting the value for D in expression 5.2, is 17.438. In consulting the chi-square distribution, it is found that for two degrees of freedom the value 9.21 corresponds to the .01 level. This means that if the null hypothesis were actually true, a chi-square this large or larger would result by chance less than one percent of the time. Since a chi-square of 17.438 was obtained, the null hypothesis must be rejected. That hypothesis stated that there is no significant difference between the actual and predicted activity sequences. Since the chi-square value for the individual represented in Table 17 was larger than the critical value of X2 at the 99 percent confidence level, the null hypothesis must be rejected, meaning that patterned variations do exist between the actual and pre- dicted sequences of activity. Hence, it must be inferred that the simulation model, based on the time-space mechanics of constraints, does not accurately predict the observed time-space behavior of the first member of pattern group I (Table 17). The decision to accept or reject the research hypothesis based solely on the results of one experiment is quite obviously incomplete. Rather, the K-S test must be applied to all experiments of a simula- tion run. Once the differences between the actual and predicted activity sequences are computed, the maximum difference statistic (0) may be calculated for each individual in a pattern group. The D and the X2 statistics are given in Appendix G for members in each of the two pattern groups. Of the 68 experiments conducted for pattern group I, only the predicted results for two individuals were significant at the .01 254 level. This means that for all but three percent of the experiments, the null hypothesis was rejected. Rejecting the null hypothesis infers that the majority of actual and predicted activity sequences were significantly different. These results relate specifically to the ability of the simulation model to predict activities and their sequencing. In looking at the ability of the model to predict activity loca- tions and their sequence, the results are slightly improved (Appendix G). For all but five experiments, the null hypothesis was rejected, meaning that significant variations exist between the actual and the predicted sequences of activity locations. Table 18 gives the overall X2 statistics for pattern group I. Since both chi-square values exceed the critical value of X2 at the 99 percent confidence level, the null hypothesis must be rejected. Thus, for the majority of experiments, actual and predicted locational sequences are dissimilar. A total of twenty-five experiments resulted from the second simu- lation run. The results are slightly more significant. For eight percent of the experiments, the null hypothesis was accepted, suggesting that the predicted types, durations, and sequencing of activity closely approximated those originally recorded in the time-space budgets. When looking at the aggregate result for group II (Table 18), however, the null hypothesis must be rejected since the departures of the pre- dicted from the observed activity sequences are too great to have occurred by chance. A second means for evaluating the model's results is to inspect the predicted sequences of activity location (Appendix G), which 255 TABLE 18. CHI-SQUARE STATISTICS FOR PATTERN GROUPS PATTERN GROUP I 2 D NA NP X Activity Sequences .244 876.8 529.4 78.542 Locational Sequences .207 933.6 721.3 69.710 PATTERN GROUP II 2 D NA NP X Activity Sequences .191 797.2 7462.9 42.743 Locational Sequences .122 958.0 776.4 25.521 incorporates type of location, time-stay at a particular place, and sequencing of movements from one location to another. Sixteen per- cent of the experiments for group 11 predicted locational sequences which were not significantly different from the actual patterns. In fact, there were additional chi-square statistics that hovered about the critical value of X2. Although the actual and predicted loca- tional sequences in these instances were significantly different statistically, the closeness of the results suggests that they should not be totally discounted. Table 18 gives the overall X2 statistics for pattern group 11. As was the case for group I, both chi-square values exceed their res- pective critical value at the 99 percent confidence level. Therefore, 256 the null hypothesis must again be rejected because the deviations of the predicted from the observed locational sequences for group II are too great to have occurred by chance. Since the model is unable to accurately predict the observed time- space paths of individuals, it is necessary to examine the weaknesses of the model. How accurate is the model in estimating activities for episodes having differing levels of commitment? Table 19 summarizes the number of correct activity assignments for episodes ranging from planned to totally unexpected. The best rates of prediction occur for those activities which are planned while the poorest predictability is associated with activities which have not been planned in advance. More specifically, those activities which are either routine or planned in advance with others obtain the best rates of prediction for both TABLE 19. SIMULATED ACTIVITY ASSIGNMENTS BY LEVELS OF COMMITMENT (Observed) (Simulated) Prediction Actual No. Correct Activity Rate GROUP 1: of Episodes Assignments (2) Arranged with Others 658 321 I 48.8 Planned Independently 583 206 35.3 Routine 744 413 55.5 Unexpected 236 67 28.4 GROUP II: .Arranged with Others 240 115 47.9 Planned Independently 186 81 43.5 Routine 281 166 - 59.1 Unexpected 124 40 32.3 L :E ‘1 I 257 simulation runs. It is important to note that activities having these types of commitment (Figures 30 and 31) are also the activities having the greatest fixity in terms of activity choice, timing, and location (Figures 28 and 29). The preceding observation raises the question, how accurate is the model in estimating activities for episodes having not only different levels of commitment, but also different degrees of constraint? Table 20 presents information pertaining to the various combinations of commitment and constraint, ranging from planned activities fixed in both time and space to unplanned activities having a low degree of constraint. Column A of Table 20 gives the sum of all observed acti- vity durations common to a particular category. The values of 3 indicate the average amount of time a member of group I engages in activities having a given combination of commitment and constraint. The final column of percentages indicate the relative proportion of time devoted to activities in each of the six categories. In the second part of Table 20, the values under P represent the sum of all predicted activity durations for each category while 5 values indicate the predicted average amount of time group I members perform activities having the various commitment and constraint characteristics. Finally, percentage values describe the simulated amount of time devoted to activities in each of the six categories. The data in Table 20 were subjected to a K-S test in order to determine whether any significant difference exists between the actual and predicted sequenced activity durations as classified by degree of commitment and level of constraint. The null hypothesis states that there is no significant difference between the observed and simulated 258 oar.¢H a x Nos. a a mz HHU< QNHHHO< QMH. I. r a’\\\/ ~\\ a \II\ a .J 3 (s a P s s III \’(III'L a‘ la \\ 0")! \\ " u a N/(\ 4 /\ ou‘ \l<\\ II‘\ II / x A II \NlrLs 1 z I m N .593 2555 ..... A .. _ enomo 2554A. I I” o. n p . — . p u m. 267 are highest obtain the greatest degree of correspondence between the. actual and predicted activity and location characteristics. Those episodes which have lower fixity ratings consistently have poorer matching of the simulated with the observed. Once again, the examples shown in Figure 39 point out the inability of the model to cope with travel time estimation. The problem of accurately predicting travel between activity locations is also evident in the aggregate results for both groups (Figure 40). Research Hypothesis The central research hypothesis of this investigation is that the critical determinant of the structure of a person's day, given environ- mental forces (i.e., the objective constraints), is the extent to which one feels constrained relative to certain activities, times, and loca- tions. A methodology has been formulated which is built around the idea that subjective constraints are the critical determinants of one's behavior in time and space. The vehicle used-to examine this hypothesis is that of simulation. Thus, an attempt has been made to simulate the unknown variables, the timing, sequencing, and location of acti- vities, in terms of the parameters or known variables, the subjective constraints. The findings from the modelling experiments demonstrate that the subjective constraints acting on the choice, timing, and location of activities are modestly important in attempting to understand the formation of activity sequences and the paths people follow through time and space. The statistical results of the model's output suggest that they are not as significant as they were hypothesized to be. 268 Based upon these tests of validation, the research hypothesis must be rejected. The rejection infers that subjective constraints alone are not the critical determinants which structure time-space paths at a daily scale. It is the conclusion of this researcher that the limitations of the methodology rest mainly with the conceptualization of constraints in relation to the phenomena to be explained-~the structuring of daily behavior in time-space. Necessary but simplifying assumptions were made as part of the modelling strategy that precluded a more direct concern for the objective constraints. This set of constraints includes the peculiarities of the environment in which action takes place. Objective constraints also vary from person to person in the form of available transport media. These differentials in transport modes for overcoming distances in time-space were not directly addressed. The distinction of subjective from objective constraints was designed to simplify the research problem and to facilitate the investigation. V The divorce of the two categories of constraint is not altogether realistic. As was suggested earlier, constraints circumscribe man's actions. In support of the chosen research methodology, it was stated that choices of activity can only be realized within the context of constraints. Therefore, the investigation of constraints, or negative determinants, might prove to be a more fruitful line of research than the focus on positive factors that lead to activity choices. This con- tention is still held. However, as a result of this investigation it is concluded that subjective constraints can only be understood in the context of objective constraints. In any further research dealing with 269 the structuring of time-space paths, it is recommended that the two categories of constraint remain distinct but not divorced from one another in the analysis. Some specific comments must be made with regard to the value of the computer simulation model. Simulation has been employed as a descriptive technique and not a predictive one. The simulation model was developed in order to answer the question-~to what extent are subjective constraints descriptive of one's behavior in time-space? Relying upon Mbnte Carlo procedures, the simulation generates time- space paths of individuals which describe the choices, timing, and location of activity. Mbreover, it permits a degree of random varia- tion to represent the uncertainty in man's actions. Monte Carlo simulation is not a powerful explanatory technique, but its simplicity, its heuristic value, and its dynamic nature have made it extremely useful in exploring the research hypothesis. Furthermore, simulation procedures are necessary in that the highly complex processes in ques- tion preclude purely analytical solutions. The results of the simulation experiments suggest that we cannot reproduce the complexities of daily behavior, seen as an unbroken sequence of actions in time and space, solely on the basis of subjective constraints. However, the model is capable of rather accurately des- cribing the type of activity chosen, its timing and location for certain periods of the day which are considered to have the most constraining effect. .Thus, activity pegs, those points around which much of the remainder of the day was organized, were most accurately described by the model. Activities having high subjective fixity ratings were also rather accurately described by the model. These are generally acti- 270 vities in which the individuals had little or no choice in their timing, duration, and location. In fact, a relationship has evolved suggesting that the more constrained the event, in terms of commitment and time- space fixity, the better the model's estimation of the activity per- formed at that time as well as its sequence and location. As with the use of most simulation models, more problems arise than can be solved. First, the problem of matching the two major dimensions of the model--the subjective fixes and the objective des- criptions (i.e., activity choice, timing, and location)--require much more thought and investigation. It is central to the present methodology that what people feel to be the major constraints upon their days should be input first and the simulation of the rest of the day built up around these. But the actual manner in which they are assigned activities, times, and locations is highly problematic. This is primarily because the way in which an episode may be regarded as constraining differs from occasion to occasion. Sometimes a con- straint is felt relative to a particular activity (e.g., a person could not have performed this activity at any other time) and some- times it is felt to be totally independent of the activity performed (e.g., a person could not have been elsewhere at that time). At other times a constraint is felt relative to a particular activity location (e.g., a person could not have performed this activity at any other place) while at many times it is felt to be totally inde- pendent of spatial location (e.g., a person could have been elsewhere at the same time). To deal with all types of constraint in exactly the same way seems somewhat unrealistic. Further, even when fixes have been established, there remains the 271 problem of simulating around them. This involves specifying the man- ner in which they affect preceding and subsequent activities. The model has proven to be particularly weak in this regard since it was observed that time periods with lower fixity ratings had the poorest prediction rates. Perhaps the most glaring weakness of the model is its inability to cape with travel episodes. Spatial locations are only treated implicitly by the model. That.is to say, exact geographical locations of activity recorded in the time-space budgets were grouped into categories of activity locations (Appendix C), for instance, work- place, one's own residence, somebody else's residence, etc. As a result, distance between activity locations was measured by the surrogate variable--trip length (i.e., travel time). Geographical distance and direction of movement were not treated explicitly due to the simplifying assumptions of the model. These and many more difficulties will have to be remedied before a full-fledged simulation can be fully operationalized. CHAPTER 6 SUMMARY AND CONCLUSIONS: RETROSPECT AND PROSPECT The principal aim of this research has been to develop a methodology for investigating how individuals' decisions about their time-space behavior interrelate. What determines where, when, for how long, and in what sequence activities are.performed? In attempting to understand more clearly the structuring of everyday behavior this research adapted a methodology based on the time-space mechanics of constraints. SUMMARY Conceptually, the behavior, or activities, of an individual can be viewed as a series of discrete episodes occurring in a sequence through some specified period of time. These episodes have meaning at different temporal and spatial scales. Those that relate to a day's activities, such as work, housekeeping, or eating, were chosen for investigation. Events in the daily sequence are seen to possess an essential order in the sense that certain types of events occur in fairly predictable cycles or routines and take place in a space of fairly predictable locus. Activities, therefore, occur in a time-space continuum: there are temporal regularities inherent in spatial regu- larities, and temporal rhythms obviously vary over space. Two general approaches to the study of time-space behavior were considered. One orientation, the more common in geographical litera- ture, focuses on choices of activity that are manifest in behavior. 272 273 A second approach, and the one chosen for this research, focuses on the constraints which circumscribe man's actions. Since choices of acti- vity can only be realized within the cOntext of constraints, the second approach is more general and more likely to lead to a discovery of how the decisions regarding one's time-space behavior interrelate. Two broad groups of constraint were identified. First, objective. constraints are those which are imposed by the environment. These con- straints tend to shape the "niches" of possible action open to an individual. The pattern of niches formed by objective constraints describes a map of potential events. These potential events can be more narrowly defined by a second group, the subjective constraints, which are to a great extent self-imposed. Such constraints arise as a result of the individual's social and physical environment. Thus, a given environmental situation is subjectively perceived by different people as constraining choice to varying degrees. Although the unique- ness of the environmental context is important, the individual's reaction to it is crucial. Any situation to which an individual reacts has at least three dimensions: a temporal position, a location in space, and a potential activities set. These dimensions interact to produce a feeling of constraint upon the individual. One's evaluation of the way in which these dimensions interrelate is complex. But as a result of the priorities and constraints per- ceived by an individual, any activity has a subjective fixity rating associated with it. The subjective fixity rating is defined by the degree of commitment to the activity and the extent to which it is con- strained in time and space. It is thought that an individual schedules a variable proportion of his day according to these ratings of fixity, 274 in order to facilitate synchronization of activities and movement in time and space.' Although much of the scheduling process is undoubtedly routine, the ordering of more complicated, unfamiliar, or crowded com- binations of activities is objectively calculated. Activities to which the individual is strongly committed and which are space- and time-fixed tend to act as points around whiCh the ordering of activities, according to their flexibility, are scheduled. The major research hypothesis is that the critical determinant of the structure of a person's day, given environmental forces (i.e., objective constraints), is the extent to which one feels constrained relative to certain activities, times, and locations. The term structure is interpreted to mean not only the types and sequences of activities, but also their location and duration.' In order to adequately test this hypotheSis it was first necessary to isolate a sample population and to develop survey instruments for obtaining the needed data. The sample population was drawn from the faculty, staff, and student populations of Michigan State University, located in East Lansing, Michigan. A systematic random sample, strati— fied by university affiliation, resulted in a target survey population of 200 persons, of which there were 138 respondents. In order to study the time-space mechanics of constraints operating on the formation of activity sequences, attention must be focused on the paths, or behavior, of the individual through time-space. The instrument chosen for collecting this type of information is the time- space budget. The time-space budget, which is very central to this research, incorporates the sequence, linkage, timing, duration, and frequency of activities, as well as the spatial and temporal coordinates 275 of one's behavior. Time-space budgets focus on two related aspects: people's overt behavior and the perceptions of their physical and social environment. The time-space budget can, therefore, record spatial behavior (moving and stationary), and it can be indirectly related to environmental perceptions (via questionnaire techniques) as these interact with overt activities. Perhaps the primary attribute of time-space budgets is that they record behavior patterns which are not directly observable due to their spatial and temporal extent. Furthermore, this method of activity accounting is unique in that, as far as their activities for the sampled time span are concerned, individuals are treated as totalities, and the entire sequence or pattern of activities can be analyzed. The time- space budget, when carefully completed, makes available data which describe the behavior of an individual as a sequence of unbroken actions in time-space. This does not mean that the activity sequences are fully explained as understood by the persons who create them, but at least the totality is there for examination. Based upon the conceptualization of constraints, a simulation methodology was developed in order to define the time-space mechanics of constraints which are thought to rule how individual paths are channeled, diverted, or even routinized. It was hoped that such an approach would shed light on how decisions about one's time-space behavior interrelate. At this stage of research into the organization of daily behavior, a negative determinants approach was considered to be the safest and, perhaps, most productive. As a first step toward the goal of developing a simulation model, an attempt was made to isolate that behavior which demonstrates recur- 276 rent patterns in time-space. Hence, a preliminary hypothesis regarding the principle of consistency in human behavior was tested using a sample of time-space diaries. The claim was made that the concept of pattern could be meaningfully applied to any set, or sub-group, of time-space budgets and that this pattern is partly defined by the sequence in which activities are performed. A second assumption of the consistency hypothesis stated that the amount of time a person devotes to an activity follows a pattern over sub-groups of individuals just as much as does the position it occupies in that person's sequence. Such simple patterns of activity are necessary conditions for the development of a simulation model. In order to test this hypothesis, an algorithm was devised which groups individuals on the basis of the amount of time which they spend on different activities and the order in which they are performed. The algorithm focuses on those points in an individual's sequence where the activity performed is the same as that performed by another individual. Consequently, it was possible to sum this total "overlap time" over individuals. The total overlap time was the linking index used for grouping; the greater the overlap time, the greater the similarity in behavior patterns. The algorithm decomposed the sample population into groups based upon the sequential behavior patterns of individuals. The groups maximize within-group similarities and between-group differ- entials. Three distinct groups emerged as a result of the grouping of tine-space budgets. Therefore, the hypothesis dealing with the con- sistency of behavior patterns was accepted. Patterned variations do exist between subgroups of the university population, where the concept of pattern is construed in terms of sequence and duration of daily 277 activities. As a result of the grouping analysis, it was discovered that "typi— cal" workdays encompassed a variety of forms. But two clearly discern- able archetypes were the highly structured day (Group II) normally in- cluding numerous work and social commitments, and a more loosely structured day (Group I), spent largely at one's residence and involving non-formalized work, leisure, and routine activities. When looking more closely atthe patterns which emerged, especially as people described the priorities and constraints which they associated with the individual episodes in their days, it was learned that basically three levels of commitment could be related to normal work activities. First, there were formalized work episodes-~classes, seminars, committee meetings, and staff duties--which were regarded as by far the most important fixing or structuring events, both with respect to specific locations and particular times of day. Next came less formalized work phases of longer duration and not normally involving interaction. They were only regarded as shaping the respondent's day in the sense that they were often tied to particular times, but flexible as to loca- tion. It seemed reasonable to conclude that they were fixed to certain times largely because they were related to future deadlines (e.g., preparing reports, assignments, or lectures). Finally, there were threshold activities which appeared to mark the end of the working day-- for instance, in the afternoon an increase in loosely structured activity was still classified as work by the majority. Leisure activities were of two dominant forms. On the one hand, there were the special occasions such as social events which were arranged in advance with others and involved a fairly high degree of 278 perceived constraint. Alternatively, there were the unplanned time- filling episodes. Two major types of punctuating activities were also evident. These were short in duration and normally peripheral to the major features of behavioral sequences. First, there were routine (personal) and domestic chores which were not felt to be of structuring significance, but were comonly considered as tied to particular loca- tions. Second, shOpping activities, for some, were also punctuations. They were normally short and undertaken by many on the way home from work or classes in the evening. Patterns of behavior, insofar as they relate to the structuring importance of certain perceived constraints, are revealed at least as significantly through the sequences of activity which people perform, as through the amount of time people devote to them. In the algorithm used, individuals were grouped together not only on the basis of the activities performed, but also on the basis of the order, or sequence, in which they were undertaken. The analytic approach permitted the identification of different groups of activity sequences nested within the basic diurnal cycle. Not surprisingly, the basic waking—sleeping and mealtime-to-mealtime cycles were found to be of overriding impor- tance in dictating the overall structure of the day. However, within, and to some extent independent of, this structure certain other impor- tant sequence patterns were found. Two of these are of particular interest in that they indicate how key episodes-ewhether formalized work, social events, or routine activities-~are integrated into the overall structure of the day. First, if they are important work acti- vities it is often the case, as suggested in Chapter 4, that they are immediately preceded by other work activities which involve preparation 279 for the major event. Second, the important fixes are dispersed through- out the working hours. A typically important activity, tied to a par- ticular time and place gives way to less important social or routine behavior which is also place-fixed, simply because it follows an acti- vity that was fixed with respect to location. This, in turn, leads to even more relaxed behavior, often merely passing time, which precedes the build-up toward the next major fix in the day. This build-up, just as the previous decline, involves activity tied to the location of the forthcoming event, but otherwise relatively unimportant. The pattern is one of oscillation (Figure 33). Thus, the working day, for members of the university population at least, may be considered as a process which oscillates between active committed high priority phases and passive uncommitted low priority ones. Can activities and their timing and location be associated with this oscillating pattern of commitment and time-space constraints throughout the day? This is the question that the simulation model addresses. In conjunction with this problem it was hypothesized that time-space constraints and levels of activity commitment are the critie cal determinants of behavior, or activity sequences, in time-space. This hypothesis derives from the studies of Hagerstrand who suggests that people's consumption of time and space is largely fixed and routin- ized. In practice most people have little choice about where or when they sleep, eat, work, or relax--at least from a day-to-day perspective. Therefore, Hagerstrand suggests that human activity may best be under- stood in terms of the constraints rather than the incentives of aeti- vity. This research in general, and the simulation methodology in particular, are seen as extensions of Hagerstrand's basic model. In 280 lieu of Hagerstrand's fixed-unfixed dichotomy of activity, this research has envisaged the individual's day as an integrated function of a more extensive range of flexibility, defined by one's level of commitment to each activity and by the time and space constraints to which one feels subject. This information which was collected in the time-space budgets is only descriptive of the constraints people experience. Nevertheless, it was used as the base information of the simulation model. Given the data pertaining to levels of commitment and time-space constraints for all activities, the model attempts to simulate, in an integrated sense, the daily behavior of a given individual. The unknowns, or the simuland, includes the types of activity performed as well as their timing, duration, and locations. The simulation performed reasonably well in estimating the acti- vities to be associated with periods of high commitment and constraint. But the model failed in its ability to accurately estimate activity sequences over the day as a whole and for periods of low commitment and constraint in particular. The inability of the model to simulate daily behavior applies not only to the estimation of activities and transitions between activities, but also to the accurate description of their location in time and space. Thus, it was concluded that subjective constraints alone are not the critical determinants of time-space behavior at the daily scale. CONCLUSIONS: RETROSPECT AND PROSPECT In retrospect, several conclusions can be drawn as a result of this research. First, the data for the research came from the members 281 of one institution, a university, in a sizeable metropolitan area. In many ways this might be expected to be a highly unique sample since those affiliated with a university are generally thought to have a greater degree of freedom governing their days than the majority of the working population. The variation between individuals in the ways they structure their days would be expected to be much larger than within other homogeneous groups of working peOple. However, clearly defined patterns of behavior among this group emerged which suggests that peOple, through habit, structure their days in a standardized form regardless of the lack of direct constraints. One general con- clusion resulting from the analysis of these behavior pattern groups is that the sequence of a day's activities is "pegged" around key struCturing episodes (e.g., work, eating, studying, etc.) that are interspersed with relaxed forms of behavior which serve to give the day's events balance and continuity. The principal aim of the present inquiry has been to develop a methodology for investigating how an individual's decisions about his time-space behavior interrelate. The approach adopted was based on the time-space mechanics of constraints. Mere specifically, the methodology is founded on the premise that subjective constraints determine the structuring of paths of behavior in time-space. The adequacy of the methodology has been evaluated on its ability to replicate observed behavior. As a result of the simulation experiments, the methodology has been judged to be inadequate. The findings of the modelling experiments demonstrate that the subjective constraints acting on the choice, timing, and location of activities do not accurately describe the activity sequences, or paths, 282 people follow in time-space. The statistical results of the model's output indicate that they are not as significant as they were hypo- thesized to be. These results cast doubt on the validity of the research methodology and modelling strategy. Hence, it must be con- cluded that subjective constraints alone are not the critical deter- minants which structure time-space paths at a daily scale. Another conclusion of this research is that the limitations of the methodology are due primarily to the conceptualization of constraints. Necessary but simplifying assumptions of the mOdelling strategy de- emphasized the role of objective constraints. In fact, it was the aspiration of the research methodology to develOp general rules con- Cerning the formation of activity sequences in time-space regardless of the peculiarities of the environment in which behavior was observed. That is, can we develop general rules of time—space constraints which underlie behavior patterns, irrespective of the particular environment in which the behavior has been observed? As a result of the research effort, this goal is still a distant one. Observed behavior is partly determined by the structure of the environment in which it occurs. Therefore, the set of objective con- straints describe the peculiarities of the environment in which action takes place. Although the distinction of subjective from objective constraints was designed to simplify the research problem and to facilitate the investigation, the divorce of the two categories is not altogether realistic. In support of the chosen research methodology, it was stated earlier that choices of activity, with its temporal and spatial rami- fications, can only be realized within the context of subjective con- 283 straints. In light of the research results, the idea is now advanced that subjective constraints can only be understood in the context of objective constraints. Thus, in future research which explores the structuring of time—space paths, it is recommended that the two cate- gories of constraint remain distinct but not divorced from one another in the conceptualization. One suggestion made in this paper is that the most fruitful course for the development of a time-space framework in geography is to try as simply and rigorously as possible to devise conceptualizations and methods of basic calculation before going on to the greater complexities of large-scale empirical applications. However, a considerable amount of methodological and empirical research will have to be conducted in order to better develop a model for describing the relationship between objective and subjective constraints and time~Space paths of individuals. Before doing so, at least three problem areas which have arisen as a result of this research need to be resolved. The first problem area relates to the way we organize our data in geographical analysis. we must have a minimum of spatial and temporal precision in the location of our data. The time-space budget is only a means toward this end; without a well organized time-geographic infor- mation system the data resulting from the budgets will be of limited value. As far as the time dimension is concerned, a useful conventional system of reference based on solar or astronomic time (e.g., days and years) exists. By contrast, when it comes to space, we have to cope with regions of all sorts of odd sizes and shapes, which is quite problematic. A solution to this problem has been suggested and described by 284 Hagerstrand, and is based on a co-ordinate square grid system covering a whole country.1 It provides for a regular and standardized framework suited to all kinds of localized data regarding population, activities and different social, economic, and natural variables. The adoption of a co-ordinate grid system would allow for data organization at various spatial scales from a micro-scale at the intra-urban level to a regional scale, however defined. This is the type of matrix needed for efficient handling of time-geographic information and for the identi- fication of objective constraints. A second problem area concerns the methods of analysis used to investigate the research problem. Simulation has been employed as a descriptive technique to illuminate the relationship between subjective constraints and individual sequences of behavior in time-space. Relying upon Mbnte Carlo procedures, the simulation has generated time-space paths of individuals which describe the Choices, timing, and location of activity. Admittedly, Mbnte Carlo simulation is not a powerful explanatoryqtechnique, but its simplicity and heuristic value have made it useful in exploring the research hypothesis. Although the model is deemed inadequate in simulating observed behavior, this is not to sug- gest that the modelling strategy should be totally abandoned. As a result of the simulation experiments, additional information has been obtained regarding the relationship between constraints and the choice, timing, and location of activities. It is recommended that the Monte Carlo procedures, which permit a considerable degree of random variation 1Torsten Hagerstrand, "The Computer and the Geographer," Transactions, Institute of British Geographers, 42 (1967), pp. 27-34. 285 to represent the uncertainty of man's actions, be replaced by a more systematic approach that can better accomodate the oscillating pattern of constraints and commitment. A goal for future simulations would be to broaden the scape of the model (e.g., better treatment of the spatial dimension and objective constraints) while at the same time relaxing a greater number of limiting assumptions. A third problem area concerns the unit of analysis used in this investigation. The research has sampled from one institution, a university. The writer has thereby been in a position to describe the daily routines and patterns of time—space constraint associated with only the members of that institution. However, the study of individuals as institution members focuses upon their lives in only one of their roles. Every person has many roles and the logical implication is to look at the individual in what for most is the primary reference group, the nuclear family or household. Other important secOndary roles will be revealed by a record of one's daily activity cycles, and the perfor- mance of any one of these is so integrated with the performance of others, that no one can be understood completely on its own. Carlstein has also made suggestions regarding the unit of analysis problem.2 His focus is at an aggregate or group level. In order to survive, work exchange information, socialize, and so on, individuals must continually come together in time and Space. It is not for the geographer to take the group for granted but it should be regarded as an assembly kit rather than as a finished product. Although most 2Tommy Carlstein, Regional or Spatial Sociology? Lund, Sweden: Department of Social and Economic Geography, University of Lund, 1972. 286 people wake up each morning among members of a household, the primary group, there are still other groups in which individuals usually participate that are still unformed as people have their breakfast. For most people, each day is begun bymaking an investment in a time- consuming movement before other groups take shape. In other words, there is a cost of forming groups which is paid by other groups which cannot be formed at the same time. The costs can be conveniently viewed in terms of subjective and objective constraints. For the geographer who takes movement as an integral component in group formation, the whole set of spatial origins and destinations in- herent in a settlement pattern becomes a very important social sub- system. The observation explains why so much emphasis has been placed on the concept of accessibility in geography, whether measured in time, distance, or monetary units. In using the concept of accessibility, geographers often look at potential group formation in the population of an area rather than at which groups are actually formed. It follows logically that many of the applications of geography to planning are a matter of facilitating group formation by improving transport and by rearranging elements in the spatial pattern. Such applications would include planning the relative location of the origins and destinations in space such as work places, shops, administrative offices, hospitals, and other public facilities. In conclusion, the methodology of time-space constraints might better be Operationalized if the unit of analysis were adjusted to the level of a group with attention given to potential rather than actual group formation. What is the nature of time-space constraints that limit the potential for group formation? What are the distributional 287 impacts of public facilities on various social groups? To what extent do perceived as well as objective constraints interact so as to actually disrupt the activity patterns of social groups? These are just a few examples of important questions which might be examined through an extension of the time-space constraint methodology developed in this research. APPENDICES APPENDIX A. THE TIME-SPACE BUDGET DIARY* *The diary presented on the following pages is only a portion of actual diary used in the survey. In order to minimize repetion, only the first six pages of the form are reproduced here. 288 289 5.3.3: 8:28 .293 .2 22:2: .2368 .... as 28.88 .0 225590 mecca-em d .5 £23.33... ESQ mosses: _\_,_ ._. . Amish >3er m 0 4n— m m. .3222... ..o 23 its... 8372:: .o 323 6: toassess. .02 33232:. .NH.n massmou cw com»aEn0mcm mes uuouoe »o: on omaomn .uoenoa Laos-q~ po»acmwmov oz» new mesmw usox «cause; such .no»» -«>L»ua moo» oemcumoo 0» some: »u Hma»ov mo ~o>o~ on» acmzocn .Ha: .coauauonoou waox no» so» neat» .caau< .Aeomm-mmn . awzv -um>_ucw one cow xav maommx» a mo comueoa a av-n .mnv woman »xu: on» consume: ouco~um ~amuom no» o»:»«»mc_ no»:meou us» »a no onov 290 co o»au»m:~gw o: .:M vougmu muumv oz» oxen v»=o: u: so; long ch ...u»o .mcmxmaz .man ..»o>muvc use .eou -cnmmaav emu ..m.ov m ceaaou cm soc»ocm o» oua»n oco souu vo~ -~O>mn» so» so; o»uumv:e .uoz»oca o» ouaen uco scum acmamo>ue» mo x»m>m»om on» use»nocon cog: .zusm mm doecouoe on v_:o;m on. x»e>e»ua o»aeea0m a ma ..u»o .guoz .mcmqaonm .ucwuoos oo»»wesou .mmaau ma zuzm x»m>m»os as some so 0» men» 4 .x»m>m»oa use» c»m: vonsfiuce on »oc visonm ecu x»m>m»ua o»a»aqo« a me x»m>m»ua as some to o» ucm~Ho>aa» »:oem use» as» »s;» useEogom .A.u»o .mcemcmm »mo: .mosoxo ..m.oe some Haeocom on» veouo» .nwouv -vm m.:om»aoo~ us» go seam »oc one so» mi .xnmeucu» no .»oue»m .:OE»OOm»o»:m »moeaoc on» o» :o«»uuo« on» o»uumvcw .n :e:»ou :_ ouafia moxn» x»w>w»us manenu-uuo sumo upon: vacuum no» cost .x»«>«»ua an commemop o» moo: so» me nosed acne ma on: omeofim .mmafiu uxo: use» «Lemon u:o;-cu.u~n: pace» 0» ousie a an ac Lumammn» a was» 0» amusem seen» on so; we no .mxoon mafinue»umm o» house 0» poo: so» Omauuon ones» can so» um .xuaenad on» c» on pox coca .zocx o» oxq~ o~soz o: .odneaxo mow .omnwmmoc ma awa»ov none as aw mum»m>m»ua mac» opwuumov ounogm .flx»m>a»ua »az» onauumov omaomm .on «we >»«>«» -ua vauuomou on» we use» sewn on» »a 0min ucmz»»ca acmov use: pox »o: so aoc»usz--o>aw assfiou cm was A.u»u .mo>«»a~om .mo»ae -mmn~o .mosunofifiou .mvcomeu ..u.oi »»m>m»un us» mo cam» on» »n som :»M1 was omfio ucoxcm »oc no us;»o:3--n30u cezfiou cm ux»m ->m»os some how comusELOUcm mcMIOAHOM on» Co»co unmade nessaou n»w»w one cue30w on» c_ ._.u»o .heomcmmmamv emu .Aeo>wuvv emu .oiuqup .ucmxfimx ..m.o_ Ho>me» no» lo: o»nuMm:w umuo»a eon»o=s 0» music oco acme unm>os one so» we .mmoevpn on» no»co macaw“ .mcoa»a00» ao:»o "as Cam .500» on» mo Cones: so com»nmuunuv a one ucmu»m:n us» no use: oz» ;»oo so»co Omaomm .o»mn xuwmuo>mca a »n x»«>m»oa an new com»nuo~ oz» u:»uo»:o can so» u— .co»» -auoq oz» mu»:o :Ea~ou seen» on» em ecu .x»e>«»ua on» communov cssfioo vacuum oz» :— .x»m>m»un an »»a»m so» some: »a one» oz» ceaaou »moe-»uu~ on» em Co»co .a: »uu so» use» on» us unmccmuom “mania oz» em couscous on vHaocn um so: van xmv ”aummx» a «cwzozm ofiaeuxo ca van nouoc once one one: .xuamv on» em -wu o» no» oxM~ v~=ox a: mo: emugmxo or 35:03.2: been -mmn . :mze meannuoou mo »coe»uanoo on» an nauseoum econ »uo»:ou omaome .vo»: no -m3 a»av >o>esn Once» :u_:3 mow summons» on» «o ou:»ac on» no xo>usm nee» »:ons cod»ue»0u:w Haco«»vaa mom .xo> .»:m on» em com»anmumuuue moo» ecu so» acne» o» axed v~aox on .o»O»meou mm xo>aam on» as soon ma noxoaumov on Him: use»ammua»eae omoz» mo moss: mo unmm on» can xmco hopes: a an momma»:uvw mm emcee comm .xmanw»cOnwu=ou mo»uom» on .oeuaou wo .HHLI momuamv ~n3vm>mmcM on» mo n»eo»:ou och .xsv on» »:onm:oug» mo«»«>m»ua on» on so» on »:o mas» -mm so» use» uneven amsox o: .xeamv on» eucan -eOu 0» map »xo: on» ~«»== co>o so .xav on» Mo was on» mw»=a »~sz man so» nu:o:»~< .m0m»m>m»ua n.xav »n:» we swam” a mammoox op add: :ox .omcun a :~ .von 0» on no» new» oz» Hm»ca mcmaeoa 0:» cm as use so» use» or» soso .»m»us on» mo "Ha ao»~ou use» mouse on» co a: saw mucus» o» som acmxna one o: .ououomozh .u« on som con: one «son: van on so» »s:3 .odommmoa no x»o»am:uoa an .mocx 0» sad» mason or man one you .x»au on» am new >»»muo>m:a on» em :»on am: so» »u;» nuwummmuam we v:«: on» new new» moo» pecan so» gums: cm ma: on» mo ou:»uAn a a: v~mao 0» Annie meg» some co«»aae0mcm on» em: 0» cam» v»:oz or A»:aceum»umm xo>e3m oz» 0h was» .somosso 291 N. NA is 2.:3 .o: no It __ O. .5513 .=c_.»uuo_. oz» 5 comuncmumov 0...» men .5530 13:35. on» em ~u>auu me one. us» use .xoceaofi a mum—Icon.— oea :0» 9.2.; ”*Tsrnlv o 5F: a?! m «.36: 2:3 as «83.02... ..8 3.0 0:235... 0500 5.! no: 3.. 9.3 was. r. t Amp»: 25.5.3.qu no: nausea u...» .3 .30»: nan-nuke so: one: 9:315 van hopes: so was: soc.— ue» 5533 2332.2. of 5 was 33:3,; 0;» .25» on» use ..Coao some to... : woo—J ml OJ C133 *1 . 3: $32 1433 1 = Ed :2 .m‘hN «$95.53 L3 9.9.0. ”surname: r. v .tpLans. 4 0+ .Liu \strCLfl ... “in: 8:83 n N 53:3 5.. .53 "spies Es as eee; 8.. £8 292 .cea~ou .:oe»auo~. 0:» cm :o_»a=_»nuv on» HUM can .:e:~ou.»»_>_»uu_o:» cm ~u>auu we once 0:» »=n .xueuaofl a mcmumouoe sun :0» cheer msmsuu.co name use: mzmvmmao was cones: no can: scan oz» .nfiawou .nmuu_m nauseo.uuo now. museums on» no .Amo»mu .comunuo~. on» :_ ace .x»m>m»ua 0:» .OEM» 0:» »:m .xeuco some use HHQ I'll-e2 2:: 3: calls woe: 2:3 6 23202: .8 30> on? 0:225 0500 2:! no; 3.. oz; wn_~_:o_mm~. n e 8:83 :33 3F n N _ 323% 2:2 ME. :95: 3.. Sue APPENDIX B. THE SUPPLEMENTAL INTERVIEW SCHEDULE 293 T (|) Your Interview Number | ‘flSU’ (2) Date (3) Length of lst Interview M SPACE Budge, survey (4) Length of 2nd Interview ( ) minutes Invesiiqoior: J. D. Stephens (5, Respondent Number _____(,2_,..) / / / /. (.7 Address of Resldsnce: Tminutesl (2i-25)//////(26-30)l/l/l/ (Tl CALL NUMBER I 2 3 4 5 . . . Interview Nunber Hour of the Day Date ‘ Day of Week Results —’ If an interview is taken, complete the items imediately below and attach this cover sheet to the Interview Schedule and the Time-space Diary. (a) Date of completed time-space diary. _/L_/_-/_/__/_/ —-‘} By observation of interviewer! (Bi-35) (9) (IO) [3 Male CI Caucasian CI Nesro (36) /__/ CI Fem” D Other [3 Don't know (37) _/ —' I f NO interview is taken, complete NONRESPONSE FORM 19401.) and on back side of page. NONTItsP"o"—T_—Ns FOR» (I) NONINTERVIEW AND NONSAHPLE FORM: (Fill out for cases where no interview was obtained) INTERVIEWER: check one: Respondent absent. Someone at university address. but respondent absent. Describe on back side of page No one at location (university address). No such address. Invalid university address. No eligable respondent. Refusal. Give detailed descriptlon on back sIde of this page. Other. No interview obtained for reason other than above. Explain fully on back side of this page. DDCICIU D -—r) INTL‘RVIEIJER: Please Supply as much of the following information as you can without making inquiries of others present. (2) Sex of probable (3) Race of probable (4) Civil status of probable respondent . respondent . respondent . U "a“ [j Caucasian 0 Single 0 Negro 0 Married 0 Female D 0 h ( If ) C] Divorced or Separated t er spec y 0 HI I d I] Don't know U Don't know 0 Don't knov I 295 _' INTERVIEB’ER: Please supply as much of the following information as you can without making inquiries of others present. (5) Age of probable respondent: (6) Location of probable respondent's address: D l9 yrs. or less D Residence Hall [J 20 - 2‘0 yrs. D Faculty or Married Housing U 25 ' 29 Y'S- D East Lansing D 30 ' 39 YFS- D Lansing D “0 ' 6“ yrs. 0 Other (specify) D 65 yrs. and over D Don‘t know D Don't know (7) University position of probable respondent: D Don't know D Faculty [3 Student [3 Full-time [3 Part-time [3 Full-time C] Part-time D Professor D Graduate Student ( ) 0 Associate Professor 0 Senior D Assistant Professor D Junior D Instructor 0 Saphomore D Other (Specify) D Freshman D Other (specify) [3 Staff E Full-time D Part-time D Maintenance/Custodial 0 Administrative D Secretarial [3 Clerical [3 Skilled tradesman D Food Service D Other (specifyL (Bl Address of non-interview location: —p INTERVIEwER: Space for cements on either interview or non-interview situation. T Interview Number Respondent Number | ‘flSL" ("l What is your age (as of your last birthday)? M ____vrs- (38-39) /_/_/ SPACE Budaef survey (l2) What is your marital status? (40) __I 0 Single D Separated investigatorl J. D. Stephens Cl Named D "mowed D Divorced (l3) Please indicate your current position at the university. (“l-53) I I I I TE Faculty D Student El Full-time U Part-time [3 Full-time [3 Part-time 0 Professor D Graduate Student ( ) D Associate Professor U Senior D Assistant Professor D Junior D Instructor D Saphomore D Other (specify) C] Freshman D Other (specifyl C] Staff [3 Y Full-time [3 Part-time [3 Maintenance/Custodial D Administrative D Secretarial 0 Clerical D Skilled Tradesman D Food Service D Other (SWCHY) (l4) For how many years have you been associated. in one way or other, with Michigan State University? yrs. mos. . ((I‘I'lfl) I___I___I__I__I —-D If NON-STUDENT, go to cub; if STUDENT, continue below with 015 (I5) For how many credits are you enrolled this term? crs. (AB-#9) I_£_/ ('6) What is your major (program of study) at the University? (50-53) I I / I I College Major (l7) Are you employed in any capacity by the University? DYes 0N0 (5") _/__I —' If response to 017 is NO, go to 018; otherwise, continue below—j 0. Please describe your position (55) I_I 5. Approximately how many hours do you work at this job per week? (56-58) I I I I hrs. -——) NOTE: For faculty and graduate teaching assistants, ask: (59-63) / / / / / / Approximately how many contact hours do you have per week? (al.-58) / / / / / / c- where is your place of work on campus? (I8) Are you employed in any job outside the University? D Yes 0N0 (69) _/__I —' If response to 018 is NO, go to 019; otherwise, continue below (I. Please describe your position (70) / / b. Approximately how many hours do you work per week? hrs. (70-73) /___/__/ _/ c. where is your place of work off-campus? 3 Card 2: (ll-25) I I I I I / (26-30) I I I I I I 297 (l9) Would you please check the box of the group on this card that indicates how much total income you (and your family, if applicable) received in the calander year l972-- that is, before taxes? _' NOTE: For students, please include your grants (scholarships, fellowships and awards), income from vacation jobs, part-time employment, allowances from parents, wife '8 earnings, etc. (3i) _/_/ C] “"5" $2.000 C] $6.000 - 7.999 C] Sl2.soo - H.999 D 52,000 - 3.999 C] 5 8.000 - 9.999 C] 515.000 - H.999 D sh,000 - 5,999 [:1 slo.000 - l2.l.99 [J 525.000 and over (20) On this card would you please check the box of the type of housing structure in which you currently reside? Own D Rent (32) L.’ Single family Faculty or married housing C] Duplex U Residence Hall C] Trai lor Sorority/Fraternity house or cooperative Small apartment building (:5 units) Other (specify) DUDDD Large apartment building (:55 units) (33) U (2|) How long have you lived at your present address? V'S- "'05- (3‘0'37) I I I I I —' NOTE: Include summer vacation period if, for example, respondent was living at the same address before and after the vacation. (22) How long have you lived in the Lansing metropolitan area (including tri-county area)? yrs. mos. (3340') I I I I I (23) Before moving to the Lansing area, where did you live? (Town/City) (State) What was the approximate size (i.e., total population) of that town/city? (“2) I__I —-’ NOTE: If rural location, ask for total population of nearest village, town or city. [:1 less than 2.500 [:1 25,000 - 50,000 [3 250.000 - 500,000 C] 2.500 - l0,000 [3 50,000 - IO0,000 C] 500,000 - l,000,000 D i0,000 - 25,000 D 100,000 - 250.000 [3 l,ooo,000 and over (24) About how many times have you changed your residence since being associated with Michigan State University? (“3'“) I I I —' NOTE: Do NOT include temporary/awner vacation residences outside of the Lansing area. (25) How many adults (>i8 yrs. of age) live at your residence (or household)? (iiS-llb) I I I NOTE: If living in RESIDENCP' HALL-~the number of adults sharing room or suite. (23) Do you live with your family? D yes D no (h7) _/_I If response to 026 is .70, go to 029; otherwise, continue below d ”ling at (27)liow many children (~ l > (32) On this card you will see four E €52 £33 3 S 8 ‘5 5 categories of travel (A through D t; 3 C -->- 76' :E '5 '53». along the top side of card) together ‘t'.’ g E: if u z ‘23 3 5'5 with nine modes of transportation 8 E 3 g a: L 8 a §E 04 (along left-hand side of card). «i 2%, m u «.29 ,,, 2 c . c '0 c You: - c u c g Uslng the frequence-of-use codes 3 g 3 3 5 3.; 2 — - 2 e >. (given at the bottom of the card), a; J: u 3.19.! UN '6' “£2 would you please indicate for each - o e v- o c u u. 8'5 u— 8' Gum‘- o.cm-- 00"- 0‘00 travel purpose (A through D) the n g a; m I” ‘3 “N: m 5 > frequencies to which you use the g 03 L 5 l6 8 3 5 3 3 5 of: various modes of transportation. g avg g'h‘t- a zu- m u u m I: l) Car (your own) t 2) Car (friend's or neighbor's) 3) Motorcycle (or motorbike) 4) Bicycle 5) Bus 6) Taxi 7) Walking 8) Hitch-hiking 9) Other (specify) Prequency-of-use codes: 0 — not applicable 3 - sometimes . 1 - never 4 - frequently 2 - seldom 5 - very frequently (33) Card2: A(60-69) II/I/IIII/I 8(70-79) III/III/II/ Card3: C(2l-30) IIIIIIIIIII DUI-(IO) III/IIIIIII From the list of activities on this card, would you please indicate about how often you engage in each. For those that are seasonal activities, indicate how often you take part in them when they are in season. Circle the hunter which best fits you: 1 - never; 2 - seldom; 3 - sometimes; 4 - frequently; 5 - very frequently. l) i 2 3 ll 5 Visiting relatives 2) i 2 3 ll 5 Visiting neighbors and friends 3) i 2 3 ‘I 5 Going to plays, concerts, or movies 4) l 2 3 il 5 Reading, studying, listening to music 5) l 2 3 II 5 Going to church or religious activities 3) l 2 3 ‘l 5 Going to classes or lectures (extracurricular) 7) i 2 3 ll 5 Hatching television 8) l 2 3 ll 5 Shopping (except for groceries) 9) l 2 3 'I 5 Going to watch sports events [0) l 2 3 ll 5 Playing cards, and other indoor games II) I 2 3 ll 5 Playing active sports )2) i 2 3 'I 5 Going to nightclubs, bars, restaurants )3) l 2 3 ‘l 5 working on hobbies, painting, music [4) l 2 3 'I 5 Participating in club meetings and club activities I5) l 2 3 Il 5 working in and around the house, yard, or building I6) l 2 3 b 5 Other (specify) I7) i 2 3 ll 5 Other (specify) Card3:(Eb-62)IIIIII/II/II/IIII/ 5 299 Interview Number Respondent Number Supplementary Questions to Time-space Diary iNTERViEwER: Briefly, but carefully, review all entries that the respondent has (i) (2) (3) recorded in the time-space diary, checking to see that: (1) the reporting of activities has been accomplished at an acceptable level of detail; (2) no portions of the respondent's day remain unreported; (3) activity locations are clearly indicated; (4) modes of travel are specified; (5) the times spent travelling to or from activities are recorded as separate activities; (6) activities and associated information are reported sequentially. If the level of reporting is too general, probe for more details about the activities. After any probing for clarification, administer the following supp lemen tary ques t ions . we would now like you to go through the diary of activities as you have recorded them. For each activity, would you please record (in column 9 of the diary) y0ur answer to the question: TO WHAT EXTENT HAS THE ACTIVITY PLANNED? You may choose from the four possible responses given on this sheet that best answers the question. - arranged with other people - planned independently - routine - unexpected AMMN (The respondent should repeat this procedure for each activity entry in the diary). Now would you please review the day's activities as you have recorded them. While doing so, try to pinpoint those activities or 'episodes' which, in your opinion, are important in the sense of having had to be done at a fixed time or location. You might think of these activities as having been 'pegs' about which you feel your day had to be organized. (a) Please indicate these activities by placing a check (or 'X') in column 7 on the lines corresponding to these activities. (b) How, in column 8 would you please rank those (N) activities in order of their importance (l,2, ... ,N). A 'I' would indicate the most important, a ‘2', the next most important, and so on. Finally, I would like to ask you four simple questions for each of the activities that you have recorded In the diary (except for sleep and travel). These questions seek only yes/no answers. (a) Could you have done anything else at this time? [Column IO) (b) Could you have done this (activity) at any other time? [Column II] (c) Could you have done this (activity) elsewhere? [Column l2] (d) Could you have been elsewhere at this time? [Column l3] (Repeat this questionning procedure for each activity entry in the diary (with the exception of sleep and travel)] APPENDIX C DATA CODEBOOK FOR INTERVIEW SCHEDULE AND TIME-SPACE BUDGET DIARY DATA RECORD 11 Card Item 2 Columns Number Data Item and Code(s) 1 Deck number (1) 2 - 4 5 Respondent number (i.e., unique number assigned to respondent by interviewer) 5 - 7 Sequence identification number (001,002, ... ,n; where n=total number of reapondents) 8 - 9 Card number (01) 10 blank 11 - 15 6 X geocoded grid coordinate of respondent's residence 16 - 20 6 Y geocoded grid coordinate of respondent's residence 21 8 Weekday of completed time-space diary: l-Sunday 3—Tuesday 6-Fr1day Z—Monday 4-Wednesday 7-Saturday S-Thursday 22 - 23 8 Day of month (1,2, ... , 31) 24 - 25 8 Month: lO-October; ll-November 26 9 Sex of reSpondent: O—no response; l-male; 2—femsle 27 - 28 11 Age of reSpondent (as of last birthday) 1Data records 1 through 3 contain information obtained from the supplemental interview schedule (Appendix B). 2Item numbers that appear under this heading correspond to those item numbers in the interview schedule (Appendix B). 300 33 35 37 39 41 29 30 31 32 34 36 38 40 42 12 13 l3 13 14 14 15 16 16 301 Civil status: O-no response 3—divorced 1-sing1e 4-separated 2-married 5-widowed University status (or position)-- University affiliation: 0-not reported l-student Z-faculty (instructional and research programs and administrative officers) 3-staff (clerical, technical, and labor classes) Subdivision of university affiliation: If student (1): If faculty (2): O-not reported O-not reported l-doctoral level l-professor Z-master's level and 2-associate Specialists professor 3-senior 3—assistant 4-junior professor 5-sophomore 4-instructor 6-freshman 5-research 7-other associate 6-administrative officer 7-other If staff (3): O-not reported 3-labor l-administrative 4-other Z-clerical/technical Part-time/full-time affiliation: 0-not reported; l—full-time; 2-part-time Length of time associated with university-— Number of years: OO-not reported; otherwise, actual number of years Number of months: OO-even number of years or not reported; otherwise, the actual number of months (01,02 ... ,12) Number of credit hours for which respondent is enrolled this term (for students only) Major program of study (for students only)-- College in which enrolled (university classifi- cation codes) Major program within college (university classification codes) 45 48 53 S9 62 67 43 44 47 52 S7 58 61 66 71 72 73 74 17 17a 17b 17c 17c 18a 18b 18c 18c 19 20 21 302 Is respondent employed by university (for students only): O-no reSponse; 1-yes; 2-no Occupational classification codes (for university- employed students only): O-not reported 3-food service/labor l-instruction and 4-other research 2-c1erica1/secretarial/ technical Number of work hours per week: For students-—hours devoted to university- employment per week; For faculty--contact hours per week; For staff-—average number of work hours per week X geocoded grid coordinate of respondent's university work location Y geocoded grid coordinate of respondent's university work location Is respondent employed outside the university? O-not reported; l-yes; 2-no Number of work hours per week (for those employed outside the university) X geocoded grid coordinate of respondent's non- university work location Y geocoded grid coordinate of respondent's university work location Annual income classification codes: O-not reported 3—4000-5999 7-12500-14999 1-under $2000 4-6000-7999 8-15000-24999 2-2000-3999 5-8000-9999 9-25000 and over 6-10000-12499 Housing classification codes: O-not reported 6-sorority/fraternity l-single family house or c00perative 2-duplex 7-residence hall 3-trailor 8-faculty or married 4-apartment housing Does reapondent own or rent residence? O-not reported; l—own; 2-rent 74 77 10 12 15 17 20 23 - 76 - 78 14 19 22 27 22 22 23 23 24 25 26 27 28 29 30 31 303 Length of time at present address-- Number of years Number of months: OO-even number of years or not reported; otherwise, the actual number of months DATA RECORD 2 Same information as coded on data record 1 Card number (02) Length of time (as resident in Lansing metropolitan area-— Number of years Number of months: OO-even number of years or not reported; otherwise, the actual number of months Size of place in which respondent lived before moving to the Lansing area: O-not applicable l-less than 2,500 5-50,000-100,000 2-2,500-1o,ooo 6-1oo,ooo-2so,ooo 3—1o,ooo-25,ooo 7-250,ooo-soo,ooo 4-25,ooo-so,ooo 8-soo,ooo-1,ooo,ooo 9-1,000,000 and over Number of changes of residence since affiliated with the university Number of adults living at respondent's address Does respondent live with family? O-not reported; l-yes; 2-no Number of children in family and living at home (for respondent living in family situation only) Is respondent's spouse, if any, employed outside the home? O-not reported; l-yes: l-no Distance (in miles) from residence to campus (distance in miles and decimal fraction of miles to two decimal places) Number of cars owned by respondent 28 - 29 31 32 33 34 35 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 32 32a 32b 32c 32d 304 Number of cars owned by others living at residence (i.e., how many are available for reSpondent's use) Modes of transportation to and from university (i.e., for classes, university employment, and work-related activities): car (respondent's) car (friend's or neighbor's) motorcycle (or motorbike) bicycle walking bus taxi hitch-hiking other Means of transportation for shopping (i.e., for food and other everyday needs): car (reSpondent's) car (friend's or neighbor's) motorcycle (or motorbike) bicycle walking bus taxi hitch-hiking other Means of transportation for local leisure activities: car (respondent's) car (friend's or neighbor's) motorcycle (or motorbike) bicycle walking bus taxi hitch-hiking other Means of transportation for non-university employment: car (reapondent's) car (friend's or neighbor's) motorcycle (or motorbike) 3Legal codes for items 32a, 32b, 32c, 32d, and 33: O~not reported; l-never; 2-se1dom; 3-sometimes; 4—frequently; 5-very frequently. 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 11 - 14 33 4 305 bicycle walking bus taxi hitch-hiking other Frequency of involvement in leisure activities: visiting relatives, neighbors, and friends going to plays, concerts, or movies reading, studying, listening to music going to church or religious activities going to classes or lectures (extra curricular) watching television Shopping, except for groceries going to watch Sports events playing cards, and other indoor games playing active Sports going to nightclubs, bars, and restaurants working on hobbies, painting, music participating in club meetings and activities working in and around the house, yard, building other DATA RECORDS 3+NA Deck number (2) Respondent number (i.e., unique number assigned to respondent by interviewer) Sequence identification number (001,002), ... ,n; where n=total number of respondents) Card number (i; where i=03,04, ... .N)5 Start time of activity (time given in hours and minutes) The_number of activity modules needed to describe the behavior of an individual over a 24-hour period varies from respondent to respon- dent. Two activity modules can be coded on one data record. Therefore, N82 (background data record) + reSpondent's total number of activity modules / 2. 5Item numbers that appear under this heading correspond to those item numbers in the time-Space budget diary. 15 19 24 29 31 18 23 28 3O 32 33 35 36 37 38 39 WU.) 10 306 End time of activity (time given in hours and minutes) X geocoded coordinate of activity location Y geocoded coordinate of activity location Primary activity code (see classification of activity codes below) Secondary activity code (see classification of activity codes below) Type of activity location (see activity location codes below) Others present or involved in activity (See social contacts codes below) Travel mode: O-not reported 5-walking 1-car (respondent's) 6-bus 2-car (friend's or 7-taxi neighbor's) 8-hitch-hiking 3-motorcyc1e (or motor- 9-other bike) 4-bicycle Subjective ranking (relative to respondent) of important activities: O-not applicable l-first important activity 2-second important activity . O O O O ' O 9-ninth important activity Degree of reSpondent'S commitment to activity: O-no response 1-arranged with others Z-planned independently 3-routine 4-unexpected Four-part subjective fixity question—- part l--activity-choice fixity part 2--temporal fixity of activity part 3—-Spatial fixity of activity part 4--Spatia1 fixity of activity O—no response; l-yes; 2-no 307 43 - 44 blank 45 - 76 1-13 Repeat data format above (columns ll-42) for data describing the second activity module to be coded in columns 45-766 THE CODING OF ACTIVITIES Activities listed in the respondents' diaries were coded into 98 activity categories using a two-digit system.7 The first digit divided activities into ten main groups: work (0), housework (1), child care (2), shopping (3), personal needs (4), education (5), organi- zational activity (6), entertainment (7), active leisure (8), and passive leisure (9). The complete two—digit activity code is presented in Table C-l together with their abbreviations. Several activities may take place Simultaneously, and provision was made in the coding procedure for the recording of two simultaneous activities. In order to accomplish this, one activity had to be desig- nated as the primary activity and the other as secondary. The dis— tinction between the primary and the accompanying or secondary activity depended on how the reSpondent described his behavior. 6Repeat the format of this data record as many times as necessary in order to describe the respondent's behavior (sequence of activity modules) over the 24-hour period. 7Perhaps the most widely recognized system, and the one adopted here with minor modifications, is the one developed by the 12-nation consortium of social scientists. See Alexander Szalai, "The Multi- national Comparative Time-Budget Research Project," The American Behavioral Scientist, 10 (December, 1966), appendix. 308 wowoouw .muvcamg wcficouw .huvosmq «a monono noovuoo Amwmnumm mo Homoomfio .xamamvamv wcwcmoau uoocuso ma mono; cmoao chaxmalomn .wcwnmma .wcwnmosmv mcwcmoao uoomcH NH anamoao How: monmfio >m3m wcwuuaa can a: wcwnmmz Ha ooom oumooum 000w mo wcwxooo can cowumumomum OH xmoz UHHmszQ mated ooh ou Hm>muh uuoamcmuu mo momma How magmams wauvsaoow .oomamxuo3 Eoum chaumu mam cu Hm>mue mo mxmouo xuoz msauxuoa wcwusw mooauma wawxuoaucoc confluumoum vow mxmoun unaswmm mo umnuo .xu03 u< xuos vow momma no mcwuumum muowmo momamxu03 man on uomam me9 no xuoz um mammz oomfiaxuo3 um Hmmz 00 non vaooom MyoS .uum .humwaaxsm .umumaomwob mo wmmamo .wofiufimz A.oum .mmawcoma mo czoo Ixnmun .owmuuonm maaaam ou moo ..w.ov xuoz aouw woumHomw on coo u“ ma mafiu wofixuoz_wcfiu:v cofiumsuuouofi Ho wcfiuama ha< co wcwuooa xuos woauoos uo cowmmSUmav mommamulxuos mo mawuuo>o oo Boom nmumaomw MHHmofiwfioomm on one u« ma mEHuum>o No oaon um xuos mac: unwaoua no mac: um Macs Hmaowwmowoua Hmeuoz Ho xuoa umfiswmm AwEo: ovfimuaov xuoz Hmcowmmomoua Hmsuoz 00 RH 09 amaomzzoo MZHH 92¢ mzHH czHMmoz monoo onHMMmm< onHmHQOma VHH>HHQ< maoo mmou MHH>HHU< HHUHQIOZH Hlo mqm<8 309 mongoosm mooom noasmaoo manmnav mo mcnmmnonsm Hm wonuoxnmz muosoona pom mooow noasmooo hmvmno>o mo mcnmmxonsm om mmoH>MMm 924 mooou mo qummnH unoamcmnu mo momma now wcwunms wsnvSHocn convannu howanoum cu Hm>mna mm xcmam . mom: uoz mm unmhnmn .nmnuo mnosuo mm nuamms .vnnnu Aconvanno wo guano; ou woumaon monun>nuom nonuo no .omnucwo no nouooo cu munmn> manosfiocnv onmo Hmonnwz om wonmmaa noonusc mxamz mom mmamw noouuso mm wcn%maa nooocH conuosnumcn Hogans mom mmSmw nooocH «N connanco.0u Mame omnoanno cunz moonummno>oou .cmnvanco ou mxooo Hoonomlcoc nonuo no mmfimu mo wcnpmmm mm xno3maoz so damn Amcommma com momwonmxmv xnos Hoocum mo conmn>noanm mm OHGU flHfiSU fiwhfiHHSU HmUHO Cu QHMU HN onmo mnmm weapon on mnmo om zmmaAHmo 0H mm was mHHnb :uHS wonamwo ..w.mv mnocuo ma nonm3 .umom nooxmatlmmHHQQSm nouns vow noom ma onmo Hmancm .wSHvanmu onmo finance .woncmonmo NH mooxan nonuo moonumnmao mac: mom mnnmmon nonuo on mooxas monuoao .oum .moonm .monuoao mo aooxao no nnmaom ma onHmmmm< onHmHQOmn NHH>HHu< maoo voacwucoollalo mqmmne unoamsmnu mo momma now wanunma wanvaaocn .monun>nuum coconuooa o>oom onu On vmuooaooo wonHHm>mnH me nonuo .mum>nnm mnonuo .oonnnummvnoo: .mmnun>nuom oum>wnm we wcnumom . ummn no amz sq amoam manumma Aamucovnoonv aooam manuhma we Amman uswnz . Aamnncmmmmv ammam unwnz we mamma ucmnsmummm mac: onu mvnmuao name: «a wxomcm .mammz @805 um mxumom new mamoz me moanwm cu mnmu xnos oaonomso: on oopsaoon no: ma .muaaom cu om>nw mnmo we Hmonvma Hmoomnmm mac: um mnmo Hmowooa Hmcomnmm Hq moonwhs Hmcomnom A.oum .omn cu wcnow .QS wcnuuowv wonmmonm .momnwmc HmGOmnmm oq mMHBH>HHo< nmmHMUmmmzoz Qz< me<>me .mMMAm .m4Hmm aqloq mon>nom .Ho>mnH unommomnu mo momma now wcnunmS wonvaaoon .mmnun>nuom mononuooe o>onm man on wouoocooo wonaam>mne am xcmam com: noz mm oon>nmm nonuo mnmSuo mm Sana an wonunmz moon>nom vow mwoow mo ommnonoa ecu now wcnososv .wcnunmz om mon>nmm nnmaom moon>nom nonuo can nnmaom mm mon>nom m>nnwnumnafiao< moonmmo .moon>nom o>numnumnofiao< am A onmo Hmonooz mac: man monouno onmo Hmoncmz mm mnmo Hmoomnom oao: man ovnmuso onmo Hmoomnom mm onHmmmm< onHmHmomma wHH>HHU< mnou vmsafiuaoollalu MAMoo omonu swan nonuo mmaHuooa cH ooHumaHoHnnmm No nmonwo we xnoz SOHnmuHomwno HonuHHoa no HmHoom n no HmHonwo nouooHo no mm nuH>Huom znmuasHo> Ho mUHuHHOQ .coHc: .oum SOHG: .munma a mo nonama mm :oHumoHoHunmm oo mmHHH>HHo< onHHHomHHoo 92¢ uH>Ho monoo >v5um .Ho>mnH . unoamsmnu mo momma now wcHnHm3 waHvoHooH .mmHuH>Hnom ooSoHnaoa o>onm man on omuooccoo wSHHHo>mnH mm monum nonno mnonuo mm oonH>nma=m .HmHnouse madam uconcmaoocH no .nonmomon .thnm oomH>noaSm .HmHnonSH mm onmmH on comm cOHuoanumcH Hmaomnoa now mxoon no mSmH>on no moHcmom om mmmum now onnHm3 nOnSu . no Human HmSOHuosnumSH mo nomads now wconoH no now moHuHmz mm SOHnmuHSmooo HmEnomoH mm Bonn uwumHomH on one uH HH .mnonuo :uHS mnmuuma omanmnlmmnSoo mo aonmoomHQ «m SOHumuHSmcoo hvonm nonou no mmmum HmSOHnuonumoH nuHa coHuSUHSwoou no wcHmeH mm xnoaoEom SOHuoznnmcHumem vow xnos nonmommn noumHon waHoSHocH .monSuomH can moonsoo ucwnmmmHo now chHomon .onan3 .hoSnm oum>Hnm ..w.mv xno3msom mm mommmHo nocuo moonsoo ononnu HoHommm no AmHHmoonmooov monouooH HmHooom on moooocmuu< Hm mommmHo vcmuum mnH>Huum oHaHooHna on» waHmn mmHooum .AuoovSum mumsmmnwlumom no mumsvmnw Inooosv mnmoHaom .monauooH .mommeu on moonwaouum oaHuIHHSm om UZHZHmmmm< ZOHHmHmomma MHH>HHU< mnoo wmsaHudoollHlo MHmmna unonmomnu mo momma now onnHma onoSHocH .moHuH>Huom omcoanoa o>onm can on monomocoo wcHHHo>mnH mm HmHoom nonuo Ho>oom ooSOanma mmonu Gaza nmsuov mooHuaooon wchcouo< mm mnmo .wwmo non .ommu mm mHmoa .xnnmm mocmHnH no cmnowmo no on omnmumo Home :uHS SoHuamuon no zunmm on mocmHnw nuH3.mcHuHmH> mvcmHnH wcHuHmH> no moooHnm mo uHmH> moH>Houom mm masons: aoHananm .asmmsz an noummnH mnoao .unooaoo .nmummcH mm moH>oz moH>oz mm mnSuHso mom: nsHoluan: .305m .onoomw .HHmsnonSE .mSUnHo Hn mucm>o mnnomm ucm>o munoam m moHocouud ow wens nwne unommomnu mo momma now waHuHm3 wcHuSHooH .meuH>Huom umooHuomE m>onm one on wouoooaoo onHHo>mnH mo :oHumNHomwno nmsuo mnoxuo we mSOHnmuHommno .omHz A.ouo . .hnmuHHHE .uconma .zHHEmmv mSOHumHoommm nonno :H SOHnmmHoHnnmm no mHHoasoo nuHmnm>HoD AmSOHmmHaaoo .moonuHaaoov mHHoozoo nuHmno>Hos mSOHnm> SH aoHumaHUHnnmm co ooHnomna mSOHwHHom moHaoaonoo moonHHmn onooouum vow moHuomnm mSonHHom mo SOHumuHomwno mSonHHom mSoHumNHomwno mSonHHon :H SoHummHoHunmm no moHuH>Huom oH>Hu AmnoonosHo> ..w.ov >nH>Huom oH>Hu o>HuomHHoo onalaoz mo onammmm< onHmHmUmmn wHH>HHo< mooo cmsaHudooslHlo mqde 313 onSmHoH .Ho>mna unoamcmnu now memos now onnHma wanSHocH .mmHnH>Huom voaoHuoos o>onw man cu kuuoccou wSHHHo>wnH om onnu .onom mnH>Hnom mHonH> o: .wchuoc wcHov .wcHaomHn .wcHxSHnu .wcHuomHmmn .wcmeHmm mm mum>Hna .mnmuumq oucoocoamonnoo oum>Hna wcHanz no SOHummno>coo :oHummnm>aoo moonaoHou waHvaHoaH .mooHummno>coo cm nmama comm nounamSmc wavamm mm ocHNmmme wmom .ouo .muoHnaama .mHmoHvOHnoa .m3mH>on wchmom am noon 23m 38a 953% mm monouon anm monoomn On onooumHH mm 5 83933 323m: 3 Ovam OHomn man on wchmanH 0m mmanmu m>HmmmnH unoamcmnu mo momma now waHnHmz onoSHoaH .menH>Hnum woooHuoms o>oom won on nouomcooo wcHHHo>mnH mm oEHumma nwzno mnmnuo mm mosmw nOHnmm mwamw >anoom mm on38 wcmez wcchHm .ucossnnmcH HmonSE m waHmmHm ow xnoz unm A.onm .mnSDmnmuHH .mnSuaHaum .wcHuchov wSOHnmono oHumHun< mm moHonoz waHaom .ono .wcHuuch .wcH3om .xnos meomz cw mmHnnom mSOHnomHHoo .mpoooL HSUHonooH mm fins m $32. 9:3, mm waHxH: .onsmHm mSOHuumHHoo .maHnosn .mconnaoxm Hm munomm o>Huo< mmHonoxo Hmonhna van unoam m oUHuomnm ow mmDmHma m>HHo< az< mHmomm mmaow ZOHHmmmm< onHmHmommo MHH>HHU< mnoo wmaaHu:00IlHIU MHmHumnumHaHavm SmanHno nonuo no uncoano: aOHnwuHamwno no mo mnoeama huHSuom .mnaovaum SOHHoH .mnoanSIOU oonmvaon m.n:uvaoamon man ovauSo vaoHnm .mo>HumHon moomvaon m.naocaommon no muHva nonuo oovaHmon m.naov:oamon no noanHno vaono a .mHnooa vonHuvaHaS home meSvH>HnaH nonuo on ~¢ In\Dl\HumnuchHawm noanHnu nonuo no mnonano: 8HnouH§mno no mo anon—awe onamvnum SbHHom no mnoxn031oo oucwnHmon m.uaovaommon man ovanSo moooHnm .mo>HuoHon monouHmon m.ncowcommon can no mnHva nonno moawvaon m.naovaommon um SoanHno omaomm ocon .3 tn~OIxtn OHNM ZZDHOO 9200mm 223400 HmmHm mmnoo HU483» % APPENDIX F THE COMPUTER SIMULATION PROGRAM" * The following Simulation is programmed in FORTRAN 6000 Extended and was implemented on a CDC 6500 computer in the Computer Laboratory of Michigan State University. 323 324 on oom DON owm usm 00m umN ocm 0mm ONN OHN new coH omH ONH 00H omH u¢H amH ONH OHH OOH on on ON on cm 0* on ON oH oo tum sz I_m tum er 2~m sz 1Hm t—m sHm mum zwm IHm :Hm sz z_m x~m t—m tum z—m {Hm tum IHm .0mu.~.m-00znu« C.ou.fi.H.QD&hu< 0H.Hufi am 00 Nb.HnH om no >Hoz_ IhIH no mwszz nzmQZUdmwx aqua anon on co .o.no.2. u— .z. mnmazn anew mzo_n:mnmnmno >nnnnmnnanazao came ouz oooHuowm \ooom.oomH.oomH.oO@.coo..mc.acm.omH..oH.om\~OH.Hufi.Afiwaso. .Hum.Haydn. qhqo \ooo..om..ON.ooH.om\.m.Hum.Hp.h. qhqo \oHHHH-owHHH.oHNHH.oNNHH.oHHmHsoNHNH..HNNH..NNNH.oHHHNH tum..NHHN..HNHN..NNHN..HHN~.oNHNN..HNNN.oNNwwxaoH.HNH..~.»aa<~hzm30wm wthH>~hu< no 20Hnmmnzn mrnn loam «an nznamnmzcu oza .nzwznnzxno .muzanxoazn no mwaomu min mnq23<>w H4. hmmzqm 44cm 0.00Hfiamoo.o+.ovxmmazayua kmmzqm «om mem MEG—z: szhwa on ZO—hu73m zauHZD 4440 o.numx_m 0.0umowo meN on mambzaoo >thHm 024 wma hum moznnzuo m+snusw m+nmunn Hfifi.~HuH..h.Hufi..p.Hvoozkuq.. HomoH.o. Dawn 0.Hu¥ o¢ mo mufifi Hum— o>¢mhzm my_h 104w mom zo—hdeo—uzou hzudmanOu wzb oz—omouma >mwmwrh .moan4d >h~xuu oz« .hzwxhuxxou .wuzHQZH mom mwdzooz >h~>~hu< owuzwaawm no hmm m>~h<4mm Gama >m omeHFZmDH ozum owa«wth mrwn quw mom Xmozm hzuqxhwzou w>_buwfim3m a mhznzou Oh bzmxamw maznnzuu o.H+mx~mumx~u n.nau.H._.noznu4 mwxum 4P~>~hu< um om 0» Do Hoomohmoao.u.OOZhuqcmu.ooNoh4.Hm.”wouzhuk~>~k04 dc muZmumDUUO dun gumxu om On Go .oom.ho.am.Hacozbu4m>-an2mw .m «u .N.H u_ nozmhqm FZmthzzuu mourn [Hm {Hm tum zwm {Hm {Hm tum z—m IHm tum ZHm tum tum am Oh do o.ou.~.~.ooinua Oh On 00 Hooooow..fiamuoozhu<. um swim aam>¢dh mo dmme mH 4<>mwhzm wxah oszDo szmOudmd >n~>Hnu< uH om an ac OoH+momaumomn om.co.um ..n.n.ooznu<. in own ma hZDOu .Omlezoz mH "oz—124m wuzHH coHH omHn u¢HH omHH oNHH oHHH poHH oooH omoH onoH oooH omoH oeoH omoH omoH oHoH oooH coo one are com IHm ynm sz sz me {Hm rmm tum sz rnm {Hm :Hm tmw sum sz zmm sz 1mm t—m rum t—m mum tum me tum rum tum tum tum th me tum duuwmd oun— 0mmn.HNQH 0mm 00 mowduowau .wwn £04m mam owZGHmmq Om4< w~ >h~>~hu< to ZO—h<004 024 zn~nHhu< win mwzmzmmhmo Home; mHH- thxowm QZHIUJAUm th warnneuo maznnzuo aznu.o.nsoozn82 ..arnnnaax_u.cm.m. ii on.numzn man no .o.~.ocznu< zn w344> xmozn wxnnm >4Hmmmnz~ mznn mow xwozu nznqmnmzou m>~numwm2m mnaazou oHH on ow .o.o.0m.~n.~.ooznu<. in xuw a we owe a mq>mwnzn mxnn m— Nn.Hu~ oHH co UUUUUL) P4 H San ULJL) U UUU UUU 328 uomH ommH upmH oomH OmmH memH ommH CNmH onH oomH uo¢H OmoH o~¢H uoeH om¢H u¢¢H OmaH ON¢H oH¢H woeH oomH ommH upmH oomH OmmH o¢m— ommH ONMH OHMH oomH ooNH omNH rnm znm rem tum zmm :Nm er er :Hm rnm znm zmm :_m r_m rpm I_m z—w sz {Hm xwm :Hm rum :_m th tum I_m xnm x—w Inm th tum rum >«aaq uoma my» 10am mwam<> ozn>mnmnmm «on mnanaummnm mznzamnmo xu.~.qm.en;nua >49 m.4HCZH m1» m0 4<>mmn2~ IhldH th or menu >h~>~hu< IHOK nib zmfimmq omH 0 no H+xuy m<>+m>oun>u oeH an no .mq>.mb.n.oz<.b>u.no.x. in mmqm amva: aum2 ZIDDUU mefigu ml» d0 mmiu IUHIE 2H Axx.4¥.x.kxmw+md>nma> mmuuczu thqmpw7cu 27¢ hZLEEHZYQu U» officxouuq >H~>~Pu< no mwuhmanmcmoma m1» wu>Ho >dua< kam Hm>fi.m<>.m. IDZM 444m cmxdoudwn >h—>Hhu< QZHZHmehmD mom mmmtnz rm07 uz~>w~¢hwm «on wHQHmUmmDm wzuzmwnmm (‘1 q LJKJ'JL.‘HLI(.)‘JL) L‘UUUdk-U UUL) ONH k3UL) 329 oHoH ocoH OOmH Ome OhmH oomH ome oewH Ome ommH onH CQmH mopH OQPH ONNH ochH oth 0¢~H OMPH ONFH SHNH CONH OOOH oon OhbH uccH omoH o¢oH omoH ONOH oHcH oooH tum z~m {Hm H+hH>Hhu< rmmwzdm zo~mmq .wzdm th mu own u— nad>mwnz~ mth FZmDOwmem xuuru sz znm tum I_m znm sz sz raw ram snm zmm znm znm smm er tum Inm znm znm I_m sz sz sz Ina tum xnm mum {mm Han .mnu.o.<~.goxno< xxuiw.an.ooinuq 4<>wmnzu Inlqw wI» 0k ZDthuwa >h~>~hu< Inl¥x wxh onmm< u~H Oh Wm H+XXH1¥ 4a>+4>Jud>m me ck ow .md>om4.x-024o4>0onwom. u“ .hx.x¥.x.u044+4<>uad> mmHhH>Hnu< mm ZCanuum mmwamcmn mu>mo >«aaq uémq ~4>Q.m«>.a. LDZm gnaw onbquoq >nH>Hhu< wZHZHZEmhmO «on mwszZ IQQde qummzwo Huxx munx onH Oh no manunzn H+mum ooH On co .nzH.wm.<~. in Hum NHunzH .“ U‘ UUUUF“ L) ..o L.) L.) k.) H UC)L)""‘ U'JU 00H omH 330 om~N ONNN oHNN CONN ooHN omHN ONHN oon omHN 04HN umHN uNHN oHHN ucHN coom omom Chow ooow Qmmw ooom omom ...r N .c.. N OHON ooom QOOH ome OFOH OOOH omoH O¢OH omoH ONOH znm tum :nm xnm znw :nm znm x—m znm znm znm znm znw «>40 m.m2C do szkuDdhm Ad>wmhzu wx_h «on wznhdd >thHu w>~huwfim3w Daron whadyuu o¢~ or Do .0.Hlomz.HH.m~_ooxhu<. m~ ”Hwy ONm 0» mo Awhohoomww TH H+m~u4~ xmu q 2~4k20u woomuam 3m? mwoo nmc>umnz~ wi—k PszmeZH 0 U 5" mrn 20 nuwmnm o¢_znamnmzmu nmma :Hmen m><1 up hImDOIn mmoOmHaw mwurh mad mwxmu min .mwwn “In 29mm z~m me tum I_m rum zmm tum IHm in tum sz sz sz sz tum sz IHm sum omo n60; .zonhdxzo .>h~>~hu< OwZHIdmhuo mwxwu win me qum we Zu—nqumm 24 mzammmq thtowm wzmxfimguu wI» w37Hn2mu .m.m.x.nuqou.o.qm moanua on an no H+4u4 CNN on De ..m.x;o.wm.zxro. mi Hum mH*wHuza:o 00H 0» mo .mHuHo.q~.aoxnu4 oow on 00 Amwa.mz..n.mH.ooxnu<. m— H+<~um~ H+omuou xxn.m.«~.ooznu< xu.~.<_.ooznua omw On 00 .oma.mz..H.k~>HhU< no zrmhqmzo QZHZHmehmo «nu mwszz :uozaa mhqwamD 4e>mwh7~ waacuama >k~>~hua mxuh Ibis— wih zu.n.m_.noznu< new on go .m.no.nw. en wjzunzuu Huhfi Hafivmhoowoxq. uH ~.Hnfi mom ho a» >b~>-uq Irux «In xukd Cunfi zonmma QxN Ch £0 #52:“! 4c>+4>0u4>u CON an 00 .m<>ommomocz«om>o.no.m. m— .Sx.zxuxuhxmu+44>umq> .m>0.m«>.o. znz~ mmcu oz~z~zumhma «Um ammznd Fuoznm mhdxm7mo H x .m.m~.mozhu.mm.omzhuq HlaxxtmeuumH umm an to .Hommodwler. ma .z~ah caq mNH Oh ZmDhmx ..wmm 4 ..wow. LmeHdehmm 7mmm >qum44 ma: hulk L20 czn>wm mnzmnxw ZCLHanQ >nH>Hhu< uH nmm on go .o.n.nm..n.m~.cuznu<. um o¢m.oom.oom .rnuxmu. in n+mnum_ xn+mnnznm >PH>HFUQ quHDm >74 IhH; moqmew>u .xm~. zcfihqafio >h—>~hu< uH wzfizgmnm: DH ucmrq ins; HuxH .H.hmozu. up omH\.xz.mDmny~ m<>awhzm Ihimm Ch zoundmao >h~>~buq Ihtxz chmmq >4m>den7wn :Hm an we H+xxnxx 44>+4>Oum>m on on on HJ<>cmn.m.02d.m>ockw.x. nu .H.xx.xvnuuoa> .mzouhddDo >h~>HhU< wmmnmoma WZHqHZOu >o.oa>.m. zzzm mono Huxx «a "D (N £37”) an L) Q L.) L: 0". «1 t l L) (J (...) I u.) UUUF". 333 OOHM ome ONHM QOHM umHm 0¢Hm omHm UNHM uHHm JOHM poem boom 050m 000m umcm O¢Om cmom mmym OHOm ucum U®TN omam Chow GOON omom u¢om 0mmm umwm 4Hum Obwm 00mm Gmmm 14m 24m sz zhm er sz sz rpm 5mm 24m :Mm :Hm znm I_w 1—m sz sz THw tum {Hm £4m IHm sz sz iwu zwu {Hm 24m new 24m sz zwm mm mz~n ha >hH>HHU4 thlx ml» 2n aH¢ On mm¢.mo¢.oo¢ .s:. a 44cm >nn>nnoq aminozq mwm. one on we H+mxumx 4<>+4>uu4>c mt .44>.m4.m.neq.4>o.nu.m4 in .nx.mx.x.um4q+4«>u4<> meann>nnua no Pennaecb wamamneo wm>io naana oHna ZQHPQUF4 >nw>wnun 07~2_ZQMhuQ >qum< u04d m1» Stag mw34q> 02H>Hwku .4>O.4d>.m. 132v 4440 Humx amm ammzne TQAZSH mnnumHm: 4unx umm On we 4meun34 4+4n4 con un no .n/H.mo.+m. em 4H4 n4nnA~ a men mnannummmm mzmzamewc nub: men on ”c Hun: Hm UU'J" if ‘J'-"K. k '3 "’3 (_) '3' 334 OHmm 00mm O0¢m ow¢m UN¢M oo¢m 0m¢m u¢¢m ur¢m 0N¢m OH¢M Wm¢m COMM ommm 05mm 00mm ommm u¢mm “mmm uNmm mHmm .J.\.. n. m pamm ummm Ohmm 40mm Qmmm o¢Nm omNm ONNM oHNm OGNM 14m yum me zmm :Hm 54v 14w rem 3pm 1pm sz znw 14¢ 14m Sww rpm 24m xwm 2mm xmm sz THm :Hw 14m xnm 24w z_m th sz {Hm sz 24m 324 ow¢.o~¢.un¢ .mnt.n:xm4.. um .N.xx.x.huh~>ubuq Ihmz m4Dcoz >h_>—kuq w.4qu~>HDzH mhdfinn Un¢ Uh A? new a no .N.nn.x:. we no: on no .em. we .«so.n.en.xo.nn.x.m4.oanm. Mann» any” .wznn.uwa41 HHYN .m.4.e.no< uTHH wha4DU4au nc¢ On 40 a.m.i540mshu4.oi.mmv ’i. ».4 mw4h~>~nuq Ch 4w4mHmw< m70~hdut4 nxmowiuHS l ! I .Y‘ “I 00¢ on 00 Hah.mfi.oflznuqoawomv n“ H+nH C24 r— mwzuh ha QmZOHmwa wmmnH>Hnuc nzuuwuuac mm< gH>tJL>¢ U (,1 '3 l (_)U L) UtJC) 335 ommm cmmm UHmm 00mm Gown Omhm chhm 005m omNm v¢~m omen umpm CHNW CONN COOM wwom Ohom COOM umom OJQM Omen CNOM Odom 000m 00mm ummm Ohmm Ucmm ummm O¢mm cmmm UNmm 1~w x_m znm xnm xnm znm znm znm tum. :Hm 2Hw 24m zww 24m 2mm 24m :Hm z~w sz zwm I_m znm zmm 24m me 24m Imm z—m sz {Hm 2—m IHm an .xuunaoa Qm4somrum23 .msz_n IQUCJ 024 c<4u 0mm.n¢m.o¢m HooleN.H.muxnU<_ 267—2Hd7ua m:» «on m .mumcxu m1» mmn<4firmw him: m 0 00¢ Oh unq— Nh.H H oooH u; a n4 .wm4qm.uo<4m 0mm 0» we wawn.nuq4m 4:;Hhuw¢~H4 hmw uH >:nu< nu zm4n mm ozHLU44uu m1» mnvmnguu Dem on no rm; mszn H+mnumw s41on.o.m4.ouznoa com on a.mHu.o.mH4en2nuq co. .xmn.ow.mcu4 um I - D I I. HHH mH4032 hue .mxu.m mn4nmxnua yu.>.mn.noznue ¥mH.HnmmH OHm uo Hlxrwuxwu O¢N 0» mo 9.00.4.m44d uxnuq mxu.m.mm.ac5ku« Z~Iou.o. m—vnnzhuc L.) U L) '0 mtJLWJnJU am mam ome 0m¢ 05¢ 336 umH¢ 0¢H¢ omH¢ UNH¢ OHH¢ ocH¢ do u¢ 0mo¢ u~w¢ ooo¢ umo¢ u¢o¢ ante ONO¢ GHC¢ 00u¢ oaom omom Upom 000m ymvm 0¢0m .om0m omom OHOm 000m 00mm owmm Opmm 00mm ommm O¢mm zmm 14m z~m th sz zwm {Hm {Hm 2mm 2am z—m 24w I_w me sz 2—m tum ZHm anw tam tmm 2mm zmw [Hm zhm 2mm tum 24m I—m Haum 1mm sz >h~>~hu< ozwxcaaou\wzuomuwaa to >h~4~04m0md zunhcuno osz a«Dc.»udhaxo.4q.x.dmnzu DOOH Oh wimp up xue4m .m.q_.emznu .>.q~.aorn >h~>~huq ma w24~hn~>~nu< w4mxmnz~ ommacmzum ozn 2mmxnmm omxnu m4 zunnqomb 2mm; umo up no .~..m_.nmsnua .m.q~.QGihUdu.w.mH4602nua 4<>xmn24 wzah oz~3044nu\wz—omuwmn no ZOHFHH>Hhuq zwmmmq cxn.o.m4.aornuq H+Hum4 umo an we Hnnumn o¢o an no Hm<4u4 u— mnuox .mH.no.Qx. an .xXHUHucx Omo.coo.uoo .muyx. m~ oHo 0» mm 4+xxuxx 4d>+4>0n4>u omo on Do .4<>.m4.m.oz<.4>o.nw.a. mH .H.xx.x.nuu44> .4>o.4<>.m4 zazw 44¢ omh¢ Ohh¢ Uc~¢ Cm>¢ O¢F¢ OMN¢ um~¢ 0Hh¢ oo~¢ o¢c¢ 0wo¢ u~w¢ uoo¢ omo¢ o¢o¢ Omo¢ GNO¢ aHo¢ uoo¢ toomé gene onae Some omme meme 0mm¢ o~m¢ onmv come oo¢¢ o«¢¢ 14m :Hm z_m 24m znm znm 24m 24m tum 2~m rnm xnm 2mm znm znm sz znm sz tum :nm znm an xnm xnm xnw z~m 24m 24m xnm me 24m zpw .mwmnomuoxa xu<®IXOO4 mo Q¢w1a0x004 02H>54a2m .hzwzomm mHIh omo On 00 aquw.m~.ooxnu« m4unx .mH.no.ox. mu .gx.mnuux om~ e» um H+xxuxx 4a>+4>ou4>u own on 00 .4q>.m4.a.o24o4>r.nw.a. u— .H.xx.mx.nuu4<> .4>c.4<>.u. 232a 44cc . HNv—v— .w.<_.ooznuqunm.m~.noinu< mxu.>.m~.nmznuq o4<>awhz~ mt—h Ch ZU—hduo4 oz< .ZCthmDo .>h~>mfiu< onmmd SHN on cm n+mxumx 4<>+4>ou4>n Own an cu .4q>om4.w.oz<.4>0onoox. m— . .n¥.x.mx.uo4q+4<>n4<> .4>c.4q>.e. 232m 44qo numx .m.qn.ooxn4~nu<_uo can on Do HbHohw.xH. uH mexxuxH com on we H+xxnzx 44>+4>ou4>u Onm on no .4<>.m4.¢.oze.4>o.no.a. an .n.xx.x.noao+4<>u4<> .4>u.4<>.a. 222a 44am >n_>~no« e: >nnnnmamcla zunnaanu cyan anx aux .o.n_.oamu.a~. zqan 44eu xz~4 >hH>Hhu< m4mdmoad hmoz ozHu Uh thzwww umo on Do .DnonoodH. u4 H+dnua~ DON on ma omo on no .Dwoh4oau. u— HIQHHQH 4m~ Q» 46 .644w4 mu b~+HHDH Hum" O- Oh no hHlHuDH HNQH on» Ch uo HGQ4¢4 C4 xne «o ems >nn>nnu< za zonsce we mangoes naxn zonnquS oza .zunnaeao .monoxo urn nwzniamnmc nHm Ham ‘4'...) U f) 0‘ L} '4 c.) N 7.6:: awn Qm~ 340 omen owem onem ooem uamm ommm onmm comm ommm oemm cmmm uwmm 04Mm cnmm oomm oowm onmm oowm ommm o¢mm ummm ommo o-m gown wonm omnm osnm oonm omnm oenm omnm omnm IHw sHm {Hm {Hm 2mm er 24m {Hm {Hm sz :Hm 24m 2—m z~m 24m 2~m z—m mum {Hm {Hm 24m 24m :Hw 1km 24m :Hv 24m 2mm znm tum 1am sz 00h Oh mm 000 0» 0w .m—mh. m— cHo on we .4Q.Oio~m.H.QUEnJ<. mu mxu4n Cum on uo H+mxnm¥ 4<>+4>uu4>u 00w Le 00 ~4<>oM4oaocz<.4>Gono.dv uh .hz.mx.z.u04q+4<>u41> .4>0.4<>.x. 2321 44am Humz 4be com on we 4*mHunrw 4+4u4 new on do .Hz4.m4odH. m4 Hn4 uHun7H >nH>HPU< mo 7U~hh—4qumuqn m>~h<4DaDu quu \ovoooxznm «new Z~m.23xu.$gzmmm muemn2H .mamfivhmmob..mucHamc.hudc..nH.m~.nu1nu< \mux PDA? u .o.oamo.u04d..¢.0H.mo.bx~u..o.o.¥2~4doamo.mbvhu Omh UNF 0H5 90% 00c ewe upo 00c omo U¢o one our 0H0 0&0 00m {mm Ohm com 9mm mom 0mm Cmm on Qmm 00¢ nzH oz. dz” a2~ azn QZH ozu dzm azH a?” ez~ ezH o2» S74 cZH QZH dzH dzH dz“ az— a2— oz” a2» arm azH oz“ dz” dz” azw QZH dz“ oz. Zathv {Hm HuhmauH. whmox mDZthUU m~.-.HH.~.zDzo.132mmm .uow.oH. whuax ONH C» aDDHan ovumu fiuHH 00 Hmoh4oaDOH. u“ oouHuHHmDoH.nnc2_ za 2“ mnaoox nnn>nnoq .c.4uw..w._4xznma. .0n.4.coxnuh~>Hhu< aszHaumia uhqo PDQFDQ #300 >un4u .d :3 <1 ;. L t! umm.m ..n" H o H .L‘ 7’5”? {YO :JF‘LJ ‘YC4ti ’Y .11— OmH C» no HOOmCoDMo24 m4 4+3"; 2 .oHN.m. acme m37unz;u u:7~ng«u H0.an..x.fi.—_UQ4<. Ho¢Nam4 oqum me HNH L)L..H.)<_J L) 0L4 OLH m¢H fimH DNH 345 \IC' (DOC) u-o—t ( vhmo (Dr-i LJQQ m¢m I l C A SH (a as on em C om ow CH (mo mmo 0H0 oco oom 0mm Onw .uow mmm co: ...m d cmm 04¢ 29m ZCm tum rum Zum 2cm Zum 729w Zum ?pm nium 37...". 2:0 Dre 3:0 32m. :zx 32m and azh aim ozh ozH az~ ozH QZH a2“ az— dzw azn azH QZH .4>u.4<>.«. 4q>umh2~ a mum HP. chzH n2~qokm7nu m>~humflan 4qhoh mIn mwhnmzuu zouw o.H+.HI.n24mx.o.chmm.uu 1320.337mwanunm \4qx ZUZLUU Enzm mzuhDDmdpw Hm“. Ammo. Ho.mm¢.me. Hoomum.xmm.eomuw.xHHv .HonuoH.rmH~ neommfiH.XnH. Ann. .o.mmm.xu. anommmm.on. 8.N.oi.:..um. u: in." a oe mu 0.3nn .moumnu .aocHno H¢4um 20Hw1m2mu mTHh Ah.um. zoum uéHFUUSTDm 97w .u... yr. novu4>u CoOHdd) m~n¢ H I CXU hdzwwu edenuu hdzmum nomh arm m4¢c4md>q u2~n wszuuhmc .uom.oan..ONH..00..uo.om¢..um..mH.ooHnom\.wH.Hup..w.amc4 «new \.oo.oun.ooN.oa ..m\.m.HNp..fi.n4 a no \amo..m¢..N¢.on..mm.ohH..w\.~.Hufi..m4wk. «ham xc.>1.o<4m 4hH>Hhuq metbmm mt—n 024 mu>h 4u>quh wkdfiwhmg Ch mWDQuu:ua umxHhh .mDo.humho¥O.4h~>~huq Oh mxuhACZq wa>h 4m>q mzHH mu or D» co omnyhou>z u¢N Uh DO .>z_ um OOH 0h 00 ..a.homw.h¢h. u~ 0m 0» um ~+4u4 Jd>+J>UNd>U 0? Ch 00 .4<>oma.a.DZdoa>Ookw.mv ma .4.amvamch+44>u4<> an; .4>O.44>.u. znzm Agcu .wmadm.u>fi >4Q we wrap OzwcaCuuq IhQZwa a~MP w2~1umbwo Oh mammxsz SLfiZQa mhuqumc 00~ Ch 00 .4<.ow..w.ax.002hu<. u“ nN O» 0@ .0.0.0mo.h.a¥.3uzhuq. m~ (_) k.) U F4 0"" L‘. a) ’17 N [J »‘, A ‘1 L) L) (-3 U 348 com Ohm gem Cmm 0¢m 0mm omm ofiw cOw 005 amp opp new um» u¢~ OMN 0N» oak web coo Omo 050 Doc omo o¢o Qmo 0N0 Odo 000 com 0mm ohm I.» z~h in» zuh twp I_h xah tub I_p :Hh I_h zwh SH» sh» xpp twp tab rah 2H» zfih zmh zHF zwb in» 1M» 1M» I_p x_» imp 2H» zap xmh 4qn.m.az.ooxpua .mx..~.m.x.»u.a~.qoxkuu u¢~.nmm.umw ..m~lhh. aw ..o.a_.oozbuqi.mdvuh¢huhh o¢~ 0» 0o .o.m.omo.o.ox.cuzbukH>HkUd ~fi-zquum~ n¢ 0k um ~+fiun J§>+J>UN4>b o» o no .4<>.w4.«.oz¢.4>o.pu.m. um .fiuwvxz~aq+44>uad> 00 wk no .H.fi_xzmaq+44>n4q> 0m Ok Us «Odqw. uH ga>m.4q>.d. znza 4440 Hnfi .¥u~ 0N Ok no ~+¥ux om 0k 00 ..x.~z~awo zk~>~k04 afi.~z«akum~ m¢ 0k mm ~+fiufi a ....k>+n.->QNJ_>( o» o no .4<>.w4.«.oz<.4>o.po.m. a” .3.H.XZ—4q+44>u44> 00 Uk U0 .~.fi~x4~44+44>u44> om Ok 00 acqaa. a— .4)O.4q>.d~ 232a anu fiufi .xHH om 0k no ~+¥ux om 0k 00 ..x.~zHawo zqak .mw~»_4.mcmoma 2cmp~m4qah gnaw; .o.k~.o«4¢.afi. gum» wzmpaoau;u 92w yuahL: .um4<¢.uxu zanhmx .msm».uxu xu.~.ax.auzkuq )'. tf‘ J\ 9N GOOD ’5 ”W N 350 0mm 0¢N umm CNN Dam com wag om“ aka cod cmH u¢~ 9mg QNH cd~ 0nd co um 0k we um o¢ on cl Ch 0mm uNm ofim mam mom 0mm Ckm uz: a2: uZ: uz: a.: a2: u3... uZD uZD ma: u7: uZD ma: uZD uz: a2: a2: a2: a2: uZD a2: m2: m2: a2: uZD Zak Zak Zak Zak zak Zak Zak H uzocomyoz. fl: Ha.q.4~ «_ .-0.q~.kuuzmua_ .quw mink 4440 mm pk cu .o.m?.x__ uh xHua~ quc .H.n_»m_xmuw a.m.mzumz .:.m4.~. u” fioDodzuuz z.du~ 0N we zu‘u4 Hus Unm: unfit .msmkonoqgu n¢ 0k 00 .044a. L_ ..fifio42aouuzn mawarnz ZOQZ