THE UNDERENUMERATION OF MICHIGAN POPULATION CENTERS BY THE CENSUS OF POPULATION AND ITS SIGNIFICANCE A Technical Report for flu Dogma of M. S. MICHIGAN STATE UNIVERSITY John E. MeIcher I977 LIBRARY Michigan State University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 6/01 clelRC/DateDue.p65.p.15 ABSTRACT THE UNDERENUMERATION OF MICHIGAN POPULATION CENTERS BY THE CENSUS 0F POPULATION AND ITS SIGNIFICANCE By John E. Melcher The variation in size, sc0pe, orientation and methods used in con- ducting community research are great and the establishment of a conceptual definition of community is Insic to most of this research. Community is often conceptualized in its most basic terms as a densely settIed center or a densely settled center plus a less densely settled hinterland. This type of conceptualization implies people and locality which is often operationally restricted to population centers. The identification and enumeration oftnpulation centers can be a means through which communities can be identified and enumerated. The enumeration of population centers, Of course, is only a first step be- cause in order for a population center to be classified as either a community center or community, additional empirical information would be required. Presently, the main source of data available for enumerating popu- lation centers is the U.S. Bureau of the Census. However, the census enumerates only those places with a population of 1,000 or more or those that are legally incorporated. Since many centers less than 1,000 in size are known to be unincorporated, it becomes obvious that the census does not enumerate many smaller centers. Research conducted throughout the country has found that the census enumerated a low percentage of the total number of population centers in given study areas. John E. Melcher This piece of research was undertaken in order to either sub- stantiate the validity or eluminate the deficiencies of the census as a data source for enumerating p0pulation centers in the State of Michigan. Numeric and percentage representations of census overenumeration and underenumeration in terms of selected place characteristics have been determined for eleven operationalizing indicators organized around eight hypothesis. This was done for the total universe of state places ex- ceeding 74 in size at sometime from 1930 to 1970 and for places exceeding 74 and less than 1,000 in size in 1970. It was determined for the universe of state places exceeding 74 in size at sometime from 1930 to 1970 that largen places, growing places, places with Ifigh proximity to larger places, and places located in urban areas are most likely to be enumerated by the census. It was revealed that the census under represents small places, declining places, places with low proximity to larger places and places located in rural areas and over represents larger places, growing places, places with high proximity to larger places and places located in urban areas. The findings related to the universe of places exceeding 74 and less than 1,000 in size in 1970 were somewhat different from those just mentioned. The census over represents larger places and under represents small declining places. The findings of this study indicate that biases do in fact exist in the census enumeration of population centers. Thus, the implications of the findings are significant for the basic researcher , the community developer , the national policy and program developer as well as any person who uses the census listings for population center enumeration. THE UNDERENUMERATION OF MICHIGAN POPULATION CENTERS BY THE CENSUS OF POPULATION AND ITS SIGNIFICANCE by John E. Melcher A TECHNICAL REPORT Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Resource Development 1977 o a, 5' 9V? [Vt-‘33 J I C. 2! ACKNOWLEDGMENTS It is with great appreciation that I extend deep felt thanks to all those who contributed to the success of this effort. I would like to express my gratitude to Dr. William Kimball, who so patiently served as academic advisor, committee member and friend, Dr. Richard Rodefeld, whose patient encouragement, advise and constructive criticisms made this effort a reality and who served as minor advisor, thesis advisor, committee member and friend, Dr. Milton Steinmuller, who lent timely support and encouragement and served as a committee member, to fellow students Dan Clay, Kevin 6055, Al Kirk and Mark Lancelle, who so often provided timely advice and encouragement over a glass of beer, to my parents and family for their loyal support, to Cindy Herfindahl, who not only spent many sleepless nights typing numerous drafts and correcting the misspelled words but who also lent encouragement and support, and to Blue who was always there. ii TABLE OF CONTENTS CHAPTER I INTRODUCTION ..... ' ................ Methodological Problems in Community Research ......... Population Centers as a Unit of Analysis ........... Evidence of Underenumeration - Literature Review ....... CHAPTER II STATEMENT OF THE PROBLEM AND RESEARCH OBJECTIVES . . . The Problem .......................... Significance of the Problem .................. Expected Relationships between Place Characteristic and Likelihood of Census Enumeration and Implications for Representativeness of Census Enumerated Places ..... Research Objectives ...................... CHAPTER III RESEARCH METHODS ................... Data Collection ........................ Enumeration Criteria ................... Procedures Used to Enumerated Michigan Population Centers. Data Analysis ......................... Phase I Analysis ..................... Phase II Analysis ..................... Phase III Analysis .................... Statistical Tests ..................... CHAPTER IV RESULTS AND ANALYSIS ................. Phase I Analysis ........................ Phase II Analysis ....................... Phase III Analysis ....................... CHAPTER V SUMVIARY AND CONCLUSIONS ................ Introduction .......................... Summary of the Problem ..................... Summary of Methods ....................... Summary of Findings ...................... Contributions of Findings ................... BIBLIOGRAPHY ............................ Table Table Table Table Table Table Table Table Table Table LIST OF TABLES The Number and Percentage of Places With Populations of 75 or More in 1970 by 1970 Size and Source of Enumeration ........................ The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By 1970 Population Size and Source of Enumeration . The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Percentage Change in Population Number From 1930 to 1970 and Source of Enumeration ................. The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Their Pattern of Population Change From 1930- 50 and 1950- 70 and Source of Enumeration ................... The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Size of Largest Place in the County and Source of Enumeration ........................ The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By 1960 Proximity to Larger Places and Source of Enumeration. . . . The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Metropolitan Status and Proximity of County to Counties with a Place of 50,000 or More and Source of Enumeration. . The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By 1965 SMSA Status of the County and Source of Enumeration ........................ The Number and Percentage of Michigan P0pulation Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Regional Location and Source of Enumeration ........ : The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Percentage of County Population Rural and Source of Enumeration ........................ iv Page 13 33 35 36 37 38 39 40 41 42 Table Table Table Table Table Table Table Table Table Table 11: 12: 13: 14: 15: 16: 17: 18: 19: 20: The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Percentage of Rural Non-Farm Residents Within the County and Source of Enumeration ................. The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By the Percentage of County Workforce Employed in Extractive Industries and Source of Enumeration ............ The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By 1970 Population Size for Census Enumerated and Total State Places .......... The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size By Percentage Change in in Population Number From 1930 to 1970 for Census Enumerated and Total State Places ........ The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Some- time from 1930 to 1970 By Their Pattern of Population Change From 1930-50 and 1950-70 for Census Enumerated and Total State Places ..................... The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Some- time from 1930 to 1970 By Size of Largest Place in the County for Census Enumerated and Total State Places . . . . The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By 1960 Proximity to Larger Places for Census Enumerated and Total State Places . . . . The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Metropolitan Status and Proximity of County to Counties with a Place of 50,000 or More for Census Enumerated and Total State Places. . . . The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By 1965 SMSA Status of the County for Census Enumerated and Total State Places . . . . The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Regional Location for Census Enumerated and Total State Places .......... Page 43 44 46 47 48 50 51 52 53 54 Table Table Table Table Table Table Table Table Table Table Table 21: 22: 23: 24: 25: 26: 27: 28: 29: 30: 31: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Percentage of County Population Rural for Census Enumerated and Total State Places ......................... The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Percentage of Rural Non-Farm Residents Within the County for Census Enumerated and Total State Places ...................... The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By the Percentage of County Workforce Employed in Extractive Industries for Census Enumerated and Total State Places ............... The Number and Percentage of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By 1970 Population Size and Source of Enumeration ........... The Number and Percentage of Michigan Places Exceeding 74 and Less Than 1000 in Size in 1970 By 1930-1970 Percentage Change in Population Size and Source of Enumeration The Number and Percentage of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size By Their Pattern of Population Change From 1930-1950 and 1950-1970 and Source of Enumeration ..................... The Number and Percentage of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By 1960 Proximity to Larger Places and Source of Enumeration ..... The Number and Percentage of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By 1965 SMSA Status of The County and Source of Enumeration ...... The Number and Percentage of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By Regional Location and Source of Enumeration .............. The Number and Percentage of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By Percen- tage of County Population Rural and Source of Enumeration. . . The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By 1970 Population Size for Census Enumerated and Total State Places .................... vi Page 55 56 57 58 59 6O 61 62 63 64 65 Table Table Table Table Table Table 32: 33: 34: 35: 36: 37: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By 1930-1970 Percentage Change in Population Size for Census Enumerated and Total State P aces ......... . ................ The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size and Less Than 1000 in Size in 1970 By Their Pattern of Population Change From 1930-1950 and 1950-1970 for Census Enumerated and Total State Places ......... The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size and Less Than 1000 in Size in 1970 By 1960 Proximity to Larger Places for Census Enumerated and Total State Places The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size and Less Than 1000 in Size in 1970 By 1965 SMSA Status of The County for Census Enumerated and Total State Places ....... The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By Regional Location for Census Enumerated and Total State Places ............. The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By Percentage of County Population Rural for Census Enumerated and Total State Places . . . . vii Page 66 67 68 69 70 71 CHAPTER I INTRODUCTION Methodological Problems in Community Research The minimal criteria for the existence of a science is a body of facts or truths systematically arranged to explain general laws. The logic for the establishment of these laws has come to be known as the scientific method. A dictionary definition states that the scientific method is "principles and procedures for systematic pursuit of knowledge involving the recognition and formation of a problem, the collection of data through ob- ‘ A servation and experiment, and the formation and testing of hypotheses." more technical examination of the logic of science indicates the following as being the major components of the scientific method: 1. The impartial observation and collection of data. 2. The use of induction to establish hypothesis or preliminary generalizations. 3. The testing and validation of generalization or hypothesis. 4. The use of deduction to make conclusions. 5. The testing of conclusions and establishment of theory. 6. The testing of theory and establishment of law. This paradigm establishes that the foundation of scientific law and theory is observable and quantifiable data. The strength and accuracy of generalizations and conclusions of scientific investigations, are directly related to the strength and accuracy of the data collected. What would the consequences be however, if the data was inaccurate or biased as a result of inadequate methods of collection or establishment of inappropriate limitations? 1 Webster's Seventh New Collegiate Dictionary, (1965), s.v. "Scientific Method". 2 The answer is obvious - inaccurate or biased data yields inaccurate theory. The development of both the natural sciences and the social sciences depends largely on the quality of data collected. Many scientific investi- gations are conducted focusing on a single phenomena, however, each investi- gation attempts to obtain data that can explain the phenomena just a bit more accurately than previous inveatigations. Methods, limitations and other components of scientific investigation continually evolve to higher levels of sophistication and specificity in order to better explain given phenomena. It has been suggested by some scholars that findings are related to methods. That is to say, in the investigation of certain social phenomena the methods employed will give unintentional bias to findings. John Walton conducted an exhaustive analysis of thirty-three studies dealing with community power structure and demonstrated that the type of method employed was highly related to the type of power structure observed.2 He stated that an analysis of variables "... reveals a developmental sequence in which discipline tends to determine the research method adopted, which in turn is related to the type of power structure identified." 3 Walton suggests that the development of comparative research methods is needed to temper or eliminate bias and aid in the development of more accurate generalizations. John Walton is not alone in suggesting that a major problem in community research is the bias of findings by research methods. Albert Reiss believes 2John Walton, "Substance and Artifact: The Current Status of Research on Community Power Structure", American Journal of Sociology, 71 (January, 1966): 430-438. 3John Walton, "Discipline, Method and Community Power: A Note On the Soiiology of Knowledge," American Sociological Review 31 (October, 1966): 68 . 3 that the methodological problems in community research are barely separable from the theoretical ones and that a major weakness is the lack of compara- tive methods. He gives three reasons for researchers failure to utilize comparative methods: 1. Community studies usually are individual case studies which lack data on factors relevant to specific theoretical comparisons. 2. Community investigators fail to examine more than the most elementary community variables in a large number of communities. 3. The validity of much community data for intra as well as intercultural comparisons is questionable. Reiss suggests that in an effort to deal with these problems first, theory is developed about community phenomena so relevant theoretical community variables can be specified and second, that the number of communities for which community parameters are available should be increased.4 Richard L. Simpson professes that a major weakness in community research is its non-cumulative nature. He suggests, as does Reiss, that this is due to the large number of case studies conducted in the past. This non-cumula- tive nature yields low degrees of comparability thus minimizing usefulness for generalizing about community phenomena.5 In case studies the community is generally viewed as a laboratory for studying some problematic aspect of the community. The methods used by researchers are extremely varied and individualized. Maurice Stein points out that: "The range of variation found among community studies in the literature is great, making the task of reconciling them 4Albert J. Reiss, Jr., "Some Logical and Methodological Problems in Community Research", Social Forces, 33 (Oct., 1954): 51-57. 5Richard L. Simpson, "Sociology of the Community: Current Status and Prospects", Rural Sociology 30 (June, 1965): 127-149. 4 difficult. Aside from obvious differences between the communities in terms of such important dimensions as age, size, economy, regional location, etc., additional problems are introduced by the fact that the field work was con- ducted at different times and that the talents, training, facilities, and interests of the field workers differed widely from study to study. Each research report is a synthesis by the author of several orders of data about a particular community, arranged according to his sense of significant social structures and processes. This synthesis rarely takes into account the relation of the material to other related studies nor do most research reports contain chapters which satisfactorily present generalized conclu- sions."6 In contrast to this type of study, the comparative analysis approach encourages standardization and replication of useful procedures most notably in the area of demographic composition, economic structure, political organization, power structure and decision making leadership values, and community public opinion.7 The need for improved methods in community research seems clear, however, there is still one area that has not yet been discussed, that is, research and action programs in community development. Community development is not a separate area of theoretical knowledge, but an area in which different areas of knowledge intersect. (Some of the more traditional sociological sub- ject matter, such as power structure, stratification and networks of inter- personal influence, seem especially relevant to community development. Com- munity development theory, however, suffers from the same methodogical pro- blems that the more basic community research does. That is to say, there is 6Maurice R. Stein, The Eclipse of Community: An Interpretation of American Studies, (New York: Harper and Row, PubTishers, Inc., 1964). 7R.J. Havighurst and A.J. Jansen, "Community Research", Current Sociology XV (1967): 21 and 31. ‘ 5 a need to strengthen comparative techniques so analysis of several action programs can yield higher levels of generality. Thus the basic and applied realms of community investigation can work together to further the under- standing of comnunity processes. "Like other practical concerns, comnunity develOpment programs need to derive ideas from basic research; but in addition, they can contribute to basic knowledge if they are conducted and reported, as the best of them are, with this objective."6 The inter- connection of the basic and applied areas of community research is vital to the evolving process of scientific growth in the area of community. Again, the importance of method, perhaps more specifically comparative method, is not only significant for basic community research but also for the more applied realm of community development. The need for accurate community data is also important in terms of establishing sound community theory. The methods employed in the problem identification, data collection, and data analysis all have effects upon results which in turn have impact on theory. A weakness in any of these areas will lead to a weakness in theory, just as a bias introduced at the data collection level will produce biased results and biased theory. Thus, the use of sound methods with high levels of comparability, complimented by accurate unbiased data will yield stronger theory with higher degrees of generality. This is true of both basic and applied community research. Population Centers as a Unit of Analysis The variations in size, scope, orientation and methods used in conducting community research are great and the establishment of a conceptual definition of community is basic to most of this research. Community is often conceptualized 8 Simpson, p. 149. 6 in its most basic terms as a densely settled center or a densely settled center plus a less densely settled hinterland.9 This type of conceptuali- zation implies people and locality which is often operationally restricted to "population centers". The identification and enumeration of population centers can be a means through which communities can be identified and enumerated. The enumeration of population centers, of course, is only a first step, be- cause in order for a population center to be classified as either a com- munity center or community, additional empirical information would be re- quired. P0pulation center, defined as "a clustering of occupied residences in space with high levels of proximity (short distances) between the residences,"11 is a useful unit of analysis for three reasons. First, population centers represent one of the two dichotomous settlement models often used to illustrate patterns of change. The models are generally defined in terms of proximity of dwelling units, with "dispersed" identified with low proximity and "clustered" with high proximity. The major portion of the U. S. popu- lation both urban and rural, live in centers or clustered dwellings. Second, population centers are the locus of economic and social institu- tions. The importance of rural centers in this respect has grown as open country schools, churches, businesses and other institutions have declined. 9 Richard D. Rodefeld, Enumerating Michigan Population Centers (places) and Determining their Size from 1930-1970; Rationale, Procedures, Problems and Results, Department of Sociology, Michigan State University, (East Lansing: By the author, 1976), p. 4. 10 Ibid., p. 3. 11 Ibid. 7 Third, population centers have been the unit of analysis in a number of studies of places that use the census as the prime data source. This is particularly true of research focusing on the growth, stability and decline of places. It also has proven to be a viable unit for research that con- ceptually restricts community to population centers. This is true of re- search concerned with community power structure, stratification, and central places.12 For the purpose of this paper, population centers have been conceptually defined as clusters of people in space and operationally restricted to those with 75 or more people. Evidence of Underenumeration - Literature Review Presently the main source of data available for enumerating population centers is the U. S. Bureau of the Census. However, the census enumerates only those places with a population of 1,000 or more or those that are legally incorporated. Since many centers less than 1,000 in size are known to be un- incorporated, it becomes obvious that the census does not enumerated many smaller centers. Leon B. Perkinson, an agricultural economist for the U.S. Deparment of Agriculture, conducted a research project concerning "Population Distri- bution in Non-Metropolitan Michigan” while stationed in the Department of Resource Development at Michigan State University. He wrote, "The number and population of unincorporated places under 1,000 population is usually con- sidered to be an unknown. Governmental services for such places are usually provided by the township or county government and since they lack legal boundaries, the responsibility of collecting information about unincorporated 12 Ibid., p. 4. 8 places is often lacking. Even the U. S. Bureau of the Census reports the population of incorporated places or unincorporated places of 1,000 people or more only. Therefore, the smaller unincorporated places could be con- sidered as 'non-places' since no information is available about such places from usual sources."13 This was found to be a problem for many researchers interested in studying the smaller places that make up an important portion of the settlement fabric of the United States. Dr. C. Luther Fry's book, American Villagers was among the first major published works concerned with 14 villages in the United States. At the time of publication in 1926, Dr. Fry was Director of the Bureau of Standards of the Institute of Social and Religious Research and Associate Director of the American Village Study. The book was intended as an attempt to answer some fundamental questions about U.S. villages and their residents. More specifically these questions dealt with: How many village residents are there? Are village populations declining? What kind of people live in villages? What do villagers do for a living? What are the distinguishing peculiarities of village populations? What functions do villagers perform? For the purpose of the study, villages were conceptually defined as population centers and operationally restricted to population centers having between 250 and 2500 inhabitants. The first topic considered in the re- search was the determination of the number of villages and villagers in the 13 Leon B. Perkinson, Population Distribution in Non-Metrgpolitan Michigan Report 145 Development and Public AffairsTTEast Lansing, Michigan State University Agriculture Experiment Station), (September, 1971), p. 7. 14 C. Luther Fry, American Villagers, (New York: George H. Doran Co., 1926). 9 United States. A major problem in securing this data resulted from the fact that "... there are no official population figures for a large proportion of the American Villages, because the Federal Census does not attempt to d."15 Nine- secure separate enumerations for places that are unincorporate teen-twenty census data indicated that of the 12,858 incorporated rural places, 2,619 were hamlets (less than 250 in size) whose total p0pulation was 461,890. Thus, the number of incorporated villages was 10,239 with a population of 8,509,659. "Large as are the census figures, they do not really give the country's total village population, since they include only the inhabitants of in- corporated places. Unincorporated centers are entirely omitted, yet it is a matter of common knowledge that the number of unincorporated villages is large. It is recognized that in certain states there are more of them than of places that are incorporated. Obviously, therefore, unincorporated places should be taken into account in an estimate of the nation's total village population."16 As a result of the incomplete enumeration by the census, Fry turned to an unofficial source to determine the number and population of American villages. The source he used was the Rand McNally Atlas. Using the 1921 edition of the atlas, Fry compiled a list of 18,381 villages with a total population of 12,858,521. This indicated that nearly one out of every eight Americans lived in a villages in 1920. All the incorporated villages were taken directly from the 1920 Census by Rand McNally, meaning that the enumeration and population estimates for unincorporated places came from unofficial sources. The atlas enumerated 8,142 unincorporated villages with an aggregate population of 4,348,862. 15 Ibid., p. 26. 16 Ibid., p. 27. 10 The general reliability of the atlas count was partially confirmed by verifying the commonly held attitude that there were a large number of unincorporated places. The figures indicated that approximately one third of the country's total village population was located in unincorporated places. The atlas figures also suggested that the larger the centers the more likely they were to be incorporated. The average size of all the atlas villages was 700 as compared to the 831 enumerated by the census. This meant that the average population of the unincorporated centers was 534 or 36% less than the average population of the incorporated villages. Fry found that the number and distribution of incorporated villages (census enumerated) was very different from the number and distribution of the universe of all villages. He found that the incorporated places only made up 56% of the total universe. The percentage enumerated by the census varied considerably from region to region with 89% enumerated in the West North.Central region, 67% in the East North Central and 7% in the North East. These discrepancies indicated that the census was an inadequate data source for the purpose of studying villages in the United States. He wrote: "Even admitting that the atlas furnishes only a rough estimate of the number of all villages, and that therefore the proportion of these that are incorpor- ated is only approximately correct, nevertheless the discrepancies between the atlas figures and those of the census are so great as to rule out census data as a basis for estimating the distribution of villages from region to region."17 It is fair to assume then, that the same problem exists not only from region to region, but from state to state and county to county. Another researcher interested in the settlement pattern of smaller places 17 Ibid., p. 32. ll . 18 was Paul H. Landis. In 1939, while an associate professor of Sociology at Washington State Uniyersity, Landis conducted a research.project intended to: 1) estimate the total number of places with fewer than 2500 people in the nation, unincorporated and incorporated, and 2) estimate the total population in these places, classifying all places with fewer than 2500' people into three units: hamlets, places with less than 250 population; small villages, places with 250-999 population: and large villages, places with 1000-2499 population. The problem of census underenumeration of places was a major problem for Landis just as it was for Fry. Landis, however, employed a different means to collect the needed data. Instead of using Rand McNally as the source for unincorporated place data, Landis turned to Bradstreet's Book of Commercial Ratings which lists all places in the United States having mercantile enterprises. This source listed both incorporated and unincor- porated place population estimates. Landis recognized that any population estimate of this nature was subject to error, so he developed a corrective procedure that was intended to make the data as valid as possible. Landis found that in 1930 there were, according to estimates, almost seventeen million (16,826,935) people in the United States who lived in places of under 2500 pe0ple, or, in other words, about 14% of the total U. S. population was living in these places. Of these people almost half (7,642,177 or 45.4%) were living in unincorporated places not identified by the census. This group may have been even greater, said Landis, since data on which findings were based were from commercial rating volumes which only listed those places with mercantile enterprises. The remaining places were 18 Paul H. Landis, "The Number of Unincorporated Places in the United States and their Estimated Populations," Research Studies of the State College of Washington 6 (December, 1938): 160-188. 12 incorporated and were therefore reported by the census. Similar research conducted in other states also found that the census enumerated a low percentage of the total number of population centers or places in given study areas. Vincent Heath Whitney reported that in a ten county subregion in north-central North Carolina, 75% of the villages and hamlets were not enumerated by the census.19 Similarly, E. Willard Miller reported that in a six county subregion in western Pennsylvania 95% of the villages were not reported by the census.20 D. G. Marshall published an article in which he indicated that for the years 1920, 1930 and 1940, the census enumerated about 5% of all U. S. hamlets (1-250 in size), 54% of all the villages (251-2,499 in size) and 15-18% of all places smaller than 2,500 in size.21 This suggests that a large number of places in the United States are not enumerated by the census and that if in fact this is the case, it is possible the characteristics of census enumerated places will not provide an accurate description for the universe of all places and centers. It also suggests that research resulting from the study of census enumerated places may not have a high level of generality for the universe of all places or for the smaller places within that universe. This will certainly be a problem in any state, region, or county where there are large numbers of unincorporated places less than 1,000 in size and suggests that a bias toward larger places may exist in the census. 19 Vincent Heath Whitney, "Notes on the Reliability of Atlases for Estimating the Populations of Unincorporated Places," Rural Sociology_10 (December, 1945): 387-393. 20 E. Willard Miller, "P0pulation and Functional Changes of Villages in Western Pennsylvania," Pennsylvania Historical Magazine 43 (March, 1960): 59. 21 D. G. Marshall, "Hamlets and Villages in the United States: Their Place in the American Way of Life," American Sociological Review 11 (April, 1946): 159-165. 13 In 1974 a research project was initiated which was designed to determine the extent, causes, and consequences of change in the size and economic base of rural population centers and communities in Michigan.22 In an effort to avoid the problems associated with census underenumeration of population centers a set of procedures were developed to enumerate the universe of Michigan population centers exceeding 19 in size at any time in the 1930-70 time period. Populations for these centers were determined and recorded by decade from 1930 to 1970. It is appr0priate to isolate several interesting figures demonstrating the problem of census underenumeration of population centers in the State of Michigan as reported in some of the initial findings of the project. All places with 1970 populations of 75 or more were classified on the basis of their size and source of enumeration. Table 1: The Number and Percentage of Places With Populations of 75 or More in 1970 by 1970 Size and Source of Enumeration. Source of Enumeration ngsggf Non-Census* Igtgl_ 1970 Size _#_ ‘7: i g i it 75 - 999 201 21.7 724 78.3 925 100.0 1000 + 397 87.4 57 12.6 454 100.0 Total 598 43.4 781 56.6 1379 100.0 *Census and Non-Census are two mutually exclusive categories. The listings under the census heading are those places exceeding 74 in size and reported by the Census. The listings under the Non-Census heading are those places exceeding 74 in size and not reported by the Census. For a detailed discussion of data collection procedures see pages 22-25. 22 Rodefeld, 1976. 14 The initial results indicate that: - The census enumerated only 43% of all places in the State of Michigan, meaning that almost 57% or 2/3 of the total places were not enumerated by the census. - The census enumerated 87% of all places with populations of 1000 or more. - The census enumerated only 22% of all places less than 1000 in population, i.e. it failed to enumerate 78% of all places with populations less than 1000 in 1970. The findings of previous research clearly indicate that a census under- enumeration of the total number of population centers does exist in the United States. They also suggest that the likelihood of enumeration by the census is strongly related to the size of the place, that is to say, that census data may contain significant degrees of bias toward larger places and against smaller places. The ramifications of these findings in terms of research projects focusing on population centers and particularly small places (less than 1000 in size) seem clear; the census may not be an adequate source of enumeration for these purposes. Further this suggests that prior empirical research which has relied on census data may have resulted in the development of biased theory. CHAPTER II STATEMENT OF THE PROBLEM AND RESEARCH OBJECTIVES The Problem The previous discussion has demonstrated the low percentage of smaller places enumerated by the census. The amount of work and effort that has been put into prior research in this area indicates a strong interest in the academic community to better understand the settlement fabric of the United States and further reveals the desire of these researchers to fill the void created by the limitations placed on the collection and publication of census data. The focus of most of this previous research has concentrated, naturally enough on smaller places (those less than 1,000 population) with their prime intent to determine the number and population of centers not enumerated by the census. While the census deficiencies in this area are well documented, little or no work has been done to determine the nature or magnitude of differences between census enumerated places and those places not enumerated by the census. Furthermore, little is known of how representative census enumerated places are for the entire universe of places. A detailed analysis of places enumerated and not enumerated by the census has yet to be done by any researcher. This has created an information void related to the ex- tent and implications of relying on census enumerated places and their characteristics to generalize about the total universe. The next logical step in understanding the underenumeration of places is to identify and assess the characteristics of census and non-census enu- merated places. Past research has found size of place to be highly related 16 to its likelihood of enumeration by the census. As a consequence the distri- bution of census places by size has probably not been very representative of the size distribution of the universe of places, i.e. the relative signi- ficance of larger places has been overestimated and that of smaller places underestimated. If this is true for size of place it may also be true for other characteristics of places as well. For instance, census enumerated places may not provide an accurate description of the number and relative significance of growing and declining places, urban and rural places and places with high and low proximity to larger places. These are the major objectives of this research. The rationale for these objectives are well grounded in both the interests of the basic researcher and the applied practitioner concerned with population centers and the closely related community center. Significance of the Problen Greater understanding of both the basic settlement fabric of this country and the limitations of data sources will provide a more stable foundation to base both future research decisions as well as public policy decision. The re- searcher relying on the census for population center data may be introducing a bias that could have a significant effect on his project. Likewise, any in- dividual using census data as an aid in the design of a deveIOpment program may run the risk of introducing misleading data that could have adverse effects upon his efforts. It is because of these potential problems that the need arises to understand the possible deficiencies, limitations and biases asso- ciated with census enumerated places. An investigation into the nature and magnitude of differences between census enumerated places and those places not enumerated by the census, will. 17 make specific contributions to any research_effort concerned with studying the total universe of places. It will make contributions to basic know- ledge by detennining empirically how representative census enumerated places are of the total universe of places. Implications of these findings could be farreaching considering the widespread use of census data. In more applied terms, such an investigation would reveal actual numbers and percentages of places with given characteristics. For example, the actual number and percentage of places that are experiencing either growth or de- cline will be determined. It is clear that the problem of census underenumeration has been a concern of scholars for sometime, which suggests that any contribution toward a better understanding of this problem will aid the basic researchers as well as those more applied oriented scholars who develop action programs directed toward the many aspects of community development. Stated even more broadly, any individual who uses the census enumeration of places for academic purposes, develOpmental purposes, or public policy decisions, will benefit to know what deficiencies or biases may exist in the data set. Expected Relationships Between Place Characteristic and Likelihood of Census Enumeration and Implications for Representativeness of Census Enumerated Places Several basic relationships are expected to exist between place charac- teristics and the likelihood of census enumeration. A relationship should exist between size of place and likelihood of census enumeration, or more specifically the census should enumerate a greater percentage of larger places than smaller places. The reason for designating this relationship is that past research has found that the larger the place the more likely it will be enumerated by the census or conversely the smaller the place the 18 less likely it will be enumerated by the census. This suggests that in comparison to the total universe of places, census enumerated places will over represent larger places and under represent smaller places. Second, a relationship should exist between the change in size of a place and the likelihood of enumeration by the census. That is, the census should enumerate a lower percentage of declining places than growing places. The reasons for this relationship are as follows. Past research has found that the smaller the size of the place the greater the likelihood it will de- cline. Since past research has also found that the census underenumerates smaller places it stands to reason that declining places will also be under- enumerated by the census. This further suggests that in comparison to the total universe of places, census enumerated places will over represent growing places and under represent declining places. Third, a relationship should exist between the proximity of a place to a larger place and likelihood of census enumeration. More specifically it is expected that places with high proximity to larger places will have higher rates of enumeration by the census than places with low proximity to larger places. Two reasons for the expected relationship are: l) as a function of population decentralization and suburbanization the places with high proximity to larger place will be disproportionately large and more likely to be growing, hence more likely to be enumerated by the census, and 2) places with low proximity to larger places are expected to be disproportionately small and morelikely to be declining, hence less likely to be enumerated by the census. This suggests that in comparison to the total universe of places, census enumerated places will over represent places closer to larger places and under represent places further from larger places. Fourth, a relationship is expected to exist between the urbanity of 19 an area within which places are located and likelihood of census enumeration. Specifically, it is expected that the more rural the area within which a place is located, the less likely its enumeration by the census. Two reasons for the expected relationship are: l) as a function of population decentrali- zation and suburbanization the places located in rural areas are expected to be disproportionately small and more likely to be declining, hence less likely to be enumerated by the census as compared to places located in more urban areas, and 2) with the change in agriculture and extractive industry technology (namely mechanization) came reductions in farm number and in- creases in farm size resulting in a loss of farm population and corresponding decline in rural place size, hence these places are expected to be dispro- portionately small and less likely to be enumerated by the census. This then suggests that in comparison to the total universe of places, census enumerated places will over represent places located in urban areas and under represent places located in rural areas. Previous research indicates that a keen interest exists among researchers in the study of smaller and/or rural places. Since the bulk of census under- enumeration exists for the smaller places, specifically those less than 1,000 in size it is expected that the previously mentioned relationships should exist not only for the total universe of places, but should also exist for the universe of smaller places (less than 1,000 in size). Research Objectives The objectives of this study are as follows: Objective 1. To determine the relationships between various place charac- teristics and likelihood of census enumeration (non-enumeration) for places 20 exceeding 74 in size at sometime between 1930 and 1970. More specifically, the objective is to test the following hypotheses: HYPOTHESIS I: HYPOTHESIS II: HYPOTHESIS III: HYPOTHESIS IV: A positive relationship exists between the 1970 size of place and their likelihood of census enumeration, i.e. larger places are more likely to be enumerated by the census than smaller places. A positive relationship exists between change in size of place from 1930 to 1970 and their likelihood of census enumeration, i.e. growing places are more likely to be enumerated by the census than declining places. A positive relationship exists between proximity of place to larger places and their likelihood of census enumeration, i.e. those places closer to larger places are more likely to be enumerated by the census than places that are further away. A positive relationship exists between the ur- banity of an area within which places are located and their likelihood of census enumeration, i.e. places located in rural areas are less likely to be enumerated by the census than places located in urban areas. Objective 2. To determine the consequences of any such relationships for reported characteristics of state places exceeding 74 in size at some- time from 1930 to 1970. Assuming Hypotheses I-IV are supported, the following hypotheses should also be supported: HYPOTHESIS V: Compared to the total universe of state places, census enumerated places will over represent larger places and under represent smaller places. 21 HYPOTHESIS VI: Compared to the total universe of state places, census enumerated places will over represent growing places and under represent de- clining places. HYPOTHESIS VII: Compared to the total universe of state places, census enumerated places will over repre- sent those places closer to larger places and . under represent places that are further away. HYPOTHESIS VIII: Compared to the total universe of state places, census enumerated places will over represent places located in urban areas and under represent places located in rural areas. Objective 3. To test Hypotheses I-VIII for census enumerated places and the universe of places with sizes of 75 to 1,000 in 1970. (The Hypotheses will be designated as follows: Hypothesis Ia, Hypothesis IIa, etc.). Objective 4. To discuss basic and applied contributions and implications of the findings. CHAPTER III RESEARCH METHODS It is an intention of this study to be able to critically evaluate the strength of census data for generalizing about the total universe of population centers in Michigan. Given the limitations of census definitions for inclusion, it is apparent that there is a lack of knowledge concerning those places less than 1,000 in size and not legally incorporated. Thus the first step in this investigation is to identify Michigan population centers and determine their sizes. This data collection phase will be discussed in terms of criteria for enumeration and procedures used to enumerate Michigan population centers. The next step is then to establish methodological procedures for arranging the data for analysis according to stated hypothesis. The data analysis is done in three phases. Phase I is designed to determine if a relationship exists between selected place characteristics and likelihood of census enumeration. If relationships are found to exist, then the extent of either over enumeration or under enumeration of places by the census in terms of the selected place characteristics will be determined in Phase II. Phase III of the analysis follows the same procedures that are employed in Phase I and Phase II of the analysis, however, restricting the universe to those places exceeding 74 and less than 1000 in size in 1970. Data Collection The major sources for enumeration and determination of population size for this study are the census of population (for legally incorporated places and/or those exceeding 1,000 in size), Rand McNally Commercial Atlas and Marketing Guide (for legally Unincorporated places less than 1,000 in size), and questionnaire responses from knowledgeable localities. The specific techniques and procedures employed here were developed by Dr. Richard D. 23 Rodefeld while serving as an Assistant Professor of Sociology at Michigan State University. A detailed description of rationale, procedures, and problems are available in an unpublished paper entitled, "Enumerating Michigan Population Centers (Places) and Determining Their Size from 23 1930-1970: Rationale, Procedures, Problems and Results". Enumeration Criteria As was mentioned earlier, a population center is defined as "a clustering of occupied residences in space with high levels of proximity (short distance) between the residences." This definition identifies two settlement patterns defined in terms of proximity, i.e., dispersed and clustered. Population centers are a dominant location of economic and social institutions, and have been the unit of analysis in many studies of places drawn from the census. Population centers also provide a means through which community population centers or communities can be enumerated. A total enumeration of Michigan population centers exceeding 74 in size at some time from 1930-1970 was made. A total enumeration allows a compara- tive approach to be undertaken in the research, as a result of excluding any biases associated with census enumerations. The generality of results can be stated with high levels of specificity since the parameters for the total universe will be known. It is intended that the applied contributions of a total enumeration will increase the comprehensiveness and specificity of knowledge on the extent, magnitude and location of changes in population centers . 23 Ibid. 24 Ibid., p. 3. 24 Procedures Used to Enumerate Michigan Popglation Centers The Rand McNally Commercial Atlas and Marketing Guide and Census of Population were used to enumerate all Michigan population centers exceeding 74 residents at any time in the 1930-70 time period and determine the populations of these centers from 1930 to 1970 by decade. All centers enumerated by the census were extracted for 1930 and 1970 along with their estimated populations. This list was compared to the Rand McNally listing of places for 1930 and all centers exceeding 74 in size that were not listed by the census were then extracted along with their popula- tions. All extracted centers (census and Rand McNally) were then compared with the 1970 Rand McNally listing and all centers exceeding 74 in size that had not previously been identified were then extracted. This last proce- dure identified those centers that were established after 1930 or those that exceeded 74 in size after 1930. The 1940, 1950 and 1960 populations for these places were then extracted. At the end of the enumeration process 1,613 centers were identified having a maximum size exceeding 74 at some- time between 1930 and 1970. Missing population data was a major problem with approximately one- quarter of the centers lacking population estimates for one or more years. A variety of procedures were established to reduce the amount of missing data, however the major procedure employed was to send questionnaires to individuals familiar with the missing data cases. Five different question- naires were developed for this purpose. A total of 725 questionnaires were sent to local postmasters, mail carriers, township supervisors, and other local contacts. As a result of the questionnaire responses and additional recoding procedures, correcting for name change, dual listings, parts of other places, size reclassifications and other factors contributing to missing data, a final total of 1,592 Michigan population centers exceeding 25 74 in size at some time between 1930 and 1970 was established with only 86 places having missing data for one or more of the years assessed. It is the data set just described that provides much of the data for this research effort. Though the reliability of the data has not been for- mally tested, it is believed to be the best available for such objectives as outlined here. It may be misleading to report that this is the best data available because in reality it is the only such data set in existence. The efforts of Dr. Rodefeld however are presently being replicated for the State of Ohio through an effort centered at Ohio State University and a similar project is also expected to be initiated for the State of Pennsylvania. Data Analysis The data analysis is carried out in three phases. The first phase is designed to determine the relationship between selected place characteristics and likelihood of census enumeration for all Michigan places exceeding 74 in size at some time between 1930 and 1970. The second phase of the analysis is designed to determine the level of representativeness of census enumerated places for the total universe of state places exceeding 74 in size at some time between 1930 and 1970. The third phase of the analysis is similar to the previous two analyses with the addition of placing controls on the size of population centers to be analyzed. Population centers will be limited to those places exceeding 74 and less than 1,000 in size in 1970. Phase I Analysis It has been established through previous research that limitations placed on census definitions have resulted in census underenumeration of population centers. Places that are less than 1,000 in size and not legally incorporated are not enumerated by the census. The obvious and most basic 26 question to be addressed is whether relationships exist between various characteristics of Michigan places exceeding 74 in size at some time from 1930 to 1970 and their likelihood of enumeration by the census. This is accomplished by operationalizing and testing four hypotheses. The hypotheses to be tested and operational indicators are as follows: HYPOTHESIS I: Indicator 1: HYPOTHESIS II: Indicator 1: Indicator 2: A positive relationship exists between the 1970 size of places and their likelihood of census enumeration. 1970 size of place - as reported by the Census of Population, Rand McNally Commercial Atlas and Marketing Guide and responses from local infor- mants and classification into one of the following categories: 1) 75-149, 2) 150-299, 3) 300-499, 4) 500-749. 5) 750-999, 6) 1,000-2,499, 7) 2,500-9,999, 8) 10,000-49,999, and 9) 50,000 + A positive relationship exists between change in size of places from 1930 to 1970 and their likelihood of census enumeration. Percentage change in population numbers from 1930 to 1970 - calculated by dividing the difference between the 1930 and 1970 population by the 1930 population for each center and classification into one of the following categories: 1) -100% to -51%, 2) -50% to -1l%, 3) -10% to 10%, 4) 11% - 49%, 5) 50% - 99%, 6) 100% - 249%, and 7) 250% + Pattern of p0pulation change from 1930 to 1970 - calculated by determining change in size of places from 1930 to 1950 and 1950 to 1970 and classification into one of the following mutually exclusive categories: 1) consistent decline (those places that declined from 1930 to 1950 and 1950 to 1970), 2) recent decline (those places that grew or remained stable from 1930 to 1950 and declined HYPOTHESIS III: Indicator 1: Indicator 2: Indicator 3: 27 from 1950 to 1970), 3) recent growth (those places that declined or remained stable from 1930 to 1950 and grew from 1950 to 1970), 4) consistent growth (those places that grew from 1930 to 1950 and 1950 to 1970). A positive relationship exists between proximity of places to larger places and their likelihood of census enumeration. Size of largest place in county - as reported by the Census of Population and classification into one of the fbllowing categories: 1) 0-999, 2) 1,000-2,499, 3) 2,500-4,999, 4) 5,000-24,999, 5) 25,000-49,000, and 6) 50,000 +. Proximity to larger places in 1960 - as reported by Dr. Perkinson in the September, 1971 Michigan State University Agricultural Experiment Station Research Report #145, entitled "Population Distribution in Non-Metropolitan Michigan, 1940-1970,“ with classification into one of the following categories: 1) further than 50 miles from a city greater than 25,000, 2) within 50nmiles of a city greater than 25,000, and 3) located in an SMSA county. Metropolitan status and proximity of county to counties with a place of 50,000 or more in 1970 - calculated from size of largest places in county and adjacent counties as reported by the Census of Population and coded in the Michigan Population Centers Study conducted by Dr. Rodefeld, Michigan State University and classification into the following categories: 1) county has no place greater than 50,000 and is not adjacent to county with place greater than 50,000, 2) county has no place greater than 50,000 and is adjacent to county with place greater than 50,000, and 3) county has no place greater than Indicator 4: HYPOTHESIS IV: Indicator 1: Indicator 2: Indicator 3: Indicator 4: 28 50,000. SMSA status in 1965 - as reported by Dr. Perkinson in Michigan State University Agricultural Experiment Station Research Report #145 and classification into one of the following categories: 1) SMSA county, and 2) non-SMSA county. A positive relationship exists between the urbanity of an area within which places are located and their likelihood of census enumeration. Regional location of place - as coded in the Michigan Population Centers Study conducted by Dr. Rodefeld, Michigan State University and classified into one of the following categories: 1) Upper Peninsula (UP) - West, 2) UP-East, 3) Lower Peninsula (LP) - NW, 4) LP - NE, 5) LP - SW, and 6) LP - SE. Percentage of county population that is rural in 1970 calculated by dividing the rural population number by the total population number as reported by the Census of Population and classification into one of the following categories: 1) O%-24%, 2) 25%-49%, 3) 50%-74%, and 4) 75%-100%. Percentage of rural non-farm residents within the county in 1970 - calculated by dividing the rural non- farm population number by the total population number as reported by the Census of Population and classification into one of the following categories: 1) 0%-24%, 2) 25%-49%, 3) 50%-74%, and 4) 75%-100%. Percentage of county workforce employed in extractive industries in 1970 - calculated by dividing the number of people employed in extractive industries by the total number of people in the workforce as reported by the Census of Population (Industry of Employed Persons table) and classified into one of the following categories: 1) O%-2%, 2) 3%-5%, 3)6%-10%, and 4) 11% + 29 Phase II Analysis If Hypotheses I-IV are supported then the characteristics of census enumerated places will not be representative of the universe of state places. Thus, the objective in this phase of the analysis is to determine the level of representativeness possessed by census enumerated places exceeding 74 in size at sometime between 1930 to 1970. This is accomplished by operationali- zing and testing hypotheses V-VIII. The hypotheses to be tested and operationalizing indicators are as follows: HYPOTHESIS V: Compared to the total universe of state places, census enumerated places will over represent larger places and under represent smaller places. (This hypothesis will be operationalized by the same indicator as identified for Hypothesis 1). HYPOTHESIS VI: Compared to the total universe of state places, census enumerated places will over represent growing places and under represent declining places. (This hypothesis will be operationalized by the same indicators as identified for Hypothesis II). HYPOTHESIS VII: Compared to the total universe of state places, census enumerated places will over represent those places closer to larger places and under represent places that are further away. (This hypothesis will be operationalized by the same indicators as identified for Hypothesis III). HYPOTHESIS VIII: Compared to the total universe of state places, census enumerated places will over represent places located in urban areas and under represent places located in rural areas. (This hypothesis will be operationalized by the same indicators as identified fbr Hypothesis IV). Phase III Analysis The final phase of the data analysis is restricted to those population 30 centers in the state exceeding 74 and less than 1,000 in size in 1970. Hypotheses I-VIII will be tested in the same manner as is done in Phase I and 11 except for the universe of smaller places. This set of hypotheses will be designated as follows; Hypothesis Ia, Hypothesis IIa, etc. Statistical Tests Statistical tests were performed at the time of creation of each table. Major reliance was placed on numeric and percentage distributions and comparisons. For Hypotheses I-IV the percentage of total places in each independent variable category (indicator category) enumerated by the census and not enumerated by the census were computed. If no re- lationship exists between the independent variable and likelihood of enumeration by the census then the percentage of places in each independent variable category enumerated by the census should be the same as the per- centage of total state places enumerated by the census. Any deviation from this will indicate a relationship between the independent variable and like- lihood of enumeration by the census. The magnitude and pattern of these deviations will indicate the strength and direction of the relationship. For Hypotheses V-VIII, the percentage of the total state places enumerated by the census for each independent variable category and the percentage of the total state places for each independent variable category were computed. If census enumerated places were representative of the total universe of state places then no deviation should exist between the two percentages for each of the independent variable categories. Any deviation from this will indicate a relationship between the independent variable and representativeness of census enumerated places. The magnitude and pattern of these deviations will indicate the strength and direction of the relationship. Gamma statistical tests were computed and reported for each crosstabulation 31 indicating the direction and strength of the relationships. CHAPTER 1v RESULTS AND ANALYSIS In order to satisfy the stated objectives of this research project, eight hypotheses were developed and tested. Each hypothesis was opera- tionalized by one or more indicators. Crosstabulations were made between each indicator and source of enumeration with the results reported in tables and written narrative. Analysis was conducted in three phases and relied heavily upon nu- meric and percentage distributions and comparisons. Phase I was directed toward the testing of Hypotheses I-IV to determine if there exists a re- lationship between the operationalizing indicators and likelihood of enu- meration by the census for places exceeding 74 in size at sometime between 1930 and 1970. Phase II was directed toward testing Hypotheses V-VIII to determine if there exists a relationship between the operationalizing indi- cators and representativeness of census enumerated places which exceeded 74 in size at sometime between 1930 and 1970. Phase III combined the procedures of Phase I and II and limited analysis to those places exceeding 74 and less than 1,000 in size in 1970. Phase I Analysis The first objective of this study is to determine the relationships between various place characteristics and their likelihood of enumeration by the census for all places exceeding 74 in size at sometime from 1930 to 1970. In an effort to accomplish this objective, four hypotheses were developed and tested by making crosstabulations between operationalizing indicators and source of enumeration. The percentage of total places in 33 each independent variable category (indicator category) enumerated by the census and not enumerated by the census were chmputed. If no relationship exists between the independent variable and likelihood of enumeration by the census thenthe percentage of places in each independent variable category enumerated by the census should be the same as the percentage of total state places in each of these categories. Any deviation from this will indicate a relationship between the independent variable and likeli- hood of enumeration by the census. The magnitude and pattern of these deviations will indicate the strength and direction of the relationship. HYPOTHESIS I: A positive relationship exists between the 1970 size of place and their likelihood of census enumeration. Indicator 1: 1970 Size of Place (Table 2) Table 2: The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By 1970 Population Size and Source of Enumeration 1] Source of Enumeration Census* Non-Census* Total 1970 P0pulation Size i y _#_ %_ _fi %_ 75 - 149 3 1.1 260 98.9 263 100.0 150 - 299 28 11.5 216 88.5 244 100.0 300 - 499 67 31.5 146 68.5 213 100.0 500 - 749 59 45.4 71 54.6 130 100.0 750 - 999 44 58.7 31 41.3 75 100.0 1000 - 2499 165 83.8 32 16.2 197 100.0 2500 - 9999 144 90.0 16 10.0 160 100.0 10,000 - 49,999 68 90.7 7 9.3 75 100.0 50,000 + 20 90.9 2 9.1 22 100.0 Total 598 37.8 781 62.2 1379 100.0 1] Missing data = 213 Gamna = -.84 *Census and Non-Census are two mutually exclusive categories. The listings under the census heading are those places exceeding 74 in size and reported by the census. The listings under the Non-Census heading are those places exceeding 74 in size and not reported by the census. For a detailed discussion of data collection procedures see pages 22-25. 34 The results displayed in Table 2 provide very strong support for Hypothesis I. As was suspected, these results are in keeping with the findings of previous research.and verifies the existence of a positive relationship between the size of place and likelihood of enumeration by the census. As shown in the table, the lowest percentage of total places enumerated by the census (1.1) is for places with the smallest size in 1970 (75-149). The highest percentage of total places enumerated by the census (90.9) is for the larger size places (50,000 +). As the 1970 size of places increased the percentage of places enumerated by the census also increased. Gamma was calculated to be -.84 indicating a very strggg association between 1970 size of place and source of enumeration. HYPOTHESIS II: A positive relationship exists between change in size of place from 1930 to 1970 and their likelihood of census enumeration. Indicator 1: Percentage Change in Population Number From 1930 to 1970 (Table 3) The results displayed in Table 3 provide strong support for Hypothesis 11. As shown in the table, the lowest percentage of total places enumerated by the census (1.5) is for places with the largest percentages of decline (-100 to -51%). The highest percentage of total places enumerated by the census (64.5) is for those places that have experienced moderate growth (50- 99%). The percentages associated with higher growth categories decline somewhat, revealing a slightly curvelinear relationship. Gamma was calcula- ted to be -.42 indicating a moderate association between the percentage change in population number from 1930 to 1970 and source of enumeration 25 For a detailed discussion of conventions used for describing gamma values see James A. Davis, Elementary SurveyyAnalysis, (New Jersey: Prentice Hall, Inc., 1971), p. 49. 35 Table 3: The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Percentage Change in Population Number From 1930 to 1970 and Source of Enumeration 1] Source of Enumeration % Change in Census Non-Census Total Population Number fi_ %_ fi_ %_ fi_ %_ Declining -100 To -51 3 1.5 203 98.5 206 100.0 -50 To -11 36 17.0 176 83.0 212 100.0 Stable -10 To 10 50 27.8 130 72.2 180 100.0 Growing 11 - 49 188 59.7 127 40.3 315 100.0 50 - 99 131 64.5 72 35.5 203 100.0 100 - 249 69 48.6 73 51.4 142 100.0 250 + 105 40.9 152 63.8 257 100.0 Total 582 37.6 933 62.4 1515 100.0 1/ Missing Data = 77 Gamma = -.42 Indicator 2: Pattern of P0pulation Change From 1930 to 1970 (Table 4) The results displayed in Table 4 provide strong support for Hypothesis 11. As shown in the table, the lowest percentage of total places enumerated by the census (12.5) is for places experiencing consistent decline. The highest percentage of total places enumerated by the census (55.0) is for places experiencing consistent growth. Gamma was calculated to be -.59 indicating a substantial association between pattern of change from 1930 to 1970 and source of enumeration. 36 Table 4: The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Their Pattern of Population Change From 1930-50 and 1950-70 and Source of Enumeration 1] Source of Enumeration Census Non-Census Total Pattern of Change i °/_o _#_ % f %_ Consistent Decline 29 12.5 203 87.5 232 100.0 (30-50-70) Recent Decline (50-70) 63 31.2 139 68.8 202 100.0 (Past Growth 30-50) Recent Growth (50-70) 40 15.2 223 84.8 263 100.0 (Past Decline 30-50) Consistent Growth 450 55.0 368 45.0 818 100.0 (30-50-70) Total 582 38.4 933 61.6 1515 100.0 1] Missing data = 77 Gamma = -.59 The percentages associated with each indicator category provide strong support for Hypothesis II. The indicators display a positive relationship between change in size of place form 1930 to 1970 and their likelihood of census enumeration. None of the results provided any evidence to contra- dict Hypothesis II. HYPOTHESIS III: A positive relationship exists between proximity of places to larger places and their likelihood of census enumeration. Indicator 1: Size of Largest Place in County (Table 5) The results displayed in Table 5 provide strong support for Hypothesis 111. As shown in the table, the lowest percentage of total places enumerated by the census (19.3) is for places located in counties that have the smallest ”size of largest place in county“ (0-999). The highest percentage of total 37 places enumerated by the census (50.6) is for places located in counties with the largest "size of largest place in county" (50,000 +). With the exception of one category (1,000-2,499), the percentage of places enumerated by the census increased as the size of the largest place in the county in- creased. Gamma was calculated to be -.16 indicating a low association be- tween size of largest place in county and source of enumeration. Table 5: The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Size of Largest Place in the County and Source of Enumeration Source of Enumeration Size of Census Non-Census Total Largest Place in County it :4 it f/9_ i % 0 - 999 11 19.3 46 80.7 57 100.0 1000 - 2499 60 45.8 71 54.2 131 100.0 2500‘ - 4999 63 32.1 133 67.9 196 100.0 5000 - 24,999 232 33.2 467 66.8 699 100.0 25,000 - 49,999 65 36.3 114 63.7 179 100.0 50,000 + 167 50.6 163 49.4 330 100.0 Total 598 37.6 994 62.4 1592 100.0 Gamma = -.16 Indicator 2: Proximity to Larger Places (Table 6) The results displayed in Table 6 provide strong support for Hypothesis III. As shown in the table, the lowest percentage of total places enumerated by the census (25.3) is for places further than 50 miles from a city greater than 25,000. The highest percentage of total places enumerated by the census (46.3) is for places located in counties which contain an SMSA. As the proximity to larger places increased, the percentage of places enumerated 38 by the census also increased. Gamma was calculated to be -.29 indicating a low to moderate association between proximity to larger places and source of enumeration. Table 6: The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By 1960 Proximity to Larger Places and Source of Enumeration' Source of Enumeration Proximity to Census Non-Census Total Larger Places _#_ jg i % i 75 Further than 50 miles from > 25,000 130 25.3 383 74.7 513 100.0 Within 50 miles of city > 25,000 222 40.5 326 59.5 548 100.0 SMSA 246 46.3 285 53.7 531 100.0 Total 598 37.6 994 62.4 1592 100.0 Gamma = -.29 Indicator 3: Metropolitan Status and Proximity of County to Counties with a place of 50,000 or More (Table 7) The results displayed in Table 7 provide strong support for Hypothesis 111. As shown in the table, the lowest percentage of total places enumerated by the census (30.8) is for places located in counties that have no place greater than 50,000 that are not adjacent to a county with a place greater than 50,000. The highest percentage of total places enumerated by the census (50.6) is for places located in counties that contain a place greater than 50,000. As the proximity to a place greater than 50,000 increased the percentage of places enumerated by the census also increased. Gamma was calculated to be -.25 indicating a low association between metropolitan 39 status and proximity and source of enumeration. Table 7: The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Metropolitan Status and Proximity of County to Counties with a Place of 50,000 or More and Source of Enumeration Source of Enumeration Metropolitan Census Non-Census Total Status and Profimi ty i %_ it %_ if 1% County has no Place 208 30.8 468 69.2 675 100.0 50,000; is not adjacent to county with place 50,000 County has no Place 223 38.1 363 61.9 585 100.0 50,000; is adjacent to county with Place 50,000 County has Place 50,000 167 50.6 163 49.4 330 100.0 Total 598 37.6 994 62.4 1592 100.0 Gamma = -.25 Indicator 4: SMSA Status (Table 8) The results displayed in Table 8 provide strong support for Hypothesis 111. As shown in the table,the lowest percentage of total places enumerated by the census (33.2) is for places located in counties that do not contain an SMSA. The highest percentage of total places enumerated by the census (46.3) is for places located in counties that contain an SMSA. As the proximity to an SMSA increased, the percentage of places enumerated by the census also increased. Gamma was calculated to be -.27 indicating a low association between SMSA status and source of enumeration. 40 Table 8: The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By 1965 SMSA Status of the County and Source of Enumeration Source of Enumeration SMSA Status of Qeg§g§_ Non-Census Total My. a a a Z [—1 SMSA 246 46.3 285 53.7 531 100.0 Non-SMSA 352 33.2 709 66.8 1061 100.0 Total 598 37.6 994 62.4 1592 100.0 Gamma = -.27 The percentages associated with each indicator category provide strong support for Hypothesis III. The indicators display a positive relationship between proximity of places to larger places and their likelihood of census enumeration. Indicator 1 displayed a slightly curvelinear percentage dis- tribution while all the other indicators displayed linear distributions. None of the results provided any evidence to contradict Hypothesis III. HYPOTHESIS IV: A positive relationship exists between the urbanity of an area within which places are located and their likelihood of census enumeration. Indicator 1: Regional Location of Place (Table 9) The results displayed in Table 9 provide substantial support for Hypothesis IV. As shown in the table, the lowest percentage of total places enumerated by the census (13.1) is for places located in the eastern region of the upper peninsula. The highest percentage of total places enumerated by the census (45.4) is for places located in the southeastern region of the lower peninsula. The table verifies expectations that a 41 lower percentage of places will be enumerated by the census in the upper peninsula and the northern portion of the lower peninsula and a higher percentage will be enumerated in the southern portion of the lower peninsula. Gamma was calculated to be -.28 indicating a low association between regional location of place and source of enumeration. Table 9: The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Regional Location and Source of Enumeration Source of Enumeration . Census Non-Census Total 8.9-. 9 ion 1‘. It .t It _#_ :79. UP - West 38 18.8 164 81.2 202 100.0 UP - East 13 13.1 86 86.9 99 100.0 LP - NW 57 42.5 77 57.5 134 100.0 LP - NE 33 33.7 65 66.3 98 100.0 LP - SW 174 40.0 261 60.0 435 100.0 LP - SE 283 45.4 341 54.6 624 100.0 Total 598 37.6 994 62.4 1592 100.0 Gamma = -.28 Indicator 2: Percentage of County Population that is Rural (Table 10) The results displayed in Table 10 provide moderate support for Hypothesis IV. As shown in the table, the lowest percentage of total places enumerated by the census (30.2) is for places located in counties with 25%-49% of the population designated as rural. The highest percentage of total places enumerated by the census (50.5) is for places located in counties with the lowest percentage of rural population (0-24%). With the exception of the lowest percentage group, as the percentage of the county rural population increased, the percentage of places enumerated by the census also increased 42 Gamma was calculated to .11 indicating a low association between percentage of county population that is rural and source of enumeration. Table 10: The Number and Percentage of Michigan P0pulation Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Percentage of County Population Rural and Source of Enumeration Source of Enumeration Percentage of County Census Non-Census Total PopuTation Rural _#_ _%_ _#_ %_ _#_ :4 O - 24 166 50.5 163 49.5 329 100.0 25 - 49 88 30.2 203 69.8 291 100.0 50 - 74 186 34.1 359 65.9 545 100.0 75 -100 158 37.0 269 63.0 427 100.0 Total 598 37.6 994 62.4 1592 100.0 Gamma = .11 Indicator 3: Percentage of Rural Non-Farm Residents Within the County (Table 11) The results displayed in Table 11 provide moderate support for Hypothe- sis IV. As shown in the table, the lowest percentage Of total places enumerated by the census (33.1) is for places iocated in counties with 25%- 49% of the population designated as rural non-farm. The highest percentage of total places enumerated by the census (50.5) is for places located in counties with the lowest percentage of rural non-farm population. (O-24%) With the exception of the lowest percentage group, as the percentage of the county rural non-farm population increased the percentage of places enumerated by the census also increased. Gamma was calculated to be .13 indicating low 43_ association between percentage of rural non—farm residents within the county and source of enumeration. Table 11: The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Percentage of Rural Non- Farm Residents Within the County and Source of Enumeration Source of Enumeration % Rural Non-farm Census Non-Census Total Population in County i %_ _#_ y i 1 0 - 24 167 50.5 164 49.5 331 100.0 25 - 49 165 33.1 334 66.9 499 100.0 50 - 74 196 33.9 382 66.1 578 100.0 75 -100 70 38.0 114 62.0 184 100.0 Total 598 37.6 994 62.4 1592 100.0 Gamma = .137 Indicator 4: Percentage of County Workforce Employed in Extractive Industries (Table 12) The results displayed in Table 12 provide strong support for Hypothesis IV. As shown in the table, the lowest percentage of total places enumerated by the census (30.9) is for places located in counties with the highest percentage of the work force employed in extractive industries. The highest percentage of total places enumerated by the census (49.0) is for places located in counties with the lowest percentage of the workforce employed in extractive industries. As the percentage of county workforce employed in extractive industries increased, the percentage of places enu- merated by the census decreased. Gamma was calculated to be .18 indicating a low association between percentage of county workforce employed in extrac- tive industries and source of enumeration. 44 Table 12: The Number and Percentage of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By the Percentage of County Workforce Employed in Extractive Industries and Source of Enumeration Source of Enumeration Percentage of County_ Census Non-Census Total Rdrkforce - Extractive ~ it % _#_ % _#_ %_ 0 - 2 202 49.0 210 51.0 412 100.0 3 - 5 181 33.5 359 66.5 540 100.0 6 -10 160 34.6 302 65.4 462 100.0 11+ 55 30.9 123 69.1 178 100.0 Total 598 37.6 994 62.4 1592 100.0 Gamma = .18 The percentages associated with the indicator categories provide strong to moderate support for Hypothesis IV. The indicators suggest that the more rural the location of a place, the less likely it is to be enumerated by the census. Conversely, the more urban the location of a place, the more likely it is to be enumerated by the census. The indicators display slightly curvelinear percentage distributions, however none of the results provide any evidence to contradict Hypothesis IV. 45 Phase II Analysis The numeric and percentage distributions and comparisons made in Phase I of the analysis revealed for the most part, strong indicator support for Hypotheses I - IV. Generally, the larger the size of the place in 1970, the greater the increase in size from 1930 to 1970, the higher the proximity to larger places, or the more urban the area of location, the greater the likelihood of enumeration by the census. These findings suggest that Hypotheses V-VIII should be supported as well. More specifically, census enumerated places should over represent the percentage of large places, growing places, places with high proximity to larger places and places lo- cated in urban areas. Phase II of the analysis is designed to determine the extent of census under representation (underenumeration) in terms of selected place character- istics or indicators. The percentage of the total state places enumerated by the census for each independent variable category (indicator category) and the percentage of the total state places for each independent variable category were computed. If census enumerated places were representative of the total universe of state places then no deviation should exist between the two percentages for each of the independent variable categories. Any deviation from this will indicate a relationship between the independent variable and representativeness of census enumerated places. The magnitude and pattern of these deviations will indicate the strength and direction of the relationship. HYPOTHESIS V: Compared to the total universe of state places, census enumerated places will over represent larger places and under represent smaller places. _ 46 Indicator 1: 1970 Size of Place (Table 13) The results displayed in Table 13 provide very strong support for Hypothesis V. As shown in the table, the census over repcesents places larger than 500 in size and under represents places smaller than 500 in size. A greater percentage of under representation by the census is found to exist as the size of the place decreases. The greatest percentage of under representation (-18.6) is found to exist for those places 75-149 in size. The greatest percentages of over representation by the census exist for those places in the 1000-2,499 (+13.3) and 2,500-9,999 (+12.5) size categories. No evidence exists to contradict the validity of Hypothesis V. Table 13: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By 1970 Population Size for Census Enumerated and Total State Places 1] Source of Enumeration Census Total Difference in 1970 Population Size fi_ %_ fi_ %_ Percentages 75 - 149 3 0.5 263 19.1 -18.6 150 - 299 28 4.7 244 17.7 -13.0 -300 - 499 67 11.2 213 15.4 -4.2 500 - 749 59 9.9 130 9.4 +0.5 750 - 999 44 7.4 75 5.4 +2.0 . 1000 - 2499 165 27.6 197 14.3 +13.3 2500 - 9999 144 24.1 160 11.6 +12.5 10,000 - 49,999 68 11.4 75 5.4 +6.0 50,000 + 20 3.3 22 1.6 +1.7 1] Missing data = 213 47 HYPOTHESIS VI: Compared to the total universe of state places, census enumerated places will over represent growing places and under represent declining places. Indicator 1: Percentage Change in Population Number From 1930 to 1970 (Table 14) The results displayed in Table 14 provide very strong support for Hypothesis VI. As shown in the table the census over represents growing places and under represents stable and declining places. The greatest percentage of under representation by the census (-13.1) exists for places that have experienced the greatest percentage decline in population num- ber (-100 to -51). All places that experienced growth from 1930 to 1970 were over represented by the census, however, as the percentage of growth Table 14: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size By Percentage Change in Population Number From 1930 to 1970 for Census Enumerated and Total State Places 1] Source of Enumeration % Change in Census Total Difference in Population Size fi_ %_ f_ %_ Percentages Declining -100 To -51 3 0.5 206 13.6 -13.1 -50 To -11 36 6.2 212 14.0 -7.8 Stable -10 To 10 50 8.6 180 11.9 -3.3 Growing 11 - 49 188 32.3 315 20.8 +11.5 50 - 99 131 22.5 203 13.4 +9.1 100 -249 69 11.9 142 9.4 +2.5 250 + 105 18.0 257 17.0 +1.0 Total 582 100.0 1515 100.0 -- 1] Missing data = 77 48 increased the percentage of over representation decreased. The greatest percentage of over representation by the census (+11.5) is found to exist for places that experienced 11 to 49 percent growth. Indicator 2: Pattern of Population Change From 1930 to 1970 (Table 15) The results displayed in Table 15 provide strong to moderate support for Hypothesis VI. As shown in the table, the census over represents consistently growing places and under represents declining and recently growing places. The greatest percentages of under representation by the census (-10.3 and -10.5) exist for places that experienced consistent decline and recent growth. The greatest percentage of over representation by the census (+23.3) exists for places that experienced consistent growth. Table 15: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Their Pattern of Population Change From 1930-50 and 1950-70 for Census Enumerated and Total State Places 1] Source of Enumeration Census Total Difference in Pattern of Change fi_ %_ f_ %_ Percentages Consistent Decline 29 5.0 232 15.3 -10.3 (30-50-70) Recent Decline (50-70) 63 10.8 202 13.3 -2.5 (Past Growth 30-50) Recent Growth (50-70) 40 6.9 263 17.4 -10.5 (Past Decline 30-50) Consistent Growth 450 77.3 818 54.0 +23.3 (30-50-70) Total 582 100.0 1515 100.0 -- 1] Missing Data = 77 49 The differences in percentage associated with.the indicator categories provide strong support for Hypothesis VI. The indicators displayed a census over representation of growing places and an under representation of de- clining places. The results provided little evidence to contradict Hypo- thesis VI. HYPOTHESIS VII: Compared to the total universe of state places, census enumerated places will over represent those places closer to larger places and under represent places that are farther away. Indicator 1: Size of Largest Place in County (Table 16) The results displayed in Table 16 provide little support for Hypothe- sis VII. As shown in the table, the census over represents two categories (1000-2499 and 50,000+) and under represents the remaining four categories. The greatest percentage of under representation by the census (-5.1) exists for places located in counties with size of largest place ranging from 5000-24,999. The greatest percentage of over representation by the cen- sus (+7.2) exists for places located in counties with the size of largest place in excess of 50,000. 50 Table 16: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Size of Largest Place in the County for Census Enumerated and Total State Places Source of Enumeration Size of Census Total Difference in [argest Place fi_ %_ fi_ %_ Percentages In County 0 - 999 11 1.8 57 3.6 -1.8 1000 - 2499 60 10.0 131 8.2 +1.8 2500 - 4999 63 10.5 196 12.3 -1.8 5000 - 24,999 232 38.8 699 43.9 -5.1 25,000 - 49,999 65 10.9 179 11.2 -0.3 50,000 + 167 27.9 330 20.7 +7 2 Total 598 100.0 1592 100.0 -- Indicator 2: Proximity to Large Places (Table 17) The results displayed in Table 17 provide strong support for Hypothesis VII. As shown in the table, the census over represents places that are within 50 miles of a city greater than 25,000 and those located in SMSA counties and under represents places that are further than 50 miles from a city greater than 25,000. The greatest percentage of under representation by the census (-10.5) exists for places further than 50 miles from a city greater than 25,000. The greatest percentage of over representation by 51 the census (+7.7) exists for places located in counties that contain an SMSA. Table 17: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By 1960 Proximity to Larger Places for Census Enumerated and Total State Places Source of Enumeration Proximity to Census Total Differences in Larger Places fi_ %_ f_ %_ Percentages Further than 50 miles 130 21.7 513 32.2 -10.5 from city ) 25,000 Within 50 miles 222 37.1 548 34.4 +2.7 of city > 25,000 SMSA 246 41.1 531 33.4 +7.7 Total 598 100.0 1592 100.0 -- Indicator 3: Metropolitan Status and Proximity of county to counties with a Place of 50,000 or More (Table 18) The results displayed in Table 18 provide strong support for Hypothesis VII. As shown in the table, the census over represents places that are located in counties or adjacent to a county that has a place greater than 50,000, and under represents places that are located in or adjacent to counties that do not have a place greater than 50,000 in size. The greatest percentage of under representation by the census (-7.7) exists for places that are not located in or are adjacent to counties with a place greater than 50,000. The greatest percentage of over representation by the census (+7.2) exists for places located in counties that contain a place 52 greater than 50,000. Table 18: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Metropolitan Status and Proximity of County to Counties with a Place of 50,000 or More for Census Enumerated and Total State Places Source of Enumeration Metropolitan Census Total Difference in Status and "“" Prox1m1ty £_ %_ fi_ %_ Percentages County has no Place 208 34.8 675 42.5 -7.7 50,000; is not adjacent to county with place 50,000 County has no Place 223 37.3 585 36.8 +0.5 50,000; is adjacent to county with place 50,000 County has Place 167 27.9 330 20.7 +7.2 50,000 ' Total 598 100.0 1592 100.0 -- Indicator 4: SMSA Status (Table 19) The results displayed in Table 19 provide strong support for Hypothesis VII. As shown in the table, the census under represents places located in non-SMSA counties and over represents places located in SMSA counties. The greatest percentage of under representation by the census (-7.6) exists for places located in counties that do not contain an SMSA. The greatest per- centage of over representation by the census (+7.7) exists for places lo- cated in counties containing an SMSA. 53 Table 19: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By 1965 SMSA Status of the County for Census Enumerated and Total State Places Source of Enumeration SMSA Status of Census Total Difference in nggty_ fi_ %_ fi_ %_ Percentages SMSA 246 41.1 531 33.4 +7.7 Non-SMSA 352 58.9 1061 66.5 -7.6 Total 598 100.0 1592 100.0 -- Generally, the differences in percentages associated with the indicator categories provide strong support for Hypothesis VII. Indicators 2-4 dis- play a census over representation of those places closer to larger places and under representation for places that are farther away. Indicator 1, provides mixed results which neither support or contradict Hypothesis VII. HYPOTHESIS VIII: Compared to the total universe of state places, census enumerated places will over represent places located in urban areas and under represent places located in rural areas. Indicator 1: Regional Location of Place (Table 20) ' The results displayed in Table 20 provide strong support for Hypothesis VIII. As shown in the table, the census over represents places located in the southern portion of the lower peninsula and under represents places in the northern portion of the lower peninsula and the entire upper peninsula. This provides support for Hypothesis VIII due to the fact that the greatest concentration of urbanization in Michigan is found to exist in the southern portion of the lower peninsula. The greatest percentage of under represen- tation by the census (-6.3) exists for places located in the western portion 54 of the upper peninsula. The greatest percentage of over representation by the census (+8.1) exists for places located in the southeastern portion of the lower peninsula. Table 20: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Regional Location for Census Enumerated and Total State Places Source of Enumeration Census Total Difference in Region fi_ __ i. %_ Percentages UP - West 38 6.4 202 12.7 -6.3 UP - East 13 2.2 99 6.2 -4.0 LP - NW 57 9.5 134 8.4 -1.1 LP - NE 33 5.5 98 6.2 -0.7 LP - SW 174 29.1 435 27.3 +1.8 LP - SE 283 47.3 624 39.2 +8.1 Total 598 100.0 1592 100.0 -- Indicator 2: Percentage of County Population that is Rural (Table 21) The results displayed in Table 21 provide strong support for Hypothesis VIII. As shown in the table, the census over represents places located in counties with the lowest percentages of rural population and under represents places located in counties with larger percentages of rural population. The greatest percentage of under representation by the census (-3.6) exists for places located in counties where 25 - 49 percent of the population is desig- nated as rural. The greatest percentage of over representation by the cen- sus (+7.1) exists for places located in counties where 0 - 24 percent of the population is designated as rural. 55' Table 21: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Percentage of County P0pulation Rural for Census Enumerated and Total State Places. Source of Enumeration Percentage_of County Census Total Difference in ngfilation Rural fi_ %_ f_ %_ Percentages 0 - 24 166 27.8 329 20.7 +7.1 25 - 49 88 14.7 291 18.3 -3.6 50 - 74 186 31.1 545 34.2 -3.1 75 - 100 158 26.4 427 26.8 -0.4 Total 598 100.0 1592 100.0 -- Indicator 3: Percentage of Rural Non-Farm Residents within the County (Table 22) . The results displayed in Table 22 provide moderate support for Hypothesis VIII. As shown in the table, the census over represents the places located in counties that have the lowest and the highest percentages of rural non-farm residents and under represents places located in counties with moderate per- centages of rural non-farm residents . The greatest percentage of under representation by the census (-3.7) exists for places located in counties where 25 - 49 percent of the population is designated as rural non-farm. The greatest percentage of over representation by the census (+7.1) exists for places located in counties where 0 -'24 percent of the population is designated as rural non-farm. 56 Table 22: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By Percentage of Rural Non-Farm Residents Within the County for Census Enumerated and Total State Places Source of Enumeration % Rural Non-Farm Census Total Difference in Population in County f_ %_ fi_ %_ Percentages O - 24 167 27.9 331 20.8 +7.1 25 - 49 165 27.6 499 31.3 -3.7 50 - 74 196 32.8 578 36.3 -3.5 75 - 100 70 11(7 184 11.6 +0.1 Total 598 100.0 1592 100.0 -- Indicator 4: Percentage of County Workforce Employed in Extractive Industries (Table 23) The results displayed in Table 23 provide strong support for Hypothesis VIII. As shown in the table, the census over represents places located in counties with the lowest percentages of the county workforce employed in extractive industries and under represents places located in counties with higher percentages of the county workforce employed in extractive industries. The greatest percentage of under representation by the census (-3.6) exists for places located in counties where 3 - 5 percent of the county workforce is employed in extractive industries. The greatest percentage of over re- presentation by the census (+7.9) exists for places located in counties where 0 - 2 percent of the county workforce is employed in extractive industries. 57 Table 23: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size at Sometime from 1930 to 1970 By the Percentage of County Workforce Employed in Extractive Industries for Census Enumerated and Total State Places Source of Enumeration Percentage of County_ Census Total Difference in Workfbrce - Extractive fi_ %_ £_ %_ Percentage O - 2 202 33.8 412 25.9 +7.9 3 - 5 181 30.3 540 33.9 -3.6 6 - 10 160 26.8 462 29.0 -2.2 11 + 55 9.2 178 11.2 -2.0 Total 598 100.0 1592 100.0 -- The differences in percentages associated with the indicator categories provide strong to. moderate support for Hypothesis VIII. The indicators generally displayed a census over representation of places located in urban areas and an under representation of places located in rural areas. The results provide little evidence to contradict Hypothesis VIII. Phase 111 Analysis Phase I and II analyses have revealed strong support for Hypotheses I-VIII. The final phase of the data analysis will be directed at testing these same hypotheses while controlling for size. The total universe of Michigan population centers is operationally restricted to those population centers exceeding 74 and less than 1,000 in size in 1970. Hypotheses Ia- VIIIa will be tested in the same manner as was done in Phase I and Phase II of the analysis however the number of indicators selected will be reduced for several of the hypotheses. HYPOTHESIS Ia: A positive relationship exists between the 1970 size of place and their likelihood of census enumeration. Indicator 1: 1970 Size of Place (Table 24) Table 24: The Number and Percentage of Michigan Population Centers Exceeding 74 and less Than 1,000 in Size in 1970 By 1970 Population Size and Source of Enumeration l] Source of Enumeration 1970 Pqulation Census Non-Census Total 5.12.8. t. it t 24. it it 75 - 149 3 1.1 260 98.9 263 100.0 150 - 299 28 11.5 216 88.5 244 100.0 300 - 499 67 31.5 146 68.5 213 100.0 500 - 749 59 45.4 71 54.6 130 100.0 750 - 999 44 58.7 31 41.3 75 100.0 Total 201 17.8 724 82.2 925 100.0 1] Missing Data = 202 Game = -.79 The results displayed in Table 24 provide very strong support for Hypothesis 59 la. As shown in the table, the lowest percentage of total places enumerated by the census (1.1) is for places with the smallest size in 1970 (75-149). The highest percentage of total places enumerated by the census (58.7) is for the larger size places (750-999). As the 1970 size of place category increased the percentage of places enumerated by the census also increased. Gamma was calculated to be -.79 indicating a very strong assocation between 1970 size of place and source of enumeration. HYPOTHESIS IIa: A positive relationship exists between change in size of place from 1930 to 1970 and their likelihood of census enumeration. Indicator 1: Percentage Change in Population Number From 1930 to 1970 (Table 25) Table 25: The Number and Percentage of Michigan Places Exceeding 74 and Less Than 1000 in Size in 1970 By 1930-1970 Percentage Change in Popula- tion Size and Source of Enumeration l/ Source of Enumeration % Change in Census Non-Census Total Population Size _#_ %_ it 1%; i 2.1 Declining -100 To -51 2 1.0 203 99.0 205 100. -50 To -11 16 8.3 176 91.7 192 100. Stable -10 To 10 16 12.0 117 88.0 133 100 Growing 11 - 49 92 42.2 126 57.8 218 100. 50 - 99 52 42.6 70 57.4 122 100. 100 -249 15 17.4 71 82.6 86 100. 250 + 8 4.7 163 95.3 171 100. Total 201 17.8 926 82.2 1127 100. Gamma = -.27 OO C 00000 60 The results displayed in Table 25 provide strong support for Hypothesis IIa. As shown by the table, the lowest percentage of total places enumerated by the census (1.0) is for places with the largest percentages of decline (-100 to -51%). The highest percentage of total places enumerated by the census (42.6) is for those places that have experienced moderate growth (50 - 99%). The percentages associated with higher growth categories decline some- what, revealing a slightly curvelinear relationship. Gamma was calculated to be -.27 indicating a low association between the percentage change in popula- tion number from 1930 to 1970 and source of enumeration. Indicator 2: Pattern of Population Change From 1930 to 1970 (Table 26) Table 26: The Number and Percentage of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By Their Pattern of Population Change From 1930-1950 and 1950-1970 and Source of Enumeration l/ Source of Enumeration Pattern of Census Non-Census Total Change [—2 a z. f—a Consistent Decline 9 4.2 203 95.8 212 100.0 (30-50-70) Recent Decline (50-70) 30 18.2 135 81.8 165 100.0 (Past Growth 30-50) Recent Growth (50-70) 22 9.0 222 91.0 244 100.0 (Past Decline 30-50) Consistent Growth 137 29.7 324 70.3 461 100.0 (30-50-70) Total 198 18.3 884 81.7 1082 100.0 1] Missing Data = 45 Gamma = -.51 -61 The results displayed in Table 26 provide moderate support for Hypothesis IIa. As shown in the table, the lowest percentage of total places enumerated by the census (4.2) is for places experiencing consistent decline. The highest percentage of total places enumerated by the census (29.7) is for places experiencing consistent growth. Gamma was calculated to be -.51 indicating a substantial association between pattern of change from 1930 to 1970 and source of enumeration. The percentages associated with each indicator category provide strong support for Hypothesis 11a. The indicators display a positive relationship between change in size of place from 1930 to 1970 and their likelihood of census enumeration. None of the results provide any evidence to contradict Hypothesis IIa. HYPOTHESIS IIIa: A positive relationship exists between proximity of place to larger places and their likelihood of census enumeration. Indicator 1: Proximity to Larger Places (Table 27) Table 27: The Number and Percentage of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By 1960 Proximity to Larger Places and Source of Enumeration Source of Enumeration Proximity to Census Non-Census Total Larger Places it :4 i _%_ i it Further than 50 miles from city > 25,000 63 14.5 371 85.5 434 100.0 Within 50 miles of city > 25,000 96 23.0 321 77.0 417 100.0 SMSA 42 15.2 234 84.8 276 100.0 Total 201 17.8 926 82.2 1127 100.0 Gamma = -.06 62 The results displayed in Table 27 provide moderate support for Hypothe- sis IIIa. As shown in the table, the lowest percentage of total places enumerated by the census (14.5) is for places further than 50 miles from a city greater than 25,000. The highest percentage of total places enumerated by the census (23.0) is for places within 50 miles of a city greater than 25,000. Gamma was calculated to be -.06 indicating a negligible association between proximity to larger places and source of enumeration. Indicator 2: SMSA Status (Table 28) Table 28: The Number and Percentage of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By 1965 SMSA Status of The County and Source of Enumeration Source of Enumeration SMSA Status gfygggi Non-Census Total of County _#_ it i % i _% SMSA 42 15.2 234 84.8 276 100.0 Non-SMSA 159 18.7 692 81.3 851 100.0 Total 201 17.8 926 82.2 1127 100.0 Gamma = .12 Table 28 provides no support for Hypothesis IIIa. As shown by the table, the lowest percentage of total places enumerated by the census is for places located in counties that contain an SMSA. The highest percentage of total places enumerated by the census is for places located in counties that do not contain an SMSA. As the proximity to an SMSA increases, the percentage of places enumerated by the census decreases. Gamna was calculated to be .12 63 indicating a low association between SMSA status and source of enumeration. The percentage associated with the indicator categories do not provide much support for Hypothesis IIIa. The indicators do not support the conten- tion that a positive relationship exists between proximity of place to larger places and their likelihood of census enumeration. HYPOTHESIS IVa: A positive relationship exists between the urbanity of an area within which places are located and their likelihood of census enumeration. Indicator 1: Regional Location (Table 29) Table 29: The Number and Percentage of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By Regional Location and Source of Enumeration Source of Enumeration Census Non-Census Total Reg ion i 25. ii. L73; IL 1 UP - West 10 6.0 157 94.0 167 100.0 UP - East 4 4.5 84 95.5 88 100.0 LP - NW 38 33.3 76 66.7 114 100.0 LP - NE 18 22.2 63 77.8 81 100.0 LP - SW 55 18.3 245 81.7 300 100.0 LP - SE 76 20.2 301 79.8 377 100.0 Total 201 17.8 926 82.2 1127 100.0 The results displayed in Table 29 provide moderate support for Hypothe- sis IVa. As shown in the table, the lowest percentage of total places enu- merated by the census (4.5) is for places located in the eastern region of the upper peninsula. The highest percentage of total places enumerated by the census (33.3) is for places located in the northwestern region of the lower peninsula. Gamma was calculated to be -.17 indicating low association 64 between regional location and source of enumeration. Indicator 2: Percentage of County Population That is Rural (Table 30) Table 30: The Number and Percentage of Michigan P0pulation Centers Exceeding 74 and Less Than 1000 in Size in 1970 By Percentage of County Population Rural and Source of Enumeration Source of Enumeration Percentage of County Census Non-Census Total Population Rural y _%_ i _% i 1%; o - 24 30 19.4 125 80.6 155 100.0 25 - 49 19 9.2 187 ‘ 90.8 206 100.0 50 - 74 71 16.8 352 83.2 423 100.0 75 - 100 81 23.6 262 76.4 343 100.0 Total 201 17.8 926 82.2 1127 100.0 Gamma = -.18 The results displayed in Table 30 provide no support for Hypothesis IVa. As shown in the table, the lowest percentage of total places enumerated by the census (9.2) is for places located in counties with 25%-49% of the popu- lation designated as rural. The highest percentage of total places enumerated by the census (23.6) is for places located in counties with the highest per- centage of rural population. Gamma was calculated to be -.18 indicating a low association between percentage of county population that is rural and source of enumeration. The percentages associated with the indicator categories do not pro- vide much support for Hypothesis IVa. The indicators do not support the contention that a positive relationship exists between the urbanity of an area within which places are located and their likelihood of census enumeration. 65 HYPOTHESIS Va: Compared to the total universe of state places, census enumerated places will over represent larger places and under represent smaller places. Indicator 1: 1970 Size of Place (Table 31) Table 31: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By 1970 Population Size for Census Enumerated and Total State Places 1] Source of Enumeration 1970 Population Census Total Difference in Size §_ %_ fi_ %_ Percentages 75 - 149 3 1.5 263 28.4 -26.9 150 - 299 28 13.9 244 26.4 -12.5 300 - 499 67 33.3 213 23.0 +8.5 500 - 749 59 29.4 130 14.1 +31.3 750 - 999 44 21.9 75 8.1 +50.6 Total 201 100.0 925 100.0 -- ;/ Missing Data = 202 The results displayed in Table 31 provide very strong support for Hy- pothesis Va. As shown in the table, the census over represents places lar- ger than 500 in size and under represents places smaller than 500 in size. The greatest percentage of under representation (-26.9) is found to exist for those places 75-149 in size. The greatest percentages of over repre- sentation by the census (+50.6) exist for those places 750-999 in size. No evidence exists to contradict the validity of Hypothesis Va. HYPOTHESIS VIa: Compared to the total universe of state places, census enumerated places will over represent growing places and under represent declining places. 66 Indicator 1: Percentage Change in Population Number From 1930 to 1970 (Table 32) Table 32: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By 1930-1970 Percentage Change in Population Size for Census Enumerated and Total State Places Source of Enumeration % Change in Census Total Difference in Population Number fi‘ %_ fi_ %. Percentages Declining -100 To -51 2 1.0 205 18.9 -17.9 -50 To -11 16 8.1 192 17.7 -9.6 Stable -10 To 10 16 8.1 133 12.3 -4.2 Growing 11 - 49 92 45.8 218 19.3 +26.5 50 - 99 52 25.9 122 10.8 +15.1 100 -249 15 7.5 86 7.6 -0.1 250 + 8 4.0 171 15.2 -11.2 Total 201 100.0 1127 100.0 -- The results displayed in Table 32 provide little support for Hypothe- sis VIa. As shown in the table, the census over represents moderately growing places and under represents all other places. The greatest percen- tage of under representation by the census (-17.9) exists for places that experienced the greatest percentage decline in population number {-100 to -51). The greatest percentage of over representation by the census (+26.5) exists for places that experienced 11 to 49 percent growth. 67 Indicator 2: Pattern of Population Change from 1930 to 1970 (Table 33) Table 33: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size and Less Than 1000 in Size in 1970 By Their Pattern of Population Change From 1930-1950 and 1950-1970 for Census Enumerated and Total State Places l] Source of Enumeration Pattern of Census Total Difference in Change fi_ %_ fi_ %_ Percentages Consistent Decline 9 4.5 212 19.6 -15.1 (30-50-70) Recent Decline 30 15.2 165 15.2 0.0 (50-70) Recent Growth 22 11.1 244 22.6 -11.5 (50-70) Consistent Growth 137 69.2 461 42.6 +26.6 (30-50-70) Total 198 18.3 1082 100.0 -- 1] Missing Data = 45 The results displayed in Table 33 provide strong to moderate support for Hypothesis VIa. As shown in the table, the census over represents con- sistently growing places and under represents consistently declining and recently growing places. The greatest percentage of under representation by the census (-15.1) exists for places that experienced consistent de- cline. The greatest percentage of over representation by the census (+26.6) exists for places that experienced consistent growth. The differences in percentages associated with the indicator cate- gories provide moderate support for Hypothesis VIa. Generally, the in- dicators displayed a census over representation of growing places and an 68 under representation of declining places. HYPOTHESIS VIIa: Compared to the total universe of state places, census enumerated places will over represent those places closer to larger places and under represent places that are farther away. Indicator 1: Proximity to Larger Places (Table 34) Table 34: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size and Less Than 1000 in Size in 1970 By 1960 Proximity to Larger Places for Census Enu- merated and Total State Places Proximity to Larger Places Further than 50 miles from city > 25,000 Within 50 miles of city > 25,000 SMSA Total Source of Enumeration Census Total Difference: in fi_ %_ fi_ %_ Percentages 63 31.3 434 38.5 -7.2 96 47.8 417 37.0 +10.8 42 20.9 276 24.5 -3.6 201 100.0 1127 100.0 -- The results displayed in Table 34 provide moderate support for Hypo- thesis VIIa. As shown in the table, the census over represents places that are within 50 miles of a city greater than 25,000 and under repre- sents places that are further than 50 miles from a city greater than 25,000 and places located in SMSA counties. The greatest percentage of under representation by the census (-7.2) exists for places further than 50 miles from a city greater than 25,000. The greatest percentage of over repre- sentation by the census (+10.8) exists for places within 50 miles of a 69 city greater than 25,000. Indicator 2: SMSA Status of County(Table 35) Table 35: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 in Size and Less Than 1000 in Size in 1970 By 1965 SMSA Status of The County for Census Enu- merated and Total State Places Source of Enumeration SMSA Status Ceg§g§_ Igtal_ Difference in of County fi_ %. fi_ %_ Percentages SMSA 42 20.9 276 24.5 -3.6 Non-SMSA 159 79.1 851 75.5 +3.6 Total 201 100.0 1127 100.0 -- The results displayed in Table 35 provide no support for Hypothesis VIIa. As shown in the table, the census over represents places located in SMSA counties by +3.6 percent and under represents places located in non-SMSA counties by -3.6 percent. This is in complete contradiction to what was expected. The differences in percentages associated with the indicator cate- gories provide very little support for Hypothesis VIIa. It appears that the census tends to under represent places located in SMSA counties which is inconsistent with the stated hypothesis. HYPOTHESIS VIIIa: Compared to the total universe of state places, census enumerated places will over represent places located in urban areas and under represent places located in rural 70 areas. Indicator 1: Regional Location (Table 36) Table 36: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By Regional Location for Census Enumerated and Total State Places Source of Enumeration Census Total Difference in Region fi_ %. f. %_ Percentages UP - West 10 5.0 167 14.8 -9.8 UP - East 4 2.0 88 7.8 -5.8 LP - NW 38 18.9 114 10.1 +8.8 LP - NE 18 9.0 81 7.2 +2.8 LP - SW 55. 27.4 300 26.6 +0.8 LP - SE 76 37.8 377 33.5 +4.3 Total 201 100.0 1127 100.0 -- The results displayed in Table 36 provide moderate support for Hypothe- sis VIIIa. As shown in the table, the census over represents places located in the lower peninsula and under represents places located in the upper peninsula. The greatest percentage of under representation by the census (-9.8) exists for places located in the western portion of the upper penin- sula. The greatest percentage of over representation by the census (+8.8) exists for places located in the northwestern portion of the lower penin- sula. Indicator 2: Percentage of County Population That is Rural (Table 37) The results displayed in Table 37 provide little support for Hypothe- sis VIIIa. As shown in the table, the census over represents places lo- cated in counties that have O%-24% and 75%-100% rural populations and under 71 enumerates places located in counties that have 25%-74% rural populations. The greatest percentage of under representation by the census (-8.8) exists for places located in counties where 25-49 percent of the popula- tion is designated as rural. The greatest percentage of over representa- tion by the census (+9.9) exists for places located in counties where 75-100 percent of the population is designated as rural. Table 37: The Number, Percentage and Difference in Percentages of Michigan Population Centers Exceeding 74 and Less Than 1000 in Size in 1970 By Percentage of County Population Rural for Census Enumerated and Total State Places. Source of Enumeration Percentage of County; Census Total Difference in Papulation RuraTT fi_ %_ fi_ %_ Percentages 0 - 24 30 14.9 155 . 13.8 +1.1 25 - 49 19 9.5 206 18.3 -8.8 50 - 74 71 35.3 423 37.5 -2.2 75 - 100 81 40.3 343 30.4 +9.9 Total 201 100 0 1127 100.0 -- The differences in percentages associated with the indicator cate- gories provide moderate support for Hypothesis VIIIa. The indicators generally displayed a census over representation of places located in urban areas and an under representation of places located in rural areas. The results do, however suggest that there exists a degree of non support for Hypothesis VIIIa. CHAPTER V SUMMARY AND CONCLUSIONS Introduction The use of accurate data and sound methods in both basic and applied areas of community research is vital to the evolving process of scientific growth in this area. As with any scientific investigation, the need for sound methods and accurate data in community research is of basic impor- tance for the development of sound community theory. The methods em- ployed in problem identification, data collection, and data analysis have significant effects upon results, just as biased data can produce biased results and ultimately lead to inaccurate or biased theory. In an effort to reduce these problems, scientific investigations are con- tinually evolving to higher levels of sophistication and employing new techniques in the quest for developing more accurate data, results and theory. One basic method employed in community research is the case study approach, in which the community is generally viewed as a laboratory for studying some problematic aspects of the community. These individual case studies usually lack data on factors relevant to specific theoreti- cal comparisons and community investigators generally fail to examine more than the most elementary community variables in a large number of communities. In an effort to minimize these problems, it has been suggested that theory should be developed about community phenomena so relevant theoretical community variables can be specified and the num- ber of communities for which corrmunity parameters are available can be increased. This suggests that the non-cumulative method employed in 73 many community studies should be altered, and new comparative methods developed to maximize their usefulness for generalizing about community phenomena. Community development theory sufferssfrom the same methodological problems that the more basic community research does. Again, there is a need to strengthen comparative techniques so analysis of several action programs can yield high levels of generality. In this way, both the basic and applied realms of community investigation can work together to further the understanding of community processes. Community develop- ment programs need to derive ideas from basic research, however they can also contribute to basic knowledge if they are conducted and reported in such a way that comparative methods are employed. Summary of the Problem The variation in size, scope, orientation and methods used in con- ducting community research are great and the establishment of a conceptual definition of community is basic to most of this research. Community is often conceptualized in its most basic terms as a densely settled center or a densely settled center plus a less densely settled hinterland. This type of conceptualization implies people and locality which is often operationally restricted to population centers. The identification and enumeration of population centers can be a means through which communities can be identified and enumerated. The enumeration of population centers, of course, is only a first step be- cause in order for a population center to be classified as either a community center or community, additional empirical information would be required. 74 Presently, the main source of data available for enumerating popu- lation centers is the U.S. Bureau of the Census. However, the census enumerates only those places with a population of 1,000 or more or those that are legally incorporated. Since many centers less than 1,000 in size are known to be unincorporated, it becomes obvious that the census does not enumerate many smaller centers. This contention is supported by several pieces of research that date back as far as 1926. It was in that year that Luther Fry published his findings and revealed that incorporated villages (or those recorded by the census) only made up 56% of the total universe of villages in the United States, indicating that 44% of the villages were not enumerated by the census. The percen- tages enumerated by the census varied considerably from region to region with 89% enumerated in the West North Central region, 57% in the East North Central and 7% in the North East. Paul H. Landis found that in 1930 there were, according to estimates, almost seventeen million people in the United States who lived in places of under 2,500 people or in other words, about 14% of the total U. S. population was living in these places. Of these people 45.4% were living in unincorporated places and thus not identified by the census. Similar research conducted in other states also found that the census enumerated a low percentage of the total number of population centers or places in given study areas. Vincent Heath Whitney in 1945 reported that in a ten county subregion in north-central North Carolina, 75% of the villages and hamlets were not enumerated by the census. Similarly, E. Willard Miller in 1960 reported that in a six county subregion in Western Pennsylvania 95% of the villages were not reported by the census. In 1946 D. G. Marshall published an article in which he indicated that for 75 the years 1920, 1930 and 1940, the census enumerated about 5% of all U. S. hamlets (1-250 in size), 54% of all the villages (251-499 in size), and 15-18% of all places smaller than 2,500 in size. The focus of most of this research has concentrated on smaller places (those less than 1,000 population) with the intent of determining the number and population of centers not enumerated by the census. It is unknown how representative census enumerated places are of the entire universe of places and no information is available to help determine the implications of relying on census enumerated places and their charac- teristics to generalize about the total universe. Past research has found size of place to be highly related to its likelihood of enumeration by the census. As a consequence the distri- bution of census places by size has probably not been very representative of the size distribution of the universe of places, i.e. the relative significance of larger places has been overestimated and that of smaller places underestimated. If this is true for size of place it may also be true for other characteristics of places as well. For instance, census enumerated places may not provide an accurate description of the number and relative significance of growing and declining places, urban and rural places and places with high and low proximity to larger places. It is the intent of this research to be able to provide data that can help determine the existence of such relationships. Summary of Methods The first step in the investigation was to identify Michigan popu- lation centers and determine their size. The major sources for this data collection phase, was the census of population (for legally incorporated places and/or those exceeding 1,000 in size), Rand McNally Commercial 76 Atlas and Marketing Guide (for legally unincorporated places less than 1,000 in size), and questionnaire responses from knowledgeable locali- ties. A total enumeration of Michigan population centers exceeding 74 in size at sometime from 1930-1970 was made. After having extracted places from the census reports they were compared to the Rand McNally listing and those places not already identified were then extracted. Missing data was then filled in from questionnaire responses, yielding a final total of 1,592 Michigan population centers exceeding 74 in size at sometime between 1930 and 1970, with 86 places having missing data for one or more of the years assessed. The data was arranged according to eight hypotheses which were operationalized by specific indicators. The hypotheses are as follows: HYPOTHESIS I: A positive relationship exists between the 1970 size of place and their likelihood of census enumeration, i.e. larger places are more likely to be enumerated by the census than smaller places. HYPOTHESIS II: A positive relathonship exists between change in size of place from 1930 to 1970 and their likelihood of census enumeration, i.e. growing places are more likely to be enumerated by the census than declining places. HYPOTHESIS III: A positive relationship exists between proximity of place to larger places and their likelihood of census enumeration, i.e. those places closer to larger places are more likely to be enumerated by the census than places that are funther away. HYPOTHESIS IV: A positive relationship exists between the ur- banity of an area within which places are located and their likelihood of census enumeration, i.e. places located in rural areas are less likely to be 77 enumerated by the census than places located in urban areas. HYPOTHESIS V: Compared to the total universe of state places, census enumerated places will over represent larger places and under represent smaller places. HYPOTHESIS VI: Compared to the total universe of state places, census enumerated places will over represent growing places and under represent de- clining places. HYPOTHESIS VII: Compared to the total universe of state places, census enumerated places will over represent those places closer to larger places and under represent places that are further away. HYPOTHESIS VIII: Compared to the total universe of state places, census enumerated places will over represent places located in urban areas and under represent places located in rural areas. Crosstabulations were made between the operationalizing indicators and source of enumeration. The analysis of the data was carried out in three phases. Phase I tested Hypotheses I-IV, Phase 11 tested Hypotheses V-VIII, and Phase LLItested Hypotheses Ia-VIIIa for places exceeding 74 and less than 1,000 in size in 1970. Major reliance was places on numeric and percentage distributions and comparisons. Summary of Findings It was suspected that the stated hypotheses would be supported by the findings. The degree of support for the hypotheses related to the uni- verse of places exceeding 74in size at sometime from 1930 to 1970 was quite high. As was suspected and stated in Hypotheses I-IV, larger places, growing places, places with high proximity to larger places, and places 78 located in urban areas are most likely to be enumerated by the census. Further testing of Hypotheses V-VIII revealed the consequences of census underenumeration in terms of the differences in percentage of places enumerated by the census and the percentage of the total universe of places for each of the specified categories. The findings clearly supported these hypotheses, revealing that the census under represents small places, declining places, places with low proximity to larger places and places located in rural areas and over represents large places, growing places, places with high proximity to larger places and places located in urban areas. The actual percentage of over and under representation by the census varied from indicator to indicator and from category to category, however, the overall support for the hypotheses was very strong. The high degree of support for Hypothese I-VIII suggested that the support would also be as high or higher when testing the hypotheses for places exceeding 74 and less than 1000 in size in 1970. The findings however, are somewhat different. The findings indicate a lower level of support for Hypothesis Ia-IVa than was indicated for the larger universe of places. Hypothesis Ia was strongly supported indicating that the larger places are more likely to be enumerated by the census than smaller places. Hy- pothesis IIa also was supported,however, not to as high a degree as was demonstrated for the universe of larger places. For the universe of small places, growing places are more likely to be enumerated by the census than declining places. Hypothesis IIIa, suggesting that places with high pro- ximity to larger places would be more likely to be enumerated by the census than places with low proximity to larger places, was found to have little support. Hypothesis IVa, suggesting that urban places are more likely to 79 be enumerated by the census than rural places, also had little support. Corresponding to these findings, Hypotheses Va-VIIIa revealed mixed support. It was found that census enumerated places did over represent larger places and under represent smaller places. Census enumeration also over represents small growing places and under represents small de- clining places. Little support was revealed for Hypothesis VIIa which suggests that the census enumerated places will over represent those places closer to larger places and under represent places that are farther away. Also little support was demonstrated for Hypothesis VIIIa, which suggests that census enumerated places will over represent places located in urban areas and under represent places located in rural areas. Again, the percentage distributions for these last two hypotheses were either extremely curvelinear or reversed from what was expected. The findings for the universe of small places indicategthat something more is happening in terms of representativeness than has been detected in this investigation. Several possible explanations could be suggested. First, it is possible that the indicators selected are inadequate for revealing relationships between proximity to larger places and census enu- meration and rurality and census enumeration. This however, seems un- likely. Second, it is possible that the census does a poor job of enu- merating places located in urban fringe areas. This is a likely possibili- ty given that new places tend to be established as a result of the high rates of decentralization and suburbanization occurring in these areas. Many smaller places do in fact exist in these areas. Table 34 indicates that 61.5% of places exceeding 74 and less than 1000 in size in 1970 are located within 50 miles of a city greater than 25,000 or in an SMSA county. Though it appears that this is a possible explanation, it must be stressed 80 that it is only speculation. Further research must be carried out to substantiate this. Contributions of Findings The scientific method, as it has evolved over the centuries has dictated the necessity of obtaining good data in any investigation that makes contributions towards the development of sound theory. The intro- duction of biased or other wise inaccurate data can have significantly negative effects upon findings and related contributions to both basic and applied knowledge. It is in this spirit of scientific investigation that this research was undertaken in order to either substantiate the validity or eluminate the deficiencies of what is perhaps one of the most widely used data sets for enumerating population centers in this country - the United States Census. The findings of this study indicate that biases exist in the census enumeration of population centers. The implications of these findings in terms of basic contributions suggest that if the concern of a piece of research is to study the total universe of places, the census does not give a very representative sample of places on the whole. These findings further indicate that for researchers interested in studying the total universe of places in terms of growth and decline, proximity of places to larger places, or rural and urban places, reliance on census data could be very midleading unless corrective procedures are incorporated into the investigation. Such procedures have not been dealt with in this study, however both numeric and percentage representations of census overenumeration and underenumeration in terms of selected place charac- teristics have been determined for the State of Michigan, which may be 81 of some help in the development of these needed corrective procedures. Contributions are also made in the furthering of comparative methods in community research. The findings presented here examine variables in a large number of population centers which yield a high degree of compara- bility and maximizes its usefulness for generalizing about certain aspects of community and place phenomena. The results and findings of this study also provide useful data for the professional community developer. If it is the desire of the CD agent to direct his efforts toward places that are declining in number, it becomes obvious that sole reliance on census data will yield misleading figures in terms of the actual number of places that have experienced either growth or decline. The productivity of these professionals would be enhanced if they were aware of the limitations of the data source. This problem of reliance on census data is even more obvious in terms of the development of national programs and policies directed toward community development. The census is the basic source of data that reports to government officials the state of the communities in this country. The negative results of introducing inaccurate data at this level could be great. To be aware of the limitations and implications of the basic data set seem invaluable not only for CD agents but also for program developers. It is hoped that the results presented in this study will aid those who use the census enumeration of places to become more aware of the limi- tations of the data set. It is also hoped that these findings may encourage the Census Bureau to reevaluate the definitional limitations now imposed on place enumerations. And lastly, it is hoped that this piece of research will become a meaningful part of the evolution of community research and will perhaps provide a theoretical framework for future research in this area. BIBLIOGRAPHY BIBLIOGRAPHY Davis, James A. Elementary Survey Analysis. New Jersey: Prentice Hall, Inc., 1971. Fry, C. Luther. American Villagers. New York: George H. Doran Co., 1926. Havighurst, R. J., and Jansen, A. J. "Community Research." Current Sociology XV (1967): 21-31. Jansen, A. J., and Havighurst, R. J. "Community Research." Current Sociology XV (1967): 21-31. Landis, Paul H. "The Number of Unincorporated Places in the United States and Their Estimated Populations." Research Studies of the State College of Washington 6 (December 1938): 160-188. Marshall, 0. G. "Hamlets and Villages in the United States: Their Place in the American Way of Life." American Sociological Review 11 (April 1946): 159-165. Miller, E. Willard. "Population and Functional Changes of Villages in Westgrn Pennsylvania." Pennsylvania Historical Magazine 43 (March 1960 : 59. Perkinson, Leon 8. Population Distribution in Non-Metropolitan Michigan. Report 145 Deve10pment and Public Affairs. East Lansing, Michigan State University Agriculture Experiment Station, (September 1971). Reiss, Albert J., Jr. "Some Logical and Methodogical Problems in Community Research." Social Forces 33 (October 1954): 51-57. Rodefeld, Richard D. Enumerating Michigan Population Centers (places) and Determining Their Size from 1930-1970; Rationale, Procedures, Pro- blems and Results. East Lansing, Michigan: By the Author, 1976. Simpson, Richard L. "Sociology of the Community: Current Status and Prospects." Rural Sociology 30 (June 1965): 127-149. Stein, Maurice, R. The Eclipse of Community: An Interpretation of American Studies. New York: Harper and Row, Publishers, Inc.,Tl964. Walton, John. "Discipline, Method and Community Power: A Note on the Sociology of Knowledge." American Sociological Review 31 (October 1966): 684-689. 83 Webster's SeventppNew Collegiate Dictionary. (1965), s.v. "Scientific Method." Whitney, Vincent Heath. "Notes on the Reliability of Atlases for Estimating the Populations of Unincorporated Places." Rural Sociology 10 (December 1945): 387-393.