nny-L up: ‘4‘“ I Viv-“fifi'V‘ u. .. v" . 1 Mi, '...L$\ . Z . rfl L «K aux-r '. a. A ‘ 2 Wang ”fie"? ‘ 0‘2". ' '5' I .' . .g £1“ ,1; —‘. (: _.',"- .4. -.'-,.- .~— - ' VI. .- .... I ,3 .. 1 .4.— . . Ira:- u? - ~ |. o~ .. . n ' “3.11:" .1 u... .. Ma'v-a H .AA .,J ' q ‘2'" Ha utm—M».l‘"‘ ' . -V ‘ tic—“:31 .o» , .— . r‘ . ‘ «.4 4 .. «J.— —:- ' 'g‘v‘ "N - .— — M...“ ‘ , """.~.. .327. .u -- 0. 1 ...._. ‘ {r ' H-._ q...«r~u . u...”- - ~ 4;- .2: "f... ¢ -._.- < ‘- unva;*,-rv I- 7—- “FA- .. “:3. "14—. ' H‘ ~a 4‘ -«n n ”go...- ”no I l fin: : II ‘ H m. L:- IiililiiiiiilliMWliiiliilliiiliiiiliiliililiiii 312930208 This is to certify that the dissertation entitled THE IMPACT OF COGNITIVE MAP-READING TASKS ON THE ABILITY TO NAVIGATE WITH A MAP presented by Amy Kathleen Lobben has been accepted towards fulfillment of the requirements for Ph 0 D o degree in Geography \i/JM‘J ”1%? fl“ C(fl'YQ / Vajor professor Date 4107/62//97 MSU is an Affirmative Action /Equal Opportunity Institution 042771 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 RECAU.ED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 0921196 THE IMPACT OF COGNITIVE MAP-READING TASKS ON THE ABILITY To NAVIGATE WITH A MAP By Amy Kathleen Lobben A Dissertation Submitted to Michigan State University In partial fulfillment of the requirements For the degree of DOCTOR OF PHILOSOPHY Department of Geography 1999 ABSTRACT The Impact of Cognitive Map-Reading Tasks on the Ability to Navigate with a Map By Amy Kathleen Lobben Understanding how people navigate with the use of a map requires researchers to examine the cognitive processes associated with map reading. This examination necessitates both identifying cognitive tasks and evaluating their relative influence. The objectives of my research were to identify the cognitive tasks associated with reading a map for navigation, to develop a testing device that would assess a person’s ability to navigate with a map, and to determine the relative influence of each task to map-reading and navigation. Five cognitive processes of map navigation were identified: object rotation, memory, visualization, “sleuthing”, and environmental navigation. An experiment was designed to measure subjects’ abilities in these five areas. A computer administered and scored test was developed to evaluate abilities for the first four processes and a within-building exercise was designed to evaluate the fifth. Subjects completed the abilities evaluation and participated in a real-world map navigation exercise evaluated on accuracy, hesitancy, and efficiency. An examination of the relationship between the scores from the testing device to the real-world map navigation exercise indicated that the Map Reading Ability Test successfully predicted map navigation ability. In addition, the relationship between individual test sections and specific parts of the real-world map navigation exercise revealed that three cognitive processes (visualization, spatial memory, and “sleuthing”) yielded variably significant influence in map reading and navigation. H 'MIO IV! EDGE MEN'S [Dc guidance and .‘i utihlilim generrmsly "fit!“ by MM”_ 4 i the course villus :lwncflmtm' 1: Appl’ttiiticd Her «my in“ ”I.” .' *3 ' though some at uni-1.. muddy Waters. H i -' V . . A: u tammam: nun-rt‘w. Richard (3er rim/uh! MM” findwsen' s friendship MM“ my WW: 4 “T my ovm ideas Wri’c ssart ie' invaluable to me ijt:\’Ul/:‘UUK m entire Phi); expertise m lcSi nit'xriv'p'i‘ m. valuation and emivm was M h. accent“ complain a: at i . . nun-n. Ttauumil in. .mL' a did \im. Amogant'. . influence mafia“ :‘ ‘ 'nlt 21mm: gi' Cdt't'slw m: -...i:\.;:I.-.~:1 pTIWNjcd by mm“ "it \C thcfi.l~fefstu)) charm-um“ Winn}. He tawny-1 : - _ m selfless support .. ' hub that!!!) ‘IV'IT-e-i'hhurj'x‘. mow. .' .. ' i, 4"? 'i‘ “I l i \f ' AMYKA’I‘HIJZEIII , a ., . “ ., ' v11 . . - - 3v," .. . " ~ - , - . ,- . ‘ up." mum. , .u'avmmmemaaum'm ACKNOWLEDGEMENTS The guidance and inspiration generously offered by Judy Olson through the course of this dissertation is appreciated. Her many insights kept me focused through some, at times, muddy waters. As a committee member, Richard Groop provided new perspectives when my own ideas were scarce. Jeff Andresen’s friendship and encouragement were invaluable to me throughout my entire Ph.D. journey. Richard DeShon’s expertise in test development, evaluation, and analysis was pivotal in the successful completion of this project. Randall Schaetzl’s and Alan Arbogast’s influence on my education was profound. They generously shared with me their vast knowledge of and enthusiasm for geomorphology and I was introduced to a field that I now truly enjoy. The thorough educational foundation provided by Borden Dent is greatly appreciated. He taught me to love the field of study that I now teach with enthusiasm. The selfless support, love, flexibility, and encouragement offered by Scott Clark made the completion of this dissertation and degree possible. Finally, to my children Marshall and Evan Clark who have accompanied me from the start down the long road and to Benjamin Clark who just joined the trip, you have my eternal and unconditional love. You make everything possible. TABLE OF CONTENTS List of Tables List of Figures Chapter 1 Introduction Research Questions Research Objectives Structure of the Dissertation Chapter 2 Literature Review Differing Approaches to Cognitive Map Reading Studies Map Reading Strategies Cognitive Processes The Cartographic Process The Map-Reading Process Thematic vs. Reference Maps The Map User as Part of the Cartographic Process Cognitive Tasks of Navigational Map Use Spatial Memory Object Rotation Symbol Identification Wayfinding Environmental Navigation Visualization “Sleuthing” Summary xi 9““ 11 ll l3 14 16 17 20 23 24 25 26 27 28 Chapter 3 Research Structure and Methods Research Structure Methods Design of the Experiment The Ability (Predictor) Test Object Rotation Visualization Sleuthing Spatial Memory Focus Group Evaluation Environmental Navigation Test Real- World Navigation Test Test Administration Preliminary Testing Subjects Chapter 4 Results and Analyses Results Map Reading Ability Test Object Rotation Visualization Sleuthing Spatial Memory Environmental Navigation Real-World Map Navigation Exercise Analyses Calculating Scores Map Reading Ability Test Computer Test Object Rotation Visualization Sleuthing vii 30 30 31 31 35 38 41 43 46 48 55 56 57 58 58 59 61 63 65 68 69 71 71 71 71 72 74 75 Spatial Memory Environmental Navigation Real- World Map Navigation Exercise Test Evaluation Reliability Test Validity Criterion Validity Construct Validity Object Rotation Visualization Sleuthing Spatial Memory Environmental Navigation Chapter 5 Discussion Predictor Test Object Rotation Spatial Memory Visualization Sleuthing Environmental Navigation Improvements on the Study Chapter 6 Summary and Concluding Remarks Significance of Research Suggestions for Further Research Bibliography viii 76 77 78 79 80 82 83 88 90 94 96 99 103 108 109 110 111 112 113 113 117 119 122 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 4.19 4.20 LIST OF TABLES Example Object Rotation results for 5 subjects Example Visualization part 1 results for 5 subjects Example Visualization part 2 results for 5 subjects Example Sleuthing part 1 results for 5 subjects Example Sleuthing part 2 results for 5 subjects Example Memory results for 5 subjects Example Environmental navigation results for 5 subjects Example Real World Map Navigation Exercise results for first 5 subjects Example Environmental Navigation scores for first 5 subjects Example Real World Map Navigation Exercise for first 5 subjects Reliability results for Visualization Reliability results for Object Rotation Reliability results for Sleuthing Reliability results for Spatial Memory Multiple Regression Factor analysis results Linear regression with predictor test dependent variable and sleuthing independent Linear regression with number of rotations as dependent variable and object rotation dependent Linear regression (rotations/object rotation) with leverages removed Multiple Regression 60 62 63 64 65 67 68 70 77 79 81 81 82 82 84 86 87 91 93 93 4.21 4.22 4.24 4.25 4.26 4.27 4.29 4.30 4.31 Linear regression with number of rotations as dependent and spatial memory independent Linear regression with number of stops as dependent variable and visualization as dependent Linear regression with locating dependent and sleuthing independent Linear regression with locating as dependent and all test sections as Independent Linear regression with duration of looks as dependent variable and spatial memory as independent Linear regression with duration looks as dependent and all test Sections as independent Linear regression with duration of looks as dependent and object rotation as independent variable Linear regression with duration of looks as dependent and object rotation as independent variable (leverages removed) Linear regression with duration of looks between location #3 and #4 as dependent variable and environmental navigation as dependent Multiple regression with look4 as dependent variable and all test sections independent Linear regression with duration of looks between location #3 and #4 as dependent variable and memory as dependent 93 95 97 98 100 101 101 102 105 106 106 2.1 2.2 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.14 3.15 3.16 3.17 3.18 3.19 4.1 LIST OF FIGURES Model of the cartographic process Model of cognitive processing First screen in the introduction to the Map-Reading Ability Test Sample screen from the object rotation section Sample screen from part 1 of the visualization section Legend explanation Sample screen from part 2 of the visualization section An example map from part 1 of the sleuthing section An example photograph from part 1 of the sleuthing section Example map with arrows Sample screen from part 2 of the sleuthing section Highlighted route in memory section Directions corresponding to travel route Focus group questions Summary of focus group responses Map layout of floor and location of numbers Environmental navigation scoring sheet Map used in the RWMNE Locations of the locating task Locations of the 4 navigation destinations RWMNE score sheet Histogram of Object Rotation Scores 15 17 33 34 36 36 37 38 39 40 41 42 43 44 45 47 48 51 52 53 54 73 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.16 4.17 Histogram of Visualization Scores Histogram of Sleuthing Scores Histogram of Spatial Memory Scores Histogram of Environmental Navigation Scores Scatterplot with visualization and sleuthing Scatterplot with RWMNE and sleuthing Scatterplot with rotations and Object Rotation Scatterplot with rotations and Object Rotation (leverages removed) Scatterplot with rotations and memory Scatterplot with stops and visualization Scatterplot with locating and sleuthing Scatterplot with looks and memory Scatterplot with looks and object rotation Scatterplot with looks and object rotation (leverages removed) Scatterplot with look4 and environmental navigation Scatterplot with look 4 and memory 74 75 76 78 85 88 91 92 94 96 98 100 102 103 105 116 Chapter I Introduction Why can some people read maps and navigate through an environment better than others? The answer to this question has eluded researchers working in both cartography and psychology. Although cognitive studies and spatial ability measures have been conducted for over 100 years by psychologists and for 30 years by cartographers, as recently as 1987, Blades and Spencer observed that “as yet there has been no analysis of the cognitive processes which are involved in using a map” (p.65). Although some studies provide insight into the cognitive processes associated with specific map reading tasks, this statement remains true. To begin to understand these processes, researchers must first identify the tasks used in map-reading and only after those tasks are identified can the more complex process of determining their relative influence be evaluated. In short, we must know the object of study before it can be studied. Much cartographic research is guided by the cartographic communication process, which includes not only the map-maker but also the map-viewer. Over the years, cartographers have sought to understand ways in which map making may be improved, enhancing the effectiveness of the entire communication process. Many would argue, however, that even if cartographic research has made significant contributions to the map design process (itself a controversial assertion), we still lack a general understanding of how a map-reader reads and cognitively processes maps. Understanding how a reader processes the perceived map information is essential, as the map user is a vital component of the cartographic process. It is this need to gain a better understanding the cognitive processes of map reading that provides the research arena for this dissertation. Research Questions With navigation maps as the specific focus of research, the following specific questions will be addressed: What are the cognitive tasks associated with reading a map for navigation? What is the relative influence of each task to map-reading and navigation? Can a testing device be developed that will predict a person’s ability to navigate with a map? The research and questions addressed in this dissertation begin to collectively focus on some potential cognitive tasks of map reading and their relative influence on navigation with a map, an important function to spatial science. All geographers study the spatial environment in some way. While some, such as geomorphologists, may restrict their studies to only the physical environment, others, such as human geographers, study people and their interaction within space. Interacting with the environment necessitates the development of maps, both mental and physical. As a result, “it is the need to communicate about spatial information and to understand behaviors taking place in environments that has caused geographers to be interested in cognitive maps” (Lloyd 1997, p.1). Research Objectives The goal of this research is to evaluate whether the ability to perform certain cognitive tasks affects ability to navigate in the real-world environment with a map. If we can use a sit-down test to successfully differentiate between map users on their abilities to perform these tasks, and if performance on these tasks correlates with actual navigation ability in a real-world environment, we should have a test that can predict navigational map-reading ability in practical circumstances. Such a test should be useable to predict whether individuals will likely be able to handle requirements of certain jobs such as those of census taker, delivery person, or police officer. But such a test would also be of considerable theoretical interest as we try to understand differential spatial abilities and what hinders some individuals from being able to perform wayfinding with maps. In the long run, it could affect how we train people in navigation and map reading. Structure of the Dissertation The remainder of this dissertation is structured into five chapters addressing literature review, methodology, analysis, discussion of results, and conclusions. Chapter 11 includes a review of previous literature as it relates to the development of and support for this research. Chapter III documents the methods employed while conducting the focus group, pilot test, and structured experiment. Chapter IV includes data and their analyses, while Chapter V discusses these analyses. Finally, Chapter VI contains concluding remarks and implications for future research. Chapter 2 Literature Review Research into cognitive processes in map reading has been conducted primarily in the fields of psychology and cartography. Researchers borrow theories and study results from one another’s discipline, but with a few exceptions, most notably the work of Downs and Lieben (1986, 1987, 1988[2], 1989[2], 1991, 1992, 1993, 1994, 1997), most of the cognitive research has been conducted either by psychologists or by cartographers but not by both in collaboration. While studies found in the literature of the two disciplines address the same topic (map-reading ability), the actual foci of the studies are quite different. In psychological research, the primary focus is on human cognitive abilities. Less attention is given to the type (and sometimes even the quality) of the maps used in the studies. In other words, they often do not differentiate between different types of maps (Lieben & Downs 1989), and “maps” are sometimes crude approximations of the street and road maps we generally use to find our way in the world. Cartographers, on the other hand, focus on the map as well as the map reader but lack a deep understanding of cognitive psychology and its theoretical constructs. With few exceptions, cartographic research has not been based on cognitive theories of spatial knowledge (Lloyd 1982, Blades & Spencer 1986, Lieben & Downs 1989, Crarnpton 1992). Blades and Spencer argue that “if cognitive cartography is to establish a firm scientific foundation it needs to be ‘theory driven’. It is important that cartographers develop theories about the processes involved in understanding a map and that these theories lead to hypotheses which can be tested experimentally” (Blades & Spencer 1986 p.7). In addition, the methods used in many cartographic experiments make validity measures difficult to obtain (Lloyd 1982). In fact, terms such as test validity, test reliability, or construct validity are rarely seen in cartographic literature. Cognitive map-related testing has been conducted sporadically over the past 100 years. In psychology, the line of study that brings maps to mind most immediately is spatial abilities assessment. Galton, in 1883, developed the first spatial abilities’ assessment tests in 1883 (Anastasi & Urbina 1997). Since then, thousands of such tests have been developed and administered (Eliot & Smith 1983), many of which have no stimuli that we would label as geographic maps. The 19705 saw an increase in experimental psychology that focused on understanding the constructs assessed by intelligence and aptitude tests (Anastasi & Urbina 1997), some of which addressed map reading ability. For example, Bronzafi et al. (1976) asked subjects to use a map of the New York City subway to navigate to a given destination; subjects’ success was characterized as either ‘acceptable’ or ‘not acceptable.’ Some researchers have used computer simulation to test the effectiveness and efficiency of route planning (Hayes-Roth & Hayes-Roth 1979, Streeter & Vitello 1986). Others addressed the influence of map orientation on route planning (Levine et a1. 1984, Shepard & Hurwit 1984). The spatial testing conducted by psychologists focuses primarily on understanding spatial cognition as part of the overall cognitive process. These tests helped to set the precedent for several cartographic cognitive studies that focus on understanding spatial cognition as part of the overall cartographic communication process, which has been a concern in the field for nearly half a century. After a period of concept development that shified cartography from questions of physical representation to questions of map effectiveness, cartographers asserted that research should focus not only on the needs of the map reader but also on their map-reading skills and abilities. In addition, cartographers recognized the importance of the map reader’s interpretation of the map information as an integral component in the communication process (Kolacny 1969, Ratajski 1973, Morrison 1976). Kolacny (1969) argues the importance of understanding the map reader’s cognitive process by suggesting that researchers should not just focus on the reader’s map-use needs but also their skills and abilities. Ratajski (1973) discussed cognitive processes of the map reader more specifically by posing that a reader’s memory affects map reading and interpretation ability. Morrison (1976) introduces even more specifics by suggesting that a map reader’s interpretation of map information is influenced not only by the cognitive processes used during the interpretation of map information but also by “the information existing previously in the cognitive realm of the map reader” (p. 96). Beginning in the 1950s, psychophysical studies focused on identifying map reader contributions by evaluating their subjective impressions of various map design elements (text, colors, symbols, and layout) (Cole 1981). Subject matter included magnitude estimation of graduated circles (Flannery 1971, Groop & Cole 1978), map symbol clustering (Jenks 1975), pattern complexity on choropleth maps (Olson 1972, Monmonier 1974) and gray scale effects (Kimerling 1975). Psychophysical studies provided results of interest for map design but provided little insight into how the reader processes the information. As attention shifted to processing, cartographers began calling for more experiments that focus on identifying the cognitive processes associated with map- reading (Olson 1979, Gilmartin 1981, Blades & Spencer 1986). This shift in focus broadened the scope of research as cartographers began to look at more components in the map reading process. Although the scope of research broadened, most research into understanding the cognitive processes of map-reading still is conducted primarily in psychology, which means that the focus of map-reading cognitive research continues to focus on cognition and not the map. Differing Approaches to Cognitive Map Reading Studies Recent cognitive map-reading experiments (the majority of which have been conducted by psychologists) may be divided into at least two categories, those that investigate strategies used by the map reader and those that focus on cognitive processes used by map readers during the task of map reading. Map Reading Strategies A map reading strategy is a specific method or tactic used by a map user in the process of reading a map. A strategy is not independent of a cognitive process. In fact, a person’s cognitive processes may influence the strategies or methods used in reading a map. For example, a person whose visual memory is not as strong as reasoning ability may not rely on memorizing the map as much as on analyzing the map, or a person possessing a strong mental object rotation ability may not physically rotate the map during navigation. A distinct difference between strategies and processes is that everyone has and uses the same processes, but at varying levels of ability or effectiveness. However, the difference in an individual’s ability levels in varying processes may dictate the strategies they use. Thus, everyone uses the same processes, but not everyone uses the same strategies. Many researchers use a ‘think aloud’ method for investigating map reading strategies (Blades and Spencer, 1986). They present a subject with a map and ask them to think aloud, or describe how they are learning the map. For example, Thomdyke and Stasz (1980) asked subjects to look at a map for a given amount of time and then reproduce the map from memory. While studying the map, subjects explained the strategies they used (memorizing one area at a time or memorizing hierarchy, for example). The researchers found that certain strategies resulted in more accurate map learning. Additionally, Gilhooly, et a1. (1988) and Crarnpton (1992) used the ‘think aloud’ protocol to investigate strategy differences between expert and novice wayfmders, asking subjects to explain how they are learning the map. Investigating strategies does not always involve the ‘think aloud’ method, however. Kulhavy, Schwartz, and Shaha (1983) studied strategies people use to remember map information by presenting subjects with three versions of a single map and then asking them to perform a recall task followed by a recognition task. The first map version contained features labeled with text only, the second version contained text in addition to mimetic symbols, and the final version contained text plus geometric symbols. Because subjects performed better on the task when faced with the text/mimetic symbols and the text/geometric symbols maps than they did on the text only version, the researchers concluded that map viewers rely on visual memory to learn and retrieve map information. Cognitive Processes Researchers investigate cognitive processes more frequently than strategy. The methods used in cognitive process experiments differ from those used in cognitive strategy experiments. Most often, the researcher has identified a specific cognitive process used in map reading and a subject’s level of task performance is assessed relative to other subjects’ performance on the same given task. These methods usually result in researchers presenting a subject with a given task, series of tasks, or series of questions, designed to reflect their ability level in a given cognitive process. For example, several researchers have designed experiments to reflect the relative influence of object rotation on map reading ability. These experiments investigate whether a person must perform a mental rotation when the map orientation and direction of travel are not aligned (Levine, Marchon, and Hanley 1984, Lloyd and Steinke 1984, Shepard and Hurwitz 1984, Aretz and Wickens 1992). The results of these studies indicate that object rotation is a cognitive task used during map reading and may influence a person’s map learning/reading speed and accuracy. This influence is especially profound with more complex maps because as maps increase in complexity, the map reader is less able to mentally rotate the entire map and the rotation process is abandoned and replaced with an analytic process (Aretz and Wickens 1992). Additionally, researchers have studied the potential relationship between cognitive ability measures and map reading ability. The results are consistent between studies and indicate that higher scores on general mental/cognitive ability tests correlate with higher scores in map reading ability tests (Thomdyke and Stasz 1980, Sholl and Egeth 1982, Lloyd and Steinke 1984, Streeter and Vitello 1986, Kovach, Surrette, and Aamodt 1988). Many of the cognitive studies conducted by geographers are carried out under the general heading of “cognitive mapping”. One of the earliest definitions of cognitive mapping was presented by Downs and Stea (1973), who define it as “a process composed of a series of psychological transformations by which an individual acquires, codes, stores, recalls, and decodes information about the relative locations and attributes of phenomena in his everyday spatial environment” (p.9). Studies such as the ones referenced thus far often focused on specific issues in an attempt to understand the nature of cognitive mapping. The Cartographic Process The Map-Reading Process Map reading is a cognitive process involving several tasks. A person faces a cognitive task any time information must be deciphered. Cognitive models include the intellectual processes used during task performance and include methods of acquisition, representation, organization, storage, retrieval, and use of information (Anastasi & Urbina 1997, Lloyd 1997). Cognitive processes individuals use to solve a given problem (such as those in map reading) will differ from task to task. The information being processed may be received via sources such as verbal or written instructions, from social interaction, from navigation maps, or from an internal source such as stored memory. While maps provide an external source of information that must then be subjected to cognitive processing, they are almost always combined with a person’s mental map. Mental maps are developed from direct exposure to an environment (repeated trips to and from a grocery store, for example) as well as from representational sources (travel guides, novels, maps, and oral directions to destinations). As a result of differing experiences, individuals process dissimilar information and whether the experiences are considerably different, such as two people taking different routes to the same destination, or only slightly different, such as one person riding in the front seat and one in the back seat of a car, they cause individuals to develop different mental models. These different models may contribute differently to map-reading ability. The term ability has been purposefully selected here. There has been a recent change in the usage of terms associated with psychological assessment. The terms aptitude and achievement were used previously, and aptitude tests “measure the effect of learning under relatively uncontrolled and unknown conditions, whereas achievement tests measure the effects of learning that occurred under partially known and controlled conditions” (Anastasi & Urbina 1997, p.475). In other words, both terms have referred to individual learning where aptitude tests reflected the cumulative learning across a lifetime and achievement tests measured accomplishment as a result of a relatively controlled learning experience such as a college course. Drawing a clear distinction between aptitude and achievement becomes increasingly more difficult, however, as term usage appears slightly muddied in test designs and applications. Some tests classified as ‘achievement’ tests may actually reflect broad educational experiences while some ‘aptitude’ tests rely on fairly controlled prior learning. Regardless of it’s categorization, any cognitive test provides some type of ability measure and, therefore, the term ability ofien is being used in place of the previous terms, aptitude and achievement (Anastasi & Urbina 1997). Thematic vs. Reference Maps Since maps provide an external source of information and imply various cognitive tasks, depending on the type of map, it is important to differentiate between the different types of maps. Thematic maps present spatial data or themes, usually one data set per map (population, corn production, deaths by cancer, or bird migration routes, but not all four on one map) or a combination of data sets related in some way (population and percentage of population over 65, education and income, price of electricity and electricity consumption). Thematic maps seldom depict individual items (for example, a population thematic map does not show locations of individual people) but rather a distribution over space. Reference maps are used to find location, direction, or relative position of specific features. They do represent individual features such as national parks, roads, railroads, and cities. Reference atlases contain whole collections of reference maps at relatively small scales and national governments normally engage in extensive production of large scale reference maps known as topographic sheets. While many maps are generally identifiable as thematic or reference, some maps lie within the continuum that exists between the two categories. A navigation map is an example as it serves a specific purpose, navigation, and, therefore, arguably may be “thematic.” However, navigation maps do not represent a distribution over space; rather they show individual items and, therefore, must be characterized as closer to reference maps than to thematic maps. The use of the map and application of the information received from the map differs considerably for the two extremes of the continuum. The thematic map requires the reader to process and analyze the presented information, drawing conclusions about the spatial patterns, resulting in a greater understanding of the spatial phenomenon presented. There is little urgency with thematic maps; a leisurely contemplation is not only acceptable but perhaps even intended. The reference map, or at least the navigation (road or street) map, requires readers to apply directly the information on the map to the real world; readers must often use the map while in the process of navigation. Processing of information must be relatively instantaneous, and it must be accurate or the reader will get lost. Since different classes of maps are used differently by the map reader, different cognitive tasks may be associated with each type. As a result, researchers must choose carefully the type (as well as the design quality) of maps used in experiments. The Map User as Part of the Cartographic Process Current models of the cartographic process include the map reader as an integral component, as illustrated in the model in Figure 2.1, adapted from Dent (1996). In this model, information is transformed at lease twice, from reality to the map (transformed by the cartographer) and then again from the map to perceived reality (transformed by the map viewer). Attaining an understanding of how the map reader processes map information may allow cartographers to develop more effective map design methods, which would be an improvement of the first transformation process (reality to map). Feedback 6— Map Legends —9 <— lnset Maps —> Data Map Author/ M M Field Cartographer —9 ap _9 ap Reader Transformation Transformation Two Map Making Map Use e———— Cartographic Process ————9 Figure 2.1 Model of the cartographic processes (alter Dent 1996) Understanding map information processing may require researchers to break apart the individual stages involved in cognitive processing (receiving, transforming, and using information), allowing the cartographer to investigate the relative importance of individual tasks used in the map-reading process. The first step, then, is to identify those individual cognitive tasks associated with navigational map reading. fid'usfi Cognitive Tasks in Navigational Map Use The previous section argued that different types of maps require that the map reader perform different tasks. But the map reader may also need to perform different tasks at different times with the same map. This section will present some of the varied tasks performed when using a navigation map. Kulhavy and Stock (1996) suggest that map learning involves two general cognitive factors, a memorial system, which determines the amount and quality of information retained from the map, and control processes, which govern the strategies the map reader utilizes in learning the map (the task they used). Both control processes as well as memory are factors included in information processing (cognitive) models. An example of such a model is shown in Figure 2.2 and is adapted from Gagne (1974). This model helps illustrate the importance of cognitive studies. In the model, the environment represents not only the physical environment in which a person navigates, but also the map. The receptors (ears, eyes...) receive the information from the environment and the response generators lead to the reaction which is carried out by the “effectors” (which allow a person to relay processed information back to the environment). The steps in between receiving and reacting are represented by processes and memory, both of which govern the transformation of information and dictate the reactions that will take place. In the case of map reading and navigation, the receptors take in the information and the processes and memory affect the response (navigation). The processes and memory will cause different people to navigate at varying levels in the same environment with the same map. Therefore, it is these mental processes that are studied in order to understand differences in map reading abilities. Response (——" (_' Efl‘ectors—lé— Generatfl Long-Tenn Memory Rece tors Processes Short-term _) p ’—') _) Memory 1 Figure 2.2 Model of cognitive processing (from Gagne 1974) Spatial Memory A person has a limited system for memorial recall, and levels of retention and recall differ from person to person (Miller 1956, Gilhooly et a1. 1988). People encode and store objects in spatial memory and they perform this process at varying levels of effectiveness (Stevens & Coupe 1978, Kulhavy, Schwartz, & Shaha 1983, Lloyd 1989, MacEachren 1992a). In fact, people may use differing strategies to encode a map while navigating. They may encode the map information either as a map image or as procedural knowledge. A map image is created through visual processing where the map is processed into memory similar to the way our visual system processes the actual map. Procedural knowledge is considered verbal information and the map information is processed in verbal terms or like written descriptions of the environment or route navigation information (Lloyd 1997). Retaining map information as either a map image or as procedural knowledge represents two different strategies used by map readers, and Lloyd (1997) also identifies two theoretical cognitive structures used for encoding spatial information. These structures differ depending on whether the information will be used temporarily (object files) or for problem solving and planning (mental models). People “open” object files in order to identify or categorize objects they encounter. Object files are often temporary and open, and then close as soon as the object is identified. Sometimes an object may be altered; in that case, a new file is not opened, but the old one merely changed to incorporate the new information. For example, an object file contains information about a friend. This information is used to recognize that friend when encountered. If, however, the friend changes hair color, a new object file is not created; the information is simply updated in the original file. If we accept the notion of object files, they must be used frequently in navigation. Say for example you are heading north on a street and you know, according to your map, that you should turn west at the town elementary school. If you see a building in the far distance, you will open an object file for that building and determine whether it is the elementary school. As you approach the building you see a sign representing the well known, trademarked, McDonald’s “golden arches”. You know the building is not the school and you reject it for the object file. You see another building and as you approach it, you see a flagpole and a playground. You determine that it could be the school. As you get closer you can read the sign identifying the building as the town elementary school (it belongs in the file) and you turn west and close the object file. A mental model, on the other hand, “is a cognitive structure that has been encoded to represent one’s spatial knowledge of an environment” (p.86). It is the brain’s storage and organization of information we have acquired about our environment. It may be created at a single moment in time, but is updated as more information is gained. Also, it may result from experience in an actual environment, from viewing a map, or from reading a description and may represent real or fictional space. The mental model allows us to solve spatial problems and provides information that allows us to plan spatial tasks. Spatial memory affects the efficiency with which information acquired from maps is learned and retrieved (Levine, Jankovic, & Palij 1982, Kulhavy, Schwartz, & Shaha 1983, Peterson 1985, Lloyd & Steinke 1986). One could argue that because spatial memory influences how well a person encodes map information, differing memorial processing abilities may affect individuals’ map-reading ability. Although how well someone can remember a map may influence map reading and navigation, map learning and map reading for navigation are separate aptitudes (Goldin and Thomdyke, 1981) and, therefore, a high ability in spatial memory (photographic recall, for example) may not necessarily result in high map reading and navigation ability. The second general cognitive factor proposed by Kulhavy and Stock, control processes, operates to guide the acquisition of information and is based on the map reader’s prior knowledge, which has accumulated in the memorial system. These processes are associated with specific map reading tasks and differ from spatial memory as they govern map-reading strategies, or how the map reader acquires and uses the information. A survey of previous research has produced no study that identifies the underlying control processes and their associated tasks. However, drawing from a number of study results, one can piece together some of the cognitive tasks that may be associated with specific constructs of navigational map-reading ability. Object Rotation Many experiments conducted in the area of spatial ability focus on determining the effects of a single cognitive ability, as it is associated with a particular map-reading task. Some researchers have geared their studies to trying to determine the relevance of object rotation as a cognitive task of map reading (Hintzman, O’Dell, & Amdt 1981, Steinke & Lloyd 1983, Lloyd & Steinke 1984, Aretz & Wickens 1992). Aretz and Wickens (1992) designed experiments to determine whether mental rotations are required in map use and whether map complexity affects rotation 20 efficiency. They showed subjects sets of two side-by-side maps and asked subjects to determine whether one of the maps was correctly rotated (as opposed to flipped or mirrored). Their results indicate that rotation is a process affecting map reading and map complexity affects the efficiency of rotation. Shepard and Hurwitz (1984) asked subjects to use computer simulated travel paths in order to determine the correct direction of travel on turns (left or right). The location of the turn was not always in an “up” location on the map (meaning that the direction of travel on a path when the turn is encountered is not “up”) and their results indicate that when the turn was not “up,” response times increased. In fact, response times increased linearly as the angle of the travel path increased from “up.” They concluded that subjects had to first mentally rotate the map to the “up” position before they could decide which direction to turn. Levine, Marchon, and Hanley (1984) investigated the affects of rotation on “you-are-here maps.” Similar to previously discussed results, they discovered that response time increased and task completion was less accurate when the maps were misaligned, or not in the forward-up position. The forward-up position is probably more effective even when the task performed involves geographical directions (Livingstone 1992). Lloyd and Steinke (1984) asked subjects to view a series of maps that had been rotated and possibly flipped and determine whether the map was “correct” or a “mirror” (flipped) version. Their results indicate that maps containing more information required longer response time when rotated from north-up. 21 Mentally rotating an object appears to be an integral process associated with reading paper maps since most paper maps are not multi-oriented, meaning that a map will maintain only one direction toward the top of the map. Cartographic tradition dictates that north is most always oriented toward the top. Since map users are familiar with this protocol, “orientation bias” occurs when they may be faced with a map that is not in north-up orientation. Studies (MacEachren 1992, Lloyd and Cammack 1995) show that maps may be learned so they are orientation free, but the accuracy and amount of time taken to encode information from multiorientation is greater as compared to similar maps that are in north-up orientation. Real life navigation will almost always require a map user to travel in a direction other than north (or whatever direction is oriented toward the top of the map). Therefore, when using a traditional north-up oriented map, readers must rotate the map mentally and maintain the actual map and its objects, including text and symbols, in its original orientation, or they must physically rotate the map and then mentally rotate the text, symbols, and other objects included on the map. Similar to the “you-are-here map” findings, studies have shown that map readers perform paper map-reading tasks more accurately and faster when the map is physically rotated so the direction of travel is oriented toward the top of the map (Aretz & Wickens 1992). That finding suggests that mental rotation of text and symbols is less difficult than mental rotation of map geometry. Navigation errors that result from incorrect object rotation are typically relative errors. These errors result when the frame of reference on a map is manipulated as a result of rotation, translation (when frame of reference is altered), or 22 scaling of the map during the mental encoding process (Lloyd 1989). These navigation errors that occur as a result of object rotation ability may be alleviated with the use of in-car navigation system maps, which are constantly and automatically rotated. Since most map readers will need to travel in a direction other than the one that is “up” on the map, they will most likely be faced with the task of object rotation. This task will not be encountered every time a person uses a navigation map but, when presented, may have a significant influence on a person’s ability to benefit from the map. Symbol Identification Since the purpose of a map is to show a scaled representation of the real world, all maps contain symbols (Robinson et a1. 1995, Dent 1996). Symbol identification is a task facing every map reader and the ability to understand symbolization relies on the ability to differentiate between symbols and understand that they represent real three-dimensional environmental objects (Bluestein and Acredolo 1979). The complexity of the map affects the encoding of symbols and information into the spatial memory system (Winn & Sutherland 1989). The number of symbols a map reader can encode varies according to the individual spatial memorial system (Winn 1991). Studies have shown that map readers of nearly any age (preschool to adults) seem to be able to decode the relatively simple symbols contained on most maps 23 (Blaut & Stea 1971 , Meyer 1973, Blaut & Stea 1974, Bluestein & Acredolo 1979, Ottosson 1988, Freundschuh 1990, Lobben 1997). Therefore, symbol identification may be a poor discriminator between better and worse map readers, but it is basic to one’s ability to effectively read and navigate with a map and is faced every time a person reads one. Wayfinding Navigation is a necessary part of nearly everyone’s life and, as a result, most people will become familiar with at least one geographic area to the extent that they become capable of route-planning and wayfmding. They will find their way through an environment using a variety of strategies and guides (such as verbal or written instructions, maps, or memory based on previous exposure) to aid them. Wayfinding is the task of developing a route or finding a way through an environment. Understanding the cognitive processes associated with the task of wayfinding is necessary if researchers are to understand the processes involved in solving spatial problems (Crarnpton 1992). People will develop either survey or route knowledge, depending on how they learn the environment. Survey knowledge provides a user with a bird’s-eye view of the geometry of the environment and is gathered through graphic representations (maps) or verbal representations (such as written or oral directions). Route knowledge results from the user ‘seeing’ the environment from an on-the-ground perspective and is gathered by traveling in the environment without a map (exploring, 24 or continuous exposure). Lloyd et al. (1995) refer to these viewing vantages as internal perspective (survey knowledge from representations) versus external perspectives (route knowledge from experience in the environment). They distinguish between external perspective as viewing an object from a “fixed vantage point” such as a map and internal perspective where the “viewer and the objects are part of the same space” (p. 5). Lloyd (1997) further suggests that the internal/extemal paradigm is not the only influence perspective has on developing cognitive maps. When viewing a static map, the reader maintains a single perspective, relating geographic space along horizontal and vertical axes as opposed to many perspectives experienced as one navigates through real geographic space. Research shows that knowledge gained by repeated exposure to an area (route knowledge), as opposed to exposure to representations (survey knowledge), results in more accurate development of a mental model of an area (Thomdyke & Hayes-Roth 1982). Additional research shows that increased exposure to an environment may lead a person eventually to develop survey knowledge, thereby enabling the person to visualize a more complete picture of the area (Golledge 1992). Environmental Navigation. Learning an environment through the use of a map and learning an environment through repeated exposure are different tasks. While both constitute wayfinding, the task of wayfmding without a map will be referred to as environmental navigation. This task does not immediately appear to have a direct 25 effect on a person’s map reading, but a person’s ability to navigate, or to develop route knowledge “on the fly” may profoundly influence how well they can navigate with a map. Although this task occurs independently of the map, if a person can quickly encode an accurate mental model, they may be able to develop a “sense of direction” or a sense of “where they are” on the map. They may have the ability to reorient themselves if confronted with a part of the environment from a different angle, and, after initial exposure to an area, they may need to refer to their map less often. Visualization. For persons required to navigate in unfamiliar territory, the survey knowledge provided by a map provides the map reader with an aerial view allowing them to ‘see’ what lies ahead. According to Crampton (1992), in order to visualize the area, the map reader must mentally transform the two-dimensional map into a three- dimensional form and “see” its characteristics and objects (morphology, streets, and buildings). This act of seeing with the mind’s eye, or developing a mental representation as a result of seeing a visual image, will be referred to as visualization. Visualization is a cognitive process and allows a person to identify patterns and impose order (MacEachren, Buttenfield, Campbell, DiBiase, & Monmonier 1992). This process lets map readers “see themselves” on the map and place themselves in the real world environment. It is a map-to-environment (or, more accurately, a ‘ " to -... ' ‘) task, meaning that the map reader (or r person hearing verbal directions or otherwise receiving spatial information) takes 26 information from the representation and applies it to the environment by visualizing and predicting areas yet unseen and recognizing those areas once confronted. Visualization can be thought of as continuous as opposed to discrete; it is a task conducted by map users while they are actively reading the map and navigating. While moving through an area (and a map) the user is constantly changing locations in the real world as well as on the map and must presumably continuously visualize what lies ahead in the environment. “Sleuthing” Although not specifically mentioned in the literature as it relates to navigational map use, a person’s “sleuth ability” also influences map-reading. In this context, the term “sleuthing” refers to a person’s ability to effectively relate the clues on the map to the represented real world. Similar to a detective solving a crime based on clues, map readers solve the problem of determining their location on the map based on recognizing real world landmarks and relationships (clues), putting those clues together, and placing themselves on the map. A person executes this task when observing objects in the environment such as a grocery store, a gas station, and a library and using them to locate their correct relative position on the map. Sometimes these features are not even on the map, but clues are pieced together from bits of previous knowledge (daily rush-hour traffic pattern as a clue to the location of downtown, street-numbering systems, commercial and residential density patterns, etc). A person is able to deduce the correct location through logical reasoning. 27 Sleuthing is similar to visualization in that both require the reader to work between the 2-D map and the 3-D world. But, while visualization is a map-to- environment task requiring the map user to refer to the map during navigation and then to visualize or predict what lies ahead, sleuthing is an environment-to-map task and is essentially a problem—solving act. Unlike visualization, which is presumably a continuous task occurring while the map reader is in the process of navigating, sleuthing is discrete and will take place at the start of navigation when locating oneself on the map, at the end of navigation, and in between at critical junctures or when “double checking” location or when correcting erroneous decisions. Summary Through this literature review, I have briefly characterized map reading ability research conducted by psychologists and cartographers and identified some of the strengths and weaknesses present in research in both disciplines. The main difference between approaches lies in the specific focus of research; psychologists seek to understand cognition, while cartographers seek to understand the potential impact of cognition on the map and the cartographic process. Although cartography may have made considerable advances in cognitive research, the criticism that experimenters neglect to incorporate cognitive theories and psychological testing methods, including reliability and validity, is still a concern (Lloyd 1982, Blades & Spencer 1986, Lieben & Downs 1989). The research conducted in conjunction with this dissertation will 28 incorporate established psychological testing methods including calculated measures of reliability and validity. A second goal of the literature review has been to establish the lack of research encompassing an understanding of cognitive tasks used in map reading for navigation. As Blades and Spencer (1986) accurately observe, these tasks have yet to be fully identified and understood. Through an amalgamation of previous research, some of these potential tasks have been identified: spatial memory, object rotation, wayfmding (including environmental navigation and visualization), and “sleuthing”. The goals of this research are to assess the relative influence of these tasks and determine the extent to which they predict map reading ability. 29 Chapter 3 Research Structure and Methods Research Structure The research objective was to evaluate the impact of each of the five cognitive processes (discussed in chapter two) on map reading and navigation and also 1 determine the extent to which a sit down test could predict a person’s map reading and navigation ability. A map reading ability test and a real world map reading/navigation exercise were completed by 44 subjects. The scores from the tasks were analyzed to determine prediction of map reading. The map reading ability test (referred to as the Predictor Test) contained five sections, representing each of the five cognitive processes addressed; these sections, together with the environmental navigation task, functioned as the predictors in the analyses. Methods The experiment was conducted with the aid of human subject volunteers, who were asked to take a computer-administered and number-scored four-part test followed by an environmental navigation test and finally the map navigation task. The data gathered were used to determine the relative influences of the five cognitive processes on the combined task of map reading and navigation. This chapter will summarize the methods used to conduct the experiment, including the designing of the experiment and testing procedures. 30 Design of the Experiment The Ability (Predictor) Test The predictor test used in this research included the computer-administered Test sections (object rotation, spatial memory, visualization, and sleuthing) and a navigation exercise (environmental navigation). One of the goals of this research was to develop a test that may be used to accurately measure navigational map-reading ability, and future possible uses of the test dictate that group testing may be the most efficient delivery method. Therefore the sit-down part of the test was designed for group testing even though for this research it was administered individually. Although group tests do impose creativity restrictions and lack flexibility for examinee’s answers, the advantages of group testing (minimal role of facilitator, simultaneous administration, and objective scoring) provide support for selecting the group testing method (Anastasi & Urbina 1997). In addition, since the proposed test was designed to evaluate only one ability or class of abilities (map-reading) instead of general intelligence, there is less need to provide freedom and flexibility for examinee’s answers (Anastasi & Urbina 1997). The computer test was composed primarily of graphics about which the subject was asked to draw conclusions. As the test was designed as both a map- reading ability assessment and a device to determine the influence of each of the tasks, test materials had to be carefully controlled and most of the test contained original map-related graphics. Since subjects were told they were participating in a 31 map-reading ability study, using map-related graphics also enhanced the test’s face validity, i.e., the subjects’ assessment of test validity (whether the subjects think the test measures what it is designed to measure). P ace validity may affect test taking motivation and performance (Chan, Schmitt, DeShon, Clause, & Delbridge 1997). The first four sections of the measurement test were created on and administered by computer. The software used to create the test included: Director 6.0, a software program that allows the designer to create interactive multimedia presentations; CorelDraw 7.0, a design program used to create computer graphics; and PhotoShop 5.0, a graphic-image editing and enhancement program. Each of the cognitive tasks was represented in a separate section. An introductory section (Figure 3.1) was designed to explain the general structure and purpose of the test and to give an estimation of the duration of this part of the testing procedure. Following the introduction, the subject navigated through the test sections on a pre-determined linear path. In other words, the subjects were granted limited control of progression forward or backward and were not permitted to take the test sections in random order. 32 Figure 3.17 First screen in the introduction to the Map-Reading AbilityTest Each section began with an explanation of the task. Following the explanation, each subject was offered the opportunity to review the instructions or to wooed with some practice questions. These practice questions were designed to resemble the questions subjects would answer on the scored questions in that section. ;; 3.," In each section, the computer assessed the correctness of each response and W correctness as well as subject response times. The responses were written to text files while the subject was in the process of taking the test. These files fimnnmedrately following the test and became that person’s record 1n later “analysis Figure .1 33 Object Rotation In the Object Rotation section, subjects were presented with pairs of side-by- side maps. In some cases, the map on the right was flipped and rotated, while in other cases, it was rotated only. The objective was to determine whether the map had been flipped and rotated or only rotated. Once subjects had made the determination, they selected either the “Flipped” or the “Same” button, where ‘same’ meant that the map had not been flipped prior to rotation. An example of this task is shown in Figure 3.2. Figure 3.2 Sample screen from the object rotation section 34 Visualization The Visualization section contained two parts. In the first part, each subject was presented with a map as well as several objects (such as police station, hospital, library) that may appear on the map. The objective was to determine what object may be seen when traveling in a particular direction from a given location indicated on the map. In figure 3.3, a sample frame from the test, the arrow indicates the location and direction of travel. The graphics along the right of the map show some objects that may appear in the line of vision of the driver. The subject’s task was to determine which object (from the four given) they would encounter if they traveled from the given location in the indicated direction. The subject clicked on one of the four objects and a new, similar question appeared. The legend and symbols were explained (Figure 3.4) before beginning the test. 35 Figure 3.3 Sample screen from part 1 of the visualization section Figure 3.4 Legend explanation In the second part of the Visualization section, subjects were given written directions beginning from a specified location and ending in a different specified location. They were not shown a map, but were, instead, required to visualize the map in their head and then answer a question after the scenario was presented. In each scenario, a heading (north, east, south, or west) was designated on a particular street and they were then asked to make turns from that and subsequent streets. They were told to assume they were turning a 90-degree angle for each turn. Following the scenario, they were asked to identify the direction they would be facing (Figure 3.5). They indicated an answer by clicking on the appropriate button. 1-. g. ‘.,. . < ‘ L, , 1’ .' . _ . . . .7 4. Figure 3-5 Samue .0 20.....f0v........ ~‘ 37 Sleuthing The sleuthing section also contained two parts. In the first part, the subject was presented a map (Figure 3.6) and a panoramic photograph (Figure 3.7) of some location in the environment. The objective was to determine where they were on the map given the photo presented. The map and the photo were not presented on the same page, but the subject could click appropriate buttons to toggle between them. Whitehills MacDonald .Park 1 Middle School School Warddiff Figure 3.6 An example map from part 1 of the sleuthing section 38 Figure 3.7 An example photograph from part 1 of the sleuthing section After determining the correct location on the map, the subject dragged the appropriate arrow to the correct position on the map (example Figure 3.8). This arrow indicated the relative position as well as the geographic direction (N,NE,E,SE,S,SW,W,NW) the subject would be facing to see the scene in the photograph. 39 Wlitehills .Park MacDonald Middle School School Figure 3.8 Example map with arrows In the second part of the “sleuthing” section, subjects were presented with a series of maps, each highlighting two locations. They were asked to choose the shortest and safest route from the start to the destination in each. The maps contained arrows indicating directions of travel on streets (example Figure 3.9). The subject had to mind the travel flow arrows (they couldn’t go the wrong way down a one way street even though it may be the shortest route). They selected the appropriate route from a list of three routes given. 40 Spatial Memory In the spatial memory section, the subject was presented several maps, each of which highlighted a travel route (Figure 3.10). Each map became increasingly more complex. Subjects were able to control the amount of time spent studying each map, and they were to study it until they felt they knew the route and then advance to a screen containing a list of directions. The objective was to determine the correct order of these directions as they correspond to the travel route shown on the map. The directions were listed in random order with white boxes next to them. A series of numbers were displayed at the bottom of the screen and the subject dragged the 41 numbers to the appropriate component of the directions, indicating the correct order (Figure 3.11). Figure 3.10 Highlighted route in memory section 42 Figure 3.11 Directions corresponding to travel route Focus Group Evaluation Before the computer-administered Map-Reading Ability Test was administered to the main subjects, a focus group evaluation of the test was conducted. Focus groups usually include 5-10 people and are moderated by a facilitator who encourages focused, in-depth discussion in the areas in which feedback is desired. This qualitative evaluation method has proven useful in evaluating cartographic products (Monmonier and Gluck 1994, Duh et al. 1998, Olson et al. 1998). The focus group discussion followed a pre-determined line of questions (Figure 3.12). The group was able to provide insightful feedback about the difficulty level, clarity of task instructions, and design of the test by providing their opinions of 43 the existing test as well as providing ideas for change, documented as Figure 3.13. Many of these suggestions were implemented; the test was edited and readied for the pilot study. The focus-group step is documented here because of its importance in creating a reasonable and meaningful test. 1. Do you have any initial reactions to the test? 2. Were the instructions clear? 3. Which parts were hardest? Easiest? 4. Do you have any comments about: A) pictures B) photos C) static maps D) animation used in introduction explanation E) text (size, color, clarity) F) navigating or proceeding through the test 5. In your own words, what do you understand this test to be about? Figure 3.12 Focus group questions 44 l. The most significant initial reaction from all group members concerned the difficulty level of the memory section; all thought that section was too difficult. The suggested changes included allowing more time on each map (especially the last one) and only including the actual directions in the list of choices (no extras). Another initial reaction included a programming problem in the sleuthing section; clicking on the middle of the map sent the viewer to the end of the section 2. They generally felt that all instructions were clear, especially in the object rotation section. Specific suggestions included: in the visualization section, specify that the user has to stay on the road (no turns) and specify how far you can look off the road; in the same section, the arrow on the last one needs to move more to the right of the hotel; in the legend, instead of using three different colors of circles, use three different geometric symbols to alleviate any problems with color blindness; for all sections, a north arrow should be included on every map. 3. They, again, felt that the most difficult section was spatial memory and the easiest section was object rotation (in this section they did seem to think there was a slight variation in difficulty level) and in the other two sections they felt that there was a range of difficulty levels. 4. They felt that all the photos, static maps, animation used in introduction explanation (they really liked the animations), text (liked size, brevity, and clarity), and navigation were all very well done, but they felt that some of the pictures in the visualization section were difficult to identify (especially the dam). 5. Initially, the group was a little confused with this question, but with further explanation, they revealed that they felt this test related to “spatial orientation”, “map understanding”, “how well you read a map”, “can you use the map”. Figure 3.13 Summary of focus group responses 45 Environmental Navigation Test The environmental navigation test was designed to determine a person’s ability to navigate in an environment without the aid of a map. The test was conducted on the Michigan State University campus in the Engineering building, a large and relatively complex building with multiple hallways. Three markers, labeled 1, 2, and 3, respectively, were affixed to the wall in three separate locations on the third floor of the building (see Figure 3.14 for locations). Subjects were led to the locations of the markers in numerical order. They were then taken to the second floor and told that the layout of the two floors was nearly identical and were asked to identify the location of each marker as seen on the previous floor. I followed behind the subject during the test, allowing them to set the pace for the task and also allowing me to evaluate them unobtrusively. The task was evaluated based on three criteria (stops, hesitancies, and correctness) and marked on a score sheet (Figure 3.15). Due to different walking speeds of each subject, the entire exercise was not timed; rather, stops were timed. Each hesitancy was noted. A location was deemed correct if a subject pointed to within 5 feet of the original location. Incorrect locations were recorded based on whether the location was in the correct hall, correct side of the hall, and the proximity (in feet) from the correct location. 46 - Common Areal Lobby . - Exit Figure 3.14 Map of layout of floor and location of numbers 47 Stops Stops 3 Figure 3.15 Stops - Duration (1,2,3) Duration (1,2,3) Duration (1,2,3) Real- World Navigation Exercise The Real World Navigation Exercise was designed to assess the validity of the Map Reading Ability Test. The environment for the exercise was established within the Engineering and Animal Science buildings; these buildings are physically attached, resulting in a continuous hallway network and the point at which one ends and the other begins is not readily apparent while inside the buildings. Because the environmental navigation test took place in the Engineering building, the participants were already in the complex and the testing session could continue without changing location. The physical testing environment required establishing a mock town street network. Street signs were created and affixed to hallway walls at every hallway Env. Nav. Hesitancies Hesitancies Hesitancies 48 Correct Correct Correct Environmental navigation scoring sheet Incorrect Correct hall Y N If N, how off Correct side Y N Proximity Incorrect Correct hall Y N If N, how off Correct side Y N Proximity Incorrect Correct hall Y N if N, how off Correct side Y N Proximity intersection using poster putty and clear packing tape. This setting offered several advantages over an outdoors street network. First, it was a fairly controllable environment; weather conditions and varying street network traffic patterns could be eliminated as potential influencing factors. Further, the labeling of street signs was controllable, allowing me to label or not label streets at will (advantages of this will become apparent). And finally, it was a safer and less costly environment for subjects; actual navigation by car was too risky and too costly. 9“ Before creating the buildings street” map, an analysis of three actual street maps (ADC’s Visitor’s Map of Washington DC, Rand McNally’s map of Atlanta, and the American Automobile Association’s map of Portland) was made for the purpose of determining the average number of map symbols per city block on a real map so the same number would be maintained on the building map. The analysis revealed an average of 0.4 symbols per block. In addition to the labeled “streets,” the final map (Figure 3.16) created for the exercise contained only two sets of symbols representing exits and lobby/common areas. The number of symbols as contained on the three city maps was closely maintained. Subjects referred to the map during the exercise, which included two locating tasks and four navigation tasks. The locating tasks were conducted specifically to determine subjects’ sleuth ability since they were tasks that relate the clues in the real world to the map in order to accurately determine self-location. For these two locating tasks, I gave the subject the map and asked them to use my pen to draw a small “x” on the map designating the location of a wall in the first task and a door in 49 the second task (see Figure 3.17 for locations). Each of the four locations of the navigation task was slightly different and represented actual scenarios that one would encounter when navigating to a destination in the real world. Location one was a street intersection. Locations two and three were both street addresses; however, the street sign was intentionally left off the building hall in location two. The fourth location was the southwest corner of a lobby. The significance of each type of location will be explained in the next chapter. For each navigation destination, I verbally stated the location and asked that they navigate to the physical location in the building (see Figure 3.18 for locations). I explained that they could use the map to assist in navigation, but I would hold the map and they could ask for it when needed. By so doing, I could easily determine the number of uses and could tap my stopwatch to measure time. 50 3.50 em I oEanomo I 33.. 32.4 85500 I 22m ban D Z .41: 3.4;; ' 7! Figure 3.16 M515 used in the’RWMNE 5.50 em I 2.3235 I >33 >32 88:50 I 22m bio D z , u... .F 6 as. .53 £1. Figure 3.17” "'L‘d'cétibhs ofthe locatingtask‘ ' ' ' 52 Z .250 an.“ gxm I 2.25680 I 3%.. .6 32¢. 52:50 I 22m Ewe D I e. e m. V m n y m 2%.... fl .. .7 a] Figure 3.18 Locations of the 4 navigation destinations 53 A score sheet (Figure 3.19) was used to evaluate each subject’s performance during the exercise. Each task was scored independently and data were gathered for four criteria: the number of times the subject rotated the map (rotations were assumed to be 90-degrees unless otherwise noted), the number and duration of stops, the number of hesitancies (times when they slowed stride but did not stop), and the amount of time spent reading the map (one mark represented a 1-5 second map reference time). Map Test Time 1" Task 1. Stops Duration (1,2,3) Hesitancies Correct Incorrect Correct hall Y N If N, how off Correct side Y N Proximity Time 2Id Task 2 Stops Duration (1,2,3) Hesitancies Correct Incorrect Correct hall Y N If N, how off Correct side Y N Proximity 3. Stops Duration (1,2,3) Hesitancies Correct Incorrect Correct hall Y N If N, how off Correct side Y N Proximity 4. Stops Duration (1,2,3) Hesitancies Correct Incorrect Correct hall Y N If N, how off Correct side Y N Proximity Figure 3.19 RWMNE score sheet 54 Test Administration I checked the entire environment before each session and replaced names and numbers as necessary. Because street signs and numbers affixed to the walls in the building were not permanently attached to the walls, they sometimes became detached. I met each volunteer at the Dairy Store since it is a well-known landmark on campus and is located in the complex in which the experiment took place. The section containing the Dairy Store, because it is so widely familiar, was not used in the experiment. Throughout the entire experiment I followed a planned, written script. I led the subject to a computer lab in which four computers were loaded with Director 6.0 and the map test. The subject sat at one of the computers at which time I asked them to read and sign a consent form. I briefly explained the experiment and asked that they hold all questions about the study as a whole until the end of the entire session. They then proceeded to take the computer-administered sections of the Predictor Test, which lasted approximately 45 minutes on average. On completion of the map test, I led them to an elevator that we took to the third floor to begin the environmental navigation (no map) wayfinding test. I explained that beyond my instructions, we could not talk for the remainder of the exercise and, again, encouraged them to ask questions at the completion of the session. After I led them to each of the three markers we took the elevator to level two. Once on the second floor, I explained that the floor layout was nearly identical to the prior floor and asked them to lead me to the locations where they had seen the three markers (there were no 55 markers on the second floor walls). They were evaluated throughout the entire exercise using the environmental navigation score sheet. After locating the final marker, I led them back to the elevator, which we took to the first floor and the start position of the Real-World Map Navigation Exercise (detailed instructions in Appendix B). Preceding the RWMN E, I explained that I had created a mock street system by labeling the halls and doors with street signs and numbers (I showed them examples). They began the map reading/navigation exercise with the first of the two self-locating tasks, each of which were timed. I then gave them the first destination and asked them to navigate there. Upon successful navigation to the first location, 4 they completed the second self-location task. I then gave them the second destination, followed by the third and fourth, which concluded the exercise. I then answered any questions they had, asked them to complete a short questionnaire, gave them their payment, showed them how to return to the Dairy Store, and, finally, thanked them for participating. Preliminary Testing A pilot study with 3 female and 2 male subjects was conducted for the primary purpose of evaluating scoring methods and overall testing procedure. This study revealed that although the testing did run smoothly, it was too long. Also, the original scoring for the environmental navigation and map reading exercises included recording times for the entire exercise; it proved to be too difficult to time stops and map reading, as well as the entire exercise. Since walking pace would influence the 56 overall time, while stops and time taken to consult the map would not, only the later two criteria were timed. These changes were made to the test administration, resulting in the methods outlined above. Using the changed procedures, I tested two additional pilot subjects and found that all ran smoothly. Testing began following this second pilot test. Scores from the pilot subjects were not used in the analysis. Subjects A total of 45 subjects participated and 44 useable results were obtained (an analysis of the power level provided by 44 subjects will be discussed in Chapter 4). Subjects ranged in age from 21 to 68 years; 26 were female while 18 were male; 29 subjects came from the university student community (undergraduate and graduate students) while 15 came from outside the university and held occupations such as retiree, city planner, and civil engineer. The volunteers were obtained through advertisement flyers posted in community areas including recreation centers, gyms, laundromats, and grocery stores in addition to verbal advertisement in introductory college classes. Each participant received $15 and a coupon for a serving of ice cream at the campus Dairy Store. Subjects were not segregated based on any self-reported map-reading expertise. Studies have indicated that “expert” map readers do not possess an advantage over “novice” map readers (Thomdyke & Stasz 1980, Gilhooly et al. 1988). The Predictor Test was designed to discriminate between ability levels and should yield better results if subjects possess a range of map-reading abilities. 57 Chapter 4 Results and Analyses This chapter presents the results gathered from the Predictor Test (object rotation, visualization, sleuthing, spatial memory, environmental navigation) and the Real-World Map Navigation Exercise. The raw data will be presented in the Data section; the calculation of the “scores” used in the analyses and the statistical tests themselves will be given in the Scores and Analysis section. The meaning and implications are addressed in Chapter 5, Discussion. Results Map Reading Ability Test The five parts of the Predictor Test each had separate scores. The computer test, administered in the Macromedia Director program, recorded both the subject’s answers and the response times, each of which were written to a text file as the test proceeded. The environmental navigation exercise was administered in a physical environment and subject responses were recorded manually by the facilitator and then entered into an Excel file. The computer test scores were then collected into the Excel file. 58 Object Rotation In this section, subjects were asked to decide whether the map was “rotated only” or “rotated and flipped”. Table 4.1 displays an example of the results as generated for the first 5 subjects. A “C” indicates that the subject answered correctly, ‘61,, and an indicates an incorrect answer. Response time for each question is recorded in milliseconds. Most subjects selected the correct answer for each question. In fact, for all 44 subjects, only 25 questions (out of a total of 704 questions) were missed. A large range of times was recorded, however, from a low of 127 to a high of 5489 with a mean of932.8. 59 Subjecttt Question1 Question1 Question2 Question2 Question3 Question3 accuracy time accuracy time accuracy time 1 C 426 C 454 C 430 2 C 559 C 601 C 589 3 C 728 C 821 C 698 4 C 4276 C 5489 C 4056 5 C 551 C 598 C 540 Subject# Question4 Question4 Question5 Question5 Question6 Question6 accuracy time accuracy time accuracy time 1 C 398 C 401 C 398 2 C 469 C 598 C 522 3 C 632 C 726 C 754 4 C 3938 C 4356 C 4087 5 C 530 C 604 C 621 Subject# Question7 Question7 Question8 Question8 Question9 Question9 accuracy time accuracy time accuracy time 1 C 456 C 354 C 165 2 C 564 C 499 C 569 3 C 777 C 501 C 726 4 C 4659 C 2109 C 4875 5 C 688 C 368 C 631 Subject# Question10 Question10 Question11 Question11 Question12 Question12 accuracy time accuracy time accuracy time 1 C 485 C 398 C 444 2 C 603 C 555 C 504 3 C 741 C 703 C 765 4 C 4989 C 3545 C 4089 5 C 570 C 498 C 465 Subject# Question13 Question13 Question14 Question14 Question15 Question15 accuracy time accuracy time accuracy time 1 C 465 C 587 C 332 2 C 634 C 674 C 431 3 C 780 C 820 C 646 4 C 4358 C 5146 C 2298 5 C 623 C 645 C 468 Subject# Question16 Question16 accuracy time 1 I 585 2 I 718 3 C 835 4 C 4685 5 C 599 Table 4.1 Example Object Rotation results for 5 subjects. There were 16 items in this section. C=correct; I=incorrect. Times are in milliseconds. 6O Visualization The visualization section contained two parts, each of which required subjects to choose the correct answer from multiple choices. Table 4.2 shows the computer- generated results from the first part of the visualization section for subjects 1-5 and Table 4.3 shows data from part two. Similar to the object rotation section, very few questions were answered incorrectly in either part. In part one, only 60 questions out of 440 were missed and in part two 11 out of 220 were missed. Time varied greatly again from 366 to 21064 with a mean of 2714.34 in part one, and from 1263 to 14453 with a mean of 4226.35 in part two. 61 Subject # Question 1 Question 1 Question 2 Question 2 Question 3 Question 3 accuracy time accuracy time accuracy time 1 C 689 C 654 C 628 2 C 870 C 790 C 812 3 C 1669 C 1401 C 1076 4 C 8254 C 8630 C 7680 5 C 1256 C 1056 C 1046 Subject # Question 4 Question 4 Question 5 Question 5 Question 6 Question 6 accuracy time accuracy time accuracy time 1 C 630 C 587 C 465 2 I 1 5065 C 690 C 550 3 I 18974 C 1644 C 1323 4 I 17741 C 8632 C 5645 5 I 1589 C 1098 C 1164 Subject # Question 7 Question 7 Question 8 Question 8 Question 9 Question 9 accuracy time accuracy time accuracy time 1 C 620 C 555 C 594 2 C 820 C 777 I 1055 3 C 1770 C 1498 C 1624 4 C 7798 I 12361 C 9241 5 C 1 11 1 I 1587 C 954 Subject # Question 10 Question 10 accuracy time 1 C 842 2 I 1259 3 C 1976 4 C 9980 5 C 1369 Table 4.2 Example Visualization part 1 results for 5 Subjects. There were 10 items in this section. C=correct; I=incorrect. Times are in milliseconds. 62 Subject # Question 1 Question 1 Question 2 Question 2 accuracy time accuracy time 1 C 2247 C 1985 2 C 5785 C 3149 3 C 8644 C 6873 4 C 2994 C 2847 5 C 4216 C 3843 Subject# Question 3 Question 3 Question 4 Question 4 accuracy time accuracy time 1 C 2436 C 1674 2 C 5365 C 4725 3 C 9721 C 8298 4 C 6080 C 3223 5 I 3385 C 3017 Subject # Question 5 Question 5 accuracy time 1 C 2613 2 C 5560 3 C 7716 4 C 3515 5 C 6034 Table 4.3 Example Visualization part 2 results for 5 Subjects. There were 5 items in this section. C=correct; I=incorrect. Times are in milliseconds. Sleuthing Because the first part of the sleuthing section required subjects to drag arrows, marking locations on the map, the computer-recorded responses were not simply correct or incorrect as in previous sections. Responses included which arrow the subject chose (correct, 45-, 90-, 135-, or 180-degrees off), the location in which the arrow was placed (in pixels from the upper left screen comer) and the response time for each question. Table 4.4 contains the results for subjects 1-5 from the first part of the sleuthing section and Table 4.5 contains results for the same subjects from part two. The first part of this test section resulted in more diversity in answers, with 63 more subjects answering incorrectly, and as in the previous two sections, time varied considerably. The range of times was 742 to 5743 with a mean time of 3 142.5. Only 12 out of 220 questions were answered incorrectly. Time in part two varied from 653 to 14022 with a mean of 2230.67. Subject# Question Question Question Question 1arrow 1location 1timer 2arrow 1 C 370.403 1 053 90 2 C 347.398 3124 C 3 C 341 .398 1924 C 4 90 705.292 5058 7 90 5 C 31 8.402 3612 C Subject# Question Question Question Question Question 2|ocation 2timer 3arrow 3Iocation 3timer 1 454.168 2356 90 648.456 3723 2 203.417 2019 C 750.385 2591 3 202.407 2975 C 729.391 2917 4 708.282 4631 135 692.332 4757 5 201.401 4033 C 741.371 3558 Table 4.4 Example Sleuthing part 1 results for 5 Subjects. There were 3 items in this section. For the arrow, C=correct placement; numbers indicate degrees from correct. Location indicated the pixel location of the arrow. Times are in milliseconds. 64 Subject # Question 1 Question 1 Question 2 Question 2 accuracy time accuracy time 1 C 2233 C 1829 2 C 2317 C 2298 3 C 2723 C 2158 4 C 1 1 159 C 9080 5 I 2442 C 1621 Subject # Question 3 Question 3 Question 4 Question 4 accuracy time accuracy time 1 C 2431 C 2603 2 C 2805 C 1985 3 C 2263 C 2028 4 C 10468 C 8941 5 I 1967 C 1 331 Subject # Question 5 Question 5 accuracy time 1 C 2488 2 C 1904 3 C 2561 4 C 12091 5 C 1784 Table 4.5 Example Sleuthing part 2 results for 5 Subjects. There were 5 items in this section. C=correct; I=incorrect. Times are in milliseconds. Spatial Memory The spatial memory part of the test asked subjects to study a map showing a route including the start and finish locations and then indicate the correct order of route directions by dragging the appropriate numbers to the boxes corresponding to each element of the directions. Results were even more complex than those for the sleuthing section because they included study time for the map, correct-incorrect response for each element in the directions (with number of elements different for each of the three maps included), and time to complete the choices in ordering the elements of the directions. 65 Table 4.6 shows the responses of subjects 1-5 in this section. Most subjects were able to order the elements correctly for the first two questions; only 2 subjects made a mistake on the first question and 5 on the second. Time varied from 552 to 5872, with a mean of 960.75 for question one and 1032 to 6145 with a mean of 2206.45 for question two. The third question appeared to be more difficult; almost half of the subjects made at least one mistake on ordering the elements. Times not only varied, but were much higher (2064 to 21805) than in the previous two questions. 66 uestion 1 Subject # Study Time Element 1 Element 2 Element 3 Task Time 1 1384 C C C 841 2 1 196 C C C 815 3 2807 C C C 993 4 8697 C C C 2313 5 1389 C C C 1140 Question 2 Subject # Study Time Element 1 Element 2 Element 3 Element 4 1 2964 C C C C 2 2639 C C C C 3 2667 C C C C 4 1507 C C C C 5 2281 C C C C Subject # Element 5 Task Time 1 C 1391 2 C 1703 3 C 2771 4 C 3998 5 C 1706 Question 3 Subject # Study Time Element 1 Element 2 Element 3 Element 4 1 4927 I I I I 2 4201 C C C C 3 2607 C C C C 4 1643 l I C C 5 3064 C I I I Subject # Element 5 Element 6 Task Time 1 l I 8701 2 C C 2630 3 C C 3421 4 C C 21805 5 I C 3480 Table 4.6 Example Memory results for 5 subjects. Study time indicates the amount of time spent studying the map and is in milliseconds. The elements indicate whether the correct order # was placed with the corresponding instruction. The task time (in milliseconds) reflects the amount of time taken for the numbers/directions matching task. 67 Environmental Navigation The data collected during the navigation task on the evaluation sheet (Figure 3.15) included both the correctness of the identified location (correctness of the hall, correctness of the side of ball, and proximity in feet from the actual location) and the number and duration of stops. Table 4.7 contains responses from subjects 1-5. Every subject correctly identified the hall and side of the hall for each of the three markers. The only variation in scores was provided by the proximity in feet from the actual location and the duration of stops. Duration of stops varied from 0 to 226 seconds. Proximity varied from 0 feet to 60 feet for marker one, 0 to 50 feet for marker 2, and 0 to 110 feet for marker 3. Marker l Marker 2 Marker 3 Stops Subject # Correct Distance Correct Distance Correct Distance Hall Side Hall Side Hall Side 1 Y Y 511. Y Y 0ft Y Y 5011 33 2 Y Y 30ft Y Y Ofi Y Y 5011 8 3 Y Y 20ft Y Y 2011 Y Y 0ft 0 4 Y Y 10ft Y Y 10ft Y Y Oil 22 5 Y Y 1011 Y Y Sfi Y Y 50ft 26 Table 4.7 Example environmental navigation results for 5 subjects 68 Real World Map Navigation Exercise The data from the Real World Map Navigation Exercise, which were to be the dependent values in the analyses, were also recorded on an evaluation sheet (Figure 3.19). Subjects were evaluated on several criteria, including: finding the three assigned locations; number of map rotations; number and duration of stops for each location; and time spent studying the map. Table 4.8 contains results for subjects 1-5. All subjects successfully navigated to every location. Variation was observed in map rotations, which ranged from 0 to 14 with a mean of 2.6, the duration of stops (measured in seconds) the subjects made, which ranged from 0 to 77 with a mean of 18 and the amount of time subjects spent studying the map, which ranged from 25 to 215, with a mean of65. Information recorded for the two self-locating tasks (Table 4.8) included correctness of hall, side of hall, and distance recorded in “blocks” from the actual location and time spent completing the task (recorded in seconds). Slightly more than half of the subjects (24) identified the correct location in the first locating task, while 36 correctly located themselves in the second task. Time ranged from 6 to 128 seconds, with a mean of 55.7 seconds for the first and 3 to 37, with a mean of 11.3 for the second task. 69 Subject# Locating Task #1 Hall Time Corr? it of halls Corr. Side? from corn? 1 Y Y 17 2 Y Y 39 3 N 2 N 47 4 N 2 Y 19 5 N l N 80 Locating Task #2 Hall Corr? # of halls ~<-<-<~<-< from corn? Corr. Side? z-<-<-<'< Time 10 l7 16 12 27 Time (map consulting) 80 55 55 95 75 Stops (seconds) Table 4.8 Example Real World Map Navigation Exercise for first 5 subjects 70 Analyses Before statistical analysis could be completed, “scores” were computed for each of the Predictor Test sections and for the Real-World Map Navigation Exercise. These scores were then used in the statistical analysis. This section will first address the computation of scores and then the statistical analyses themselves. Calculating Scores Map Reading Ability Test The calculation methods used in determining the scores from the computer test portion of the Predictor Test were fairly consistent but were very different from the method used to calculate scores in the environmental navigation portion. Therefore, even though the computer test and the environmental navigation exercise combined to function as the Predictor Test, the scoring methods will be addressed separately here. Computer Test Determining a scoring method depends on the type of test administered. According to Anastasi and Urbina (1997), two distinct types of tests exist. A pure speeded test contains a large number of unifome low difficulty items (all items well within the ability of persons to whom the test is administered); the time limit allotted 71 for taking the test is set so low that no one can complete all items. The test is evaluated based purely on the number of items attempted. A pure power test on the other hand consists of items of steeply graded difficulty so that the items toward the end of the test are too difficult for anyone to solve. Power tests are evaluated based on number of correct responses. Anastasi and Urbina suggest that in actual practice, most tests are not purely speeded or power, but rather a combination of the two. The test used in this research was not a pure speeded nor a pure power test. Tasks were generally low enough in difficulty for subjects to answer correctly, but instead of a time limit being set, the time taken to answer questions was recorded. Some questions were more difficult than others, but they were not ordered in such a way that a maximum difficulty level could be determined for subjects and only a few questions were difficult. Because many subjects answered all test questions correctly and those who did not, missed only a few, the computer test sections were scored in a manner based, but not identical to, on methods used in speeded tests. A score for each section was calculated based solely on response time for correct answers. Object Rotation. In this section, subjects decided whether a map was flipped and rotated or only rotated. Scores ranged from 263 to 4025, with a mean of 771. A histogram and a K-S test of the final scores shows a reasonably normal distribution (Figure 4.1). 72 Number of Subjects -3 SD -2 SD -1 SD x +1 SD +2 SD +3 SD Object Rotation Score (in thousands) Figure 4.1 Histogram of Object Rotation Scores. In a K-S test for normality, p=.659 (the closer p is to 1.0, the more perfect the fit). 73 Visualization. The scores for each of the two visualization sections were calculated in the same way as the object rotation scores, based on mean response time. Total scores for the section were calculated by averaging the scores from both parts. Scores ranged from 490 to 3765, with a mean of 1336. A histogram of the final scores illustrates a normal distribution (Figure 4.2). Number of Subjects -3 SD -2 SD -1 SD x H SD +2 SD +3 SD Visualization Score (in thousands) Figure 4.2 Histogram of Visualization Total Scores. In a K-S test for normality, p=.672. 74 Sleuthing. The results of the first part of the sleuthing section indicated which arrow the subjects chose, the placement (pixel location) of the arrow and the response time. The second part of the sleuthing section involved selection of shortest feasible routes, scores were recorded and calculated exactly as in the object rotation and visualization sections. Scores for the entire sleuthing section ranged from 784 to 4752 with a mean of 2541. A histogram of the final scores illustrates a normal distribution (Figure 4.3). Number of Subjects -3 SD -2 SD -1 SD x H SD +2 SD +3 SD Sleuthing Score (in thousands) Figure 4.3 Histogram of Sleuthing Scores. In a K-S test for normality, p=.723. 75 Spatial Memory. Scores from the memory section were calculated using the same method as used in previous computer test sections (mean response time). Scores ranged from 857 to 9793 with a mean of 4171. A histogram of the final scores illustrates a reasonably normal distribution (Figure 4.4). Number of Subjects -3 SD -2 SD -1 SD x H SD +2 SD +3 SD Memory Score Figure 4.4 Histogram of Spatial Memory Scores. In a K-S test for normality, p=.624. 76 Environmental Navigation The Final Score for environmental navigation was calculated by adding the number of feet (proximity) to the total duration of stops. Table 4.9 shows calculation of scores for subjects 1-5. Total Scores ranged from 2 to 220 with a mean of 57.43. A histogram distribution and K—S test for normality both indicate that the environmental navigation scores were not normally distributed (Figure 4.9). Marker I Marker 2 Marker 3 Stops Subject # Correct Distance Correct Distance Correct Distance Hall Side Hall Side Hall Side Total 1 Y Y 5ft. Y Y Ofi Y Y 50ft 33 88 2 Y Y 30ft Y Y 011 Y Y 5011 8 88 3 Y Y 20ft Y Y 20a Y Y ' 0ft 0 4o 4 Y Y 1011 Y Y 10ft Y Y 0ft 22 42 5 Y Y 1011 Y Y 511 Y Y 5011 26 91 Table 4.9 Example environmental navigation scores for 5 subjects 77 Number of Subjects -3 SD -2 SD -I SD x H SD +2 SD +3 SD Sleuthing Score (in thousands) Figure 4.5 Histogram of Environmental Navigation Scores. In a K-S test for normality, p=.142. Real World Map Navigation Exercise While data were recorded on several criteria in the RWMNE exercise, only stops and map study time were included in the formula devised to determine a score for the main analysis. Because all subjects successfully navigated to every location, accuracy was not a factor. The sum of the amount of time spent studying the map and the duration (in seconds) of stops were added for the final Real-World Map Navigation Exercise score (the Location Tasks scores did not factor into the RWMNE 78 score). Table 4.10 shows the calculation of scores for subjects 1-5. Scores range from 7 to 120 with a mean of 52.04. Subject# Looks Stops RWMNE Total Score 1 80 6 86 2 55 0 55 3 55 l 56 4 95 3 98 5 75 0 75 Table 4.10 Example Real World Map Navigation Exercise for first 5 subjects Test Evaluation Any test administered in research involving human subjects should be tested for reliability and validity as they both affect the ability to confidently report results as well as develop strong conclusions. Before the main analysis, then, such tests were performed on the Predictor Test, which is the instrument under scrutiny in this research. Researchers may utilize several methods for testing the reliability and validity of their measurement devices, depending on potential error of the test and how the test scores will be used. 79 Test Reliability Test reliability refers to the consistency of scores that may be obtained by a person when tested on different occasions. In other words, test reliability is determined by how consistently a test measures a given trait or ability. Several analyses for test reliability may be conducted. Researchers select a reliability method depending on what they identify as the potential sources of error. For example, if the same test is designed to be administered to the same subject repeatedly, then the potential source of error is the test/re-test reliability, or if more than one individual will rate, or score, the test, then the potential source of error in inter-rater reliability. For the purposes of this research, the alpha coefficient, which is designed to evaluate the internal consistency of a test (how consistently the questions measure an ability), was calculated. This calculation is based on the number of items (questions), the average item covariance, and the sum of all elements in the covariance matrix of items. The calculated value (a proportion) identifies the reliability of the test (see Anastasi & Urbina 1997 for a more detailed discussion of alpha coefficient). Because the alpha coefficient is a measure of internal consistency and because the map reading test is relatively heterogeneous (each section presumably measures a different construct), the reliability measure was calculated for each section of the test. The results of the analyses (Tables 4.11, 4.12, 4.13, and 4.14) identify three of the four computer test sections as strongly reliable. The analysis of the fourth section, spatial memory, resulted in an unacceptably low reliability. This low measure may have resulted because the memory section contained only three questions and each 80 question increased in difficulty; therefore, chances are very low that a person would maintain a consistent performance throughout that test section. The repercussions of such a low reliability are that any results or conclusions based on those results are questionable and will have to be noted with caution. Internal consistency results Split—half correlation 0.911 Spearman-Brown Coefficient 0.953 Guttman (Rulon) Coefficient 0.940 Coefficient Alpha - all items 0.899 Table 4.11 Reliability Results for Visualization Internal consistency data Split-half correlation 0.997 Spearman-Brown Coefficient 0.999 Guttman (Rulon) Coefficient 0.999 Coefficient Alpha - all items 0.998 Table 4.12 Reliability Results for Object Rotation 81 Internal consistency data Split—half correlation 0.998 Spearman-Brown Coefficient 0.999 Guttman (Rulon) Coefficient 0.926 Coefficient Alpha - all items 0.987 Table 4.13 Reliability Results for Sleuthing Internal consistency data Split-half correlation 0.430 Spearman-Brown Coefficient 0.601 Guttman (Rulon) Coefficient 0.101 Coefficient Alpha - all items 0.261 Table 4.14 Reliability Results for Spatial Memory Test Validity Both criterion validity (the extent to which a test accurately predicts a variable’s outcome) and construct validity (the extent to which a test measures the theoretical construct) were measured. For this research, the criterion validity indicates the extent to which the map-reading ability test (as a whole, and individual sections) measures and predicts a person’s ability to perform the map navigation task, while the construct validity measures how well the test sections measure the 82 constructs they were designed to measure. In each section, the null and working hypotheses will be stated before discussing the results. Linear regression was chosen as the statistical method for the analyses because it provides an indication of the prediction power of an independent variable. The Predictor Test was designed to predict how well a person navigates with a map in a real environment, while the Real-World Map Navigation Exercise provides an indicator of the actual map navigation ability. In the linear regression analyses, the RWMNE (as a whole or in parts, depending on the specific analysis) functions as the dependent variable, while the Map Reading Ability Test (as a whole or in parts) fimctions as the independent variable. Criterion Validity The null hypothesis was that the map reading ability test does not predict a person’s ability to navigate with a map; the alternative hypothesis was that the test does predict that. In order to determine criterion validity, the prediction test (multi-section Predictor Test in this case) must be compared to subjects’ performance in the task that is being predicted, the Real-World Map Navigation Exercise. All subjects’ Total Scores from the five parts of the ability test (independent variables) were regressed against their Total Scores from the map reading/navigation exercise (dependent variable). The results (Table 4.15) indicate that the overall model is significant. However, only three independent variables (visualization, 83 memory, and sleuthing) are identified as statistically significant contributors. Tolerance is relatively low for visualization (tolerance .425) and sleuthing (tolerance .379). A simple correlation revealed that these two variables were highly correlated (correlation coefficient .697) and a scatterplot showed the linear relationship visually (Figure 4.6). A factor analysis also reflected the relationship between these two variables as they both loaded on factor 1 (Table 4.16). Because of the significant contributions, despite relatively low tolerances, both variables were left in the regression model. Dep Var: MAPSCORE N: 44 Multiple R: 0.946 Squared multiple R: 0.895 Adjusted squared multiple R: 0.881 Standard error of estimate: 9.968 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT -11.529 4.664 0.000 . —2.472 0.018 ENVNAV —0.047 0.032 -0.083 0.847 -1.456 0.154 VIS 0.012 0.002 0.415 0.425 5.144 0.000 OR 0.001 0.002 0.031 0.714 0.498 0.621 MEM 0.002 0.001 0.171 0.848 2.997 0.005 SL 0.016 0.002 0.587 0.379 6.868 0.000 Table 4.15 Multiple Regression. The RWMNE total Score fiinctioned as the dependent variable, all five sections of the Predictor Test functioned as the independent variables (ENVNAV, VISUALIZ, Obj Rot, MEMORY, SLEUTH). 84 5000 1 T I l 4000 ° ~ 3000 ,. . — l Sleuthing 2000 l‘ . . .o “ 1000 *- - o I i I 0 1 000 2000 3000 4000 Visualization Figure 4.6 Scatterplot with Visualization and Sleuthing 85 Component loadings ENVNAV VISZ 0R2 MEMZ SL Variance Explained tact 0R0) l 2 0.491 0.072 0.788 0.458 0.237 -0.783 0.381 -0.670 0.906 0.049 by Components 1 2 1.884 1.280 Percent of Total Variance Explained 1 2 37.671 25.600 Factor Loadings Plot -0.756 0.229 0.428 -0.405 0.269 1.044 20.871 Table 4.16 Factor analysis results 86 Because sleuthing was such an influential variable in the multiple regression model, a simple linear regression was run using sleuthing as the only independent variable regressed against RWMNE Total Scores as the dependent variable (Table 4.17) (scatterplot Figure 4.7). The resulting model is, itself, a good predictor of performance on the Real-World Map Navigation Exercise, accounting for 80% of the variance vs. 89.5% for the full model. Dep Var: MAPSCORE N: 44 Multiple R: 0.894 Squared multiple R: 0.800 Adjusted squared multiple R: 0.795 Standard error of estimate: 13.095 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT -8.478 5.076 0.000 . -1.670 0.102 SL 0.024 0.002 0.894 1.000 12.943 0.000 Table 4.17 Linear Regression with predictor test dependent variable (RWMNE) and sleuthing (sleuth) independent. 87 RWMNE Figure 4.7 Scatterplot with RWMNE and sleuthing. 150 100 I 50L :3 :g.’ l l l l 0 0 Sleuthing Construct Validity Several potential cognitive processes of map reading have been proposed and discussed in chapter 2. When research involves the creation of a test that is administered to human subjects for the purpose of yielding analyzable results, the researcher cannot merely assume that the created test measures the mental constructs it was designed to measure. Therefore, in this research an analysis of construct validity was conducted to determine whether the five parts of the Map Reading Ability Test adequately measure the five constructs (object rotation, visualization, spatial memory, sleuthing, and environmental navigation). Because navigation with a 88 1000 2000 3000 4000 5000 map is a complex task presumably affected by the five cognitive processes under investigation, the construct validity analyses requires the individual test sections to be compared to parts of the navigation exercise that are assumed to be influenced by each process. Unlike criterion validity analysis, which used regressions of the total scores of each of the five parts of the map reading ability test against the total scores from the entire map reading/navigation exercise, the construct validity is determined by analyzing the relationships between each of the five parts of the map reading ability test (MRAT), which represent the theoretical constructs, and the scores from individual parts of the map reading/navigation exercise (RWMNE). The individual parts (each discussed further below) provide a measure of a person’s ability in each of the five constructs (environmental navigation, object rotation, spatial memory, visualization, and sleuthing). In other words, a construct validity analysis would determine the extent to which the object rotation test section measures a person’s ability to mentally rotate maps while navigating. The analysis involves comparing subjects’ performance on the object rotation section of the MRAT with the number of times they rotated the map during the RWMNE. The method of construct validity analysis for object rotation as well as the other four Test sections will be discussed in more detail below. Identifying the validity of the constructs (processes) was achieved through regression analyses and the results of each will be addressed individually. 89 Object Rotation. The null hypothesis was that the object rotation section of the map reading ability test does not predict whether a person rotates the map while navigating with a map; the alternative hypothesis was that the object rotation section does predict map rotation. At the start of the experiment, the assumption was that if a person could mentally rotate maps, then they would not need to rotate the map during the map reading/navigation exercise. The validity of the object rotation construct involved comparing how often a subject rotated the map during the RWMNE with the subject’s score from the object rotation section of the Test. The nmnber of times each person rotated the map during the RWMNE functioned as the dependent variable, while the score from the object rotation section of the map reading ability test was the independent variable. The results (Table 4.18) (scatterplot Figure 4.8) indicate no significant relationship between the two variables and the null hypothesis could not be rejected. However, three observations exerted large leverage; they were removed and the new model (Table 4.19) (scatterplot Figure 4.9) indicates a significant relationship. The removed observations were the three highest (worst) scores on the object rotation section of the MRAT. they were sufficiently out of range from the rest of the scores, and sufficiently close to one another, that they raise a question of whether some people are simply in a different category on this ability. All test sections were regressed against rotations to determine the predictability of any additional independent variables (Table 4.20). The results indicated that memory exerts a significant influence in the model. A simple linear 90 regression further identifies that memory is a significant predictor of rotations (as tested in the MRAT) (Table 4.21) (scatterplot Figure 4.10). Dep Var: ROTATIONS N: 44 Multiple R: 0.184 Squared multiple R: 0.034 Adjusted squared multiple R: 0.011 Standard error of estimate: 8.767 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT 9.360 1.812 0.000 . 5.166 0.000 0R2 0.002 0.002 0.184 1.000 1.213 0.232 Table 4.18 Linear Regression with number of rotations as dependent variable (rotations) and the object rotation of the Predictor Test independent (0R2). 30 I I I I O. 20 .. .o' — (I) o c i . m 0 O 10 —oo o — .0 0 —5L I 1 I l 0 1000 2000 3000 4000 5000 Object Rotation Figure 4.8 Scatterplot with rotations and OR. 91 : 0.425 Dep Var: ROTATIONS N: 41 Multiple R: 0.652 Squared multiple R Adjusted squared multiple R: 0.410 Standard error of estimate: 6.796 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT -5.931 3.283 0.000 -1.807 0.079 0R2 0.030 0.006 0.652 1.000 5.364 0.000 Table 4.19 Linear Regression (Rotations-dependent, OR-independent) with High leverages removed 30 I I I I I I I 20— . - ' - U) o o C .2 o 8 . . m C O 10— o o o . o - O—o—Iio-O-J—Jee 1 l 1 l ¢°°®°utb°e°°e°°19°e°°e°°®°° Object Rotation Figure 4.9 Scatterplot with rotations and OR (leverages removed). 92 Dep Var: ROTATIONS N: 41 Multiple R: 0.797 Squared multiple R: 0.635 Adjusted squared multiple R: 0.583 Standard error of estimate: 5.711 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT -9.169 3.222 0.000 . -2.846 0.007 ENVNAV -0.022 0.019 -0.l31 0.875 -1.199 0.239 VIS2 -0.003 0.001 -0.321 0.398 -1.985 0.055 0R2 0.016 0.006 0.344 0.658 2.736 0.010 MEM2 0.002 0.001 0.458 0.693 3.738 0.001 SL 0.003 0.001 0.354 0.365 2.098 0.043 Table 4.20 Multiple Regression. Rotations functioned as the dependent variable, all five sections of the Predictor Test functioned as the independent variables. Dep Var: ROTATIONS N: 44 Multiple R: 0.689 Squared multiple R: 0.475 Adjusted squared multiple R: 0.462 Standard error of estimate: 6.465 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT -1.782 2.272 0.000 . -0.784 0.437 MEM2 0.003 0.000 0.689 1.000 6.160 0.000 Table 4.21 Linear Regression with number of rotations as dependent (rotations) and the spatial memory section of the Predictor Test as independent. 93 Rotations 30 fT I I I 20— o o o 10— oo o o. o o o 0 on o C 0 I ".5 5 l I l l l °snssssesns Memory Figure 4.10 Scatterplot with rotations and memory Visualization. The null hypothesis was that the visualization section of the test does not predict a person’s ability to recognize an environment during the map reading and navigation exercise; the alternative hypothesis was that the visualization section does predict environmental recognition. During the RWMNE, the amount of time a subject spent stopping and “getting their bearings” was recorded (these were stops that did not include map consultation). The assumption made was that if a person could look at the map and visualize the environment, then they would recognize the actual real world environment when 94 confronted with it during the navigation exercise and not have to stop to orient themselves. The construct validity of visualization was assessed by comparing the visualization Test section to the duration of stops in the RWMNE. The results from the regression (stops in the RWMNE as dependent and the Total Score for the visualization test section, independent) does not support the rejection of the null hypothesis (Table 4.22) (scatterplot Figure 4.11). Dep Var: STOPS N: 44 Multiple R: 0.170 Squared multiple R: 0.029 Adjusted squared multiple R: 0.006 Standard error of estimate: 73.348 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT 93.630 18.161 0.000 . 5.156 0.000 VISZ -0.012 0.011 -0.l70 1.000 -1.115 0.271 Table 4.22 Linear Regression with the number of stops as dependent variable (stops) and the visualization section of the Predictor Test as dependent (visualiz). 95 300 I I I 200 — " — g. : .' o .9 . ° (0 3 o o o 100 I— . . O . _‘ o. 0 O a. ' . . . 0 M1 .L. I l e ’ e 0 1000 2000 3000 4000 Visualization Figure 4.11 Scatterplot with stops and visualization Sleuthing. The null hypothesis was that the sleuthing section of the map reading ability test does not predict a subject’s ability to locate oneself on the map during the RWMNE; the alternative hypothesis was that the sleuthing section does predict locating ability. The construct of sleuthing should reflect people’s adeptness at using the clues in the real world environment to locate themselves on the map. The pre-experiment prediction was that the score from the sleuthing section of the computer test would predict how accurately and quickly a person performed the two locating tasks in the 96 RWMNE. A score on the locating task was achieved through combining the accuracy of the location with the time (number of seconds taken to complete the task). The accuracy generated a multiplier (based on proximity to the true location), by which time was adjusted to obtain a score for the task. A comparison of the sleuthing section of the Predictor Test and the location tasks provided an assessment of the construct validity of sleuthing. The scores from the computer test (independent variable “sleuth”) were regressed against the scores from the RWMNE (dependent variable “locating”). Because the overall model was very significant (Table 4.23) (scatterplot Figure 4.12) the null hypothesis was rejected. As before, all other test sections were regressed against “locating” and while visualization and object rotation were significant, they explained less in the model than did sleuthing (Table 4.24). Dep Var: LOCATING N: 44 Multiple R: 0.854 Squared multiple R: 0.729 Adjusted squared multiple R: 0.722 Standard error of estimate: 50.360 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT -41.975 16.225 0.000 . -2.587 0.013 SLEUTH 0.005 0.001 0.854 1.000 10.624 0.000 Table 4.23 Linear Regression (locating-dependent, sleuth-independent) 97 Locating Task 70000 60000 I 50000 I I 40000 30000 20000 — .3 C 10000 —" ' I m o o l 0 0 200 Sleuthing 300 400 Figure 4.12 Scatterplot with locating and sleuthing Dep Var: LOCATING N: 44 Multiple R: 0.861 Squared multiple R: 0.741 Adjusted squared multiple R: 0.707 Standard error of estimate: 51.732 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT —64.791 24.205 0.000 . -2.677 0.011 ENVNAV —0.124 0.166 -0.067 0.847 -0.746 0.460 VIS 0.030 0.012 0.323 0.425 2.551 0.015 OR 0.022 0.011 0.193 0.714 1.977 0.055 MEM -0.000 0.004 -0.006 0.848 -0.069 0.946 Sleuth 0.050 0.012 0.566 0.379 4.223 0.000 Table 4.24 Linear Regression with the locating task as dependent variable and all sections of the Predictor Test as dependent. 98 Spatial Memory. The null hypothesis was that the spatial memory section of the map reading ability test does not predict the amount of time each subject spends consulting the map during the map reading/navigation exercise; the alternative hypothesis was that the spatial memory section does predict map study time. As presented in Chapter 2, the construct of spatial memory asserts that a person’s memory may influence how quickly a map navigation task is accomplished. Assessing the validity of the spatial memory construct should show how well the memory Test actually measured a person’s memory in map navigation. Comparing the Test section to the amount of time each subject spent studying the map would reveal the relationship and give an indication of the validity of spatial memory as a cognitive process of map reading. The amount of time (in seconds) that each subject spent consulting the map in the RWMNE was recorded. The score from the memory section of the map reading ability test (independent variable “memory”) was regressed against the score from the map navigation exercise (dependent variable “looks”). The results indicate a significant model (Table 4.25) (scatterplot Figure 4.13) and the null hypothesis was rejected. Again, a multiple regression including all Test sections was run to determine whether any other section was a significant predictor of the dependent variable. Object rotation was a significant predictor (Table 4.26). A simple linear regression was run (Table 4.27) (scatterplot Figure 4.14) and after removing three high 99 leverages, the resulting model revealed a significant relationship with a slightly lower multiple square R than memory (Table 4.28) (scatterplot Figure 4.15). Dep Var: LOOKS N: 44 Multiple R: 0.728 Squared multiple R: 0.529 Adjusted squared multiple R: 0.518 Standard error of estimate: 38.227 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT 1.627 13.437 0.000 . 0.121 0.904 MEM 0.020 0.003 0.728 1.000 6.874 0.000 Table 4.25 Linear Regression with duration of looks as dependent variable (looks) and spatial memory section of the Predictor Test as independent (mem2). 250 I I I I I I I I I 200 — - . ~ 0 150 — ° - 3’" : § ' - ' . 100- .o ‘ ~ 0 .. 3 0 ° ' ' 50 __ . .. g . . _ a ' ' o l l l l l l l l l 0 «90 49$ '5 5. 6°00 09$ 1°°° 59°“ 99610900 Memory Figure 4.13 Scatterplot with looks and memory 100 Dep Var: LOOKS N: 44 Multiple R: 0.819 Squared multiple R: 0.671 Adjusted squared multiple R: 0.628 Standard error of estimate: 33.605 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT —12.505 15.724 0.000 . —0.795 0.431 ENVNAV -0.078 0.108 -0.073 0.847 -0.724 0.474 V182 -0.027 0.008 -0.514 0.425 -3.599 0.001 0R2 —0.020 0.007 -0.302 0.714 -2.745 0.009 MEM2 0.020 0.003 0.724 0.848 7.168 0.000 SL 0.028 0.008 0.550 0.379 3.638 0.001 Table 4.26 Linear Regression with duration of looks as dependent variable (looks) and all sections of the MRAT as independent. Dep Var: LOOKS N: 41 Multiple R: 0.402 Squared multiple R: 0.161 Adjusted squared multiple R: 0.140 Standard error of estimate: 50.648 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT 21.402 24.466 0.000 . 0.875 0.387 0R2 0.114 0.042 0.402 1.000 2.740 0.009 Table 4.27 Linear Regression with duration of looks as dependent variable (looks) and object rotation as independent. 101 250 I I r I 200 e 0 o — O . O m 150 — ' — x . ’ .8. 100 ~ :1 . O :7 O ‘ . 50 — .I o . ‘ J" ' O O l I I l 0 1000 2000 3000 4000 5000 Object Rotation Figure 4.14 Scatterplot with looks and object rotation Dep Var: LOOKS N: 41 Multiple R: 0.402 Squared multiple R: 0.161 Adjusted squared multiple R: 0.140 Standard error of estimate: 50.648 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT 21.402 24.466 0.000 . 0.875 0.387 OR 0.114 0.042 0.402 1.000 2.740 0.009 Table 4.28 Linear Regression with duration of looks as dependent variable (looks) and object rotation as independent (leverages removed). 102 250 I I I I I I I O 200 l- 0 o — 0 w 150 — ° 4 "‘8‘ ' '- - _l . o . 100 — . o o . a oo. o. . . 50 '- 2 0 ° . 1 . O. '0'. 0 O 1 1 1 l I 4° 5° 3° ,3 «0° 5° 90° 5° Object Rotation V350 Figure 4.15 Scatterplot with looks and object rotation (leverages removed) Environmental Navigation. The null hypothesis was that the environmental navigation test does not significantly explain the variance in the environmental navigation section of the RWMNE (identified below); the alternative hypothesis was that the environmental navigation test does explain the variance. The construct of environmental navigation refers to the extent to which a person learns and remembers an area by working within it without a map. An evaluation of this ability was obtained in a section of the Real-World Map Navigation 103 Exercise by measuring how well a subject navigated through the building without referring to the map. Recall that the map navigation exercise required each subject to locate four positions in the building and the positions were verbally given to the subject upon their successful navigation to the previous one; the route from position two to position three required the subject to pass position four. Therefore, a subject’s ability to remember the environment and navigate from position number three to number four without having to consult the map provided a measure of their ability to mentally encode the physical environment. The amount of time (in seconds) spent studying the map between location three and four (“look4”) constituted the environmental navigation portion of the RWMNE, dependent variable. An analysis of the environmental navigation Test section and the “look4” results from the RWMNE should indicate how well this Test section measured the construct of environmental navigation. The environmental navigation score from the ability exercise was regressed against “look4” (Table 4.29) (scatterplot Figure 4.16) and the results of the analysis indicated that the relationship was unlikely to have occurred by chance (p=.05 7), but the amount of explanation is very low (squared multiple R=.058) and unlikely to be of any practical significance. Next, all five parts of the test were regressed against “look4”; this multiple regression identified memory as the only significant independent variable in the model (Table 4.30). A simple linear regression of memory (independent) against “look4” is a reasonable model (Table 4.31) (scatterplot Figure 4.17). 104 Dep Var: LOOK4 N: 44 Multiple R: 0.241 Squared multiple R: 0.058 Adjusted squared multiple R:0.036 Standard error of estimate: 4.453 Effect Coefficient Std Error Std Coef Tolerance t (2 Tail) CONSTANT 4.125 1.021 0.000 . 4.040 0.000 ENVNAV 0.021 0.013 0.241 1.000 1.609 0.115 Table 4.29 Linear Regression with the duration of looks between location #3 and #4 as the dependent variable (Look4) and the Environmental Navigation section of the Predictor test as independent(envnav). 20 I I I I o 15— ~ ‘- a . o . X 10— . coo a 9 (I oo o o 5— 000 .0 o . a o - oo o o. o o DOM—4% 0 50 100 150 200 250 Environmental Navigation Figure 4.16 Scatterplot with look4 and environmental navigation 105 Dep Var: LOOK4 N: 44 Multiple R: 0.806 Squared multiple R: 0.649 Adjusted squared multiple R: 0.603 Standard error of estimate: 2.858 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT —3.599 1.337 0.000 . —2.691 0.011 ENVNAV 0.001 0.009 0.007 0.847 0.064 0.949 VIS -0.002 0.001 -0.371 0.425 -2.520 0.016 OR -0.001 0.001 -0.155 0.714 -1.363 0.181 MEM 0.002 0.000 0.694 0.848 6.644 0.000 SL 0.002 0.001 0.490 0.379 3.136 0.003 Table 4.30 Multiple Regression (Look4-dependent, All test sections-independent) Dep Var: LOOK4 N: 44 Multiple R: 0.744 Squared multiple R: 0.553 Adjusted squared multiple R: 0.542 Standard error of estimate: 3.068 Effect Coefficient Std Error Std Coef Tolerance t P(2 Tail) CONSTANT -1.658 1.078 0.000 . —l.537 0.132 MEM 0.002 0.000 0.744 1.000 7.207 0.000 Table 4.31 Linear Regression with the duration of looks between location #3 and #4 as the dependent variable (Look4) and the Memory section of the Predictor test as independent (memory). 106 20 I I I I I I T I I 15* ~ o o V o o x 10»- o o oo o o — 5 oo o o o 5— oo o o o — ooo o o ooo oo O—J—o-I—oo-LGJe 1 l 1 I L Q K®° '1'qu 590% “009 (3060 @000 1&0 $630 QQQQNQQQQ Spatial Memory Figure 4.17 Scatterplot with look4 and memory The analyses of the data resulted in some unexpected but pertinent discoveries. The significance of all of the findings revealed above will be discussed in the following chapter. 107 Chapter 5 Discussion The previous chapter presented the data, scores, and results of the statistical analyses. This chapter will discuss these results including the significance of the MRAT as a Predictor Test and the implications of the findings for each individual cognitive process and will conclude with suggestions of how the study might have been improved. Predictor Test The Predictor Test was shown to be effective in foretelling a person’s ability on the Real-World Map Navigation Exercise. However, not all sections were equally effective and some were not statistically significant at all. The results suggest that the Predictor Test should be revised; some suggestions for improving the Test will be addressed in the section on Improvement on the Study. While it is possible to predict a person’s map reading and navigation ability using the Predictor Test, suggesting performance levels for the test is complicated by the idea that navigation ability requirements may differ from one application to another. The determination of a “good” or “bad” score is dependent on the acceptable level of navigation ability, which is determined by the content of the application. For example, using the test to select people with “good” navigation ability in conjunction with employment will be dictated by the importance of that 108 navigation ability to job performance. Census takers, delivery persons, and police officers travel to many new locations and are required quite frequently in their jobs to navigate using maps. With other occupations, such as mail carriers, who are required to navigate and read maps (at least initially), but maintain the same route consistently, map reading and navigation ability is not as important in the function of their job. Therefore, the requirements of the test administrator (such as the employer) would determine acceptable test performance levels. For practical applications of a MRAT, such as those discussed above, an alternative scoring method may be more appropriate. For the purposes of this research, it was not necessary to identify high- or low-scoring individuals, as the objective was only to look for relationships. Therefore, scores based on response times alone were calculated. However, for purposes such as employment selection, accuracy of responses may be an important criterion and could be considered in the determination of level of fitness. A certain level of error may disqualify an applicant regardless of how quickly the person answered for correct responses. Object Rotation Regression analyses determined that both the object rotation and the memory sections of the Predictor Test were adequate predictors of whether a person rotates a map during map reading and navigation. Although object rotation was a significant predictor of map rotation, the memory test section proved to be more significant. The results of this regression analysis along with the results of the factor analysis, in 109 which the object rotation and the memory test sections loaded together, prompted a search to locate a possible explanation of the relationship. According to Lezak (I 995), the source of human spatial memory and object rotation ability is the parietal section of the brain, sometimes causing an observable overlap in the processes. In fact, she notes that due to the close relation between the two processes, impairments in spatial memory and mental rotation ability are often observed concurrently in specific types of brain damaged patients. Further, the Spatial Orientation Memory Test, developed by Wepman and Turaids, specifically addresses memory and object rotation by showing subjects a card containing a rotated object, then removing the card and asking them to identify from a set of figures a second figure in the same stage of rotation as the first. Spatial memory may be required in mental rotation processes because as one mentally rotates an object, one must presumably remember the original location and be able to make continuous shifts in that memory as the mental rotation occurs. This evidence provides explanation for the unexpected appearance (in both the factor and regression analyses) of the relationship between the spatial memory and object rotation sections of the map reading ability test. Spatial Memory Regression analysis identified the spatial memory section of the Predictor Test to be a significant predictor of memory during navigation. A person’s ability to remember routes and the map environment influences how much time they spend studying the map. While spatial memory may not affect whether someone can 110 accomplish a map reading and navigation task, it may affect how quickly they can accomplish it. These results along with results from the previous section, indicate that memory is a valid cognitive process influencing a person’s map navigation ability. The relationship between object rotation appeared again, as the object rotation section of the MRAT was a significant predictor of the memory assessment in the RWMNE. Visualization While the visualization section of the Predictor Test was a statistically significant predictor (p=.032) of stops, or time spent getting bearings without a map, it failed to explain most of the variance. Three possible explanations exist for the low squared multiple R: first, visualization may not be a cognitive process used in map reading; second, using the number of stops to indicate visualization ability may not be valid; third, the visualization section of the map reading ability test may be faulty. Alternatively, the explanation may lie in some combination of the three. It is possible that the relationship between visualization and sleuthing reveals the answer. While the constructs are different conceptually, perhaps they use the same ability; or, perhaps there is a significant amount of cognitive overlap, just as there is with spatial memory and object rotation. The problem with the validity measure would then lie with the methodology of the research and the visualization section of the Predictor Test, in particular. Because of the correlation of the visualization and sleuthing lll sections of the test, it is possible that the visualization section assessed sleuthing ability rather than visualization ability. Or, perhaps what I have termed visualization is merely what results in the combined use of spatial memory and sleuthing. In fact, as the concept has been discussed, it is spatial memory that may impact how well a person recognizes an environment represented on a map and it is sleuthing that may impact how well they relate the clues in that environment to their memory of the map. Regardless, a re-structuring of the visualization test section to better represent the concepts as defined could lend some insight into whether it should be considered a separate and unique cognitive process of navigational map reading. Sleuthing The sleuthing section of the Predictor Test significantly predicted a person’s “sleuth” ability during navigational map reading, or their ability to relate the clues in the environment and locate themselves on the map. In fact, not only was the sleuthing Test section the most significant predictor of overall map reading and navigation ability, but the Test section also most effectively predicted its own act during the RWMNE. In other words, the sleuthing Test section predicted sleuthing in the navigation exercise better than the memory Test section predicted memory in the RWMNE and better than the object rotation section predicted map rotation. These results indicate that sleuthing may be a the most significant construct of map reading ability. 112 Environmental Navigation The environmental navigation score from the Predictor Test was regressed against “look4”, the amount of time (in seconds) spent studying the map between location three and four. The results indicate no significant relationship between the variables. However, when all five sections of the Predictor Test were regressed against “look4”, memory was shown to be a significant predictor. This relationship may be explained by the idea that the environmental navigation construct is a person’s ability to learn and remember an environment through experience in that environment. The inability of the environmental navigation section of the ability test to predict the “look4” score is probably due to poor construction of that test section. In the test, without exception, every subject was able to identify the correct hallway and the correct side of the hallway; the only difference in scores was in distance from the markers. The environment was too simple and, therefore, the environmental navigation section of the test does not measure a person’s ability to learn an environment without the aid of a map, but may actually measure a person’s distance estimation ability. Improvements on the Study The entire study could be improved by increasing the sample size, which would add power to the analyses, i.e., it would reduce the likelihood of failing to discover a relationship that exists. Also, the environmental navigation section of the 113 map reading ability test needs a complete revision. For ease of administration, the environmental navigation section could be included as a section in the computer test by creating a ‘virtual maze’ requiring the subjects to navigate through without a map. But, regardless of how the environmental navigation section is administered, the environment included in the test should be made much more intricate. One of the causes for the failure of the environmental navigation section in this research was that the environment was much too simple, with very few turns and few “streets”. Another test section that should undergo substantial revisions is visualization, which was shown to be more a measure of sleuthing than visualization ability. As used in this research, the section showed subjects a location on a map and four drawings of objects (hotels, parks, office buildings...) that could be located in the environment; the subject was asked to decide which object they would encounter first as they traveled on a road. The design may be improved by again indicating a location on a map, but then showing the subject photos of locations (including buildings, street...) in different directions and asking the subject to identify which image they would see first. Alternatively, the entire visualization section could be deleted from the Test altogether, leaving a much simpler and possibly just as effective Predictor Test. If it is not a separate process, that alternative would be the logical one. The object rotation section may beimproved by adapting traditional pen and pencil mental rotation tests to the computer test. These traditional tests show a subject a two- or two-and-a-half- dimensional object and then present a series of 114 objects at differing rotations. The subject must identify the figure that represents the original object in a rotated state. Finally, the spatial memory section could be improved by including a test such as the one used by Montello et al. (1999) and Lobben (1996) in which a group of objects is shown in a scene for a limited time, after which the subjects must place the same objects in a blank scene, identifying the correct relative and metric placement of the objects. This type of test can be adapted to a computer environment by showing objects in a scene for a given amount of time, then providing a list of objects across the bottom (or side) of the screen, which they must drag (using the mouse) to the correct location as seen in the previous scene. The computer could record the response time for each question (as was done in the Predictor Test used in the research presented in this dissertation), providing an advantage over the traditional pencil and paper tests, which can only time an entire section. These suggestions could improve the existing research, maintaining the same research goals. But, research questions could be revised to include new goals, as addressed in the next chapter. 115 Chapter 6 Summary and Concluding Remarks Research in cartography and psychology journals identified several cognitive processes that may influence a person’s map reading and navigation ability. Five processes (object rotation, visualization, sleuthing, spatial memory, and environmental navigation) were isolated as research foci. A map reading ability test assessed a person’s ability in each of the five cognitive processes. An experiment was designed not only to test the predictability of map reading ability from the sit- down test, but also to assess the relative influence of each of the cognitive processes in map reading and navigation. Overall, the multiple regression analysis provided evidence that the map reading ability test provided strong predictability of a person’s performance on the Real World Map Navigation Exercise. The five individual sections of the Test varied in their ability to predict performance; sleuthing was overwhelmingly the best predictor followed by spatial memory. Object rotation and environmental navigation were not statistically significant and visualization was eventually excluded due to tolerance problems with sleuthing. The validity analyses were designed to determine whether the five test sections measured the abilities they were designed to measure. The results varied. The sleuthing, memory, and object rotation sections were shown to measure the 116 intended abilities, while the construct validity of environmental navigation and visualization could not be supported by statistical analysis. Significance of Research With very little attention paid to research into cognitive processes of map reading, this research begins to identify the impact of understanding the role cognitive processes play in map reading and navigation. The identification of potential cognitive processes provided three significant results. First, it led to a synthesis of current as well as past cartographic and psychological cognitive mapping/spatial cognition literature, which could prove to be a valuable resource for anyone who wishes to continue with the research presented in this dissertation. Equally important, the amalgamation of cognitive mapping literature identified areas in which research methods could improve both in cartography (including incorporating analyses of reliability and validity of the testing instrument) and psychology (careful consideration of the type and quality of map used in research). Second, the synthesis provided the subject of study (specific cognitive processes) for the map reading ability assessment test. While the objective of this research was not to create a map navigation test for actual employment selection, results did demonstrate that a test designed to predict map reading/navigation ability in a real environment can be developed and administered with relative ease. This discovery could have a profound impact for those with a need to identify people with map reading/navigation ability (employers, for example), but for whom cost and ll7 practicality do not allow them to perform real-world assessments of the ability. Also, the identification of the relative influence of the cognitive processes allows researchers to better understand the thought processes behind navigational map use. The recognition of influential cognitive processes impacts researchers, for whom the discoveries may provide foundation for further research in similar areas, and educators. Recognizing the complexity of navigational map reading, educators have not been entirely successful in developing a method to satisfactorily teach students how to read a map and navigate through an environment. Identifying influential cognitive processes may result in more effective teaching of map reading and navigation. Sleuthing, for example, was identified in this research as being the most influential process in map reading. Map reading education may be streamlined by concentrating on improving sleuthing skills, which could result in improvement in the overall task of navigational map use. Finally, as discussed in Chapter 2, Literature Review, several researchers have investigated the object rotation (Hintzman, O’Dell, & Amdt 1981, Steinke & Lloyd 1983, Levine, Marchon, & Hanley 1984, Lloyd & Steinke 1984, Aretz & Wickens 1992, Livingstone 1992). Generally, these researchers indicate that object rotation may play a significant role in both the speed and accuracy of map navigation. Research presented in this dissertation reveals that object rotation is most likely a process used in map reading, as was observed in the RWMNE and predicted by the MRAT. However, while the process of object rotation plays a significant role in navigational map use, it is not a predictor of accuracy or efficiency of the RWMNE. 118 Additional research that investigates the influence of an object rotation predictor test on a real world map exercise may be warranted. Suggestions for Further Research As this research progressed, additional questions came to light. These questions could lead to new research projects, not only supporting, but also building on the results found here. First, this research investigated the relationship between scores on an administered map reading ability test and a person’s ability to navigate with a map. While a strong relationship exists, the relationship between a person’s general intelligence and success on the map reading ability was not known. Therefore, administering an intelligence test such as the Stanford Binet along with the map reading ability test may provide some insight into the relationship between intelligence and map reading/navigation ability. Second, strategies used with the separate processes may be studied. Only a very slight distinction was made between the cognitive abilities and the strategies people use in map reading/navigation. Some research has investigated general map reading and navigation strategies (Thomdyke and Hayes-Roth 1982, Golledge 1999) while some studies examine strategies used with specific cognitive processes during map reading and navigation, including spatial memory strategies (Gilhooly et a1. 1988, Winn and Sutherland 1989) and self-locating strategies (Peruch et al. 1986). From the research of this dissertation, we can begin to gain an understanding of the exertion of influence of specific processes of navigational map reading. Identifying 119 the strategies used in conjunction with sleuthing, for example, could identify successful techniques. If we can identify these successful strategies and if future studies support the findings here regarding the impact of sleuthing on map navigation ability, then we may be able to teach map-reading skills more effectively. Third, many paper and pencil tests have been developed and administered in attempts to gain a better understanding of spatial ability. Some researchers (Montello et al. 1999) doubt the ability of these paper and pencil tests to capture “the entire spectrum of skills that might reasonably be thought to involve spatial ability (p.517)” and the ability to identify specific processes used in map reading and navigation. Because the Map Reading Ability Test used in this research has, at least initially, predicted map reading and navigation ability, a correlation between the widely used paper and pencil and the MRAT administered in this research may help identify the extent to which paper and pencil tests do account for map reading ability as considered a component of spatial ability in psychological research. The objectives of this study were to determine whether the ability to navigate with a map is predictable and to determine the influence of several cognitive processes on map reading. The results provide initial evidence that map reading and navigation is predictable and variably influenced by cognitive processes. Only five processes were included in the test and investigated for relative influence. Thus, while the inclusion of sleuthing, as defined in this dissertation, may very well be influential on map reading and navigation, additional processes not included here may exert additional influence. Additional research will not only improve the credibility 120 of these results but also expand what we know about the relationship between cognitive processes and map reading. 121 BIBLIOGRAPHY 122 BIBLIOGRAPHY Anastasi, A. and S. Urbina. 1997. Psychological Testing. Upper Saddle River, NJ, Prentice-Hall, Inc. Aretz, A. J. and C. Wickens. 1992. “The Mental Rotation of Map Displays.” Human Performance. Volume 5(4), Pages 303-328. Blades, M. and C. Spencer, 1986. “The Implications of Psychological Theory and Methodology for Cognitive Cartography.” Cartographica. Volume 23(4), Pages 1-13. -----1987. “How Do People Use Maps to Navigate Through the World.” Cartographica. Volume 24(3), Pages 64-75. Blades, M. 1991. “Wayfinding Theory and Research: the need for a new approach.” In DM Mark and AU Frank, Eds. Cognitive and Linguistic Aspects of Geographic Space. Kluwer Academic Publishers. Pages 137-165. Blaut, JM and D. Stea. 1971. “Studies of Geographic Learning.” Annals of the Association of American Geographers. Volume 61, Pages 387-393. ----l974. “Mapping at the Age of Three.” The Journal of Geography. Volume 73(7), Pages 5-9. Blaut, JM. 1997. “Children Can.” Annals of the Association of American Geographers. Volume 87. Pages 152-58. Blaut, JM. 1997. “Piagetian Pessisism and the Mapping Abilities of Young Children: A Rejoinder to Liben and Downs.” Annals of the Association of American Geographers. Volume 87. Pages 168-77. Bluestein and Acredolo. 1979. “Developmental Changes in Map-Reading Skills.” Child Development. Volume 50, Pages 691-697. Bronzaft, A., S. Dobrow, and T. O’Hanlon. 1976. “Spatial Orientation in a Subway System.” Environment and Behavior. Volume 8, Pages 575-594. Chan, D., N. Schmitt, R. P. DeShon, C. S. Clause, and K. Delbridge. 1997. “Reactions to Cognitive Ability Tests: The Relationships Between Race, Test Performance, Face Validity Perceptions, and Test-Taking Motivation.” Journal of Applied Psychology. Volume 82(2), Pages 300-310. 123 Cole, D. G. 1981. “Recall vs. Recognition and Task Specificity in Cartographic Psychophysical Testing.” The American Cartographer. Volume 8(1), Pages 55-66. Crampton, Jeremy. 1992. “A Cognitive Analysis of Wayfinding Expertise.” Cartographica. Volume 29(3&4), Pages 46-65. Dent, Borden. 1996. Cartography: Thematic Map Design. Wm. C. Brown Publishers. Dubuque, IA. Downs, R. M., L. S. Lieben, and D. Daggs. 1991. “On Education and Geographers: The Role of Cognitive Developmental Theory in Geographic Education ” Annals of the Association of American Geographers. Volume 78(4), Pages 680-700. -----, -----, and ----. 1988. “On Education and Geographers: The Role of Cognitive Developmental Theory in Geographic Education.” Annals of the Association of American Geographers. Volume 78. Pages 680-700. Downs, R. M., and L. S. Liben. 1986. “Children’s Understanding of Maps.” Cognitive Processes and Spatial Orientation in Animal and Man, Volume 2, Neurophysiology and Developmental Aspects, ed. P.Ellen and C.Thinus-Blanc. Pages 202-19. Dordrecht, The Netherlands: Martinus Nijhoff. ---- and . 1987 . “Children’s Understanding of Maps.” In P. Ellen and C. Thinus-Blanc, eds., Cognitive Processes and Spatial Orientation in Animal and Man. Boston: Martinus-Nejhoff. ---- and ----. 1988. “Through a Map Darkly: Understanding Maps as Spatial Representations.” Genetic Epistemologist. Volume 16. Pages 11-18. ---- and . 1991. “The Development of Expertise in Geography: A Cognitive-developmental Approach to Geographic Education.” Annals of the Association of American Geographers. Volume 81. Pages 304-27. ---- and ----. 1993. “Mediating the Environment: Communicating, Appropriating, and Developing Graphic Representations of Place.” In Development in Context: Acting and Thinking in Specific Environments, ed. RH Wozniak and K.Fischer. Pages 155-81. Hillsdale, NJ: Lawrence Erlbaum Associates. ----- and ----. 1997. “The Final Summation: The Defense Rests” Annals of the Association of American Geographers. Volume 87. Pages 178-80. 124 Downs, R and Stea, D., eds. 1973. Image and Environment. Chicago: Aldine. Eliot, J. and I.M. McFarlane-Smith. 1983. An International Directory of Spatial Tests. Oxford: NFER—Nelson Publishing Company. Flannery, J. J. 1971. “The Relative Effectiveness of Some Common Graduated Point Symbols in the Presentation of Quantitative Data.” The Canadian Cartographer. Volume 8(2), Pages 96-109. Freundschuh, S. 1990. “Can Young Children Use Maps to Navigate.” Cartographica. Volume 27(1), Pages 54-66. Gagne, R. 1974. Essentials of Learning for Instruction. New York: Dryden Press. Gilhooly, K. J., M. Wood, P. R. Kinnear, and C. Green. 1988. “Skill in Map Reading and Memory for Maps.” The Quarterly Journal of Experimental Psychology. Volume 40A(l), Pages 87-107. Gilmartin, P. P. 1981. “The Interface of Cognitive and Psychophysical Research in Cartography.” Golledge, R. G. 1992. “Place Recognition and Wayfinding: Making Sense of Space.” Geoforum. Volume 23 (2), Pages 199-214. Groop, R. E. and D. G. Cole. 1978. “Overlapping Graduated Circles: Magnitude Estimation and Method of Portrayal.” The Canadian Cartographer, Vol. 15. No. 2 (Dec), pp.114-122. Hayes-Roth, B. and F. Hayes-Roth. 1979. “A Cognitive Model of Planning.” CognitiveScience. Volume 3, Pages 275-310. Herrmann, D. and L. Williams Pickle. 1996. “A Cognitive Subtask Model of Statistical Map Reading.” Visual Cognition. Volume 3(2), Pages 165-190. Hintzman, D. L., C. S. O’Dell, and D. R. Arndt. 1981. “Orientation in Cognitive Maps.” Cognitive Psychology. Volume 13, Pages 149-206. Jenks, G. R. 1975. “The Evaluation and Prediction of Visual Clustering in Maps Symbolized with Proportional Circles,” in John C. Davis and Michael J. McCullagh, eds. Display and Analysis of Spatial Data, John Wiley & Sons, New York. Kimerling, A. J. 1975. “A Cartographic Study of Equal Value Gray Scales for Use With Screened Gray Areas.” The American Cartographer. Vol. 2, pp.119-127. 125 Kitchin, R. M. 1994. “Cognitive Maps: What Are They and Why Study Them?” Journal of Environmental Psychology. Volume 14, Pages 1-19. Kolacny, A. 1969. “Cartographic Information: A Fundamental Concept and Term in Modern Cartography.” The Canadian Cartographer. Volume 6, Pages 47-49. Kovach, R. C. Jr., M. A. Surrette, and M. G. Aamodt. 1988. “Following Informal Street Maps: effects of map design.” Environment and Behavior. Volume 20(6), Pages 683-699. Kulhavy, R. W., N. H. Schwartz, and S. H. Shaha. 1983. “Spatial Representation of Maps.” American Journal of Psychology. Volume 96(3), Pages 337-351. Kulhavy, R. W. and W. A. Stock. 1996. “How Cognitive Maps are Learned and Remembered.” Annals of the Association of American Geographers. Volume 86, Pages 123-145. Levine, M., I. N. Jankovic, and M. Palij. 1982. “Principles of Spatial Problem Solving.” Journal of Experimental Psychology: General. Volume 1 1 1(2), Pages 157-175. Levine, M., I. Marchon, and G. Hanley. 1984. “The Placement and Misplacement of You-Are-Here Maps.” Environment and Behavior. Volume 16(2), Pages I39-157. Liben, L. S. and R. M. Downs. 1989. “Understanding Maps as Symbols: the development of map concepts in children.” In HW Reese, ed. Advances in Child Development and Behavior. New York: Academic Press. ----- and . 1991. “The Role of Graphic Representations in Understanding the World.” In Visions of Aesthetics, the Environment, and Development: The Legacy of Joachim F. Wohlwill, ed. R.Downs, L.Liben, and D.Palermo. Pages 139-80. HillsdalezErlbaum. ----- and . 1992. “Developing an Understanding of Graphic Representations in Children and Adults: The Case of GEO-graphics.” Cognitive Development. Volume 7. Pages 331-49. ---- and ----. 1994. “Fostering Geographic Literacy from Early Childhood: The Contributions of Interdisciplinary Research.” Journal of Applied Developmental Psychology. Volume 15. Pages 549-69. ----- and ----. 1997. “Can-ism and Can’tianism: A Straw Child” Annals of the Association of American Geographers. Volume 87. Pages 159-67. 126 Lloyd, R., R. Cammack, and W. Holliday. 1994. “Learning Environments and Switching Perspectives.” Lloyd, R. and R. Cammack. 1995. “Constructing Cognitive Maps with Orientation Biases.” In M. Portugali, ed. The Construction of Cognitive Maps. Dordrecht: Kluwer Academic Publishers. Lloyd, R. and T. Steinke. 1984. “Recognition of Disoriented Maps: the cognitive processes.” Cartographic Journal. Volume 21, Pages 55-59. ----—-1986. “The Identification of Regional Boundaries on Cognitive Maps.” The Professional Geographer. Volume 38, Pages 149-159. Lloyd, R. 1989. “Cognitive Maps: Encoding and Decoding Information.” Annals of the Association of American Geographers. Volume 79(1), Pages 101-124. Lobben, A. 1997. “An Analysis of Pre-School Children’s Mapping Abilities in a Familiar Environment.” Paper presented at the annual of the North American Cartographic and Information Society. Lexington, Kentucky. MacEachren, A. M. 1992a. “Application of Environmental Learning Theory to Spatial Knowledge Acquisition from Maps.” Annals of the Association of American Geographers. Volume 82(2), Pages 245-274. -----1992h. “Learning Spatial Information From Maps: Can orientation-Specificity Be Overcome. " The Professional Geographer. Volume 44(4), Pages 431-443. MacEachren, A.M., B.P. Buttenfield, J.B. Campbell, D.W. DiBiase, and M. Monmonier. 1992. “Visualization.” In RF Abler, MG Marcus and JM Olson, eds. Geography 's Inner Worlds: pervasive themes in contemporary American geography. NJ: Rutgers University Press. Meyer, J. 1973. “Map Skills Instruction and the Child’s Developing Cognitive Abilities. ” Journal of Geography. September, Pages 27-35. Miller, G. A. 1956. “The Magic Number Seven, Plus or minus Two: some limits on our capacity for processing information.” The Psychological Review. Volume 63(2), Pages 81-97. Monmonier, M. 1974. “Measures of Pattern Complexity for Choropleth Maps.” American Cartographer. Volume 1(2), Pages 159-169. Morrison, J. L. 1976. “The Science of Cartography and its Essential Processes.” International Yearbook of Cartography. Volume 16, Pages 84-97. 127 Olson, J. M. 1975. “Autocorrelation and Visual Map Complexity.” Annals of the Association of American Cartographers. Volume 65(2), Pages 189-204. ----- 1979. “Cognitive Cartographic Experimentation.” The Canadian Cartographer. Volume 16(1), Pages 34-44. Ormrod, J. E., R. K. Ormrod, E. D. Wagner, and R. C. McCallin. 1988. “Reconceptualizing Map Learning.” American Journal of Psychology. Volume 101(3), Pages 425-433. Ottosson, T. 1988. “What Does It Take to Read a Map?” Cartographica. Volume 25(4), Pages 28-35. Peterson, M. P. 1985. “Evaluating a Map’s Image.” The American Cartographer. Volume 12, Pages 41-55. Ratajski, L. 1973. “The Research Structure of Theoretical Cartography.” International Yearbook of Cartography. Volume 13, Pages 217-227. Robinson, A. H., J. L. Morrison, P. C. Muehrcke, A. J. Kimerling, S. C. Guptill. 1995. Elements of Cartography. John Wiley & Sons, Inc: New York. Shepard, R., and S. Hurwitz. 1984. “Upward Direction, Mental Rotation, and the Discrimination of left and Right Turns in Maps.” Cognition. Volume 18, Pages 161-193. Sholl, J. J. and H. E. Egeth. 1982. “Cognitive Correlations of Map-Reading Ability.” Intelligence. Volume 6, Pages 215-230. Steinke, T. and R. Lloyd. 1983. “Images of Maps: a rotation experiment.” The Professional Geographer. Volume 35, Pages 455-461. Stevens, A. and P. Coupe. 1978. “Distortions in Judged Spatial Relations.” CognitivePsychology. Volume 10, Pages 422-437. Streeter, L.A. and D. Vitello. 1986. “A Profile of Drivers’ Map-Reading Abilities.” Human Factors, Voltune 28, Pages 223-239. Thomdyke, P. W. and B. Hayes-Roth. 1982. “Differences in Spatial Knowledge Acquired from Maps and Navigation.” Cognitive Psychology. Volume 14, Pages 560-589. Thomdyke, P. and C. Stasz. 1980. “Individual Differences in Procedures for Knowledge Acquisition from Maps.” Cognitive Psychology. Volume 12, Pages 137-175. 128 Winn, W. D. and S. W. Sutherland, 1989. “Factors Influencing the Recall of Elements in Maps and Diagrams and the Strategies Used to Encode Them.” Journal of Educational Psychology. Volume 81(1), Pages 33-39. Winn, W. D. 1991. “Learning from Maps and Diagrams.” Educational Psychology Review. Volume 3(3), Pages 211-247. 129 "1111111111111111“