THE EGOCENTRIC MAP PERSPECTIVE IN THEMATIC CHOROPLETH MAPS by MATTHEW E. MILLETT A THESIS Presented to the Department of Geography and the Graduate School of the University of Oregon in partial fulfillment of the requirements for the degree of Master of Arts September 2010 11 "The Egocentric Map Perspective in Thematic Choropleth Maps," a thesis prepared by Matthew E. Millett in partial fulfillment of the requirements for the Master ofArts degree in the Department of Geography. This thesis has been approved and accepted by: ;)3 AL{JlA..S+ dOlO Date Committee in Charge: Dr. Amy Lobben, Chair Dr. Dan Gavin Accepted by: Dean of the Graduate School © 2010 Matthew E. Millett III IV An Abstract of the Thesis of Matthew E. Millett in the Department of Geography for the degree of to be taken Master of Arts September 2010 Title: THE EGOCENTRIC MAP PERSPECTIVE IN THEMATIC CHOROPLETH MAPS Approved: Dr. Amy Lobben Choropleth maps are a popular way of depicting spatial data. The map communication model, which theorizes that geographic information is transmitted from the cartographer to the map user via a map, suggests that cartographers are responsible for clearly conveying spatial data in a way all map users can understand. Map users, however, come from different places and may harbor certain regional biases. This thesis investigates whether map users tend to focus on data patterns within their home regions during the visual-search and decision-making processes when reading classed choropleth maps, thereby exhibiting an egocentric map behavior. Seventy-one subjects took a computer-based test asking them to identify various phenomena on a series of choropleth maps of the lower 48 states. The results show a weak positive effect of egocentric map behavior; subjects who lived in a particular state longer were slightly more likely to choose states nearby their home region. CURRICULUM VITAE NAME OF AUTHOR: Matthew E. Millett PLACE OF BIRTH: St. Cloud, Minnesota DATE OF BIRTH: July 23,1971 GRADUATE AND UNDERGRADUATE SCHOOLS ATTENDED: University of Oregon, Eugene, Oregon University of Wisconsin-River Falls, River Falls, Wisconsin DEGREES AWARDED: Master ofArts, Geography, 2010, University ofOregon Bachelor of Science, Geography, 2008, University of Wisconsin-River Falls AREAS OF SPECIAL INTEREST: Geographic Information Systems (GIS) PROFESSIONAL EXPERIENCE: Graduate Teaching Fellow, Department of Geography, University of Oregon, Eugene, Ore., 2008-2010 Intern, Land Management Information Center, St. Paul, Minn., 2007 Student Worker, Department of Geography, University of Wisconsin-River Falls, River Falls, Wis., 2006-2007 v VI GRANTS, AWARDS AND HONORS: Department of Geography Research Grant, The Egocentric Map Perspective in Qualitative Classed Choropleth Maps, University of Oregon, 2010 Graduate Teaching Fellowship, University of Oregon, 2008-2010 Excellence of Scholarship Award, Association of American Geographers and National Council for Geographic Education, 2008 Excellence for Academic Achievement Award, University of Wisconsin- River Falls, 2008 Outstanding Cartographer Award, University of Wisconsin-River Falls, 2008 Department of Geography Scholarship, University of Wisconsin-River Falls, 2007-2008 Animated Map of the Year, Estimating Population Change in Africa, 1950-2050, University of Wisconsin-River Falls, 2008 Map of the Year, African-American Population Growth in Minneapolis, 1950-2000, University of Wisconsin-River Falls, 2007 Map of the Year, Estimated World Population Change, 2005-2050, University of Wisconsin-River Falls, 2006 Vll ACKNOWLEDGMENTS In addition to my committee members Amy Lobben and Dan Gavin, I would like to thank Derek Miller, Robert Pickett, Lindsay Naylor, Nick Martinelli, Rene Kladzyk, Tom Ptak, and Alexandra Marcus for their ideas and assistance. The completion of this thesis, however, would not have been possible without the loving support of Antra Renault. This thesis is dedicated to Calvin, Camilla, and Marigold. Vlll IX TABLE OF CONTENTS Chapter Page I. INTRODUCTION 1 II. MAP COMMUNICAnON, MAP COGNlTION, AND EGOCENTRIC MAP BEHAVIOR................................................................................................................ 4 Map Communication......... 4 The Communication Process 6 Map Cognition 10 Map Reading and Visual Search 10 Mental Maps 12 Map Knowledge 14 Egocentric Map Behavior..................................................................................... 16 III. METHODS AND PROCEDURES....................................................................... 17 Thematic Choropleth Maps............... 17 Test Instrument..................................................................................................... 18 Map-Reading Task.......................................................................................... 19 Qualitative/State Maps 20 Quantitative/County-Cluster Maps 23 Fictitious Phenomena...................................................................................... 26 Map Quiz......................................................................................................... 27 xPage 27 29 29 30 31 31 37 37 40 41 44 47 47 Test Construction .. Survey . Participants and Recruitment . Test Procedure . Chapter IV. ANALYSIS AND RESULTS .. Independent Variables . Dependent Variables .. Mean Pixel Distance . State and County-Cluster Selections . Regression . V. DISCUSSION AND CONCLUSIONS .. APPENDICES . A. MAPS IN THE MAP-READING TASK . B. ZONE AND COLOR SCHEMES, AND CATEGORY ASSIGNMENTS FOR MAPS IN THE MAP-READING TASK................................................ 65 C. SURVEy.......................................................................................................... 71 D. RECRUITMENT SCRIPT 72 E. CONSENT TO PARTICIPATE....................................................................... 73 F. MAP-BY-MAP RESlJLTS............................................................................... 74 REFERENCES............................................................................................................ 76 Xl LIST OF FIGURES Figure Page 1. City of Constantinople, 1526... 5 2. The Map-Model Cycle 6 3. The Map Communication Model.......................................................................... 7 4. Visual Search Example With One Target and 17 Distracters 11 5. Anchor Points May Aid in the Formation of Mental Maps 13 6. Binomial Coefficient Equation and Results.......................................................... 19 7. State Zones for Qualitative/State Maps 21 8. Cluster Zones for Quantitative/County-Cluster Maps 24 9. Distribution of Quiz Scores and Corresponding t-Test......................................... 34 10. Oregonians' Quiz Scores 35 11. Non-Oregonians' Quiz Scores 36 12. Comparing Oregonians' and Non-Oregonians' Quiz Scores................................ 36 13. Pixel Grid Used to Determine State Centroid Locations...................................... 38 14. Pixel Grid Used to Determine County-Cluster Centroid Locations 38 15. Scatterplot and Regression Line of Rome State and Mean Pixel Distance 43 xu LIST OF TABLES Table Page 1. Comparing Self-Reported and Calculated Measures ofHome State.................... 33 2. XY Pixel Coordinates of County-Cluster Centroids............................................. 39 3. XY Pixel Coordinates of State Centroids , 39 4. Comparing Mean Pixel Distance Between Oregonians and Non-Oregonians 40 5. Percentage of Clicks on Oregon When Oregon Was in the Target Category 41 6. Comparing Mean Number of Oregon Selections.................................................. 42 7. Linear Regression Coefficients for Independent and Dependent Variables......... 43 1CHAPTER I INTRODUCTION Since the adoption of cartography as a communication science in the 1960s, researchers in the cartographic community have frequently focused their efforts on improving production and symbolization methods in order to facilitate the communication of spatial information from the cartographer to the map reader (MacEachren 1995). The academic literature is replete with such research, from Flannery's investigation on the perception of graduated symbols in 1956 to Jenks' 1967 study on data classification to Harrower and Brewer's research on color schemes in 2003. More recently, however, researchers have begun to investigate another component to map communication - how map readers contextualize what they are viewing based on their own experiences and sensibilities. This thesis contributes to the more-recent map-use investigations by focusing on how map readers' individual perspectives influence the exploration and decision-making tasks associated with viewing quantitative and qualitative classed thematic choropleth maps. To that end, the following research questions were posed: • Do map readers exhibit egocentric map behavior during the visual-search and decision-making processes when viewing classed thematic choropleth maps of the United States? 2• Do variables such as age, sex, and prior geographic knowledge significantly affect how such maps are read and discerned? Maps are ubiquitous. Not only are they found in traditional places like atlases, textbooks, and newspapers, but more and more they are appearing in digital format on portable devices such as car-navigation systems, cellular telephones, laptops and netbooks (Peterson 2008). Understanding how maps are read is essential. Local decisions in both the public and private sectors are influenced by the spatial and cartographic perceptions of a few key people (Gould 1973). In Chapter II, a review of selected literature researching and discussing map cognition, map communication and egocentric map perspectives reveals that while numerous articles and books have been published on map cognition and communication, relatively little has been investigated with regard to egocentric map behavior save for the efforts of Saarinen (1999), Gould and White (1986), and Gould (1975). Among other things, their research found that people from different places view and think about the world in different manners, and they have different sets of guidelines from which they derive certain preferences for places. These are the ideas that inspired this research. The methods used to investigate the research questions is discussed in Chapter III. The research questions were investigated through the design and implementation of a computer-based map-reading task and map quiz, and a paper-based demographic survey administered to a group ofparticipants recruited from the University of Oregon. Chapter IV details how the data were analyzed and reports the results of the analysis. Independent variables including age, sex, major, home state, and quiz score 3were related to the dependent variable - the participants' locational choices in the computer-based map-reading task - with some promising but nonetheless mixed results. Lastly, Chapter V offers concluding remarks, and discusses the how the results might relate to previous cartographic and behavioral research. 4CHAPTER II MAP COMMUNICATION, MAP COGNITION, AND EGOCENTRIC MAP BEHAVIOR Because this research is concerned with behavior associated with reading thematic choropleth maps, it is necessary to discuss: 1) How maps have been generally considered historically and how they communicate spatial information (map communication); 2) How people encode and decode spatial information acquired from maps (map cognition); and 3) The ways map readers form cognitive or mental maps and how their frame of reference may affect their spatial knowledge acquisition when viewing maps of familiar and unfamiliar regions (egocentric map behavior). Map Communication According to Muehrcke (1998), "a map has many ingredients of a painting or a poem" (p. 17). Up until the middle of the 20th century, maps were mostly thought of in this light - judged on their artistic and aesthetic qualities. How maps communicated spatial information to the reader was not usually considered (Lloyd 2000). A 1526 map from a Turkish sea atlas shows a detailed view of the city of Constantinople - complete with mosques, homes, walls, other buildings, and trees (Figure 1, Brown 1949) while Nicolas Germanus' work, created circa 1460 and based on Ptolemy's Geographia, depicts 5"-( ;' , \ \ ... / \) Figure 1. City of Constantinople, 1526 (Brown 1949). the known world with such artistic embellishments as cherubic faces representing the 12 winds (Virga 2007). These images are more impressionistic representations rather than the functional maps we are now used to seeing. Two developments in the latter half of the 20th century, however, had a major effect on how maps are considered (MacEachren 1995). First was the publication of Robinson's seminal work "The Look of Maps" in 1952. In his book, Robinson called on researchers to scientifically study and develop principles and practices that encourage good cartographic design with the goal of improving the overall map communication process. Robinson believed that maps should be functional and logically designed to serve some useful purpose (Robinson et al. 1995, Robinson 1952). What resulted was a 6plethora of studies in symbolization and design which looked at objective methods for depicting spatial information and relationships on maps; some studies were also linked to psychophysical research in psychology (Montello 2002, MacEachren 1995, Gilmartin 1981). The second important development, according to MacEachren (1995), was the adoption of the idea of cartography as a communication science. The Communication Process The idea of maps as communication devices developed in the 1960s (MacEachren 1995). Board's (1967) "Map-Model Cycle" (Figure 2) is a flow diagram - albeit a rather Mapping Analysis Sifting Outcome Comparing model with real world The model of the real world Speculative idea Appropriate level of investigation Generated by map reader No .. • Process of reduction Process of selection Data processing Cartographic, design : The model 0;- -------p-=·-=--=--=--=-=-=-=------------.I ~=====::::••••• the real world ~Fiducial scale (with regard to -distance, orientation, area, time)-- Figure 2. The Map-Model Cycle (Based on Board 1967). 7complicated one - depicting the process of how spatial information in the "real" world is relayed by the cartographer through the processes of data selection, scale reduction, and cartographic design, to the map reader where it is then filtered, digested, and more or less learned and understood (Figure 3). - Cartographer's Interpretation -.- Recipient Figure 3. The Map Communication Model (Based on MacEachren 1995). Essentially, certain spatial information is collected and gathered by and transmitted from the mapmaker to the map reader via the map; the spatial information and patterns encoded in the map by the cartographer are then decoded and stored by the map reader in the form of a mental or cognitive map (Lloyd 2000, MacEachren 1995, Board and Taylor 1977, Board 1967). Fearing (1953) noted four basic components of human communication in which all problems of communication lie (as cited in Dent 1972): • The communicator (the sender of the message) • The interpreter (the message's receiver) • The communication content (the information needed to be communicated) • The communication situation (the need to communicate or be communicated to) Relating it to cartography, Dent (1972) describes the communicator as the mapmaker, the interpreter as the map reader, the communication content as the map, 8and the communication situation as assumed since it is clear that information is being communicated via the map. However, successful communication can only occur through mutual comprehension of words and symbols (Fearing 1953 as cited in Dent 1972); understanding occurs when the map reader is able to assign meaning to what he or she is viewing (Dent 1972). Though Robinson himself never explicitly proposed a communication model for cartography, MacEachren (1995) argues that it is clear that Robinson believed that maps must have a pre-determined purpose and that some particular map knowledge is to be discerned from it by the map reader, not necessarily constructed by the map reader. MacEachren (1995) raises three objections to treating maps simply as communication devices. First, there are many ways people read and use maps - e.g., finding a location, navigating, determining land ownership, etc. - and all map readers are not alike. As Balchin (1972) found (as cited in Board and Taylor 1977), it is possible that we are not all born map readers. In addition, some maps - such as topographic maps, for example - may have no explicit theme or topic at all and are therefore difficult to objectively study, while others may communicate information other than what the cartographer originally intended. The cartographic message is often not explicit and can be formed only in the map reader's mind (Dent 1999, Dent 1972). (Board and Taylor (1977) note that although mapmakers cannot consciously incorporate more information into a map than what they are already aware of, map readers may gain further spatial knowledge through the process of induction.) The map message can also be intentionally or unintentionally manipulated by the cartographer to send a distorted representation of 9reality (Monmonier 1996). Second, the communication model has no ability to recognize the importance of art and aesthetics in cartography. And third, how can one objectively study maps if there are those who believe that maps do not objectively represent the real world? Geographers should be concerned with how spatial information is coded, stored, reconstructed and processed in memory (MacEachren 1995, Lloyd 1982, Gilmartin 1981). Further study on spatial decision-making requires that the cartographic community understand how people build and employ cognitive structures (Lloyd 1982). It is necessary, therefore, to examine how map readers construct, process, store and recall spatial information acquired from maps in their own minds. Determining map effectiveness has relevance for anyone using thematic maps in a classroom situation for teaching, or to illustrate their results in a research paper (MacEachren 1995, MacEachren 1982, Gilmartin 1981). Board and Taylor (1977) assert that it is wrong to state that a map contains no information not put into it when it was made. Map readers do not necessarily share the mapmaker's knowledge of cartographic language and symbols. It is important, therefore, for cartographers to be more aware ofmap readers' requirements. Understanding the entire cartographic process from both the cartographer's and the map reader's points of view is key (Lloyd and Steinke 1977). 10 Map Cognition Map cognition - a term coined by Tolman (1948) in his study of rats and their navigation of mazes - is the process in which an individual "acquires, codes, stores, recalls, and decodes information about the relative locations and attributes of phenomena" in his or her spatial environment (Downs and Stea 1973, p. 9). Eysenck et al. (1972, as cited by Gilmartin 1981) described several cognitive processes including perception, discovery, recognition, imaging, judging, memorizing, learning, and speech. Cognitive maps may contain information about location and attributes (MacEachren 1991), each of which are theorized to be stored in separate parts of the brain (Levine et al. 1985 as cited in Rittschof and Kulhavy 1998). Research based on understanding how maps are read and recalled could be more relevant than perceptual research on symbol detection, discrimination, and interpretation (MacEachren 1991). However, it is important to understand that such a process is not observable directly, rather conclusions must be drawn based on results which are observable (Olson 1977). Map Reading and Visual Search Map reading involves searchingfor and identifYing regions, symbols, themes, text, and the geographical order or hierarchy of elements on the page (Bertin 1983). Identification is concerned with matching symbols with the legend, discerning patterns, and assessing the map's internal structure; it is often achieved through the recognition of map features, shapes, and spatial organizations (Morrison 1974 as cited in Board 11 and Taylor 1977). But the way maps are read and understood, and the level of spatial knowledge individuals possess varies considerably from reader to reader (Lobben 2004, Gould and White 1986, Gould 1973). Expertise, culture, sex, age, ethnicity, socioeconomic status, and other variables could affect one's map-reading ability (Slocum et al. 2001). Map readers use specific map-reading strategies to complete map-reading tasks, and each map reader's individual abilities may dictate which strategies he or she may employ (Lobben 2004). Concerning the investigation of an egocentric map perspective, several processes - including visual search, visual attention and orienting - may come into play. Visual search (Figure 4) involves the active scanning of an environment (such as a map) for some particular object (the target) while filtering out extraneous features called distracters (Trick and Enns 1998). Visual search is a fundamental activity in map reading (Lloyd 1977). Visual attention, o o o o o 0 00 00 o o o o o o o o or where the map reader is looking, begins as a kind of spotlight of a certain diameter - one with a wide range of field with a low resolution - and then narrows into a smaller range of field with a high resolution; there is thus a narrowing Figure 4. Visual search example with one target and 17 distracters (Based on Trick and Enns 1998). of attention over time (Humphreys and Bruce 1995). Orienting is simply 12 aligning one's attention with infonnation stored in his or her memory (Posner 1980). For example, a map reader might search for a familiar spot on a map - say, a well-known intersection - to get his or her bearings. All of these processes are the beginning steps in the fonnation ofmap readers' mental maps. Mental Maps Mental maps - the spatial infonnation and patterns encoded in the map by the cartographer, and then decoded and stored by the map reader - are in essence a reference system used by the brain (Board and Taylor 1977). People's mental maps are fonned not only by direct experience in their environment, but also by images from books, radio, television, newspapers, and the Internet (Peterson 2008, Bryant and Tversky 1999, Rittschof and Kulhavy 1998, Kulhavy and Stock 1994, Gould and White 1986). Recent technological changes even allow almost anyone to make maps (Krygier and Wood 2005) and spatial language by itself can also be converted into a mental image (Bryant and Tversky 1999). When a mental image is fonned by studying a tangible map (paper or digital), infonnation about the map's structures (directional relationships between locations) and features (visual infonnation such as the distribution of ink or pixels themselves and other visual variables) are represented in the mental map image (Rittschof and Kulhavy 1998). Physical and cultural landscapes, climate, social attitudes, language, etc., may also be represented in one's mental map (Gould and White 1986). People's behavior often reflects images they have 13 formed from the social and physical environment they perceive rather than the true environment (Gould 1975). Lynch (1960) suggests that navigators or wayfinders form and understand their mental maps through paths (such as roads and sidewalks), edges (boundaries), districts (regions), nodes (focal points), and landmarks (readily identifiable objects). Golledge and Spector (1978), on the other hand, postulate Shopping ~ Home 8 Work that wayfinders construct their mental maps by encoding familiar locations, features, areas, and landmarks as anchor points, .. Shopping,' .'--'"'.'-1b;=:::::::::::;b··l~:::·~"~.._>i:....: H~~·~\·\.. .................. Figure 5. Anchor points may aid in the formation of mental maps (Golledge 1999). which are then used to organize and recall other spatial information. New locations are learned and imposed on the mental map based on their relative locations to the original anchor points (Figure 5). Just as in tangible maps, however, not all the aspects of the environment are represented in mental maps, rather, the spatial information is schematized (Tversky 2000) and reduced into a simpler, more organized form (Kosslyn and Pomerantz 1977) or stored as abstract constructs (Dent 1999). In any case, mental maps affect our 14 spatial behavior and decision-making abilities (Dent 1999, Golledge 1999, Thorndyke and Stasz 1980, Gould 1915). Images of maps, then, could have behavioral effects similar to the reading of tangible maps and mental maps may in fact be functionally equivalent to tangible maps in form if not in content (Kosslyn and Pomerantz 1977 as cited in Lloyd 1982). Mental maps extend beyond map readers' knowledge of spatial relationships and contain social and environmental knowledge as well. This additional information encoded by map readers shapes their individual attitudes toward the world and affects their decision-making behavior (Kitchin 1994, Gould 1973). Map readers come from different places and may have their own opinions about other parts of the world. They might also have an emotional attachment to where they live and thus tend to exaggerate about their home towns. This emotional attachment could wane with distance, and faraway places might often receive less consideration (Gould and White 1986). Their map knowledge, therefore, may have great influence -on their map behavior. Map Knowledge Kulhavy and Stock (1996) discuss two kinds of map knowledge: general and specific. In Western society, general map knowledge develops at an early age due to an exposure to the maps in books, newspapers, magazines and on television, phones, and the Internet (Peterson 2008, Kulhavy and Stock 1996). General map knowledge includes the ability to understand aerial photographs, recognize symbols on maps, have a general 15 understanding of distance and direction, and so forth. Essentially, general map knowledge encompasses all abilities to perceive images as "map-like" (Kulhavy and Stock 1996, p. 124). In addition to general map knowledge, however, map readers learn varying degrees of specific map knowledge, and thus have greater or lesser degrees of familiarity with particular maps. These differences influence the way map readers construct mental images after viewing maps, and further map reading is in turn influenced by information already encoded in memory (Kulhavy and Stock 1996). Familiar shapes encountered on maps can be encoded into memory into what Lloyd (1994) terms as prototypes. Prototypes represent a category of an object stored in the map reader's memory as an abstraction, and that abstraction captures those features that are typical ofthe object. Lloyd's study found that the more often participants were exposed to maps with prototypical characteristics, the more often they learned and used spatial prototypes. Golledge (2002) suggests that geographic knowledge levels change considerably when people are exposed to fundamental geographic principles such as "location, place, connectivity, interaction, distribution, pattern, hierarchy, distance, direction, orientation, reference frame, geographic association, scale, region and geographic representation" (p. 10). It is therefore possible that map readers with a great deal of exposure to maps and geographical concepts may perform differently in map-reading tasks than those who have not received such exposure. 16 Egocentric Map Behavior Few studies have addressed the question central to this research, that is the idea that map readers view small-scale thematic choropleth maps through the lens of their home regions and thus exhibit an egocentric map behavior. Saarinen's (1999) study asked participants from around the globe to sketch a map of the world on a blank piece of paper. The results indicated a strong propensity for drawing European-centered maps, and Europe's relative size was often greatly exaggerated in comparison to the rest of the world; more European place names were also included. Saarinen attributes this phenomenon to the prevalence of Eurocentric world maps and textbooks, and the Eurocentric instruction of world history and geography. In addition, he discovered that participants' home areas were often drawn in much greater detail while less space was devoted to distant or unknown places. Though Gould (1975) and Gould and White's (1986) studies focused on participants' preferences for certain places, they were nonetheless able to conclude that people in certain regions appear to share their spatial images; their mental maps seem to be very similar. Whether or not map readers exhibit egocentric map behavior, it is possible that numerous cultural and locational factors may affect people's map-reading and mental- map-building abilities. 17 CHAPTER III METHODS AND PROCEDURES In order to determine the existence of an egocentric map perspective by map readers, an experiment investigating the use of thematic classed chorop1eth maps of the coterminous United States was designed and administered. The ubiquity of thematic ehorop1eth maps made them a good choice for the experiment since participants would have likely already been familiar with these maps. Participants were recruited both in person and via e-mail from classes and e-mail lists at the University of Oregon and then asked to perform a map-reading task on a laptop computer which consisted of 35 thematic choropleth maps. On each map, participants were asked to read a statement about a particular phenomenon, identitY the location where that phenomenon was occurring, and click the target with the mouse. After viewing all 35 maps, a map quiz asking participants to identitY each of the lower 48 states was administered. Last, participants were asked to fill out a short demographic survey. Thematic Choropleth Maps Thematic choropleth maps capture a single distribution or relationship and depict that distribution or relationship by manipulating the visual variables of hue, value, and chroma to represent a particular quantity within an enumeration unit such as a country, 18 state or county (Robinson 1995). The use of such maps is widespread because they can be made to represent almost any phenomenon visible or invisible, and they can easily depict spatial patterns and relationships (Tyner 1992). Government agencies such as the U.S. Census Bureau, and newspapers, magazines, television, and the Internet all make extensive use of thematic choropleth maps (Harrower and Brewer 2003, Monmonier 1989), therefore they were used to test egocentric map behavior due to their inherent familiarity. Factors such as visual complexity and amount of information presented influence map effectiveness (MacEachren 1982), thus it was important make the maps simple and efficient for the reader to understand. Graphical excellence, as Tufte (2001) calls it, minimizes the burden on a map reader's working memory. The intent ofthe research, after all, is to examine egocentrism in map use, so it was important to limit any distractions introduced from overly detailed or complicated maps. Bertin (1983) sums it up simply: Understanding means simplifying. Test Instrument To construct a test instrument that effectively investigated how map readers from different places and with different levels of geographical place-name knowledge view the same thematic classed choropleth maps, several factors were considered: 1) The duration of the test (i.e., it had to be long enough to gather sufficient data to draw conclusions, but short enough to prevent unnecessary participant fatigue); 2) The visual variables of shape, size, value, and hue as they pertained to state polygons; 3) The effect of Oregon's 19 location in the lower 48 states and its relatively long distance away from other states; 4) Map readers' pre-conceived notions of geographical phenomena and previously learned ideas about states and regions; 5) Participants' choices (which had to be easily captured, stored and output for later analysis). Map-Reading Task Because ofperceptual limitations, fewer classes are better when specific information for particular locations is needed to be portrayed (MacEachren 1982). Bertin (1983) suggests that between three and seven categories is optimal, while Tyner (1992) prefers four to 10 classes. In order to keep as few classes as possible without running the risk of over-generalizing, a four-class qualitative color scheme was chosen for the map- reading task. To keep the duration of the map-reading task relatively short but also have a variety of hues from which to choose for filling the different enumeration units, the binomial coefficient equation (Figure 6) was used to calculate the ideal number of n n! (k) = k!(n-k)! Where n = 7 (number oftotal colors) and k= 4 (number ofcolors needed for each map) k=4 5 5 6 7 n=7 15 35 8 70 9 126 Figure 6. Binomial coefficient equation and results. 20 maps. Seven hues for a four-class scheme yielded 35 total maps. Eight hues would have· produced too many maps while six would have yielded too few. Hues for the qualitative color scheme were based on those recommended on ColorBrewer.org (Brewer 2010). A series of 35 qualitative choropleth maps were therefore created for the map- reading task, 18 of which depicted four classes ofphenomena with states as enumeration units, and 17 of which depicted similar phenomena with concentric rings of county clusters as enumeration units. After creating the county cluster maps, however, it was determined that the maps appeared to look more quantitative in nature; it did not seem logical for qualitative phenomena to occur in such a pattern, so the data and color schemes were changed to reflect quantitative phenomena. Each map contained several states in a target category (determined by hue for the state maps and value for the county cluster maps) from which participants were asked to choose a location. Since it was likely that many participants recruited for the experiment would be from Oregon, targets were created in numerous different locations on the maps. Qualitative/State Maps To achieve an even distribution of states and county clusters near and far from Oregon, a distance scheme was created. States were divided into state zones which corresponded to how many states away a particular state was from Oregon (Figure 7). For example, map readers using Oregon to orient themselves must visually cross through four states to reach any of the states in State Zone 4 (Minnesota, Iowa, Missouri, Kansas, Oklahoma or Texas). 21 2 3 2 3 3 Ilt'-llf' 71)1)(- '1 -- \\11\(" r\',','t1y hllHI... " S[,H~ 1111li;' '; 3 ;; d $(,lh..... J',':.-ly {101'l1 Or"90n 3 3 4 4 . 8 7 10 10 Figure 7. State zones for qualitative/state maps. This admittedly complicated scheme was designed to ensure that no states (other than Oregon) were either over- or under-represented. The idea was to weight the maps with more target states farther west and closer to Oregon. The distribution of target states was further enhanced by keeping the number of target states per zone relatively equal in the state zones that actually contained target states. The randomization process also ensured that no state was either over- or under-represented in non-target categories as well. So for each qualitative/state map, Oregon was either in the target category or not in the target category, and some randomly selected combination of states from different state zones were also in the target category. (See Appendix A for all maps in the experiment.) Since states eastward of State Zone 4 are smaller and farther away from Oregon, and to make the scheme easier to use, State Zones 5-11 were compressed into one large 22 zone simply called State Zone 5. States were randomly assigned a "1," a "2," a "3" or a "4," with a "I" being the target category and the others being non-target categories. (See Appendix B, Table B-1, Table B-2, and Table B-3, for map-by-map breakdowns of zone and color schemes, and category assignments for each state.) Oregon was in the target category in nine of the 18 qualitative/state maps. In four of those nine, Oregon was paired up with a target state in State Zone 1 - Nevada (Appendix A, Figure A-7), California (Appendix A, Figure A-II), Idaho (Appendix A, Figure A-23), and Washington (Appendix A, Figure A-35). In each of these maps, there was at least one randomly selected target state in each zone eastward of State Zone 2, with a few zones claiming two or three randomly selected target states; only one state was present in State Zone 1 in each of the four maps. In two of the nine maps, Oregon was paired up with a target state in State Zone 2 - Montana (Appendix A, Figure A-15) and Wyoming (Appendix A, Figure A-32). In these two maps, no target state was present in Zone I, rather the target states were randomly distributed throughout State Zone 3 and eastward; only one target state was present in State Zone 2. This process continued with Oregon being paired up once with a State Zone 3 target (Colorado, Appendix A, Figure A-26), once with a State Zone 4 target (Kansas, Appendix A, Figure A-5), and once with a State Zone 5 target (Tennessee, Appendix A, Figure A-30). No target state (other than Oregon) was present anywhere westward of the aforementioned target states in State Zone 3, State Zone 4 and State Zone 5. 23 This same zonal scheme was applied to nine other maps in which Oregon was not in the target category. Four of the maps featured a State Zone 1 state as the most westward target state (California (Appendix A, Figure A-34), Washington (Appendix A, Figure A-I), Nevada (Appendix A, Figure A-I8), and Idaho (Appendix A, Figure A-28)); two maps had a State Zone 2 target state as the most westward (Utah (Appendix A, Figure A-I4) and Arizona (Appendix A, Figure A-I7)); and there was one map each for State Zone 3, State Zone 4 and State Zone 5 with a similar scheme. (The most westward target states were South Dakota for State Zone 3 (Appendix A, Figure A-13), Oklahoma for State Zone 4 (Appendix A, Figure A-20), and Illinois for State Zone 5 (Appendix A, Figure A-3)). Although a random qualitative color scheme of four hues (picked randomly from seven qualitative color choices) were assigned to each map, the schemes were designed so that no one hue appeared more times as either the target or non-target categories. The order of the legends was also randomized to make sure that the target category was not always appearing first in the legend. Thus, hues were evenly represented in terms ofwhat they represented (target or non-target states) and where they were placed in the legends (first, second, third or fourth positions). Quantitative/County-cluster Maps For the quantitative/county-cluster maps, 14 county clusters of roughly the same size and shape were created in order to minimize those visual variables during the decision-making process (Figure 8). Each county cluster was comprised of three rings: 24 _" C1u~aer Zone :; I "ng 2nd ring 3rd rtng Figure 8. Cluster zones for quantitative/county-cluster maps. the first ring (or core ring) was the smallest with a diameter of roughly 30 pixels but was designed to represent the greatest amount of a particular phenomenon. The second ring, located just outside the core, had a total diameter of 70 pixels and represented the second- greatest amount of the same phenomenon. And the third ring had a 11 O-pixel diameter and represented the third-highest value of the phenomenon. Everything outside of the three-ring clusters was considered to have the lowest value of the phenomenon. One county cluster was placed in north central Oregon, and 13 others were placed at regular distance intervals from its center. Each concentric zone was 135 pixels away from the Oregon cluster and were placed as follows: • Cluster Zone 1 (135 pixels away from Oregon cluster): Montana and Nevada • Cluster Zone 2 (270 pixels away): Arizona, Colorado, and South Dakota 25 • Cluster Zone 3 (405 pixels away): Kansas, Minnesota, and Texas • Cluster Zone 4 (540 pixels away): Arkansas and Illinois • Cluster Zone 5 (675 pixels away): Georgia, New York, and Virginia The distance of 135 pixels was chosen in order to allow for proper spacing between each zone and to ensure that the entire map was covered. Each of the 17 county-cluster maps contained a total of three county clusters, chosen at random, with nine of the maps featuring the Oregon cluster in the target and eight of them with no Oregon cluster. The schemes featured different combinations of clusters in nearby and faraway zones. Nine of the 17 county-cluster maps had the Oregon cluster in the target category; of those, three maps contained a cluster from Cluster Zone 1, two contained a cluster from Cluster Zone 2, two contained a cluster from Cluster Zone 3, and one contained a cluster from Cluster Zone 4 as the next-most westward cluster on the map. In the other eight county-cluster maps in which the Oregon cluster was not in the target category, five of the maps contained a cluster from Cluster Zone 1, two contained a cluster from Cluster Zone 2, and one contained a cluster from Cluster Zone 3 as the most westward cluster on the map. Again, just as with the qualitative/state maps, the idea was to weight the maps with more target states farther west and closer to the Oregon cluster. Five different graduated-color ramps of four classes each were used, each based on changes in value (from light to dark). They were: red, blue, orange, green and violet and based on colors retrieved from ColorBrewer.org (Brewer 201 0). Each color ramp was 26 randomly assigned to each map with red and orange used four times each, and blue, green and violet used three times each. Which ring participants were asked to choose was also randomized. Six times, participants were asked to identify phenomena in the first ring, while target phenomena was placed in the second and third rings five times each. (See Appendix B, Table B-4, for a map-by-map breakdown of county-cluster locations, color schemes and target-ring assignments.) Fictitious Phenomena In order to eliminate the possibility of participants equating particular states or regions with certain activities (such as Iowa = com production, or Michigan = car manufacturing), it was necessary to generate fictitious phenomena for the themes of each map. For example, if participants were shown a four-classed qualitative choropleth map of the coterminous U.S. depicting predominant crop production in each state (the choices could be wheat, com, soybeans and cotton) and asked to choose a state in which cotton production is predominant, participants might choose states they think are major cotton producers (states in the South, for instance), instead of choosing states they see depicting high cotton production on the map. Using fictitious preferences for the state maps worked well with the qualitative color schemes, while depicting varying degrees of fictitious phenomena worked better with the qualitative county-cluster maps (Appendix B, Table B-5 and Table B-6). The color schemes, decision statements (e.g., "Click a region with the highest degree of preference for belt sanders" or "Click a state with a preference 27 for satin gowns"), legend order, and phenomena degree were all randomized. Careful consideration was given to the topics of the decision statements to make sure that the phenomena were not easily connectable to any particular states or regions. The topics, in fact, bordered on the inane in order to ensure that participants would simply read the maps rather than try to draw on previous knowledge or suspicions about certain places. Map Quiz U.S. state-name knowledge was tested through the creation of a computerized map quiz designed to follow the map-reading task. The quiz map was the same size as those in the map-reading task, and each state was shaded with the same hue (light blue). After a brief instructional screen, participants were shown a map of the lower 48 states with a statement asking them to identify a particular state. They were instructed to click the location of the state in the statement to move on to the next screen. All 48 states were included in the quiz and were placed in random order. The color and size of the map was the same for each question. Test Construction Using a U.S. base map from ESRI with an equal-area projection, the national, state and county borders were all simplified in ArcMap 9.3 for easier reading, smaller file sizes and faster rendering time. A base map of state borders only was exported to Adobe Illustrator where each state polygon was converted into a separate movie clip. This was done so that actions could be applied to them once they were exported to Adobe Flash. 28 The same base map was copied and exported to Illustrator for the creation of the 18 qualitative/state map images. Once the color schemes (and thus the category assignments for each of the state maps) were applied, the images were saved as portable network graphics (PNGs) for importation into Flash. The process was repeated for the 17 county-cluster maps, though instead of applying colors to each state, three of 14 different county clusters were placed in their proper locations on the map and given their proper graduated color ramp. These images were also saved as PNGs. The states-only base map was then exported to Flash where the alpha (or transparency) was set to zero, rendering it invisible. Using ActionScript 3, actions designed to measure what was clicked on (including both the state name and the XY pixel coordinate) were created so the participants' mouse clicks could be stored and captured for later analysis. One by one, the state and county-cluster PNGs were imported onto the Flash timeline in their previously determined randomized order, all appearing directly over the invisible states-only map layer with the mouse-click-capture actions applied to it. This invisible mesh overlay was created so that actions would not have to be reapplied to each of the 35 maps, thereby reducing file size and allowing for the application to run more smoothly. File size was further reduced by using PNGs rather than filling in all of the hues in Flash itself. One single map and set of actions was created for the quiz, and only the text changed from frame to frame. ActionScript 3 code captured each participant's data into a text file. Data included which states were clicked on during the map-reading task, and the corresponding XY pixel location of that click (the origin was a point just northwest of the state Washington 29 - X values grew larger with eastward mouse movement; Y values grew larger with southward mouse movement). Survey The final part of the test instrument was a short demographic survey (Appendix C). It was originally intended to be administered via computer, but time constraints and programming difficulties made it untenable. The alternative, then, was to administer it on paper and input the data into a spreadsheet manually. Participants were asked to identify their age, sex, country of citizenship, major or degree, current occupation, and to list the all the places they had lived (including the years they had lived there and the duration of their stay). Lastly, participants were asked to identify one state they would consider to be their "home state." Participants and Recruitment Seventy-five students from the University of Oregon were recruited via e-mail or in person for the experiment (Appendix D); each signed a consent form (Appendix E) and was paid $10 for his or her participation. Participation was open to anyone over 18 who was not colorblind. People who are colorblind, which constitutes 11 percent of the population (Krygier and Wood 2005), were excluded because it would not be possible for them to discern differences between greens and reds, two colors which were used in various combinations in the experiment (Krygier and Wood 2005, Brewer et al. 1997). 30 A colorblind-friendly scheme was not possible due to the number of maps and color combinations needed to effectively investigate the research questions. Test Procedure Following their signed consent, participants were assigned three-digit random identification numbers for privacy, and the consent forms with their identifying numbers (and names and signatures) were kept in a secure location. The consent forms were the only materials, then, that contained both the participants' names and their identification numbers. Identification numbers were then used on all subsequent testing materials, including the survey. Each of the text files exported from Flash was given the participant's corresponding three-digit identifications number as well. Each participant took the computer portion of the experiment on the same laptop computer at the same desk in the Spatial and Map Cognition Research Lab at the University of Oregon. The order of the images for the map-reading task and the map-quiz task were the same for everyone. Following the computer-based test, which took participants between 10 and 20 minutes to complete, a paper survey with the corresponding identification number already attached was completed on a nearby table. It took participants between five and 10 minutes to fill out the survey. Once participants were done with the computer-based test, the data window capturing the state names and XY locations of his or her mouse clicks in Flash was first saved to a text file, then to a Microsoft Excel spreadsheet. All of the data were eventually copied into a master spreadsheet for later analysis. 31 CHAPTER IV ANALYSIS AND RESULTS Analysis of the data collected from the map-reading task, map quiz, and survey was completed using linear regression models to ascertain how the various dependent and independent variables related. A t-test was administered in order to determine which participants were Oregonians and which ones were not was statistically similar, as well as to determine whether the results from the map-reading task were statistically different between Oregonians and non-Oregonians. A total of 75 people participated in the experiment, however, the results of four of them were not considered in the final analysis because they reported their citizenship to foreign countries in the survey. Since the map-reading task was focused on U.S. states and the quiz portion of the test focused on U.S. state-name knowledge, it did not seem appropriate to include their data in the analysis. Independent Variables Four of the five independent variables used in the analysis were extracted from the survey data: age, sex, major, and home state. The participants were relatively young (the mean age was 22 and 92 percent of the sample was 25 years or younger), majority male (59 percent), and came from several different areas of study. Geographers, which 32 included those who reported geography or GIS as their major, made up 32 percent of the sample. Of the non-geographers, 35 percent were business, business administration, economics, accounting or marketing majors. Participants were asked to identify all of the different states or countries they had lived in and to report the duration they had spent in each place; they were also asked to identify one home state. The time spent in Oregon was calculated as a percentage by taking the number of years they lived in each place and dividing by their age. Thus time became the independent variable "percent of life lived in Oregon." Thirty-seven participants (52 percent) lived in Oregon more than half of their lives - 65 percent of whom reported Oregon as the only state they had ever lived in. However, 44 participants (62 percent) chose Oregon as their home state. For the seven people who reported Oregon as their home state but actually lived in the state less than half of their lives, their collective mean time in Oregon was only 15 percent. It was thus necessary to determine which method to use for determining who was an Oregonian and who was a non-Oregonian for the statistical comparison - self-reported home state or percent of life lived in Oregon. If individual attitudes are shaped by the types ofmaps to which people are exposed as suggested by Kitchin (1994) and Gould (1973), it is not unreasonable to assume that where a person actually grew up is more important than with what state a person identifies. A series of t-tests (Table 1) using the independent variables of age, sex and major was performed in order to ensure that deriving the independent variable of percent of life 33 lived in Oregon was statistically indistinguishable from the self-reported independent variable of home state. Equal variances in each t-test were confirmed through a corresponding series ofF-tests. In comparing the samples of self-reported Oregonians and the derived subset of those participants who reported living in Oregon more than half of their lives, the results clearly show that the means of age, sex and major are all indeed statistically indistinguishable. Thus, percent of life lived in Oregon was used as the independent variable for home state. Table 1. Comparing self-reported and calculated measures of home state. t-tests (assuming equal variances) AGE SEX (1 =male) MAJOR (1 =geog.) OR (rep] OR (pct] OR (rep] OR (pct] OR (rep) OR (pct) Mean 21.818 20.459 0.568 0.595 0.273 0.216 Variance 37.082 5.644 0.251 0.248 0.203 0.174 Observations 44 37 44 37 44 37 Pooled Variance 22.756 0.250 0.190 Hypothesized Mean Dift. 0 0 0 df 79 79 79 tStat 1.277 -0.237 0.581 P(T<=t) one-tail 0.103 0.407 0.281 t Critical one-tail 1.664 1.664 1.664 P(T<=t) two-tail 0.205 I 0.813 I 0.563 I t Critical two-tail 1.990 1.990 1.990 OR (rep) = Oregonians (self-reported by participant) OR (pet) = Oregonians (calculated as percentage of life lived in Oregon) Since P > .05, the null hypothesis that the means are equal is accepted for all variables The last independent variable came from the second part of the computer portion of the experiment: the 48-state map quiz. Participants were scored based on the percentage of correct answers. In reviewing the results, a problem of lag time was discovered with the Flash interface. Some participants apparently clicked a state more than once when the frame refused to advance. Unfortunately, their second mouse-click was recorded on the subsequent frame, thereby registering an erroneous result in the 34 data output file. These obvious double-clicks (and in a few cases, triple-clicks), were disregarded and the percentage of correct answers was adjusted accordingly; only 0.85 percent of all of the quiz frames were disregarded. The overall mean score for the map quiz was 76 percent correct. When divided into two groups, Oregonians and non-Oregonians performed slightly differently but the difference was not significant (Figure 9). Oregonians had no trouble locating states along the West Coast (Washington, Oregon and California), in central west and n011hern Plains (Idaho, Nevada, Utah, Montana, North Dakota, and South Dakota), and those with unique, jutting shapes along the edges of the map (Texas, Florida, and Maine). They faltered, however, with many states in the central part of the country (particularly Oregonians 10 8 G 6c Q) :J 4CT ~ lL 2 0 50 60 70 80 90 100 Quiz Score (Mean=75) Non -Oregonians 12 10 >- 8u c Q) 6:J CT ~ 4 lL 2 0 50 60 70 80 90 100 Quiz Score (Mean=77) t-Test: Two-Sample Assuming Equal Variances Quiz scores OR NOR Mean 74.626 77.353 Variance 388.628 288.538 Observations 37 34 Pooled Variance 340.759 Hypothesized mean diff. 0 df 69 t Stat -0.622 P(T<=t) one-tail 0.268 t Critical one-tail 1.667 P(T<=t) two-tail I 0.536 I t Critical two-tail 1.995 OR = OregonIans NOR = Non-Oregonians Figure 9. Distribution of quiz scores and correpsonding t-test. 35 Arkansas, Indiana, and Missouri), and confused Colorado and Wyoming, and, to a lesser degree, Arizona and New Mexico (Figure 10). 13 78 ,l 18 1'1 so 78 " "S7 t' ;' .•? ", i T2. 16 100 Percent of Oregoni;ms' Correct Responses 81 '00 11 II f' 61 /0 16 W .D Second·highes< o Thlrd·highes< rY eD Lowe'S[ ''} Figure A-6. Oregon cluster in target; South Dakota (Cluster Zone 2) cluster next nearest target. Click a state with a preference for German shepherds ti---j{--___ 1'\ fL-J ~ J -J ~7' it" I --7- ~ ~tr 'f\/1-- \ \r~\ ~f--t---F~~>~--}~ ~II l~l! \~J ',-- ~ 7 Dog breeds \ ~ o Labr.dorrelriever ~~ ).J-~·v"",' \ D German shepherd - \ 4 J D ROllweiler \ (~ '\., D Boxer ~\ ... Figure A-7. in target; Nevada (State Zone 1) next nearest target. 51 Clicl< a region with the second-highest number of tweezers l'Zr--- _____f- / --- f'\ ( f ~- -; J---------I ~/J_,L· t-----I f ( ! r- f--; \ ~ \ ~}---7----f=' ~}.yq~ \.~'" L -.J \y) Tweez::h:::county ~ ~~\ )l~,.;z" v """'\~\ D Second-highesc :.-_ G, \ D Thir-d·highes< r '\ ) ( ~...D Lowes( -..\ Figure A-S. Oregon cluster in target; Colorado cluster (Cluster Zone 2) next nearest target. Figure A-9. Oregon cluster not in target; Montana cluster (Cluster Zone 1) nearest target. 52 Click a region with the highest number of coaxial cables F/---j{-- --- I'\~ I '~--) L 0 /( \ y7, /-- \ \;'~-,-Lf0 .\' ') £- ' ~ i -, ro~~~'/~"-"- I I 1,'-_____ t \" / '--- --.L.r--,,--J L r Coaxial cables per county ", \ ,~ Highest ~r J---'"'--..,~ V"" \o Second-highest \ y) ~ \ o Thlrd·highest \ ( ~) o Lowe.st "'-..\ Figure A-lO. Oregon cluster not in target; Nevada cluster (Cluster Zone 1) nearest target. Click a state with a preference for slotted-style screw heads rZr---J(----- r\ I ~J \ W ~'7f' '))( ~r \1"f-- 1 \#--Ar--- ( y; ~r _J \l-XV~ ;- -,' ) ~) ~.~ y f \ I )- " TJ~~\ r--L . 1'( j~,.r~~ \ ~ 'i /~'~r-7~v~"'-- I I l-~ . \ \; ;;-:,::"',.. ~~ )l~ v1.·,'~~o Philiips o Square r \\., D~ ~ ~ Figure A-H. Oregon in target; California (State Zone 1) next nearest target. 53 Figure A-12. Oregon cluster in target; Nevada cluster (Cluster Zone 1) next nearest target. Figure A-B. Oregon not in target; South Dakota (State Zone 3) nearest target. Figure A-14. Oregon not in target; Utah (State Zone 2) nearest target. Figure A-IS. Oregon in target; Montana (State Zone 2) next nearest target. 54 Figure A-16. Oregon cluster not in target; Colorado cluster (Cluster Zone 2) nearest target. Figure A-17. Oregon not in target; Arizona (State Zone 2) nearest target. 55 Click a state with a preference for Burmese cats\' r----A,2 f--~ ( I --- { ILr-J ---j-L -J 'f ( rT r-------.-..-J \ \f--t---,~ W _,-_~,l / " ~ r Ca' breeds '-- ) IA\~ D Persi." \/__\ > _~~ V"" \ D Siamese ,. ~ J ,/} \ .....D Abyssiman , '-. o Bunn~ ,I' •\ \\)--t----F (,)~ ,>-~--~)~! - J;L--~ J;' }- <:~'r"-, I i --~ \ 'y) Stitch types ''-~ ) ~~ D FuJl·cross ~ ~. \./ "" \ D Hllf-cross ~) Is. J D Qu,rter-cross \ ( \'-, DMinl,c~1 ~ ~ Figure A-32. Oregon in target; Wyoming (State Zone 2) next nearest target. Figure A-33. Oregon cluster not in target; Montana cluster (Cluster Zone 1) nearest target. 63 Click a state with a preference for couplets !~~~----- -" / (L,---- ~,J\t~/-LfJ I ',,> 2-3? f( I I -·n ~\ \J--+---p~~ ;.J ~ \-1 I I~· Ll \ Y Poetry forms "'--~ \ ~,~ D Couplet ~ -.J- \./ ') \ O "'-..f'..) ~. \Sonnet \ y ~ D Qu""in \ ( '\.....J D Sesuna ") Figure A-34. Oregon not in target; California (State Zone 1) nearest target. Figure A-35. Oregon in target; Washington (State Zone 1) next nearest target. 64 APPENDIXB ZONE AND COLOR SCHEMES, AND CATEGORY ASSIGNMENTS FOR MAPS IN THE MAP-READING TASK Table B-1. State map color schemes and legend order. Map Color ORin Nearest Nearest Legendtarget non-OR # scheme* target? zone** target order* 1 OBRV N 1 WA OBRV 3 BWRO N 5 KY MORB 5 YGRW Y 4 KS RMYG 7 OVGY Y 1 NV GOVY 11 RVYG Y 1 CA RYVG 13 RGWB N 3 SD GBRM 14 OGVR N 2 UT VRGO 15 YWOB Y 2 MT BYOM 17 BGYW N 2 AZ BGYM 18 RWOG N 1 NV MOGR 20 BRYO N 4 OK YROB 23 GVBY Y 1 ID GVBY 26 WROY Y 3 CO ORMY 28 GBWV N 1 ID BGMV 30 VBYO Y 5 TN OYVB 32 WVOY Y 2 WY YVMO 34 VRWO N 1 CA VMOR 35 GRVB Y 1 WA RVBG * B=blue; G=green; O=orange; R=red; V=violet; W=brown; Y=yellow ** Zones represent how many states away a state is from Oregon 65 Table B-2. Category assignments for state maps with Oregon in the target category. 66 Map #--> 5 7 11 15 23 26 30 32 35 States State Zone tarl!:et--> 4 1 1 2 1 3 5 2 1 away Color scheme --> YGRW OVGY RVYG BWOY GVBY WROY VBYO WVOY GRVB 0 Orei!:on (OR) 1 1 1 1 1 1 1 1 1 1 California (CAl 3 3 1 4 4 3 3 3 2 1 Nevada (NY) 4 1 3 3 3 2 4 2 3 1 Idaho (IDl 2 2 4 2 1 4 2 4 4 1 WashinR\:on (WA) 4 4 2 2 2 3 3 4 1 2 Wyomini!: (WY) 4 1 2 4 3 2 4 1 3 2 Utah ruTl 3 2 3 3 1 3 3 4 2 2 Arizona (AZ) 3 3 4 3 4 2 4 2 1 2 Montana MT) 2 4 1 1 2 4 2 3 4 3 Nebraska fNEl 3 3 2 4 1 3 2 2 4 3 New Mexico (NM) 2 2 1 2 3 4 2 1 2 3 Colorado (CO) 2 4 4 2 4 1 3 3 1 3 North Dakota (ND) 4 3 3 1 2 4 4 2 1 3 South Dakota (SD) 4 1 2 4 3 2 4 4 3 4 Kansas (KS) 1 2 1 3 2 2 4 2 2 4 Minnesota (MNl 2 2 3 3 1 1 3 3 4 4 Iowa (lA) 4 1 4 2 4 2 3 2 3 4 Texas (TX) 3 4 3 4 1 3 2 1 2 4 Oklahoma (OK 3 4 2 2 4 4 3 3 1 4 Missouri (MO 2 4 1 1 2 3 2 4 3 5 Louisiana (LA 2 3 2 1 1 3 2 2 4 5 Kentucky fIT 1 3 1 4 3 3 4 3 2 5 Illinois (IL) 3 1 4 3 2 1 4 3 4 5 Tennessee (TNl 3 1 3 4 4 4 1 4 3 5 Wisconsin (WI 4 3 4 2 3 2 2 1 1 5 Arkansas (AR) 2 2 1 3 3 4 3 4 3 6 West Viri!:inia (WV) 1 1 3 2 2 2 4 2 4 6 Virginia (VA 4 4 2 3 4 1 3 2 1 6 Indiana (IN) 2 2 1 4 1 4 4 1 2 6 Ohio (Om 3 4 4 3 4 1 1 3 1 6 Michii!:an (Mil 4 2 4 1 1 4 2 4 2 6 Mississippi (MS) 1 3 3 4 2 1 3 2 1 6 Alabama fAL) 3 3 2 2 1 3 2 1 4 6 North Carolina (NC) 1 2 1 2 4 3 1 4 3 6 Geori!:ia (GA) 2 1 3 1 3 2 3 3 4 7 South Carolina SC) 1 4 2 3 2 4 1 3 1 7 Maryland (MD) 2 1 4 1 4 2 4 4 3 7 Florida (FL) 4 4 2 4 1 1 2 1 3 7 Pennsylvania (PA) 4 3 1 3 3 4 1 1 2 8 New York (NY) 3 1 3 1 2 3 3 3 4 8 New lersev (Nil 1 2 4 4 3 4 4 2 2 8 Delaware (DEl 4 3 3 2 2 1 2 4 2 9 Connecticut (CT) 3 2 4 2 4 2 1 1 3 9 Vermont fVT 2 2 2 1 3 3 4 2 4 9 Massachusetts (MAl 4 4 2 4 4 2 3 1 1 10 Rhode Island fRll 2 1 3 3 2 3 2 3 4 10 N. Hampshire (NA) 3 4 1 4 3 1 3 4 2 11 Maine (ME) 1 3 4 1 1 4 4 2 3 * 1 = target category; 2,3 & 4 = other qualItative categories 67 Table B-3. Category assignments for state maps with Oregon not in the target category. Map #--> 1 3 13 14 17 18 20 28 34States State Zone target --> 1 5 3 2 2 1 4 1 1 away Color scheme --> OBRV BWRO RGWB OGVR BGYW RWOG BRYO GBWV VRWO 0 Oregon (OR 3 3 2 2 3 4 2 2 4 1 California (CA) 2 2 3 4 4 2 3 3 1 1 Nevada (NV) 4 4 4 3 2 1 4 4 2 1 Idaho nO) 2 4 4 4 3 3 2 1 3 1 Washington (WA) 1 2 3 3 4 4 4 2 3 2 Wyoming (wy) 3 3 2 2 2 3 3 4 1 2 Utah ruT) 1 4 2 1 4 2 3 3 4 2 Arizona (AZ) 4 2 4 3 1 1 4 3 2 2 Montana(Mn 3 3 3 4 2 2 2 1 4 3 Nebraska (NE) 4 3 2 1 3 4 3 1 3 3 New Mexico (NM) 1 2 3 2 2 3 4 2 4 3 Colorado (CO 2 4 4 3 2 4 2 4 1 3 North Dakota (NO) 4 2 3 2 1 4 4 1 2 3 South Dakota [SO) 3 4 1 4 4 1 3 2 2 4 Kansas (KSl 2 3 1 3 3 2 2 1 3 4 Minnesota (MN 2 3 1 2 4 1 3 3 4 4 Iowa (IA) 3 4 4 4 1 3 2 4 1 4 Texas (TX) 4 2 2 4 1 3 4 1 4 4 Oklahoma raK 3 4 2 2 3 2 1 4 1 4 Missouri (MO 1 3 4 1 3 3 4 2 2 5 Louisiana (LA) 4 2 1 3 2 4 1 1 3 5 Kentuckv (KY) 1 4 3 2 4 2 1 3 2 5 Illinois (lL) 2 1 4 1 2 4 2 3 1 5 Tennessee (TN) 1 2 3 1 4 3 3 2 4 5 Wisconsin fWJl 2 3 1 3 3 1 4 4 3 5 Arkansas fAR 3 4 2 4 1 1 3 1 2 6 West Virginia (V TV) 1 2 3 2 1 2 2 3 4 6 Virginia (VA) 4 3 2 1 3 1 1 4 3 6 Indiana (IN) 2 3 1 3 4 4 3 1 3 6 Ohio (Om 3 4 4 4 2 1 4 2 1 6 Michigan (MI 1 1 4 4 4 2 2 3 2 6 Mississippi [MS) 4 2 1 3 1 3 2 1 4 6 Alabama (AL) 3 1 2 1 1 1 4 2 4 6 North Carolina (NCl 2 4 3 1 2 1 3 4 1 6 Georgia (GA) 1 1 3 2 3 4 1 4 3 7 South Carolina [SC) 4 1 1 3 3 2 4 1 2 7 Marvland (MOl 3 2 2 4 4 3 1 2 1 7 Florida (FL) 1 3 4 2 2 4 2 3 1 7 Pennsylvania (PA) 2 2 4 2 1 1 1 4 3 8 New York [NY] 1 4 2 1 1 2 3 3 4 8 New Jersev (N]) 4 1 1 4 4 3 2 2 3 8 Delaware roE 4 2 3 3 2 4 4 3 1 9 Connecticut [Cn 2 4 1 3 3 2 3 4 2 9 Vermont(m 3 3 2 2 1 3 4 1 2 9 Massachusetts (MAl 3 3 4 3 2 4 1 2 1 10 Rhode Island (Rn 4 2 3 4 3 1 3 3 2 10 N. Hampshire [NA) 2 4 3 1 4 2 1 4 3 11 Maine (ME) 1 1 4 1 2 3 2 2 4 Table B-4. County-cluster map zone schemes. Map Color Zone Cluster 1 Cluster 2 Cluster 3 Target# Scheme ring 2 Red 3-4-5 MN(3) AR(4) NY (5) 2 4 Blue 0-3-5 OR (0) KS (3) NY (5) 1 6 Blue 0-2-4 OR (0) SD (2) IL (4) 3 8 Red 0-2-3 OR (0) CO (2) MN(3) 3 9 Orange 1-3-5 MT(l) KS(3) NY (5) 2 10 Red 1-2-4 NV(l) SD (2) AR(4) 1 12 Green 0-1-3 OR (0) NV(I) MN(3) 1 16 Blue 2-3-4 CO (2) TX (3) IL (4) 2 19 Green 1-4-5 MT(l) AR(4) GA (5) 3 21 Green 0-4-5 OR (0) AR(4) VA (5) 2 22 Red 0-2-5 OR (0) AZ (2) GA (5) 3 24 Orange 0-1-2 OR (0) MT(l) SD (2) 1 25 Violet 1-3-4 NV(l) TX(3) IL (4) 1 27 Orange 0-3-4 OR (0) TX(3) IL (4) 1 29 Violet 0-1-4 OR (0) NV(l) AR(4) 3 31 Orange 2-4-5 AZ (2) IL (4) VA (5) 2 33 Violet 1-2-3 MT(l) AZ(2) KS (3) 2 68 Table B-5. Fictitious phenomena for county-cluster maps. Map County Color "Click a region with the [insert # clusters scheme degree] preference for ... " Degree*present 2 MN-AR-NY Red curling irons Second-highest 4 OR-KS-NY Blue dry erasers Highest 6 OR-SD-IL Blue belt sanders Third-highest 8 OR-CO-MN Red metronomes Third-highest 9 MT-KS-NY Orange tweezers Second-highest 10 NV-SD-AR Red coaxial cables Highest 12 OR-NV-MN Green coping saws Highest 16 CO-TX-IL Blue oscilloscopes Second-highest 19 MT-AR-GA Green calculators Third-highest 21 OR-AR-VA Green paper clips Second-highest 22 OR-AZ-GA Red capacitors Third-highest 24 OR-MT-SD Orange bracelets Highest 25 NV-TX-IL Violet fountain pens Highest 27 OR-TX-IL Orange stencils Highest 29 OR-NV-AR Violet spectrometers Third-highest 31 AZ-IL-VA Orange lace curtains Second-highest 33 MT-AZ-KS Violet magic wands Second-highest * Degree described as amount per county in all county-cluster map legends 69 Table B-6. Fictitious phenomena for state maps. 70 Map Color "Click a state Legend with a preference Choice #1 Choice #2 Choice #3 Choice #4# scheme for ... " header 1 OBRV narrow-ruled Notebook Narrow Wide Medium College notebooks ruling 3 WORB console-style Piano Spinet Upright Grand Consolepianos styles 5 RWYG plastic buttons Button Brass Steel Plastic Leather materials 7 GOVY German Dog breeds Labrador German Rottweiler Boxer shepherds retriever shepherd 11 RYVG slotted-style Screw- Slotted Phillips Square Allen screw heads head styles 13 GBRW desk calendars Calendar Wall Flip Desk Pockettypes , Gown14 VRGO satin gowns fabrics Chiffon Velvet Silk Satin 15 BYOW slipknots Knot types Square Slipknot Overhand Sheet 17 BGYW pillar-style Candle Pillar Votive Taper Filled candles styles 18 WOGR Burmese cats Cat breeds Persian Siamese Abyssinian Burmese open-face Motorcycle Half-20 YROB motorcycle Full-face Flip-up Open-fact helmets helmets helmet 23 VGBY foam insulation Insulation Fiberglass Foam Polystyrene Cellulose types 26 ORWY dangle-style Earring Stud Hoop Dangle Huggy earrings styles 28 BGWV lever-type locks Lock types Mortise Lever Cam Rim 30 OYVB Michelin road Road AAA Rand Michelin Frommer's atlases atlases McNally 32 YVWO quarter-cross Stitch Full-cross Half- Quarter- Mini- stitches types cross cross cross 34 VWOR couplets Poetry Couplet Sonnet Quatrain Sestinaforms 35 RVBG the board game Board Monopoly Life Sorry! ClueClue games APPENDIXC SURVEY 71 Please answer the following questions: Age: _ Sex (M/F/Other): _ IDNumber 0 u.s. citizen? (YIN): If no, which country? . _ Major/degree (circle one and state the subject of your major or degree): Current occupation(s): _ List all of the places you have lived in chronological order starting with the most recent and going back in time. Include the years you lived there. If you have lived in more than 10 places, just list the mo~1 recent 10. See the example below: EXAMPLE State or countrv Years Duration Orell.on 2009-present 6mos China 2009 2mos Orel!.on 2008-2009 I vr California 1995-2008 13 vrs Washinll,ton 1994-1995 1vr Iowa 1991-1994 3 yrs State or country Years Duration What one state would you consider to be your home state? ----::=-__----: -,----_ (It doesn't necessarily have to be where you're living now) (Choose only one state.) APPENDIXD RECRUITMENT SCRIPT I am seeking volunteers for a study that is investigating map use. Participants will take a computer-and-paper-based test designed to measure how people read maps. Participants will view a short series of maps on a computer screen and answer questions about them; an even shorter paper survey will follow. The entire testing session is expected to take around 20 minutes and will take place at a time convenient for you (1 will be scheduling testing sessions at different times of the day and most days of the week). Testing will take place in the Spatial Map and Cognition Research Lab (SMCRL) in 160 Condon. Volunteers must be at least 18 years old and must not be colorblind. Participants will receive $10 for their participation. Thank you for your consideration, Matthew E. Millett Graduate Teaching Fellow Spatial and Map Cognition Research Lab Department ofGeography University of Oregon millett@Uoregon.edu 72 APPENDIXE CONSENT TO PARTICIPATE NumberD Informed Consent - Map Use Study You are invited to participate in a research study conducted by Matt Millett from the University of Oregon Department of Geography. The purpose of the study is to explore how map users read maps. If you decide to participate, you will be required to take a 30-minute computer-based test in the Spatial and Map Cognition Research Lab (SMCRL) located in 160 Condon Hall. There are likely no reasonable foreseen risks, discomforts or inconveniences associated with your participation in this project. Your participation in this project will help me better understand how well people use choropleth maps of the United States. However, I cannot guarantee that you personally will receive any benefits from this research. Any information that is obtained in connection with this study and that can be identified with you will remain confidential and will be disclosed only with your permission. To keep subject identities confidential, I will use a numeric code to associate subject responses with identities. No information will be released to any other party outside of the research group for any reason. Ifyou choose to participate you will be paid $10 for your participation. Your participation is voluntary. Your decision whether or not to participate will not affect your relationship with University of Oregon. If you decide to participate, you are free to withdraw your consent and discontinue participation at any time without penalty. If you have any questions, please feel free to contact my adviser, Amy Lobben, Department ofGeography - University of Oregon, (541) 346-4566. If you have questions regarding your rights as a research subject, contact the Office for Protection ofHuman Subjects. This Office oversees the review of the research to protect your rights and is not involved with this study. Office ofHuman Subjects Compliance, University of Oregon, Eugene, OR 97403, (541) 346- 2510. You have been given a copy of this form to keep. Your signature below constitutes your consent to participate, that you willingly agree to participate, that you may withdraw your consent at any time and discontinue participation without penalty, that you have received a copy ofthis form, and that you are not waiving any legal claims, rights or remedies. 73 Participant Name (printed) Researcher Signature Signature Date APPENDIXF MAP-BY-MAP RESULTS Map 1 Map 2 Map 3 Map 4 ZN ST OR NOR ZN CL OR NOR ZN ST OR NOR ZN CL OR NOR 1 WA 41 29 3 MN 43 50 5 IL 51 44 0 OR 57 27 2 NM 38 47 4 AR 41 32 6 AL 27 29 3 KS 35 56 3 UT 21 9 5 NY 16 18 6 GA 8 12 5 NY 8 18 4+ 9st -- is 6 MI 5 6 7+ 3 st 8 9 MapS Map 6 Map 7 Map 8 ZN ST OR NOR ZN CL OR NOR ZN ST OR NOR ZN CL OR NOR 0 OR 57 44 0 OR 49 22 0 OR 46 24 0 OR 35 24 4 KS 38 47 2 SO 40 44 1 NV 16 24 2 CO 46 47 5 KY -- -- 4 IL 11 34 2 WY 22 21 3 MN 19 29 6+ 6 st 5 9 3+ 9 st 16 32 Map 9 Map 10 Map 11 Map 12 ZN CL OR NOR ZN CL OR NOR ZN ST OR NOR ZN CL OR NOR 1 MT 62 36 1 NV 27 32 0 OR 41 9 0 OR 51 32 3 KS 35 48 2 SO 54 44 1 CA 22 32 1 NV 11 18 5 NY 3 15 4 AR 19 24 2 MT 11 24 3 MN 38 50 3+ 9 st 27 35 (NM) (11) (18) Map 13 Map 14 Map 15 Map 16 ZN ST OR NOR ZN ST OR NOR ZN ST OR NOR ZN CL OR NOR 3 SO 33 15 2 UT 44 29 0 OR 50 26 2 CO 49 50 4 KS 36 26 3 NE 14 21 2 MT 8 24 3 TX 24 24 4 MN 14 21 4 MO 8 15 3 NO 6 9 4 IL 27 26 5+ 7 st 17 38 5+ 8st 33 35 4+ 8 st 36 41 (LA) (12) (ME) (11) (MO) (17) (24) Map 17 Map 18 Map 19 Map 20 ZN ST OR NOR ZN ST OR NOR ZN CL OR NOR ZN ST OR NOR 2 AZ 24 30 1 NV 25 15 1 MT 62 71 4 OK 69 61 3 NO 16 18 2 AZ 25 38 4 AR 24 21 5 KY 6 -- 4 IA 14 12 3 SO 14 15 5 GA 14 9 5 LA 17 15 4 TX 41 33 4+ 9 st 36 32 6+ 6st 8 24 5+ 7 st 6 6 Map 21 Map 22 Map 23 Map 24 ZN CL OR NOR ZN CL OR NOR ZN ST OR NOR ZN CL OR NOR 0 OR 41 45 0 OR 49 55 0 OR 31 21 0 OR 46 24 4 AR 43 33 2 AZ 27 21 1 ID 11 9 1 MT 19 38 5 VA 16 21 5 GA 24 24 2 UT 11 18 2 SO 35 38 3+ 9 st 46 53 (TX) (37) (26) Map 25 Map 26 Map 27 Map 28 ZN CL OR NOR ZN ST OR NOR ZN CL OR NOR ZN ST OR NOR 1 NV 36 41 0 OR 44 16 0 OR 43 35 1 ID 35 21 3 TX 25 32 3 CO 31 44 3 TX 24 21 2 MT 5 12 4 IL 39 26 4 MN -- 3 4 IL 32 44 3 NO -- -- 5+ 7 st 25 38 3 NE 5 21 (FL) (13) 4+ 8 st 54 47 (TX) (321 (21) 74 Map 29 Map 30 Map 31 Map 32 ZN CL OR NOR ZN ST OR NOR ZN CL OR NOR ZN ST OR NOR 0 OR 38 22 0 OR 57 56 2 AZ 64 62 0 OR 19 9 1 NV 22 31 5 TN 24 29 4 IL 25 32 2 WY 6 3 4 AR 41 47 6 OH 3 6 5 VA 11 6 3 NM 22 44 6 NC 3 6 4+ 8 st 53 44 7+ 3 st 14 3 (TX) (42) (28) Map 33 Map 34 Map 35 ZN CL OR NOR ZN ST OR NOR ZN ST OR NOR 1 MT 39 18 1 CA 67 38 0 OR 22 6 2 AZ 25 50 2 WY 11 18 1 WA 3 3 3 KS 36 32 3 CO 8 24 2 AZ 25 48 4+ 9 st 14 21 3+ 9st 50 42 (CO) (22) (12) ZN = zone number; CL = county cluster; ST = state; OR = Oregonians; NOR =Non-Oregonians States in parentheses are those with notable percentages in the grouped-state results. 75 76 REFERENCES Balchin, W.G.v. (1972). Graphicacy. Geography, 57, 185-195. Bertin, J. (1983). Semiology ofgraphics. (W.J. Berg, trans.). Madison, WI: University of Wisconsin Press. (Original work published 1967). Board, C. (1967). Maps as models. In RJ. Chorley & P. Haggett (Eds.), Physical and information models in geography (671-705). Worcester, United Kingdom: Methuen. Board, C., & Taylor, RM. (1977). Perception and maps: Human factors in map design and interpretation. Transactions ofthe Institute ofBritish Geographers, 2(1), 19- 36. Brewer, C.A. (2010). ColorBrewer 2.0. Retrieved from http://www.colorbrewer2.org. Brewer, c.A., MacEachren, A.M., Pickle, L.W., & Herrmann, D. (1997). Mapping mortality: Evaluating color schemes for choropleth maps. Annals ofthe Association ofAmerican Geographers, 87(3),411-438. Brown, L.A. (1949). The story ofmaps. New York, NY: Bonanza. Bryant, D.J., & Tversky, B. (1999). Mental representations ofperspective and spatial relations from diagrams and models. Journal ofExperimental Psychology: Learning, Memory and Cognition, 25(1), 137-156. Dent, B.D. (1972). Visual organization and thematic map communication. Annals ofthe Association ofAmerican Geographers, 62(1), 79-93. Dent, B.D. (1999). Cartography: Thematic map design (5th ed.). Boston, MA: WCB/ McGraw-Hill. Downs, R.M., & Stea, D. (Eds.). (1973). Image and environment: Cognitive mapping and spatial behavior. Chicago, IL: Aldine. Eysenck, H.J., Arnold, W., & Meili, R (Eds.). (1972). Encyclopedia ofpsychology. New York, NY: Herder & Herder. 77 Fearing, F. (1953). Toward a psychological theory of human communication. Journal of Personality, 22(1), 71-88. Gilmartin, P.P. (1981). The interface of cognitive and psychophysical research in cartography. Cartographica, 18(3), 9-20. Golledge, RG., & Spector, A.N. (1978). Comprehending the urban cognitive environment: Theory and practice. Geographical Analysis, 10, 403-426. Golledge, RG. (1999). Human wayfinding and cognitive maps. In RG. Golledge (Ed.), Wayfinding behavior: Cognitive mapping and other spatial processes (5-45). Baltimore, MD: Johns Hopkins University Press. Golledge, RG. (2002). The nature of geographic knowledge. Annals ofthe Association of American Geographers, 92(1), 1-14. Gould, P.R (1975). People in information space: The mental maps and information surfaces ofSweden. Lund, Sweden: The Royal University of Lund/CWK Gleerup. Gould, P.R (1973). On mental maps. In RM. Downs & D. Stea (Eds.), Image and environment: Cognitive mapping and spatial behavior (182-220). Chicago, IL: Aldine. Gould, P.R, & White, R (1986). Mental maps (2nd ed.). Boston, MA: Allen & Unwin. Harrower, M., & Brewer, C.A. (2003). ColorBrewer.org: An online tool for selecting colour schemes for maps. The Cartographic Journal, 40(1),27-37. Humphreys, G.W., & Bruce, V. (1995). Visual cognition: Computational, experimental, and neuropsychologicalperspectives. Hove, United Kingdom: Lawrence Erlbaum Associates. Jenks, G.F. (1967). The data model concept in statistical mapping. International Yearbook ofCartography, 7, 186-190. Kitchin, RM. (1994). Cognitive maps: What are they and why study them? Journal of Environmental Psychology, 14, 1-19. Kitchin, RM., & Freundschuh, S. (Eds.). (2000). Cognitive mapping: Past, present and future. New York, NY: Routledge. Kosslyn, S.M., & Pomerantz, J.R (1977). Imagery, propositions, and the form of internal representations. Cognitive Psychology, 9(1),52-76. 78 Krygier, J., & Wood, D. (2005). Making maps: A visual guide to map designfor GIS. New York, NY: Guilford. Kulhavy, R.W., & Stock, W.A. (1996). How cognitive maps and learned and remembered. Annals ofthe Association ofAmerican Geographers, 86(1), 123-145. Levine, D.N., Warach, J., & Farah, M.J. (1985). Two visual systems in mental imagery: Dissociation of "what" and "where" in imagery disorders due to bilateral posterior cerebral lesions. Neurology, 35, 1010-1018. Lloyd, R. (1982). A look at images. Annals ofthe Association ofAmerican Geographers, 72(4),532-548. Lloyd, R. (1994). Learning spatial prototypes. Annals ofthe Association ofAmerican Geographers, 84(3),418-440. Lloyd, R. (1997). Visual search processes used in map reading. Cartographica, 34(1), 11-32. Lloyd, R. (2000). Understanding and learning maps. In R.M. Kitchin & S. Freundschuh (Eds.), Cognitive mapping: Past, present andfuture (84-107). New York, NY: Routledge. Lloyd, R., & Steinke, T. (1977). Visual and statistical comparison ofchoropleth maps. Annals ofthe Association ofAmerican Geographers, 67(3),429-436. Lobben, AK. (2004). Tasks, strategies, and cognitive processes associated with navigational map reading: A review perspective. The Professional Geographer, 56(2),270-281. Lynch, K. (1960). The image ofthe city. Cambridge, MA: MIT Press. MacEachren, AM. (1982). The role ofcomplexity and symbolization method in thematic map effectiveness. Annals ofthe Association ofAmerican Geographers, 72(4), 495-513. MacEachren, AM. (1991). The role of maps in spatial knowledge acquisition. The Cartographic Journal, 28, 152-62. MacEachren, AM. (1995). How maps work: Representation, visualization, and design. New York, NY: Guilford. Monmonier, M.S. (1996). How to lie with maps (2nd ed.). Chicago, IL: University of Chicago Press. 79 Monmonier, M.S. (1989). Maps with the news: The development ofAmerican journalistic cartography. Chicago, IL: University of Chicago Press. Montello, D.R. (2002). Cognitive map-design research in the twentieth century: Theoretical and empirical approaches. Cartography and Geographic Information Science, 29(3),283-304. Morrison, J.L. (1974). A theoretical framework for cartographic generalisation with emphasis on the process of symbolisation. International Yearbook ofCartography, 14, 114-127. Muehrcke, P.C. (1998). Map use: Reading, analysis, and interpretation (4th ed.). Madison, WI: IP Publications. Olson, J.M. (1979). Cognitive cartographic experimentation. The Canadian Cartographer, 16(1), 34-44. Peterson, M.P. (2008) Trends in Internet and ubiquitous cartography. Cartographic Perspectives, 61,36-49. Posner, MJ. (1980). Orienting of attention. The Quarterly Journal ofExperimental Psychology, 32(1),3-25. Robinson, A.H. (1952). The look ofmaps: An examination ofcartographic design. Madison, WI: University of Wisconsin Press. Robinson, A.H., Morrison, J.L., Muehrcke, P.C., Kimerling, AJ., & Guptill, S.C. (1995). Elements ofcartography (6th ed.). New York, NY: Wiley. Rittschof, K.A., & Kulhavy, R.W. (1998). Learning and remembering from thematic maps of familiar regions. Educational Technology Research and Development, 46(1), 19-38. Saarinen, T. (1999). The Eurocentric nature of mental maps of the world. Research in Geographic Education, 1(2), 136-178. Slocum, T.A., Blok, C., Jiang, B., Koussoulakou, A., Montello, D.R., Fuhrmann, S., & Hedley, N.R. (2001). Cognitive and usability issues in geovisualization. Cartography and Geographic Information Science, 28(1),61-75. Thorndyke, P.Y., & Stasz, C. (1980). Individual differences in procedures for knowledge acquisition from maps. Cognitive Psychology, 12(1), 137-175. 80 Tolman, E.C. (1948). Cognitive maps in rats and men. Psychological Review, 55, 189- 208. Trick, L.M., & Enns, J.T. (1998). Lifespan changes in attention: The visual search task. Cognitive Development, 13, 369-386. Tufte, E.R. (2001). The visual display ofquantitative information (2nd ed.). Cheshire, CT: Graphics Press. Tversky, B. (2000). Levels and structure of spatial knowledge. In R.M. Kitchin & S. Freundschuh (Eds.), Cognitive mapping: Past, present andfuture (24-43). New York, NY: Routledge. Tyner, J.A. (1992). Introduction to thematic cartography. Englewood Cliffs, NJ: Prentice Hall. Virga, V. (2007). Cartographia: Mapping civilizations. New York, NY: Little, Brown and Company.