Residents' spatial knowledge of neighborhood continuity and form

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/229893075 Residents' Spatial Knowledge of Neighborhood Continuity and Form Article in Geographical Analysis · September 2010 DOI: 10.1111/j.1538-4632.1990.tb00213.x CITATIONS READS 25 30 2 authors, including: Stuart C. Aitken San Diego State University 89 PUBLICATIONS 1,188 CITATIONS SEE PROFILE All content following this page was uploaded by Stuart C. Aitken on 29 September 2014. The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original docum and are linked to publications on ResearchGate, letting you access and read them immediately. zyx zyx Stuart C.Aitken and Rudy Prosser zyxwvu zyxwv zyxw Residents’ Spatial Knowledge of Neighborhood Continuity and Form zyxw zyx This paper discusses residents’ cognition of neighborhood fm in terms of linearbased and areal-based knowledge structures. Cognitive and behavioral data are used to create su7faces of residents’ familiarity with, and experience of, a neighborhood in San Diego, Cal$mia. The complexity of the data required the topological and relational sophistication of an ARC / INFO-based geographic information system. Cellular-based data were collected to ident$y place-spec@c measures of residents’familiarity and experience within the community. The cellular data facilitated aggregation of residents’ cognitive sugaces as absolute spaces, and also relative to their homes. Spatial autocorrelation and directional autoregression techniques are used in association with standard cognitive mapping to establish the continuity and fm of residents’familiarity and experience with their neighborhood. The findings suggest that there is a structural diflerence in the spatial familiarity of residents who perceive their neighborhood as an area and those who perceive it as a network. North-south and east-west street networks played an important part as the basis for both linear-based and areal-based knowledge, but noncardinal directions were more prominent in the familiarity surjkces of those residents who had an areal-based perception of neighborhood fm. In terms of theories of spatial knowledge acquisition, the findings suggest that (i) there may not be a direct sequential link between linear-based and areal-based knowledge structures, (ii) knowledge of a complex network may not be suflcient to provide areal-based knowledge, and (iii) an areal-based knowledge structure does not necessarily comprise an understanding of survey procedures. zyx Neighborhood has much in common with geographic concepts such as environment, landscape, and place in that it is often an important spatial variable yet it is hard to define precisely. The concept of neighborhood emerged as an important academic and planning concern when proponents such as Jacobs (1961) and Keller (1968) popularized it as a panacea for a wide range of inner-city problems. Within the planning profession, the idea of neighborhood-based local government emerged as the hallmark of a nationwide movement beginning in the Stuart C. Aitken is assistant professor of geography and Rudy Prosser is a master of geography candidute at San Diego State Uniuersity. Geographical Analysis, Vol. 22, No. 4 (October 1990) 0 1990 Ohio State University Press Submitted 6/89. Revised version accepted 11/89. 302 zyxwvutsr zyxw zyxwvu / Geographical Analysis 1960s (Silver 1985). An academic backdrop for this movement was provided by Gans’ (1962) sociological study of life spaces and neighborhoods of immigrant groups, and by Lynch‘s (1960) study of cognitive, perceptual, and symbolic meanings of the city. Over the last two decades, a social and behavioral mapping of the city has emerged, not only in terms of its obvious demographic and functional aspects, but also in terms of its rich range of human meaningfulness. Nonetheless, discussion of neighborhood-level behavior and geography remains fuzzy. For example, in the planning literature neighborhoods have been defined as “geographic units within which certain social relationships exist, although the intensity of these relationships and their importance in the lives of individuals vary tremendously” (Downs 1981, p. 15). The current study was provoked in part by this type of glib reference to the behavioral and geographical nature of neighborhoods. The paper discusses residents’ cognition of neighborhood in terms of simple spatial forms. Cognitive and behavioral data are structured within a geographic information system to create surfaces of residents’ familiarity with, and e .perience of, a neighborhood in San Diego, California. In addition, spatial autocorrelation and directional autoregression techniques are used to establish the continuity and form of residents’ familiarity with their neighborhood. We raise three issues in this paper. The first relates to the renewed interest within the social sciences and planning professions in microlevel approaches to urban living, and geography’s unique position to contribute to this research emphasis. In this study, we use methods that are inextricably geographic to address variations in neighborhood cognition. The second issue relates to the study of the form of neighborhood cognition. Stanton (1986, p. 325) casts doubt on the complexity previously associated with explanations of neighborhood form based upon residents’ cognitive representations. She suggests that residents rely a great deal on habit and, as a consequence, neighborhood form can be best understood by investigating residents’ habitual experience of place. This study confirms some of the implications of Stanton’s (1986, 1989) work. Finally, we contend that the qualitative changes which take place in spatial learning as one proceeds from network knowledge to areal knowledge may be much more complex than has been acknowledged in previous theories of spatial cognition (Golledge 1978; Smith, Pellegrino, and Golledge 1982; Golledge et al. 1985; Stem and Leiser 1988). According to existing theory, knowledge of a complex network should be enough to provide an areal knowledge structure at appropriate boundaries (Golledge 1987). Our results suggest that this is not necessarily so. As a consequence, we advise that an important distinction should be made between areal knowledge (propositions and facts) and survey knowledge (ability and skill to traverse an area). Each of these three issues are dealt with below in some detail before turning to the empirical study. zyxw MICROLEVEL ANALYSIS OF URBAN LIVING In both North America and Europe, microlevel approaches to neighborhood and community living have begun to receive renewed attention (Ahlbrandt 1984; Banerjee and Baer 1984; van Vliet-et al. 1987; Altman and Wandersman 1987; Pickus and Gober 1988; Pacione 1990; Aitken 1990). This interest is derived in part from an increasing dissatisfaction with structuralist perspectives (Lebas 1982; Gans 1984; Szelenyi 1986), and in part from sustained advances in behavioral theories and methods over the last decade. Structuralist perspectives provide important insights into the abstract functioning of the urban system and they create a theoretical tension that can inform other perspectives, but the quality of people’s daily lives is not the principal concern of these macro- and meso-level approaches. At about the same time that social zyxw zyx zy zy Stuart C. Aitken and Rudy Prosser / 303 geographers were turning to structural theories (cf. Harvey 1973; Pickvance 1974), a new field of “environment and behavior”-as it came to be called-was developing on quite different grounds. While Marxist social scientists castigated traditional urban theories, environment and behavior researchers set out to explore “what urban ecology had left in the dust” (Michelson 1976). Researchers who embraced this new field began to explore a coalescence of concerns about the behavioral and social dimensions of planning and design within the physical contexts of human experience (Aitken and Sell 1989). This research broadened as social scientists were impressed by the role played by the physical environment on human experience. Environment and behavior has been, from its inception, an interdisciplinary field at the confluence of many parts of the social sciences and design professions. Geographers offer much to this field in terms of understanding underlying concepts of spatial meaning and behavior. It is clear that more and more psychologists, environmental sociologists, landscape architects, planners, and designers who are interested in neighborhoods are writing on topics that are essentially geographic. Although geography’s intellectual associations with these cognate disciplines have been extensively nurtured in the past, some reviewers suggest that this interdisciplinary exploration has been all but abandoned (Goodey and Gold 1985; Goss 1988). And yet behavioral research in geography has sustained important theoretical and methodological advances over the last decade (Golledge and Stimson 1987; Aitken et al. 1989). Geographers have learned a great deal about concepts of meaning such as spatial cognition, environmental learning, personal space, and territoriality. Problems with measuring spatial cognition and behavior have been overcome with varying degrees of success using a variety of fairly sophisticated procedures such as the repertory grid (Aitken 1987), preference modeling (Veldhuisen 1988), contextual analysis (Preston 1986), multidimensional scaling (Mackay and Zinnes 1988; Aitken 1990), discrete choice modeling (Wrigley, Longley, and Dunn 1988), and qualitative data analysis (Leitner, Nijkamp, and Wrigley 1985). Although measurement problems are slowly being dealt with, until quite recently spatial data structure problems were either ignored or were considered intractable by behavioral researchers. In recent years, the technology surrounding computer-based geographic information systems has increased to such an extent that the complexity of managing, analyzing, and displaying spatial information is now significantly reduced. Spatial data such as population characteristics, housing attributes, service locations, and transportation flows are now readily transformed into geographic information systems (GIs). To date, however, there has been little development of behavioral data bases which can be incorporated into a GIs. The spatial nature of data such as perceptions and attitudes is beyond doubt, but how these data can be managed spatially is not a principal concern of most environment and behavior researchers. In addition, some of the issues surrounding the analysis of behavioral data sets-such as spatial autocorrelation-have received scant attention (Golledge and Stimson 1987, p. 199). The current study explores the spatial autocorrelation in a neighborhood-scale, behaviorally based, geographic information system. zyxwvu EVALUATING RESIDENTS’ SPATIAL KNOWLEDGE OF NEIGHBORHOODS Clearly neighborhoods are made up of behavioral, social, political, economic, and physical environments. The behavioral environment relates to the amount of time residents spend in the area, the tendency to shop locally, or the frequency with which residents use neighborhood recreational facilities. Subsumed within the concept of a neighborhood behavioral environment is the notion of local activity space. This may be defined as the set of neighborhood locations with which an 304 zyxwvuts zyxw zyxwvu / Geographical Analysis individual has direct contact within a particular time period. The proximity of friends and relatives or affiliations with nearby clubs and social organizations are indicators of residents’ social environments. The political cohesion of a neighborhood is often directly associated with social homogeneity (Smith 1985). Lee (1968, 1978) was one of the first environment and behavior researchers to emphasize that residents synthesize the social and territorial aspects of neighborhoods into a coherent cognitive representation, or “socio-spatial schema.” He reported that although each resident’s “ . . . constellation of experience and action is apparently unique,” there is some evidence of “norm-formation” (Lee 1968, p. 248). Lee differentiated three common types of schemata based upon female residents’ cognitive maps, friendship patterns, club membership, and shopping patterns. First, the “social acquaintance neighborhood’ represents the physical extent of social interaction. In the second sociospatial schema, the “homogeneous neighborhood,” social interaction is relatively low and cognitive factors play a large part. Residents delimit boundaries based upon factors that relate to “people like us.’’ Finally, the “unit neighborhood’ approximates Perry’s (1929) planning ideal, containing a range of shops, institutions, and amenities. Geographers have studied social networks as a basis of neighborhood integrity (for example, Walker 1977; Smith and Smith 1978), but there has been surprisingly little microlevel investigation of the importance of the built environment in prescribing sociospatial schemata. Lee’s (1968) typology encompasses the social and economic factors that help determine how individuals identify with place, but it does not account for the effects of neighborhood morphology (Stanton 1986, p. 304). Nor does Lee’s typology account for an individual’s cognitive shaping of the landscape (Silver 1985, pp. 170-71). It does not acknowledge that the physical environment provides a setting for activities that contribute to neighborhood integrity in pervasive ways. Golledge and Stimson (1987, pp. 64-70) suggest that the cues which identify neighborhoods appear to be predominantly physical. These cues include land-use, street-pattern, house-types, density, and identifiable boundaries. Everitt and Cadwallader (1981) contend that the structure of a local area has a strong influence on residents’ cognitive delimitation of their neighborhood. They question, however, the internal cohesion of these areas: “The home area undoubtedly remains an area with some meaning, but it remains to be seen whether this meaning is consistent” (1981, p. 34). An investigation of this kind of consistency is a point of departure for the current study. Of particular interest in the study of neighborhood consistency is the role of distance and space in stimulating or retarding activity (Metton 1969; Irving 1975; Smith and Smith 1978; Cadwallader 1981). Generally, geographers contend that activity space is more intensive near one’s home and declines in use with increasing distance from that base. Distance decay is the exponential decline of an activity with increasing distance from its point of origin. Stanton (1986, p. 302) contends that the spatial form of a resident’s home area does not exhibit such spatial consistency. She suggests that the directly experienced neighborhood may be linear rather than areal, beginning and ending at particular points on a path rather than having a continuous distance decay to identifiable boundaries. Neighborhood form, Stanton asserts (1986, p. 319), can be greatly influenced by factors such as street morphology and block length. ACQUISITION OF SPATIAL KNOWLEDGE: LANDMARK, ROUTE, SURVEY AND AREA INFORMATION Environment and behavior research has demonstrated that in the course of learning about the urban environment, an individual’s spatial knowledge changes quantitatively in terms of the amount of information learned and qualitatively in zyxwv zy Stuart C. Aitken and Rudy Prosser / 305 terms of how this information is stored and processed (Siege1 and White 1975; Evans 1980; Thorndyke and Hayes-Roth 1982). Much evidence has been accumulated to suggest that people learn places through a progression of stages which involve processing “landmark’ information first of all, then “route” information, and lastly, “survey” information (Golledge 1978; Smith, Pellegrino, and Golledge 1982; Golledge et al. 1985; Stem and Leiser 1988). A fourth knowledge structure which relates to “areal” information comprises a knowledge of places, and possibly spatial configurations, in an area. The general assumption is that an individual’s knowledge structure provides a basis for interpreting places in an external environment. In theories of information processing, a knowledge structure is generally viewed as a set of symbol structures representing certain aspects of an individual and the individual’s environment (Golledge 1987, p. 138). It follows that spatial knowledge structures are a subset of an individual’s knowledge of the environment. Declarative knowledge (knowing what) is the cognitive process that encompasses the individual’s understanding of places and things. It is generally assumed that declarative knowledge structures process landmark information and frames of reference (Minsky 1975; Kuipers 1978). Other researchers have broadened the concept of “frames” to encompass “stereotypes and schemata” which are related, in part, to Lee’s (1968, 1978) typology (Bobrow and Norman 1975; Kuipers 1980, 1982). Procedural (knowing how) knowledge structures encompass, at a minimum, the individual’s understanding of how to get from one place to another. Route knowledge comprises a series of procedural descriptions involving some record of landmarks and paths. While it is recognized that declarative knowledge is fundamental to human activities in an urban setting, it is also recognized that a great deal of human knowledge is procedural in nature. Gale et al. (1990) suggest that route knowledge can be quite parsimonious. They found that successful navigation along a route does not require extensive declarative or areal knowledge of scenes along the way. The bulk of the empirical and theoretical work to date has concentrated upon the integration of landmarks and routes (Golledge 1987; Doherty et al. 1989; Gale et al. 1990). Configurational knowledge (knowing where, what, and how) is less clearly defined but it generally refers to the individual’s ability to traverse complicated configurations of paths and nodes within some external (not the body axes) frame of reference, including an ability to find new routes between nodes without getting lost. The exact nature of configurational, or survey knowledge and its relationship to areal, or map-like knowledge appears to be largely unknown. Survey knowledge incorporates the concept of a good “sense of direction” (Kozlowski and Bryant 1977), but areal knowledge refers more to a familiarity with places, and possibly route configurations, within an area. More empirical research is needed to investigate these connections, for the results have some implications for theories of spatial knowledge structures. Much of the literature suggests that there is a progression from declarative to procedural to configurational knowledge representing the successive integration of landmark, path, and survey representations (Golledge et al. 1985, p. 129). Several researchers report that this progression leads to an increasingly complex cognitive representation (Downs and Stea 1973; Kaplan and Kaplan 1982; Thorndyke and Hayes-Roth 1982). For example, according to existing theory, knowledge of a complex network should be enough to provide an areal knowledge structure at appropriate boundaries (Book and Garling 1980; Pick and Acredolo 1981; Golledge 1987). It has been suggested also that spatial knowledge at the metropolitan scale is generally at the procedural rather than configurational level since most people’s activity spaces are limited (Golledge and Spector 1978; Stem and Leiser 1988). We suggest that these findings need not necessarily hold true for spatial knowledge structures at the neighborhood scale. Golledge (1978) contends that the concepts 306 zyxwvuts zyxw / Geographical Analysis of neighborhood and community develop through a spread effect that occurs in the vicinity of nodes and paths. This spread effect constitutes areal, and possibly survey, knowledge of a place. There is limited information on how this type of spatial knowledge is structured. Stanton (1986) found that although her sample could be considered to possess fairly complete areal knowledge due to the length of time they had lived in their neighborhoods and due to the small scale of her study areas, the form and consistency of respondents’ home area representations were highly variable. She asserts that this is because residents rely a great deal on habit when traveling through their home area. The concept of a cognitive representation involving complex configurational knowledge as a sophisticated guide is not necessarily appropriate to describe residents’ movement through places in which they live. Aitken and Bjorklund (1988, p. 59) express these mental representations of local areas in terms of habitual behavior applied to ordinary, everyday environments. For daily transactions with an environment, individuals develop and use knowledge structures that process “areal” information, but in a partial way because there are a limited number of associations that relate cognition to behavior. In some ways, this is related to what Kuipers (1983) calls a “commonsense knowledge structure” to the degree that intentions and expectations are not made explicit with this kind of areal knowledge. An “extraordinary” change (Aitken and Bjorklund 1988) in the local environment will result in some heightened acuity on the part of the resident which may raise expectations and intentions but, on the whole, residents’ day-to-day transactions with their local area are habituated by the familiarity of an unchanging environment.’ This concept relates also to the work of Tversky (19811, who argues that memory representations are schematized simply because this makes it possible for people to take in large amounts of information. As a cognitive process, habituation softens the complexity of any environment but, in our local environment more than any other, it encompasses an innate need for order and homeostasis. Thus, spatial knowledge at the neighborhood scale need not require complex cognitive representations to interpret what is for residents a relatively familiar environment. As a consequence, normative neighborhood forms can be investigated through fairly simple measures of place familiarity and experience. zyxwvuts DEVELOPING A BEHAVIORAL GEOGRAPHIC INFORMATION SYSTEM A sample of respondents from Hillcrest, a central San Diego neighborhood, were interviewed concerning their level of association with the local environment. These data are derived from a larger survey of residents’ perceptions of neighborhood identity and change (see Aitken 1990). Sixty-five percent ( n = 51) of the respondents in the original study were questioned on their level of experience and familiarity with their residential area. The sample is stratified such that length of stay in the community, age, income, and tenure variables closely correspond with those documented for the neighborhood as a population. The Haines’ Criss-Cross Directory fm San Diego (1987) was used to ensure appropriate areal coverage. Respondents were randomly selected from the five census tracts (3.00-7.00) that the Planning Department delimited as Hillcrest in 1976 (Fig. la; San Diego 1976). In reality, the boundaries of the neighborhoods in this area are overlapping and indistinct. Boundary defeasibility was exemplified in 1988 when the city planners redefined Hillcrest as a smaller area (Fig. Ib; San Diego 1988). Territorial coherence for Hillcrest is provided by a large neighborhood sign which marks the zyxwvut zyxwvu ‘Aitken (1990) explores some of the relationships between the scale of a neighborhood change, its proximity to a resident, and the resident’s perception of that change. V 308 zyxwvutsr zyxw / Geographical Analysis B zyxwvutsrqp zyxwvutsrqpon zyxwvutsrq MI l e s n FIG. Ib. Official Delimitation of Uptown Neighborhoods in 1988 spatial extent. When this kind of information is being sought, residents are usually asked to draw lines on maps or aerial photographs (cf. Lee 1968; Golledge and Zannaras 1973; Everitt and Cadwallader 1981; Pacione 1978, 1983; Bishop 1984). We gave the respondents a 1 : 4800 aerial photograph that covered the five census tracts which designated Hillcrest in 1976. Aitken and Ginsberg (1988, p. 74) have shown that even young children have little difficulty comprehending the vertical images. presented by instruments. such as aerial photographs provided that the zyxwv zyx zy Stuart C.Aitken and Rudy Prosser I FIG.2. Residents’ Perception of Neighborhood Boundaries / 309 zyxw scale is small enough. This notwithstanding, we provided an alternative instrument to all our respondents in the form of a map at the same scale as the aerial photograph. The respondents were asked to name their neighborhood and draw its boundaries on either the aerial photograph or the map. Only one respondent elected to use the map. The ARC/INFO GIS was used to develop information from the cognitive mapping test into a geographic data base. Figure 2 represents the street pattern of the study area with the residents’ perceived neighborhood boundaries superimposed. The center of Hillcrest is marked on Figure 2 by the symbol “H.” Only 4 percent of the respondents identified the planners’ 1976 Hillcrest boundaries zyxwvuts 310 zyxwvuts zyxw zyxwvu / Geographical Analysis as their neighborhood. All but one of the respondents in the University Heights area to the northeast named and identified a neighborhood that corresponded to the planners’ perceptions in 1988 (Fig. Ib). Discounting the residents of University Heights, Figure 2 suggests that residents’ perceptions of the boundaries of Hillcrest differ somewhat from those of the planners in 1988. Eighty percent of the respondents who lived in the officially designated Park West or Medical Complex neighborhoods (see Fig. Ib) stated that they lived in Hillcrest, and 5 percent identified with Mission Hills to the west. Street patterns and the presence of canyons clearly influence the form of perceived neighborhood boundaries. And yet there is little consensus; 50 percent of the respondents identified Hillcrest with a north-south alignment, only 25 percent agreed with the planners east-west alignment, and 25 percent recognized both types of alignment. Researchers have identified many problems associated with simple cognitive mapping tests (see Pacione 1983), not least of which are the demand characteristics of the task of drawing boundaries which render it highly unlikely that the respondents should refuse to perform it. To alleviate this and other problems, investigation beyond the simple cognitive mapping analysis probed residents’ familiarity with, and experience of, their residential area. Familiarity and Experiential Surfaces Early empirical work by geographers on how residents acquired knowledge of urban areas exploited the concepts of spatial familiarity and experience. Golledge and his colleagues cite work by Rivizzigno (1976) and Spector (1978) which focused on the information sources that residents use to acquire knowledge about locations in the urban environment (Golledge and Spector 1978; Golledge and Stimson 1987; Golledge 1987). The locations used in these experiments were chosen on the basis of general population familiarity (Spector and Rivizzigno 1982, p. 48). Multidimensional scaling analyses were used on respondents’ estimates of distances between locations in order to recover familiarity surfaces (Golledge, Rayner, and Rivizzigno 1982). These surfaces-representing the whole of the city of Columbus, Ohio-tended towards greater accuracy as the respondent spent more time in the urban area. Golledge (1987, p. 140) reports on Spector’s (1978) use of individual regression analysis to show that frequency of interaction (that is, experience) is the greatest factor influencing the accuracy of an individual’s familiarity surface. To our knowledge, the concepts of familiarity and experiential surfaces have not been applied to neighborhood-scale knowledge structures and, in addition, spatial autocorrelation methods have not been used to interpret the fundamental geographic associations underlying these surfaces. Respondents in our study were presented with two more aerial photographs of the study area. These photographs had regular cellular grids superimposed. Each cell encompassed approximately a two-block area. Respondents were asked to indicate on the first photograph their level of familiarity with each grid cell on a scale from 0 (unfamiliar) to 10 (very familiar). On the second photo they were asked to indicate their experience of the place in terms of how many times they had passed through each cell in the past week. The choice of a two-block square scale was somewhat arbitrary, it encompassed the largest number of grids (eighty per aerial photograph) that we felt the respondents could consider without fatigue. Of course, the choosing of a spatial scale in a study of this kind is fraught with difficulty. We asked respondents to represent their average familiarity with each grid cell but we have no way of knowing how influential familiar nodes within cells were in the respondents’ evaluations. Moreover, what would have happened if our cells encompassed half a block or ten blocks? Such discussion takes us into the geography of fractals (Goodchild and Mark 1987), and beyond the scope of this zyxw zyxw zyxw zy Stuart C . Aitken and Rudy Prosser / 311 paper. In defense of the two-block scale of resolution, however, Aitken (1990) has shown that a “territorial” step-boundary occurs in Hillcrest at about the two-block level. In particular, residents of Hillcrest are particularly sensitive to changes within two blocks of their homes. We might speculate from this finding that, on average, Hillcrest residents’ familiarity finds some kind of resolution at about two blocks. In an attempt to validate the experiential data we asked respondents where they went (nearest intersection) for sixteen different activities (for example, shopping, banking, and various social activities), and how often they went there in an average week. We asked also whether or not they felt that this activity was conducted in their neighborhood. This latter question was asked at the end of the interview to avoid, as far as possible, errors due to the “context effect” between related questions (Schuman and Presser 1981; Wilson 1985; Converse and Presser 1986). In addition, we expected that it would be very difficult for respondents to maintain a conscious consistency between the neighborhood boundaries that they had delimited at the beginning of the interview and the later questions concerning neighborhood activities. Given these assumptions, we found that the responses to the cognitive mapping test and the neighborhood activities questions were surprisingly consistent. Less than 8 percent of the activities (816 all together) that respondents indicated as being neighborhood-oriented lay outside of the boundaries that they had used to delimit their neighborhood; and less than 1 percent of the activities that respondents stated were outside of their neighborhood actually fell within the neighborhood boundaries they drew as part of their cognitive mapping exercise. A comparison of respondents’ experiential surfaces and the frequency data on neighborhood activities was less encouraging. The frequency data was tabulated by cell location for each individual and compared with their experiential data. In only four cases (7.8 percent of the respondents) was there a significant relationship between the two data sets (t-test, p < 0.05). Although disappointing in terms of the association between the experiential and activity data, these results are consistent with the way we framed our questions. The experiential data were derived by asking respondents to indicate how often they passed through a particular grid cell, not how often they had stopped there for a particular activity. This latter kind of data would be more appropriate in an analysis of knowledge structures based on nodes. The familiarity and experiential data were structured into the GIS. Two structures were created for each data base. The first used an absolute coordinate system taken directly from the aerial photograph. This structure was superimposed onto the street network (Figures 3a and b). The second used a relative coordinate system centered on the residents’ homes (Figures 4a and b). Figure 3a suggests that residents have a strong familiarity with the center of Hillcrest and its predominant business district. The residential areas around the north-south/east-west street axes and in University Heights to the northeast are also noted by a high degree of familiarity. In addition, residents display a relatively high degree of familiarity with Balboa Park to the southeast. Figure 3b indicates the obvious importance of the prominent north-south/east-west street network with regard to people’s experience of the area. The arms of this network hinge around the central part of Hillcrest. Figure 4a displays a classic distance decay of residents’ familiarity away from their homes. The plateau effect out to about six or seven cells from the home position is a measure of the average distance of each respondent to the area that has the greatest familiarity to most respondents, that is, the central business area of Hillcrest. The relative experiential surface is confined to a small area around the home (Fig. 4b). As one would expect, the experiential surfaces represented in Figures 3b and 4b have several local maxima which describe places visited during a typical week such as shops and parks. zyxwvuts zyxw 312 / Geographical Analysis A zyxwvu zyxwvut zyxw zyxwvutsrq Mi l e s b 8 0 to 2 ed2.l t o 4 m4.1 View t o f r o m I 172 I1816.1 18.1 t o t o 8 10 L e v e l o f 0 Low F a n i ~ t a r i t y 10 = H i g h G Southwest FIG.3a. Behavioral CIS Structured from an Absolute Coordinate System Taken Directly from the Aerial Photograph Showing the Aggregated Familiarity Surfaces Stuart C. Aitken and Rudy Prosser B zyx zy / 313 zyxwvut zyxwvut s zyxwv zyxwvut Mi l e s b 6 9 ? o 4 172 2 zyxwvutsrqp m6.l 8. 1 €734.1 ? o View f r o m t o t o 8 10 L e v e l 0 5 of L o w Experience 10 H , g h 6 Southwes+ FIG. 3b. Behavioral GIS Structured from an Absolute Coordinate System Taken Directly from the Aerial Photograph Showing the Aggregated Experiential Surfaces 314 zyxwvut zyxwvuts zyxw / Geographical Analysis A I €to t o 2 m4.1 f o 6 V i e w f r o m zyxwv zyxwvut zyxw zyxwvu M6.1 to 8.i t o 8 10 Levei o f 0 = Low F a m i i iarity 10 = H i g h Southwest FIG.4a. Behavioral GIS Structured from a Relative Coordinate System Centered on Each Resident’s Home Showing the Aggregated Familiarity Surfaces zyxw zyxwvu zyxwv zyx z Stuart C. Aitken and Rudy Prosser B zyxwvuts zyxwv €I9 t o 2 M6.1 t o 8 a2.1 t o 4 8 . 1 ! o 19 m4.1 t o 6 / 315 Legel 0 = o f Low Experience 10 High FIG. 4b. Behavioral GIS Structured from a Relative Coordinate System Centered on Each Resident’s Home Showing the Aggregated Experiential Surfaces 316 zyxwvutsr zyxw zyxwvuts / Geographical Analysis ANALYSIS OF NEIGHBORHOOD FORM Stanton (1986) concludes that there are several forms of home area. Her original study investigated the incidence of different types of home area forms amongst 157 residents in Brooklyn Heights, New York City. Collectively, these residents described several morphologies with principal distinctions between those that were areal, those that were linear, and those that were a composite of linear and areal elements. Each of these three categories comprised roughly one-third of the home areas of her sample. Stanton speculates that “it is not implausible that there are two major systems (one linear and one areal) for skills such as navigation and place cognition: normal brain functioning exhibits many other instances of overlapping, partially redundant cognitive systems” (1989, p. 3). Figure 5 illustrates the categories of neighborhood form that Stanton originally devised. Stanton’s (1986) respondents-who had never been asked to think in such terms before-were easily able to distinguish between categories. In addition, her respondents reported to a highly significant degree that they had the same home area form at their previous adult residence (chi-square, p < 0.0001). She suggests that this may be an indication of these neighborhood categories’ importance to how residents make sense of their surroundings. Not content with these results, Stanton (1989) conducted a five-year follow up study of fifty-three of the original sample of residents. Of these, only five respondents (8 percent) had a shift in the category of their home area form. Interestingly, in each case, the shift was from areal to linear configurations. While the sample size is too small to achieve significance, some support is garnered for the persistence of Stanton’s home area forms as enduring cognitive structures. Stanton’s (1986, 1989) results suggest that individuals may have a preference for a particular home area morphology if the landscape empowers it. Alternatively, the habitual nature of lived experience may impose a relatively immutable pattern on the individual’s mental landscape. zyxw I zy A SINGLE PATH I LAN AREA zyxwvutsrqponm I A NETWORK OF INTERCONNECTED PATHS A NUMBER OF SEPARATE PATHS FIG. 5. Diagrams of Possible Neighborhood Forms (source: Stanton 1986) zyx zyx zy zyxwvutsr Stuart C. Aitken and Rudy Prosser TABLE 1 Neighborhood Forms Identified by the Hillcrest and the University Heights Residents Home itself A Single path An area A network of interconnected paths A number of separate paths / 317 zyx z Hillcrest U.Heights 1 4 15 15 1 1 1 7 4 14 2 37 As a consequence of Stanton’s (1986) original study, each of the respondents in the current study was presented with the four neighborhood forms depicted in Figure 5. They were asked to identify which of these best represented their residential area. A fifth category comprised the home itself. The interview design was structured to minimize the effect of the respondents’ creation of familiarity and experiential surfaces on their identifying a home area form. First, respondents were asked to identify the form of their neighborhood early on in the interview. Second, we anticipated that the task of creating familiarity and experiential surfaces was too labor intensive and thought-provoking to be influenced by any of the respondents’ prior answers. In any case, numerous experiments on interview design have shown that there is rarely any context effect between related questions (see Schuman and Presser 1981; Wilson 1985; Converse and Presser 1986). It was expected that residents living in University Heights would view their neighborhood as an area. Although it does not have a dominant center, University Heights is a relatively homogeneous residential area, bounded by canyons on its northern and western sides and by two major roads on its southern and eastern sides. University Heights has much more clearly defined boundaries than Hillcrest. In addition, curved streets and cul-de-sacs tend to limit the through traffic in the area. Table 1 reveals, however, that only one of the respondents from University Heights perceived their neighborhood as an area. On the other hand, 40 percent of the Hillcrest residents perceived their neighborhood as an area. Sixty-five percent of those who perceived their neighborhood as networks or paths also mentioned that their neighborhood had a boundary. One can surmise then, that although bounded, University Heights is generally not perceived as a spatial area but as a street network. A possible explanation of this lies in the activities of residents. As mentioned earlier, respondents were asked to locate where they pursued a series of common activities such as grocery shopping, banking, eating at restaurants, going for walks, biking, et cetera. Residents of Hillcrest pursued significantly more activities in their neighborhood than residents of University Heights (t-test, p < 0.001). University Heights is primarily residential, offering only a few convenience shops at its periphery. Hillcrest, on the other hand, has a fairly large pedestrian-oriented business district. Residents of Hillcrest spend more time on activities in their neighborhood. These results tentatively support the notion that the more people have direct experience of an area, the more likely they are to develop areal-based knowledge as opposed to linear-based knowledge. Merely asking people whether they feel their neighborhood is an area or a network does not adequately test the efficacy of knowledge structures. The following section describes the use of spatial autocorrelation to establish the association between respondents’ perceived neighborhood form and their familiarity with the area. Measuring and testing for spatial autocorrelation is a central theme of geographic research. Spatial autocorrelation exists whenever a variable exhibits a regular zy 318 zyxwvuts zyxw zyxwvutsr / Geographical Analysis TABLE 2 Moran’s I Coefficients for Hillcrest and University Heights Residents and for Those with Areal-based Perceptions and Those with Linear-based Perceptions zyxwvuts MOWN’S I COEFFlClENT NEIGHBORHOOD: Hillcrest (n = 37) University Heights (n = 14) zyxwvutsrqp FORM: Areas (n = 16) Networks (n = 35) ‘not si ificant **signi%nt at the 0.001level 0.4813* 0.3553 0.6218** 0.3668 pattern over space in which its values at a set of locations depend upon values of the same variable at other locations (Odland 1988, p. 7). It is evident that the neighborhood familiarity surfaces exhibit important spatial regularities. It was posited that a spatial autocorrelation coefficient would provide some measure of this regularity. Cliff and Ord (1981) emphasize that the Moran I coefficient (Moran 1948) is the most useful for the evaluation of spatial autocorrelation. The index is based upon the covariation of juxtaposed map values. Similar to a classical correlation coefficient, as similar values tend to be in juxtaposition with one another, then I + 1. Alternatively, as dissimilar values tend to be in juxtaposition with one another, then I --t - 1. And for a random relation, I + 0. Griffith (1987) describes spatial autocorrelation and the nature of a map pattern in terms of spillover effects from one spatial area to another. Positive spatial autocorrelation exists in map patterns if spillover from one areal unit causes values to change correspondingly in juxtaposed areas, at least more than would be expected by chance. Griffith and Jones (1980) found that the variation in distance decay over a set of urban places is partially predictable from measured levels of spatial autocorrelation for geographic distributions. In the current study, Moran I coefficients are used to determine the effects of distance decay and the influence of the physical environment on residents’ cognition of neighborhood form. The regular grid pattern upon which the familiarity data were set offers the additional advantage of “spatial neighborhood stationarity” (Tobler 1979). In other words, each cell has the same number of neighbors and those neighbors have the same distance relations. A Moran I coefficient was calculated for each respondent’s familiarity data. A “Queen’s Matrix” (Clark and Hosking 1986, p. 383) approach was used so that the data in the cells on the diagonals are included as neighbors as well as those on the adjacent rows and columns. This introduces some variation in the distances to neighbors, but their number and the set of distance relations remain uniform across all cells that are not on the boundaries (Odland 1988, p. 23). In the following analyses boundary effects are partially countered by focusing on the interior cells of the familiarity surface. As one might expect, 98 percent of the familiarity surfaces had positive Moran I coefficients, and 96 percent of these were significant (normality significance test, p < 0.05). Table 2 reports no significant difference between the Moran I coefficients of the Hillcrest and the University Heights residents (t-test, p < 0.05). These two sample sets of familiarity data can be considered to originate from the same population. Attention may be turned instead to residents’ perception of neighborhood form. Table 2 reveals that residents who perceived that they lived in a spatial area had significantly higher Moran I coefficients than those who reported that they lived in a network or a series of paths (t-test, p < 0.001). It seems clear zyxwvu zyxwv zyxw zyx zy Stuart C. Aitken and Rudy Prosser / 319 zy that this simple measure of spatial familiarity represents accurately the form of neighborhood perceived by residents. Residents who felt that they lived in an area had high spatial autocorrelation coefficients on their familiarity surfaces with a mean Moran I of 0.6218. This confirms that residents with areal-based perception exhibit a significant degree of spatial regularity in their neighborhood familiarity. A mean Moran I of 0.3668 for those residents who perceived their neighborhood in terms of the street network indicates less consistency in their familiarity surfaces. It suggests that for them, neighborhood form may be related to the street network and, perhaps, procedural knowledge. Analysis of activity spaces revealed that area-oriented residents pursued significantly more activities in their home area than network-oriented residents (t-test, p < 0.05). These residents were also more likely to walk in the neighborhood than to drive a car (t-test, p < 0.05). Given the developmental nature of place learning, one might expect residents with areal knowledge to have lived longer in the neighborhood than those with linear knowledge. The data supports this suggestion but the association is not highly significant (t-test, p < 0.25). Given the predominant north-south/east-west direction of streets in the Hillcrest area, one would expect the network-oriented residents to have discontinuities in familiarity at the interstices of the cardinal axes. To test this supposition, Odlands (1988, p. 26) regular grid spatial autoregression model was used to identify directional influences. Odlands model explains the value at a location based upon the values of neighboring locations. In the current context, the directional influences are extended to encompass the eight surrounding values of the Queen's Matrix. For example, the value of a variable X at any coordinate location V,V may be related to the eight surrounding values by the equation zyxwvu xu,u= a + 1 x u - 1 , 0 + 1 +4xu-l,o +b,xu,,-1 + + bZXU,"+l b5Xu+l,o + + + b3Xu+l,u+l bGXu-l,o-l bsxu+l,u-l where a is a constant and the various b are regression coefficients. Variation in the regression coefficients indicates spatial directionality. The regression model was tested on thirty cells that encompassed the predominant north-south/east-west axes in the study area. The model was run on each cell for all the network respondents and all the area respondents. Figures 6a and b represent all sixty regressions with the direction of each regression's significant coefficients and R-square values displayed in each cell (the "H" symbol identifies the cell that contains the central point of Hillcrest). As one would expect from a regression between a cell and its contiguous neighbors, the R-square values are all relatively high with a mean of 0.78 for the cells of the area-oriented residents and a mean of 0.83 for the cells of the network-oriented residents. The two sets of R-square values can be considered to originate from the same population (t-test, p < 0.001). Taken as a whole, the directional regression coefficients in each cell suggest that there is a slightly greater north-south rather than east-west trend in familiarity. North-south regression coefficient weights are somewhat higher than the east-west weights (t-test, p < 0.25). This provides evidence to suggest that the planners' east-west axial delimitation of Hillcrest in 1988 does not represent the views of all residents. Of consequence to the developmental theories of spatial knowledge are the number of noncardinal negative regression coefficients in the network-oriented residents' cells (Figure 6b). This suggests that the diagonal cells do not vary directly with the dependent cell, that is, their influence is negative. The primary zy zyxwvuts zyxw zyxwvu - 320 / Geographical Analysis R2=.77 R2=.82 1.o R2=.61 & .96 R2 R2=.73 =.54 -94 I -98 .72 .41 H R2=.86 R2=.84 R2=.70 \ R2=.91 r3 c l.O - R2=.85 . 8 2 .97 .40 .32 zyxwvutsrq 3 t 1 f-: R2=.71 R2=. 60 ~2=.98 t .18 .43 '48 R2=.79 .67 t .84 l'O R2=.84 R2=.83 R2=.85 R2=.94 R2=.59 .48 x . 3 R2=.69 zyxw zyxwvuts I I 1.2 .87 R2=.94 1'3 .82 R2=.69 t .86 t .49 -90 J 1.o R2=.83 R2=.79 R2=.61 R2=.78 R2=.96 t .90 .25 .82 R2=.82 t 3 1.1 R2=.67 '= .41 t .95 FIG. 6a. Directional Autoregression for Thirty Cells Which Encompass the Center of the Hillcrest and the Predominant Street Axes: Residents with Areal-based Knowledge orientation of the familiarity surfaces of the network-oriented residents is the north-south/east-west street network. Alternatively, diagonal cells have no negative influence on those with an areal orientation, and several show a positive influence (Figure 6a). Several researchers have shown that over time landmark information is integrated with procedural rules to facilitate movement behavior (Garling et al. 1981; Golledge et al. 1985; Gale et al. 1990). In addition, Gale et al. (1990) found that route knowledge is parsimonious. They suggest that successful navigation does not require extensive knowledge about scenes along the route. Less is known of the transition from linear knowledge to areal knowledge. Gkling et al. (1981) suggest that a cognitive representation of the spatial layout of a residential area can be acquired very quickly. Scale is perhaps an important consideration. To date, z zyxw zyx zy zyxwvuts zyxwvuts r r37 / 321 Stuart C . Aitken and Rudy Prosser 12=.88 R2=.77 R2=.52 .65 R2=.76 t2=.85 -29 R2=.93 R2=.90 .36 t2 t: . =.90 =.92 R2=.55 R2=.76 -43 .32 -31 .68 75 .49 R2=.89 R2 =.86 %# .71 R 2 =.92 R2=.91 .66+ IR2=.89 R2=.67 t .45 .56 1 .o R2=.93 zyx 8 f ‘1: f ‘1 .3gt+.5; 22=.95 -24 34 .24 .63 t2 -86 -60 .59 - “ O 7 .59 R2=.83 zyx -”1 .30 I R2=.84 .38 .68 FIG. 6b. Directional Autoregression for Thirty Cells Which Encompass the Center of the Hillcrest and the Predominant Street Axes: Residents with Linear-based Knowledge neither spatial knowledge structures nor developmental sequencing principles have been established for neighborhood scale cognition. The current study suggests that knowledge of a complex network may not be enough to provide an areal knowledge structure. Alternatively, areal knowledge need not comprise an understanding of survey procedures. We suggest that areal knowledge and survey knowledge might be overlapping-perhaps partially redundant-cognitive structures. Furthermore, little evidence is gained from the current study to support the notion that areal-based knowledge of small-scale urban environments is related to how long an individual has resided in the area. We suggest rather that areal-based knowledge at this scale is gained through activities such as walking in the neighborhood, that is, knowledge that goes beyond that of the basic neighborhood street network is gained by exploring neighborhood 322 zyxwvutsr zyxwvu zyx / Geographical Analysis parks and canyons, and by taking shortcuts through alleyways and across vacant lots. Those residents who perceive their neighborhood as an area create familiarity surfaces that not only have cardinal directionality but also some noncardinal influence. Alternatively, we show that residents who perceive their neighborhood as a network of streets demonstrate a familiarity based solely upon the area’s principal cardinal street directions. Any noncardinal influence is negative. CONCLUSIONS Research in the last decade has provided solutions to many of the problems associated with measuring and analyzing cognitive and behavioral data. Until lately, the spatial structure and form of these data have not been a major concern of researchers in the field of environment and behavior. In the current study we were concerned with creating a geographic basis for analyzing residents’ perception of their home area. The creation of the empirical data set that powered the study was premised upon the fact that perception of neighborhood form is related to the habitual behavior of residents within their local area. Simple measures of spatial familiarity and experience were used to create cognitive surfaces for each resident in the sample. The spatial complexity of these data required the topological and relational sophistication of an ARC/INFO geographic information system. Spatial autocorrelation coefficients provided an indication of the consistency of residents’ familiarity with their home area. Spatial autoregression models provided evidence of the form of neighborhood familiarity in terms of predominant directional influences. We draw attention to the importance of understanding the structure of geographic learning in small-scale environments. Evidence from the study suggests that there is a structural difference in the spatial familiarity of residents who perceive their neighborhood as an area and those who perceive it as a network. Street networks played an important part as the basis for both linear-based knowledge and areal-based knowledge, but noncardinal associations were more prominent in the familiarity surfaces of those residents who perceived an areal neighborhood form. Those residents with areal knowledge pursued significantly more activities in the neighborhood and were more likely to walk in the neighborhood than drive. Theories of spatial knowledge acquisition are still at an embryonic stage. Although some considerable advances have been made in understanding the transition between landmark and linear knowledge, we have little understanding of the relationship between linear knowledge and areal knowledge. On the basis of this study, we suggest that (i) there may not be a direct sequential relationship between linear-based and areal-based knowledge structures, (ii) knowledge of a complex network may not be enough to provide an areal knowledge structure, and (iii) an areal knowledge structure does not necessarily comprise an understanding of survey procedures. Although this study supports the contention that experience, familiarity, and environmental transactions help construct and maintain spatial knowledge structures, there is still considerable speculation concerning how information is selected, structured, stored, and represented in the mind. Our findings are at best suggestive. Although past studies have shown that nodal information is integrated with procedural rules in the development of linear knowledge structures, such a sequence may not hold for higher-order knowledge structures. Areal and survey knowledge might be processed by overlapping, reticulate cognitive structures that incorporate a high degree of redundancy. It is clear that we must develop a stronger theoretical basis from our empirical studies. This basis would provide a clear distinction between different types of spatial knowledge structures, and it would identify how declarative, procedural, and survey cognitive processes relate to landmark, route, and areal information. zyxwv zyxw zyxw zy zy Stuart C. Aitken and Rudy Prosser / 323 The current study makes a modest addition to the literature and theory that concerns itself with residents’ cognition of neighborhood form and consistency. LITERATURE CITED Ahlbrandt, R. S. (1984). Neighborhoods, People and Community. New York: Plenum Press. Aitken, S. C. (1987). “Evaluative and Social Distinctions amongst Renters’ Residential Search Procedures.” The Canadian Geographer 31, 114-26. (1990). “Local Evaluations of Neighborhood Change.” A n d of the Association of American Geographers 80,247-67. Aitken, S. C., and E. M. Bjorklund (1988). “Transactional and Transformational Theories in Behavioral Geography.” The Professional Geographer 40, 54-64. Aitken, S. C., S. L. Cutter, K. E. Foote, and J. L. Sell (1989). “Contemporary Research in Environmental Perception and Behavioral Geogra hy.” In Geography in America, edited by C. Willmott and G. Gaille, pp. 218-38. London: M e m i Publishing. Aitken, S. C., and S. P. Ginsberg (1988). “Children’s Characterization of Place.” Yearbook of the Association of Pacijlc Coast Geographers 50,69-86. Aitken, S. C., and J. L. Sell (1989). “People and Paradigms.” In Changing Paradigms, edited by G . Hardie, R. Moore, and H. Sanoff, pp. 68-74. Selected Papers and Proceedings of EDRA 20. Madison, Wis.: Omnipress. Altman, I., and A. Wandersman (1987). Neighborhood and Community Environments. New York: Plenum Press. Banerjee, T., and W. C. Baer (1984). Beyond the Neighborhood Unit: Residmtial Enuironments and Public Policy. New York: Plenum Press. Bishop, J. (1984). “Passing in the Night: Public and Professional Views of Milton Keynes.” Places 1, 9-16. Bobrow, D. A,, and D. A. Norman (1975). “Some Principles of Memory Schemata.” In Representation and Understanding, edited by D. A. Bobrow and A. Collins. New York: Academic. Biiijk, A,, and T. Gkling (1980). “Processing of Information about Location durin Locomotion: Effects of Concurrent Task and Locomotion Patterns.” Scandinavian Journal of Psycho%gy 21, 185-92. Cadwallader, M. (1981). “Toward a Cognitive Gravity Model: The Case of Consumer Spatial Behavior.” Regional Studies 15, 275-84. Clark, W. A. V., and P. L. Hosking (1986). Statistical Methods fw Geographers. New York: John Wiley and Sons. Cliff, A., and J. Ord (1981). Spatial Processes: Models and Applications. London: Pion. Converse, J. M., and S. Presser (1986). Survey Questions: Handcrafting the Standardized Questionnaire. Beverly Hills, Cal.: Sage. Doherty, S., N. Gale, J. Pellegrino, and R. Colledge (1990). “Children’s versus Adults’ Knowledge of Places and Distances in a Familiar Neighborhood Environment.” Chddren’s Envfronments Qu~rterly 6, 65-71. Downs, A. (1981). Neighborhoods and Urban Development. Washington: Brookings Institute. Downs, R. M., and D. Stea (1973). lmage and Environment: Cognitive Mapping and Spatial Behavior. Chicago: Aldine. Evans, G. W. (1980). “Environmental Cognition.” Psychobgical &b!etin 88,259-87. Everitt, J., and M. Cadwallader (1981). “Husband-Wife Role Variation as a Factor in Home Area Definition.” Geografiska Annaler 63B, 23-34. Gale, N., S. Doherty, J. W. Pellegrino, and R. G . Golledge (1985). “Towards Reassembling the Image.” Children’s Enuironments Quarterly 2, 38-43. Gale, N., R. G. Colledge, J. W. Pellegrino, and S. Doherty (1990). “The Ac uisition and Inte ation of Route Knowledge in a Familiar Neighborhood.” Journal of Enuironmenta? Psychology. Fo8coming. Gans, H. (1962). The Urban Villagers. New York: Free Press. (1984). “American Urban Theories and Urban Areas: Some Observations on Contemporary Ecological Models and Marxist Paradigms.” In Cities in Recession: Critical Responses to the Urban Policies of the New Right, edited by I. Szelenyi. Beverly Hills: Sage. Gkling, T., A. Bijiik, A. Linberg, and T. Nilsson (1981). “Memory for the Spatial Layout of the Eve day Physical Environment Factors Mecting the Rate of Acquisition.” Journal of Experimental Psyc%logy 1, 263-77. Golledge, R. G. (1978). “Learning about Urban Environments.” In Timin Space and Spacing Time, Vol. 1, edited by T. Carlstein, D. Parkes, and N. Thrift.London: EdwarcfArnold. (1987). “Environmental Cognition.” In Handbook of Environmental Psychology, edited by D. Stokols and I. Altman, pp. 131-71. New York: John Wiley. Golledge, R. G., J. N. Rayner, and V. L. Rivizzigno (1982). “Com aring Objective and Cognitive Representations of Environmental Cues.” In Proximity and Prejrence: Problems in the Multidimensiaal Analysis of Large Data Sets, edited by R. G. a l l e d g e and J. N. Rayner. Minneapolis: University of Minnesota Press. zyxwvu zyxw zyxw zyxwvuts zyxw zyxwvu zyxw 324 / Geographical Analysis S. P. Marshall (1985). “A Conceptual Golledge, R. G., T. R. Smith, J. W. Pellegrino, S. Doh of Spatial Knowledge. Journal of Model and Empirical Analysis of Children’s Ac Environmental Psychology 5, 125-52. Golledge, R. G., and A. N. Spector (1978). “Comprehending the Urban Environment: Theory and Practice.” Geographical Analysis 10, 403-26. Golledge, R. G., and R. J. Stimson (1987). Analytical Behaoioural Geography. London: Croom Helm. Golledge, R. G., and G. Zannaras (1973). “Cognitive Approaches to the Analysis of Human Spatial Behavior.” In Environment and Cognition, edited by W. H. Ittelson, pp. 59-94. New York: Seminar. Goodchild, M. F., and D. M. Mark (1987). “The Fractal Nature of Geographic Phenomena.” Annals of the Association of American Geographers 77(2), 265-78. Goodey, B., and G. Gold (1985). “Behavioral and Perceptual Geography: From Retrospect to Prospect.” Progress in Human Geography 9, 585-95. Goss, J. (1988). “The Built Environment and Social Theory: Towards an Architectural Geography.” The Professional Geographer 40, 392-403. GriEth, D. (1987). Spatial Autocorrelation: A Primer. Resource Publications in Geography, Washington, D.C.: AAG. GriEth, D., and K. Jones (1980). “Explorations into the Relationship between Spatial Structure and Spatial Interaction.” Enoironment and Planning A 12, 187-201. Haines and Company, Inc. (1987). Criss-Cross Directory for South San Diego County. North Canton: Ohio. Harvey, D. (1973). Social Justice and the City. London: Edward Arnold. Holahan, C. J., and A. Wandersman (1987). “The Community Psychology Perspective in Environmental Psychology.” In Handbook of Enoironmental Psychology, edited by D. Stokols and I. Altman, pp. 827-61. New York: John Wiley. Huttmann, E., and W. van Vliet-, eds. (1987). Handbook of Housing and the h i l t Enoironment in the US. Westport, Conn.: Greenwood Press. Irving, (1975). “Distance, Intensity, Kinship: Key Dimensions of Social Interaction.” Sociology and Social Research 60, 70-86. Jacobs, J. (1961). The Death and Life of Great American Cities: The Failure of Town Planning. Penguin Books. Kaplan, S., and R. Kaplan (1982). Cognition and Environment, New York: Praeger. Keller, S. (1968). The Urban Neighborhood: A Sociological Perspectioe. Random House. Kozlowski, L. T., and K. J. Bryant (1977). “Sense of Direction, Spatial Orientation, and Cognitive Maps.” Journal of Experimental Psychology: Human Perception and Performance 4, 590-98. Kuipers, B. (1978). “Modeling Spatial Knowledge.” Cognitioe Science 2, 129-53. -(1980). “A Model of the Acquisition of Spatial Knowledge.” In Models of Dialogue: Theory and Application, edited by L. Bowl. London: McMillan. (1982). “The ‘Map in the Head Metaphor.” Enoironment and Behaoior 4, 202-20. (1983). “The Cognitive Map: Could It Have Been Any Other Way?” In Spatial Orientation: Theory, Research and Application, edited by H. L. Pick and L. Acredolo. New York: Plenum. k b a s , E. (1982). “Urban and ReQonal Sociology in Advanced Industrial Societies: A Decade of Marxist and Critical Perspectives. Current Sociology, Spring, 30. Lee, T. (1968). “Urban Neighborhood as a Socio-Spatial Schemata.” Human Relations 21, 246-68. (1978). “A Theory of Socio-Spatial Schemata.” In Humanscape: Environments for People, edited by S. Kaplan and R. Kaplan. N. Scituate, Mass.: Duxbury Press. Leitner, H., P.,Nijkamp, and N. Wrigley (1985). ‘‘ ualitative Spatial Data Analysis: A Compendium of Approaches. In Measuring the Unmeasurable, e%ed by P. Nijkamp, H. Leitner, and N. Wrigley, pp. 1-28. Dordrecht: Martinus Nijhoff. Lynch, K. (1960). The Image of the City. Cambridge, Mass.: MIT Press. Mackay, D. B., and J. L. Zinnes (1988). “Probabilistic Multidimensional Scalin of Spatial Preferences.” In Behaoioural Modelling in Geography and Planning, edited by R. G. Gollejge and H. Timmermans, pp. 198-222. London: Croom Helm. Metton, A. (1969). “ k Quartier: Etude Geographique et Psycho-Sociologique.” The Canadian Geographer 13, 299-315. Michelson, W. (1976). Man and His Urban Enoironment, Reading, Mass.: Addison-Wesley. Minsky, M. A. (1975). “A Framework for Representing Knowledge.” In The Psychology of Cumputer Vision,edited by P. H. Winston. New York: McGraw-Hill. Moore, G. T., D. P. Tuttle, and S. C. Howell (1985). Enoironmental Design Research Directions. New York: Praeger. Moran, P. (1948). “The Interpretation of Statistical Maps.” Journal of the Royal Statistical Society 10B, 243-51. Odland, J. (1988). Spatial Autocorrelation. Beverly Hills, Calif.: Sage. Pacione, M. (1978). “Information and Morphology in Cognitive Maps.” Transactions of the Institute of British Geographers 3, 548-68. (1983). “The Temporal Stability of Perceived Neighborhood Areas in Glasgow.” The Professional Geographer 35, 66-73. zyxwvutsrqp zyxwv zyxwvu - - zyxwvut - zyx z z zyxwvutsrq Stuart C . Aitken and Rudy Prosser / 325 -(1990). “Urban Livability: A Review.” Urban Geography 2, 1-30. Perry, C. (1929). “The Neighborhood Unit.” Neighborhood and Community Plunning, Vol. 7, Regional Plun of New York and Its Environs. New York: Russell Sage Foundation. Pick, H. L., and Acredolo, L. (1981). Spatial Orientation and Spatial Representation. Hillsdale, N.J.: Erlbaum. Pickus, J., and P. Gober (1988). “Urban Villages and Activity Patterns in Phoenix.” Urban Geography 9, 85-97. Pickvance, C. (1974). “On a Materialist Critique of Urban Sociology.” Sociological Review 22, 203-20. Preston, V. (1986). “A Case Study of Contextual Effects and Residential Area Evaluation in Hamilton, Canada.” Environment and Planning A 18, 41-52. Rivizzigno, V. L. (1976). “Cognitive Representations of an Urban Area.” Ph.D. dissertation. Department of Geography. Ohio State University. Rivlin, L. G. (1987). “The Neighborhood, Personal Identity, and Grou Affiliations.” In Neighborhoods and Community, edited by I. Altman and A. Wanderman. New Yorg Plenum Press. San Diego, City of, (1976). Uptown Community Plan. San Diego, City of, (1988). Uptown Community Plan. Schuman, H., and S. Presser (1981). Questions and Answers in Attitude Surveys: Experiments on Question F m , Wording, and Context. New York: Academic Press. Siegel, A. W., and S. H. White (1975). “The Development of Spatial Representations of Large-scale Environments. In Advances in Child Development and Behavior (vol. 10) edited by H. W Reece. New York: Academic. Silver, C. (1985). “Neighborhood Planning in Historical Perspective.” ]ournal of the American Planning Association, Spring, 161-74. Smith, C. A., and C. J. Smith (1978). ‘ b a t i n g Natural Neighbors in the Urban Community.” Area 10, 102-10. Smith, R. (1985). “Activism and Social Status as Determinants of Neighborhood Identity.” The Professional Geographer 37(4), 421-32. Smith, T. R., J. Q. Pellegrino, and R. G. Golledge (1982). “Computational Process Modeling of Spatial Cognition and Behavior.” Geographical Analysis 14, 305-25. Spector, A. N. (1978). “An Analysis of Urban Spatial Imagery.” Ph.D. dissertation. Department of Geography. Ohio State University. Spector, A. N., :nd V. L. Rivizzigno (1982). “Sampling Designs and Recovering Co itive Representations. In Proximity and Preference: Problems in the Multi-dimensional Analysis o;;“hrge Data Sets, edited by R. G. Golledge and J. N. Rayner, pp. 47-79. Minneapolis: University of Minnesota Press. Stanton, B. (1986). “The Incidence of Home Grounds and Experiential Networks.” Environment and Behavior 18(3), 299-329. (1989). “The Persistence of Home Ground Forms.” Paper presented at the Annual Meetings of the Association of American Geographers, Baltimore. Stern, E., and D. Leiser (1988). “Levels of Spatial Knowledge and Urban Travel Modeling.” Geographical Analysis 20(2), 140-55. Stokols, D., and I. Altman (1987). Handbook of Enuironmental Psychology. New York: John Wiley. Szelenyi, I. (1986). “The Last of the Marxist Urban Sociologists?” Contemporary Sociology 15, 707-10. Thorndyke, P. W., and B. Hayes-Roth (1982). “Differences in Spatial Knowledge Acquired from Maps and Navigation.” Cognitive Psychology 14, 560-89. Tobler, W. (1979). “Cellular Geography.” In Philosophy in Geography, edited by S. Gale and G. Olsson, pp. 379-86. Dordrecht D. Reidel. Tversky, B. (1981). “Distortions in Memoty for Maps.” Cognitive Psychology 13, 407-33. van Vliet-, W., H. Choldin, W. Michelson, and D. Popenoe, (1987). Housing and Neighborhoods. New York: Greenwood Press. Veldhuisen, J. (1988). “Utilities, Preferences and Choice.” In Behavioural Modelling in Ceography and Planning, edited by R. G. Golledge and H. Timmermans, pp. 223-44. London: Croom Helm. Walker, G. (1977). “Social Networks and Territory in a Commuter Village.” The Canadian Geographer 21, 329-50. Warren, D. (1978). “Exploration in Neighborhood Differentiation.” Sociological Quarterly 19, 310-31. Wilson, T. D. (1985). “Questionnaire Desi n in the Context of Information Research.” In The Research Interview: Uses and Approaches, edited f y M. Brenner, J. Brown and D. Canter, pp. 65-77. London: Academic Press. Wrigley, N., P. Longley, and R. Dunn (1988). “Some Recent Developments in the Specification, Estimation and Testing of Discrete Choice Models.” In Behavioural Modelling in Geography and Planning, edited by R. G. Golledge and H. Timmermans, pp. 96-123. London: Croom Helm. Zube, E. H., and G. T. Moore, eds. (1987). Advances in Environment, Behavior, and Design. New York: Plenum. zyxwvuts zyxwv zyxw zyx