-1MEANING IN THE ENVIRONMENT AS CONSTRUED BY ENVIRONMENTAL DESIGNERS AND LAYMEN BY PAUL STEVEN DEUTSCH Submitted in Partial Fulfillment of the Requirements for the Degree of Bachelor of Science at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY May, 1972 Signature of Author Department of Urban Studies and Planning, .May 12, 1 9A 7 2. Certified by. .. ....... sis Supervisor Accepted by... , Cha a De,;'tmental Committee on Theses Rotch JUL 24 1972 -2Table of Contents 5 7 7 8 Introduction........................................page The Problem.....................................page Prior Studies..................................page Personal Construct Theory......................page 10 11 The Research: Summary..........................page page 14 .................. Methodology ................... The Repertory Grid Technique.............. The Repertory Grid Test in ....... page 14 ......... page 16 Detail.... Methods of Analysis..............................page 18 Observations.*0@* ....................... Conclusions ............... . .. .. .. .. .page ....... 26 page 38 38 pae ........... ........... Theoretical ................. Methodological ................ . .. .... ....... pge 40 Pratical .............................. a9 41 Bibliography ......................... ae 47 49 Appendix A.................... ... pae Appendix B......................ae 50 Appendix C........................ae 51 Appendix 9......................ae 65 U -3Figures and Tables Table 1 General Environments used in this Study...............page 15 Table 2 Summary of Research Design................. ....a page 20 Table 3 Tables Depicting Three Different Distributions of Construct Pairs and their H and T Values.... ....... page 21 Table 4 Table of Observations............................... ...... page 28 Table 5 Total Percentage Variance Accounted ....... page 30 ....... page 33 Responsive vs. Representational Constructs..... ....... page 35 for by Factors ................................ Table 6 EDG and LG FactorLabes.......... Table 7 Table 8 Responsive vs. Representational Factors........ ....... page 36 Figure 1 Total Variance Accounted for by Factors.................... . ....... page 32 -4Acknowledgements The author would like to acknowledge the following people: Dr. H. Stephen Leff for his invaluable aid and theoretical help as the advisor for this thesis and developer of the wider applications of personal construct theory; Leonard and Suzanne Buckle for their resourcefulness in securing funding for the study and constant encouragement; Professor Mary C. Potter for her useful suggestions and criticisms; Professor Kevin Lynch for enriching the scope of the study. *1 Abstract A version of George Kelly Is (1955) Role Construct Repertory Grid Test was administered to ten graduate students of environmental design (Environmental Design Group--EDG) and ten graduate students in unrelated fields (Lay Group-LG) to determine possible differences in used to comprehend the physical environment. categories Subjects were required to generate and apply their own categories (personal constructs) to a set of environments selected by them..Data was analyzed on a subject-by subject basis usirig techniques adapted from multivariate informational analysis and factor analysis. It was determined that: (1) subjects in EDG generated either significantly fewer or significantly more constructs than subjects in the LG (p less than .002); (2) the personal construct systems of subjects in the EDG contain more significantly correlated (p less than .05) pairs of constructs than those in the LG (p less than .05). (3) subjects in the EDG had a percentage variance accounted for by their primary factors greater than subjects in the LG (p less than .2); (4) no significant difference exists between the two groups as to number of factors extracted or factor content; (5) subjects in the EDG employed significantly more responsive constructs which were highly loading on their factors than those in the LG especially for the primary factor; (6) differences in both the number and content of factors found exist between this study and previous studies. It is suggested that designers tend to oversimplify their environ- J -6mental images along subjective lines, whereas laymen tend to have a more balanced, objective image of the environment. 07Introduction The Problem Architects, planners, and designers shape the physical world in which we live. The physical world, in turn, helps to shape our behavior and moods because it molds the pattern a large extent. It is of everyday life to therefore essential for urban designers to. con- struct man-made environments which complement human behavior and emotion. Research to determine how people cognitively view their physical environment must play a vital role in helping designers create spaces which wants of the users of those spaces. (raik, true that design at present does not reflect human are attuned to the needs and 1968). If it is goals ( Lynch and Rodwin, 1958), and client, then it and that a gap exists between architect is useful to determine the nature of the difference between the professional's and the layman's environmental perception towards an understanding of the causes of this misfit, with a methodology which would reduce the problems of past attempts. The systems of categories that people use to comprehend environments determines to a large extent their feelings and behavior in those environments. These systems of categories organize the individual's image of the environment, i.e., the individual's "mental representation of the parts of reality known to him by any kind of experience. 1971). " external ( Harrison and Sarre, Studies of individuals are desirable because findings at the group level often cannot be applied to the individual level, whereas individual studies are validly generalizable to groups. (Robinson, 1950). This study examines the systems of categories environnental designers and laymen employ to characterize the environment. I -8Prior Studies Lynch (1960) showed that people in cities keep a mental image of their environment in their heads to order the physical characteristics of urban places. He obtained individual's notions of the visual form of cities through interviews, walking with his subjects through the city itself, and having his subjects draw maps of the city. The work was pioneering, but lacking rigor. The methodology has been used extensively throughout the world since 1960, often with less success and applicability to design criteria than when Lynch's skill is brought to bear upon the data produced. ( Wurman, 1971). More recent studies have tried to utilize methodologies wherein data is generated which can more easily be analyzed. Semantic differen- For this method subjects rate a standard set of of scales supplied by the investienvironments using a standard list gators. Data for subjects are then pooled for the puropse of discovering tials are often used. the superordinate dimensions of meaning or factors underlying environmental comprehension. Hershberger (1969) pre-architects, showed a group of architects, and laymen a series of slides of buildings. The subjects were asked to rate the pictured structures on a semantic differential list, from which he drew conclusions concerning differences between the laymen and the others in their perception of the environment. A number of problems are concomitant with the use of slides and semantic differentials. This method relies on unfamiliar pictured environments rather than real ones; only the sense of vision is involved. Subjects have no knowledg~e of the pictured building- in terms of levels and types of -9activity, emotional feelings toward the place, history of the place, etc. In addition, it is difficult to represent much more than a single building through the medium of slides; images of large-scale urban environments therefore remain inaccessible. Howard( 1972) has found that responses to represented environments depend a great deal upon the exact nature of the representation. His research showed, for example, marked differences in responses to slides of buildings taken on sunny as opposed to cloudy days. Although semantic differential lists are developed by pretesting, and are refined considerab3y, they still of ways of viewing reflect only a limited number the environment. Subjects cannot use their own words ifr describing how they construe the environment, nor can they omit any item on the list they might feel to be irrelevant to architectural meaning. Johnson (1968) eliminated some of the inherent difficulties when he applied the semantic differential methodology to students' perceptions of residence halls. In-that case, subjects had emotionally experienced a physical place and had been familiar with it, as to merely seeing pictures. Hershberger's opposed more recent work (1972) has also utilized real environments familiar to subjects. However, that very familiarity, while eliminating the problem of media representation, accentuates the shortcomings of the semantic differential; subjects could not communicate their personal organization of the physical world in the way they might have liked , since such descriptions were given to them. Any new and potentially useful descriptions cannot be found using a prepared listi only through trial and error could they be -10discovered, The methodology also leaves open the question of how each individual organizes his own image of the environment, data is since all subjects' pooled. Personal Construct Theory George Kelly's (1955) work on personal construct theory offers a new approach when applied to physical environments instead of its usual context of person perception. He postulated that behavior follows from knowledge, and that one's knowledge all lies within a conceptual framework which is built up over time by categorizing information. Any new knowledge or event is placed within the existing pattern of that framework, if possible. Individuals accomplish this placing through the use of characterizations of the environment, which Kelly refers to as "constructs". Specifically, By construing, he meant "placing an interpretation upon." " In construing, the person notes features in a series of elements which characterize some of the elements and are particularly urncharacteristic of others." The constructs, therefore, are bipolar, as g6d-bad friendly-unfriendly. such The Repertory Grid Test developed by Kelly makes the network of an individual's constructs visible and mathematically analyzable. The technique lends itself to factor analysis, which uncovers dimensions of meaning, major abstractions along which lines individuals construe their environment. The suitability of Kelly' s theories for environmental image study has been discovered independently by Harrison. and Sarre (1971). Their discussion of the theoretical aspects of personal construct theory, along with Honikman' s (1972), necessity for a more exhaustive discussion here. obviates the The Research: Summar A sample of ten graduate students in architecture or environmental design was obtained, along with ten graduate students in fields not related to design. The subjects were all presented with twenty of environments; they selected suggestions twenty places meeting the suggestions which were familiar to them. Subjects were then required to generate and apply their own categories ( personal constructs) to the set of twenty. Subjects gave a 114.11 to constructs which were true of an environment and a "-" to constructs which were not true of an environment. Past studies utilizing the Repertory Grid Test have found variations correlated with IQ and socio-economic status . A crude IQ test and a set of questions indicating socio-economic status were administered to the subjects to control for this variable. It was found that there were no differences between the environmental design group (EDG) lay group (LG) and the great enough to raise doubts about the validity of the results. The test yields a large amount of data for each subject which lends itself to a variety of analytic techniques. Each subject was instructed to generate characterizations until he could not think of significant additional ones. Therefore, the number of characterizations (constructs) for each subject was tabulated. It was found that subjects in the EDG generated either significantly more or significantly fewer constructs than those in the LG. Information theory has been applied to the study of cognitive systems (Attneave, 1959) to characterize two important aspects of those systems. F -12The variable H is a measure of distribution; in this s -udy, it measures essentially the degree to which subjects divide the rated environments equally into +'s and -'s according to the constructs with which they are rating the environments. H can be conceptrialized as measuring the degree of discrimination of an individual's environmental cognitive system, the extent to which an individual uses different categories to discriminate among differnet environments. No significant differences were found between subjects in the EDG or LG in this measure of discriminatory power. The variable T is a measure of organization!; in this study it measures the correlation between every If a subject used the constructs construct pair for each subject. t modern"t and Iflike", for example, and he likes modern places, T would be close to +1 -1, a perfect negative correlation, and +1, ( T varies between a perfect positive correlation.) T can be thought of as a measure of the extent to which an individualIs environmental It constructs are systematically associated with each other. was found that subjects in the EDG had more significantly correlated pairs of constructs than those in the LG. Factor analysis was then performed upon the data to determine whole groupings of constructs which were inter-related. Such groupings (factors) correspond to major abstractions along which lines individuals organize their environmental images, i.e., dimensions of meaning. The number of factors yielded for each subject ranged from three to seven. No significant difference was found in the number of factors per subject between the designers and laymen, nor in the types of factors in k -013the two groups. Designers did, however, have a significantly higher average percentage variance for the first (most important) factor than subjects in the lay group. In other words, designers tended to have one dimension of environmental image organization in which more of their total environmental cognition fell. The dimensions of meaning portrayed by factors can be thought of as a mountain range, with the peaks representing the descrete factors. The height of the peaks corresponds to the importance of that particular factor. Continuing the analogy, the range of the subjects in the EDG would include one clearly dominant peak, with a series of smaller ones. The range of the subjects in the LG would look like a more uniform series of peaks. Hershberger (1972), posed a basic list working with semantic differentials, has pro- of twenty factors important to environmental meaning. The author found significant differences between Hershberger's list and the factors generated by the subjects in this study. At least one important factor not included in the prior study was determined to be of significance (,an urban-rural factor), and many factors suggested by Hersh- berger were not culled at all from subjects in either the EDG or the LG. Methodology The subjects in this study consisted of ten graduate students in the disciplines of environmental design or architecture and ten graduate students in fields unrelated to design. The selection of graduate students rather than undergraduates was thought to be necessary because of the more specialized pre-professional training given at the graduate level, especially for design students. Some of the subjects in the EDG were already practiced in the profession of architecture'. The sample size of twenty was somewhat small, but the author believes that the Repertory Grid technique yields a rich enough field of data for each individual to render his findings significant. The Repertory Grid Technique Two basic problems common to most attempts at measuring environmental images are the selection of the environments to be described by the subjects and the generation of descriptions of those environments. It is the author's contention that merely attempting to characterize an environment by its physical attributes must inevitably omit much of that environment's profoundest meaning. The kinds of activities occurring there, the clientele, the emotional feelings, the sense of character, the flavor and depth of history, and the social graciousness of a place cannot be captured without actual familiarity with that place. It was therefore deemed essential that subjects in this study be familiar with the environments they would characterize. Some measure of control over those environments was also considered necessary to facilitate comparisons of diverse, suggested between subjects. It was dbcided to utilize a list types of environments (Table 1) from which the subjects would then select specific places with which they were familiar. Table 1 General Environments Used in This Study 1. A House 2. An airport 3. A park 4. A vacation place 5. A library 6. A city plaza 7. A restaurant 8. An apartment 9. A down-town 10. Harvard Square 11. A shopping center 12. A typical suburb 13. A hotel A European city 15. A subway 16. The worst place you can think of 17. BackBay, Boston 18. An office 19. A church or synagogue 20. A city street -16The author- believes Kelly' s original method of construct elicitation to be most effective, and duplicated his procedux - in this study. Subjects had in front of them twenty cards with the names of twenty specific places they had selected to match the author's suggested types of places. They were then instructed to select groups of three places, two alike in some important way and the third different in that same way. The characteristics ( and opposites) used to sort the environments correspond to Kelly's notion of bipolar constructs, and each sorting produces a verbal description in the subjects's own words of one way in which he organizes his environment. Subjects were told to continue sorting the environments until they could think of no new significant attribute with which to characterize the environments. The range of constructs thus elicited corresponded to the different properties which the individual subject uses to organize his image of the environment. The Repertory Grid Test in Detail Subjects were tested individually, in a room with large desk space on which they found twenty cards with the author's suggested environments printed. They were instructed to: 1. Go through the cards one by one and note on each card a specific place familiar to them which fitted the description on that card' If they came to a card describing a type of place they were unfamiliar with, subjects were told to picture their stereotype of that place and to write " stereotype" on that card. 2. Pick out a group of three cards, two of the cards in the group describing two places alike in some important way, the third card describing a place different from the other two in that same way. Inform the investigator ( the author) of the characteristic used. -173. Continue picking out groups of three in this manner until they could not think of a single new characteristic with which to describe the places. Subjects were told that the descriptions could des. cribe the environments physically, by the kinds of activities that occur there, or by the emotional feelings the place gives to them. Cards could be used more than once in forming the groups of three, but the descriptions used had to be different for each sort. 4. Rate each environment in the light of each characteristic they had mentioned. The author had been writing down the constructs generated by the subjects on separate rating sheets ( Appendix A). The sheets were then given the subjects, each sheet containing one of his constructs and its opposite in parentheses ( to help define the construct). A column of boxes labelled from one through twenty corresponded with the twenty cards with environments, also numbered. Subjects were asked to take each characteristic and apply it to each environment. If a characteristic applied to an environment and was true of that place, subjects were told to put a plus (+) in that appropriate box. If a characteristic applied to an environment but was not relevant to the environment', and the subject would not use that description in thinking about the place, they were asked to place a zero (0) in that box. After all the environents had been rated for each construct, subjects were asked to go back to those boxes in which they had placed a zero and place either a plus or minus depending upon what they would have indicated about the environment with regard to that specific characteristic if they had been forced to choose initially between only a plus or a minus. The author added three constructs to those mentioned by all the -subjects and asked them to rate the environments in the light of those three as if they had mentioned the construct themselves. The additional -18-. three constructs were added incorder to obtain some measure of communality between subjects and to elicit responses to the new constructs themselves, since they represented three basic dimensions of meaning. The constructs added were " I like this place, to obtain affective response; I A place I would judge to be good," to obtain an evaluative response; and " An easy - to- utilize place," to obtain a response based on the dimension of function. The author then administered the tests of IQ and socio-economic status (SES) to the subjects. The Repertory Grid Test required an average of approximately one and one-half hours per subject toccomplete, including about ten to fifteen minutes for the IQ and SES tests. Time consumed by the test was most directly dependent upon constructs generated the number of by the subjects. The subject with the least number of constructs generated completed the test in fifty minutes, while the subject generating the most constructs took two and one half hours. Methods of Analysis The methods of analysis performed upon the data extracted from subjects are summarized in Table 2. If the grid obtained from subjects' ratings of environments is thought of as reflecting the individuals' environmental cognitive system, then it is necessary to obtain a measure of the extent to which constructs used in characterizing environments are distributed. To the degree that the constructs are not distributed equally (with the -19maximum possible distribution) among the environs characterized, system's discriminatory power is the limited. As Bannister and Mair note: i It can be argued that a construct ... is(like a ChiSquare) maximally useful the nearer it approximates to dividing up the world half-and-half, and presumably becomes totally useless when all elements occupy one pole. The construct then discriminates nothing from anything else. If all men are bad, point of thinking of men then there is little as good or bad, since no differential predictions are generated and no differential implications are available to explore." It is possible to measure the amount of discrimination and distri- bution by means of the information theory statistic H. When computed from data derived from the grid, H will vary between .00 and 1.00 and be a direct function of the degree to which a given construct divides the environments to which it is relevant equally into + 's and -'s. Tables 3a, 3b and 3c are examples of three correlation tables which might appear in an individual' s environmental cognitive system grid, in this case comparing the constructs "good" and "modern". The total in the plus marginals indicate the number of environments characterized either as "good" or "modern", of environments characterized those in the as either - marginals, the number "not good" or "not modern". As the distributions of the constructs go from a maximum to a minimum distribution of zero and sixteen the values of H drop from 1.00 to .00 (Leff, 1969). The amount of organization in an individual's environmental cognitive system can be thought of as the degree of relation between the system elements ( constructs). The information theory statistic T, I -20defined as the amount of information shared by two variables (Attneave, 1959), is one measure of the correlation between pairs of constructs. T varies from a minimum of .00 to a maximum of 1.00. In Table 3a, the linkages of "good" and "modern are maximally distributed, and T takes on its minimum value of .00. Table 3b depicts a situation wherein the joint occurences of "good" and "modern" are significantly more restricted, and T is equal to .20. However, in the case where either of the H values of a pair of constructs is equal to .00 (Table 3c), T will always be equal to zero, because if cannot be reliably related something is not discriminable, it to anything else.( Leff, 1969). Because the degree to which a system may be organized is constrained by its degree of discrimination, a T for a given pair of H's can never exceed the smallest H of the pair. Table 2. Summary of Research Design Samples Control Variables Independent Variables 1. Environmental Designers IQ # Constructs 2. Non-environmental designers SES H (discriminatory power) T (degree of organization) # Significant T # Construct pairs (amount of organization) # Factors (complexity of organization) Factor Content -21Table 3 Tables Depicting Three Different Distributions of Construct Pairs and their H and T values 3a. (Leff, 1969). modern * Hgood - good 8 H modern 8 3b. 8 16 Tgood-modern = 1.00 1.00 = 0.00 modern +0 good 2 10 10 3c. 4 6 2 14 good H ~~ 0.54 Hmodern 0.96 T good-modern 0.20 16 modern * Hgood 0.00 good Hdern Tgood-modern 0.00 = 0.00 -22In analyzing the grid data from this study, then, the assumption is made that the higher the number of significant T's (i.e., the more closely an individual's environmental constructs are the more organized is knit to each other), that individual's image of the environment. One It potentially troublesome danger exists in making that assumption. is conceivable that a subject's constructs could be so simplistic that many of them actually represent the same or closely related lines of thought. The average T for such a subject would then be inflated, not because of meaningful because relationships between different constructs, but the constructs merely overlap in meaning. In this instance, a high average T, far from indicating a greater amount of organization in an individual's complex construction of the environment, would indicate simply the expected high correlations between constructs which were similar in meaning. Fortunately, as will be discussed later, an examination of the major dimensions of meaning extracted by factor analysis helps in the differentiation between organization and overlap. Because the Repertory Grid Test in this study was open-ended, the number of constructs generated by each subject is significant. Pre- sumably, the grater the number of constructs generated by an individual, the greater the number of ways which that individual uses to order his environmental image. The number of ways an individual perceives the environment is therefore, in turn, a measure of his sensitivity to the environment and the complexity of his environmental thinking. For this reason, the number of constructs for each subject was examined in relation to the number for the other subjects. -23The heart of the information extracted from the data was revealed by performing factor analysis upon the data content. This is whereby salient clusters of similar constructs are di.scerned. analysis is a method Factor commonly used to reveal essential dimensions of meaning contained within a large amount of data. In the context of this study, these dimensions of meaning (factors) correspond to major abstractions along which lines individuals organize and imagine their everyday physical surroundings. The number of factors present in an individual's environmental cognitive system can be taken as an indicator of the complexity of that individual's environmental organization. If, for example, one individual's data analysis revealed him to have three factors of environmental cognition and another individual's analysis showed him having seven, it can be inferred that the latter's environmental cognitive system, because of its greater number of superordinate organizational dimensions of meaning, is more complex and varried. The number of factors per subject and the differences between the number of factors per subject between the EDG and the.LG, was therefore examined. An examination of the number of factors for individual subjects was also useful in determining whether any high average T value reflected amount of organization or merely overlap in meaning. If a subject relied upon certain narrow, simplistic, or overlapping constructs, thus producing an inflated T value, the number of factors extracted from his constructs would be less than expected if the high T had actually -24depicted a high amount of organization. if To take a hypothetical situation, a subject had achieved a high T because his constructs were organized around the basic dimensions of aesthetic appeal, social appeal and familiarity, factor analysis would reveal these three basic dimensions and no others which would be significant. Another subject might have had a lower T value, but have had seven factors. It could be inferred that the latter subject's T was truly representative of environmental cognitive complexity as corroborated by the larger number of factors, while the first subject's high T was " exposed " as being inflated because it reflected only the expected high amount of correlation between constructs organized around similar lines. Additional analysis of the factors extracted was carried out to determine if: (1) the facotrs are similar to the ones found when semantic differentials are used; (2) there are significant differences in the types of factors used by subjects in the EDG as opposed to the LG; (3) there are certain factors which are dominant in the environmental cognition of either the EDG or the LG; (4)there are significant differences in the number of responsive (subjective) vs. representational (objective) factors found between the EDG and the LG. The two variables measured in this study were intelligence and socic-economic status (SES). (IQ) IQ was measured using the Shipley Hartford tests of verbal and abstract intelligence (Shipley, 1940). SES was measured using the Hollingshead (1965) two factor index of socio-economic status. Because the size of the sample used in this study was small and the distributions of the dependent variables was not known, non-. parametric statistics were used to compare groups. U Test ( Siegel, 1956) was utilized The Mann-vhitney to determine significance levels of the data. Because of the distributional uncertainty, all tests were two-tailed, providing a more rigorous measure of significance. The H and T statistics were computed using a special program for the computation of multivariate informational analysis (IaA) statistics from grid data ( Leff and Rees,1972). Factor analysis was performed using the Datatext program on the IBM 360 system, (Armor, factors 1972). This program yields all significant (to X root = 1 in this study) and computes the percentage of total variance accounted for by each factor. rotated, a procedure The factors were then which essentially strengthens the loadings of the highly correlated variables and diminishes the loadings of the lower correlated variables to present a clearer picture of the content of each subject's factors. --26- Observations Table 4 summarizes the observations for the simpler correlations and other methods of analysis. Subjects were assigned numbers from one to twenty based on the chronological order of testing .Table 4 separates subjects in the Environmental Design Group ( top group of ten) from those in the Lay Group (bottom) in order to facilitate comparison of the two groups with regard to the variables measured. The second column, labelled # Constructs," gives the number of constructs generated by each subject, tanging from a low of seven to a high of 32. Subjects in the EDG generated an average of two more constructs each than those in the LG (15.4 constructs/subject vs. 13.4 constructs/subject). An examination of the ranking of subjects according to number of constructs elicited yielded a more significant result. It was found that subjects in the EDG generated either more or fewer constructs than those in the 1G. In other words, the students in fields unrelated to design tended to have a dramatic clusteringtoward the center, generating, with only one exception, eleven to seventeen constructs. Environmental designers clustered around the extremes in this regard, with one group of five having from seven to eleven constructs, the other half having from eighteen to 32. The probability of this distribution occurring by chance is less than .002. -27No significant difference was found between the two groups for their average H values (column three). The next column gives the average T value for all of the construct pairs for each subject. Column five gives one of the most useful statistics, the number of significant T values for each subject divided by the number of construct pairs per subject. The computer printout of T values also tabulated the number of significant T correlations for each subject. - This is of little information, however, meaning unless compared with the total amount of construct pairs correlated for each subject. Therefore, column five represents the nearest measure of amount of relationship between individual's constructs. The subjects in the EDG were found to have significantly more highly correlated pairs of constructs in their environmental cognitive systems than in those of the LG. The probability of the distribution found occurring by chance is less than .05. 'was deemed significant if the probability of its ( A T value occurrence by chance was less than .05.) The average value was taken among all significant T's; the difference between the two groups that was found was samll enough to have occurred by chance (column six). The number of significant factors extracted for each subject is shown in the final column of Table h. No significant difference was found for subjects in the EDG and the ID; average number of -28. Table 4 1 2 3 4 Subject # Con- Number structs Ave. H Ave. T 1 32 .897 .087 5 11 .922 6 10 7". 5 7 Sig. T/ construct 6 Wve. value sig. T. 7 No. of Factors .256 .436 7 .053 .132 .383 4 .891 .131 .436 .487 3 9 .798 .042 .061 .495 4 8 7 .929 .063 .222 .323 h 9 18 .900 .052 .133 .406 7 10 9 .911 .092 .333 .420 4 11 18 .897 .055 .138 .425 6 12 22 .775 .044 .087 .359 7 17 18 .865 .109 .381 .450 5 2 16 .858 .041 .058 .399 7 3 17 .899 .051 .126 .373 6 4 13 .828 .027 .008 .276 6 13 12 .892 .058 .124 .429 4 14 14 .875 .060 .162 .376 5 15 12 .962 .035 .057 .486 5 16 17 .91 .054 .147 .350 6 18 11 .783 .050 .055 .465 5 19 13 .905 .52 .092 .414 5 20 9 .917 .103 .333 .428 3 ____To EDG -29factors for subjects in the EDG was Table 5 5.1, for subjects in the LG, 5.2. depicts the percentage of to.tal variance for each factor extracted for all subjects.This percentage is a measure of the number of different constructs -related to a specific factor; if, for example, the percentage is 33.31S, of to.tal variance accounted for by the first factor then one-third of that individual's total environmental cognition can be said to be organized along that one dimension of meaning. It was found that subjects in the EDG have a higher average percentage variance accounted for by their first factors than those in the 1L. Although this finding only approaches significance (the probability of the distribution occurring by chance is less than .20), Figure 1, a graph of the information in Table 5, clearly shows a substantially larger variance for the first factors in the EDG group, while the remaining factors of the EDG group account for less variance than those in the LG because the latter's are more evenly distributed. As discussed in the introduction, Hershberger (1972) has formulated a list of twenty scales with which to measure environmental meaning, drawn from various investigators' differentials. (Appendix B). research with semantic In labelling the factors extracted from subjects' data in this study, the author made a conscious attempt to apply Hershberger's discriptors according to the factor structures presented by him.It was felt that this method would provide the fairest test of the viability of Hershberger's twenty factors as to ~30Table 5 Total Percentage Variance Accounted for by Factors Factors (figures expressed in percentages) 6 2 3 20.3 7.3 5.9 5 41.6 35.7 17.0 12.5 11.1 6 57.8 11.2 7 27.5 13.4 25.7 10.1 9.5 8 . 47.4 18.6 11.9 10.5 9 33.0 12.4 7.5 10 47.0 15.9 18.3 11.1 11 35.5 12.9 11.4 9.4 7.7 12 27.4 20.2 17 51.4 12,5 9.7 9.7 7.0 EDG mean hD.4 17.5 10.7 8.4 Subject Number 1 1 8.7 4.8 3.9 3.6 6.3 5.7 4.9 7.0 6.1 5.6 5.8 4.9 5.6 LG 2 27.6 16.4 12.5 10,8 3- 29.8 21.9 10.2 8.7 4' 23.1 17.6 11.2 10.1 13 14 33.9 21.8 12.5 7.4 36.2 26.5 20.9 17.7 11.0 12.8 7.2 11.1 34.2 16.9 10.9 8.3 9.4 7.4 18 32.7 18.1 12.3 9.8 7.7 19 16.8 12.5 9.8 8.7 20 32.9 51.8 11.9 9.7 LG men n 32.9 18.0 -1.6 9.1 7.4 15 16 7.0 7.5 7.5 6.5 6.3 6.7 6.2 5.8 -31-. their actual employment by subjects in this study. The complete list of factors and highly correlated constructs per subject comprises Appendix C. Appendix D represents the factor names for each factor by subject. Table 6 summarizes the findings. Factors are listed in decreasing order of use by subjects, with the columns indicating the number of subjects in the EDG and the LG using the specific factors in their environmental image organization. Factors included in Hersh- berger's list are marked with an asterisk (*). Four dimensions of meaning emerged as significant, being employed by one-half or more of the subjects in this study. factor, representing such constructs as The "friendly" warm and familiar, hospitable, personal, and human, was found to occur among all subjects save two. An aesthetic factor, representing physical beauty or liking, and a utility factor, representing the ease with which an environment may be used, were also found to be important, with twelve subjects utilizing each. The aesthetic and friendliness factor were also found by Hershber- ger to be the two most important scales, but he placed the utility scale last in importance. The significant presence of that scale is probably inflated in this study because of the inclusion of the author's -"easy to utilize" construct; nevertheless, the simple inclusion of that one scale did not imply that twelve subjects would organize one of their basic dimensions of meaning around it, its so that importance cannot be dismissed. I -32Figure 1 Total Variance Accounted for by Factors iLd cz LLJ 0 G m f %VARIANCE ACCOUNTED FOR -33Table 6 EDG and LG Factor Labels Factor # EDG Using # LG Using Total # dubjects using Friendly * 8 10 Aesthetic* 7 5 18 12 Utility * 6 12 Urban-rural 6 6 Space * 2 6 Time* 3 S Organization * Noise * 2 2 0 Residential 0 10 Freedom 0 Public Acess 1 4 3 3 3 3 3 3 Ornate * 2 2 2 2 0 2 2 2 Privacy * Activity level Size * Individuality Consumeristic 2 Good Neatness * 0 Lighting * 1 0 Activity type Natural 0 1 1 10 Sensual 0 0 Familiarity 1 0 Working-class Young Work 0 0 * This factor included in Hershberger's list (1972), Half of the subjects had an urban-rural factor. Hershberger does not include this descriptor in his list, but the organization of constructs such as urban, in the country, greenery, and natural would seem to qualify it as an important dimension. While the factors of space, ornateness, neatness, size, lighting privacy, noise, and time also appeared in this study's subjects, none were present to the degree of importance suggested by Hershberger. In addition, his important potency dimension, along with coloring, temperature, shape, ventilation, rigidity, formality, and texture, are not represented in Table 6. No significant differences were found in the content of factors employed by the EDG as opposed to the LG. Both groups utilized the major dimensions about equally in their environmental organization. A noteworthy difference was found between the environmental designers and the laymen with regard to employment of representational versus responsive constructs (Table 7) and factors (Table 8). A representational construct or factor is more objective, such as noise, lighting, urban, sprawling, etc. Examples of responsive (subjective) factors or constructs include liking, well-designed unfriendly, etc. i.e., one that implies a value judgement. Among all highly correlated constructs for factors, subjects in the EDG had a significantly higher percentage of responsive ones than subjects in the LG. In fact, slightly over half of such constructs were responsive for the EDG, Table 7 Responsive vs. Representational Constructs 13ubj ec t Number (Figures express # of highly loading Rep. constructs/# of highly loading Res. constructs) Factor 1 2 3 4 5 6 7 Totals 1 0/15 3/6 1/3 0/2 0/1 1/1 1/1 6/29 5 3/3 0/8 1/6 2/0 3/0 1/1 1/0 1/2 8 2/0 0/3 3/0 1/1 3/0 1/0 1/1 3/0 0/2 1/2 9 3/3 3/0 2/0 3/1 0/1 10 2/2 2/1 0/3 0/1 11 12 1/7 1/1 3/0 1/3 1/1 1/1 0/6 2/3 2/1 2/1 2/0 1/1 17 6/3 2/0 3/0 3/0 6/0 Totals 17/50 20/18 18/8 15/11 10/7 h/h 5/2 91/99 2 2/2 0/h 2/1 3/0 1/0 0/1 1/0 9/8 3 h/2 3/1 4/0 1/0 1/3 1/0 h 2/1 1/1 3/0 1/0 1/1 1/0 13 h/0 3/2 2/0 0/2 14 1/4 2/1 1/1 1/0 2/0 1/2 8/8 15 3/0 2/2 1/1 2/0 2/0 0/1, 10/4 16 0/4 h/0 2/3 1/1 2/1 0/3 9/12 18 2/1 2/0 0/2 3/0 1/1 8/4 19 2/2 2/1 2/0 1/2 h/0 11/5 20 1/5 2/1 1/2 Totals 21/21 21/13 18/10 14/6 h/8 94/63 EDG 6 7 8/9 4/11 9/3 6/6 1/1 1/0 15/6 h/7 8/13 3/1 13/12 20/3 LG 13/5 3/7 14/6 3/1 h/1 12/4 9/h -36Table 8 Responsive vs. Representational Factors Subject Number Factor Total 5 6 rep rep Rep/Res F. 1 2 3 1 res res res res rep rep res, rep 6 res rep rep res 7 8 rep rep rep res res 3/2 res res rep rep res 2/3 9 res rep rep rep res 10 res res 11 res rep rep rep res, rep res rep 12 res rep res rep rep res 17 res rep rep rep rep Total Rep/Res. F. 1/9 7/3 8/2 6/h 3/4 1/2 1/1 29/25 rep rep rep rep 5/2 rep rep res rep rep rep res rep rep rep res res rep res 13: rep res rep res 14 res, rep rep rep rep res h/2 15 rep res rep rep rep res 16 res rep rep res res res h/2 2/4 18 19 res rep res rep res 2/3 res rep rep res rep 20 res rep res 3/2 1/2 Total Rep/Res F. 4/6 6/4 8/2 6/3 5/3 EDG rep 2/2 res res rep 3/4 1/3 h/2 rep h/3 h/l I0 3 h/3 2/2 51 rep 5/2 2/2 2/h 2/0 33/22 -37while only 63 out of 157 were responsive for is most strikingly manifested in the first the LG. The difference (most important ) factors; here, 75% of the highly loading constructs for the EDG were responsive, compared to 50% for the LG. In addition, nine out of ten of the EDG's primary factors were responsive, versus six out of ten for the LG. -38Conclusions Theoretical The author believes Personal Construct Theory to be an extremely promising method in the measurement of environmental images and organization. The method allows subjects to give, in their own words, the elements of environmental thought which they consider to be most important to them. This is not to imply that construct theory is without its pitfalls. The use of information theory in the ananysis of social science data has been criticized as measuring merely syntax rather than actual meaning ( Harrison and Sarre, 1971). Care must be excercised in the selection of techniques of analysis of the grid and the meaning attached to them. The grid test yields an unusually large amount of data for each individual subject; the organization of this data and the theories underlying that organization must be examined carefully to determine just what is being measured. This study presents only a few of the statistical measures which can be utilized with the grid technique. The very complexity of the information yielded through construct theory, though,,is an encouraging sign. The personal image of the environment is also very complex; the essential assumption of this study is that constructs generated by subjects can be mapped onto their total cognitive representation of the environment and thus provide access to that representation for the investigator on an individual by individual basis. With further, more sophisticated -39applications of the grid technique, investigators should be able to discover, by comparing individual' s environmental images, the kinds of internal organization which leads to conflicting images, e.g. why one person will consider a room to be " threatening " while another considers it to be comfortable. Construct theory permits a visual display of linked constructs, so that, in this example, the significant constructs connected with the individual's construction of a threatening place can be studied. Many researchers have tried to specify a definitive number of dimensions of meaning (factors) representing the individual's maximum capacity to process information. Osgood (1957) has attempted to utilize three factors as sufficient; Miller (1956) believes the average to be seven, plus or minus two; Hershberger (1971) insists that twenty such scales are "almost just right". The subjects in this study were found to have from three to seven factors representing their image of the environment. It would seem, therefore, that some sort of balance between Osgood's three and Miller's seven is The author suggests that three is indicated. too low to adequately capture all vital dimensions, and that too much conslidation of potentially useful discrete factors is involved in trying to reduce the complexity of information organization to those three. At the opposite extreme, twenty factors seems much too large a number for any one individual to employ; as demonstrated in this study, subjects actually used only a fraction of Hershberger's scales. If presented with all twenty scales, subjects will of course rate environments on the basis of each -40scale, but the author suggests that many if not most o' them are coparitively insignificant in terms of the individual' s total cognition. Methodological The Repertory Grid Test is a time-consuming method compared to the semantic differential technique. Approximately one and one-half hours are required to test each subject (as opposed to perhaps ten or fifteen minutes for the marking of a semantic differential list), and subjects must be tested individually ( semantic differentials are written, and any number of subjects may rate the given environments at the same time). In addition, factor analysis for the grid data (and other statistical measures ) are computed on a subject-by-subject basis, rather than on a pooling of all data for all subjects. The author believes, however, that the advantages of the methodology more than compensate for its comparative difficulty in administering it. S'ubjects, rather than having to rely upon descriptors of the investigator's choosing, use their own words to convey significant attributes of -the environment. eliminating the common Their descriptions may be entire sentences, dissatisfaction with 6ne w word adjectives and the possible ambiguities associated with such terse characterizations. Subjects may also omit rating a specific environ- ment on the basis of a specific construct if construct is not they feel that the relevant to their characterization of the place. (In this study, constraints of analysis sophistication prohibited allowing subjects not to rate any environment.) The grid method provides a much finer grain of information, since analysis is done separately for each individual. Conclusions may then be more validly generalizable to the population, whereas the pooled semantic differential data cannot be used to provide conclusions on an individual basis. Practical The conclusions presented in this section must be tempered by the awareness that only twenty subjects were used in this study, a somewhat small number from which to generalize. In addition, some of the analysis is not open to clear-cut conclusions; the author offers5 suggested interpretations of these results. Table 4, each subject. column '2 depicts the number of constructs elicited from Subjects in the EDG generated either significantly more or significantly fewer constructs than those in the LG. This finding can possibly be interpreted as indicating that there are two types of environmental designers. about environments, utilizes major The first type, when thinking abstractions into which details are pre-fit. The constructs of the first type, therefore, be factors in themselves. would Essentially, these designers can describe more with less. This conslusion is somewhat substantiated by the high ratio second of factors for these subjects to number of constructs. type of designer, on the other hand, is detailed characterizations. The concerned more with Dimensions of meaning for this type of designer can only be uncovered through factor analysis. An interesting .42. study might compare the designs of the two groups in order to discover the possible ramifications of these two types of thought processes upon the output of architects. The lay subjects were clustered in the middle of the construct scale, indicating that individuals not trained in thinking about the physical environment have a fairly constant number of important elements in their environmental cognition space. Table 4, column 5 portrays the level of organization found in the construct systems of subjects. It expresses the ratio of highly correlated construct pairs to the total number of construct pairs. Architects and designers were found to have more significnat correlations between their constructs than laymen. If a personal construct is thought of as being a part of an individual's cognitive field, this finding means that more parts of the field are related to other parts; in other words, there is more organization in the field. Thus, while designers tend to group certain constructs, laymen tend toward viewing each construct as a more separate entity. This is not an unexpected finding, since architects are trained to organize their conception of physical space so that they can translate the laymen--client's descrete needs and desires into general design criteria. Possible danger lies in over-generalizing these separate desires so that some of them are unintentionally sacrificed, not because of economic or other practical reasons, but because of certain assumed inferences in the designer' s mind. For example, an architect might link order with utility, and assume that, given a sufficient amount of -43order in an environment, ease of utilization will follow. His client, however, might conceive of order and utility as two more distantly related characteristics, and find that the highly ordered environment produces more monotony than ease of utilization, and that simply clarity would have sufficed and allowed for a looser degree of order. The H values for the EDG and the LG are comparable. This indicates that the differences found between the two groups as to their organization ( T values) is significant, does not merely reflect since it a relative lack of discrimination among constructs. H can also be thought of as representing the degree of individuation (Leff, 1969), or the tendency to "lump" constructs together, What this implies in terms of this study is that the T correlations did in fact represent meaningful associations between construct pairs and was not a spurious result It of subjects'. lumping together constructs by coincidence. appears that architects and designers employ the same number of dimensions in their total environmental cognition as do laymen (Table 4, column 7). Moreover, as Table 6 shows, no significant difference exists between subjects in the EDG and the LG interms of factor content. Environmental designers, then utilize very similar dimensions as laymen in thinking about the environment, in spite of their training. What could be the source of the observed misfit, then, between man and the physical environment? Aside from the possibility of differences in cognitive organization discussed above, a closer inspection of the factors extracted from the data might provide a clue. It was found that subjects in the EDG have a higler 'percentage variance accounted for by their first factor than those in the LG. In other words, architects and designers have a greater tendency than laymen to organize their constructs systems along one dominant dimension. Figure 1 represents this dominance graphically; each peak represents a factor, the height of the peak being determined by Although the inter- the number of constructs linked to that factor. pretation of this finding is not clear-cut, the author suggests that it may, in part, reflect a tendency of designers to oversimplify their image of the environment. one Laymen, on the other hand, do not allow_ dimension to dominate as murch, and thus have a more balanced system of categories with which to comprehend environments than architects and designers. One more finding must be discussed in order to bring the meaning of all the data into focus. Tables 7 and 8 depict the number of representational (more objective) and responsive (more subjective) constructs and facotrs for both groups. The EDG had significantly more responsive highly loading constructs included in their factors than the LG. The ratio of representational to responsive factors did not reveal such a striking difference between the two groups. However, most noteworthy are the results for the first (primary) factor. For subjects in the EDG, only seventeen out of sixty-seven highly loading factors on their primary constructs were representational, whereas for the LG, half of such constructs were representational. In fully addition, only one out of ten primary .factors of the EDG were representational, as compared to four for the LG. This finding is especially significant in the light of the finding that the primary factor represents more of the designer's total thought about the evnironment (40.4% to 32.9%). In short, environmental designers in this study think about the environment more in terms of personal value judgements than the laymen, who tended to image the environment more objectively. What general conclusions can be drawn from this information? The author suggests the following hypothesis as worthy of further research; Either because of their training or because of certain common inherent traits, environmental designers organize their environmental cognitive systems differently than laymen. The difference does not lie in the extent or specific environmental dimensions of meaning employed, but in the extent of organization. Designers tend to oversimplify the range of attributes they consider important, and they do this along the lines of value judgements concerning environmental characteristics rather than along the lines of the characteristics themselves. Their thinking about the environment leads them to view whole groups of attributes in a subjective way.( e.g. good or bad, desirable or undesirable, etc.), and to assume certain relationships which may not hold true for specific clients. If architects are to design spaces more in line with the needs of their users, they should be made aware of the possibility that their subjective generalizations run counter to the laymen's more balanced, objective image of the environment. -46Toward this goal, it designers, is strongly recommended that architects, and planners employ the Repertory Grid Test for themselves and for their clients to determine specific differences in their respective systems of important environmental categories before projects even enter the planning stage. The author believes that, with sophisticated analysis, the two hours spent taking the Repertory Grid Test could prove more fruitful than weeks of give-and-take discussion. Ideally, some kind of experimental situation could be set up wherein architects using the test in their professional-client relationship could be compared with a control group of architects not using the test. A comparison of the inputs and final outputs and evaluations between the two groups might be an important step toward the construction of environments which better complement human behavior and emotion. Bibliography Datatext, Armor, D., forthcoming, 1972. Attneave, F., Applications of Information Theory to Psychology, New York, Holt Rinehart, and Winston, 1959. Bannister, D., and Mair, J., The Evaluation of Personal Constructs, London, Academic Press, 1968. Craik, K., "The Comprehension of the Everyday Physical Environment," Journal of the American Institute of Planners, 34:1, Jan., 1968. Harrison, J., and Darre, P.,"Personal Construct Theory in the Measurement of Environmental Images:Problems and Methods," Environment and Behavior, 3:4, Dec.1971. Hershberger, R. , "A study of Meaning in Architecture," in Sanoff, H., (ed.) EDRA 1, 1969. , "Toward a Set of Semantic Scales to Measure Meaning of Architectural Environments", in Mitchell, J., (ed.), Environmental Design: Research and Practice, University of California at Los.Angeles, 1972. Hollingshead, A., The Two Factor Index of Social Position, unpublished manuscript, New Haven, 1965. Honikman, B., "An Investigation of the Relationship Between the Construing of the Environment and its Physical Form", in Mitchell J., (ed.),Environmental Design:Research and PracticeUniversity of California at Los Angeles, 1972. Howard, R., Mlynarski, F., and Sauer, G.Jr.,"A Comparitive Analysis of Affective Responses to Real and Represented Environments," (ed.),Environmental Design: Research and in Mitchell, J., Practice, University of California at Los Angeles, 1972. Ittelson, W., Rivilin, L., and Proshansky, H., "The Use of Behavioral Maps in Environmental Psychology, in Environmental Psychology,- New York, Holt, Rinehart, and Winston, 1970. Johnson, " Student Perceptions of Residence Hall Environments", unpublished Doctoralthesis, Oklahoma State University, 1968. Kelly, G., The Psychology of Personal Constructs, New York, W.W. Norton and Co., 1955. Leff, H.S., "Context, Psychiatric Diagnosis and Language," application for research grant, Massachusetts Mental Health Center, 1969. , and Rees, J., MJA: A Program for MJA of Grid Data, Boston 1972. -48Lynch, K., The Image of the City, Cambridge, M. I. T. Press, 1970. Miller, G., "The Magical Number Seven plus or minus Two: Some Limits on our capacity for information precessing," Psychological Review, 63:180. Osgood, C., Suci, G., and Tannebaum, P., The Measurement of Meaning, Urbana, University of Illinois Press, 1957. Perin, C., With Man in Mind, Cambridge, M.I.T. Press, 1970. "Ecological Correlations and the Behavior of Individuals' Robinson, A., American Sociological Review, 15:251. "A Self-Administering Scale for M.easuring Intellectual Shipley, W., Impairment and Deterioration," The Journal of Psychology, 1940, 9: 371. Siegel, S., Nonparametric Statistics for the Behavioral New York, McGraw-IUll, 1956. Wurman, R., Making the City Observable, Cambridge, Sciences, M.I.T. Press, 1971. -49Appendix A- Sample Rating Sheet Characterization #: 17 A clean bright lace (vs. a dirty, dim place) ENVIRONXENT NO, RATING I + 2 4+ 5 + 6 + 8+ 9 10 it + 12 + 13 14 + 15 . 16+ + / 18 + 19 + 20 - -50Appendix B Hershberger's ( 1972 ) Descriptors of Architectural Meanin and Representative Scales 1.AESTHETIC: Unique-Common 2.FRIENDLINESS: Friendly-Hostile 3. ORGANIZATION: Ordered-Chaotic 4.POTENCY: Rugged-Delicate 5.SPACE: Loose-Compact 6.ORNATE: Ornate-Plain 7. COLORING: Colorful-Subdued 8.NEAT1ESS: Clean-Dirty 9. SIZE: Large-Small 10. TEMPERATURE: Warm-Cool ll.LIGHTING: Light-Dark 12.PRIVACY: Private-Public 13.SHAPE: Angled-Curved 14.VENTILATION: Drafty-Stuffy 15.NOISE: Quiet-Noisy 16.RIGIDITY: Rigid-Flexible 17.PORMAL: Formal-Casual 18.TEXTURE: Rough-Smooth 19. TIME: Old-New 20.UTILITY: Useful-Useless -51Appendix C Highly Loading Constructs For Factors (By Subject) Subject #1 Factor 2 Factor 1 -949 warm and familiar 929 like to take people here - 929 pleasing color 949 feel solitary here 902 like to look up here (at arch) 912 my friends frequent here 894 only go here when necessary 867 think of here as a place rather than a person -748 don't recognize people here 886 always learn something here 564 place with my own personal 801 like to go heregood feelings 544 pleasant noises 781 interesting things to look at - - 779 cold, unemotional space. - 520 interesting things to look at 517 like to talk to people here visually and sensually 778 photogenic 773 interested in learning history and planning of this place -632 resent the way this place is 602 pleasant noises 564 like to walk here 556 enjoy eating here. Factor 3 827 like to live here 818 place represented by physical objects in my home 694 like to talk to people here 559 enjoy eating here Factor 4 891 think of possible improvements in this place when I'm here 516 resent the way this place is Factor 5 Factor 6 853 can sing without feeling self-conscious here 807 think of my family here -555 interested in learning history of this place Factor 7 - 880 banks prominent here physically -529 well-designed walking space Subject #2 Factor 1 Factor 2 901 people"put on a show" here to impress other people 932 place with "class"- cool -727 architecturally interesting 844 like place 820 excitement here 775 good place 709 shopping area -588 planned area Factor 3 Factor 4 761 office business done here 781 tall, massive 757 serious, intellectual 729 institutional 690 one only goes here sometimes -743 easy to utilize Factor 5 Factor 6 648 transport place 648 easy to utilize -53Factor 7 634 youth place- politics and style Subject #3 Factor 1 933 noisy, .worldly activity 893 busy place - 840 clean air Factor 2 49 young people here -837 sterile, non-cultural 764 like this place 511 product here new- 794 don't trust people here 690 get excercise here 679 feel "cooped-up" here Factor 3 852 work done here Factor 4 -855 old atmosphere 814 foreign-type place 571 "big-city" atoms., problems 545 culturally rich Factor 5 779 good place 680 can learn and think here 652 feel at ease here 606 easy to utilize Factor 6 -912 height, above ground -54Subject #4 Factor 1 Factor 2 854 good place 815 might find tourists here 834 frequented by higher 536 easy to utilize economic status -730 place "seen better days" Factor 3 Factor 4 -911 place created by public financing 925 standard conservative architecture -654 local landmark 555 quiet Factor 6 902 easy place to get lost in -577 place didn't look 885 exchange of $ integral to places function anything like it does now twenty years ago Factor 7 -809 little greenery -634 auto traffic 555 quiet Subject #5 Factor 1 Factor 2 -927 designed A Priori unalterable place 897 easy to utilize 827 place took shape over long time -733 work place 561 like -777 myth of happiness via consumption here 706 good urban env. 694 like 694 good 561 good P -55533 personal env. -547 auto-shaped place Factor 3 Factor 4 -874 public transport place 906 outdoor, non-urban -792 people share activities here- collectivism tolerated 671 auto-shaped Rotated Subject 46 Factor 1 Factor 2 885 functionally acceptable 922 people live here -885 dislike place 815 24-hour place -885 degeneration of human expression -879 little consideration for people's needs and wants 859 pedestrian comfort -837 unpleasant in off-hours 805 architecturally good 626 old, personal, human Factor 3 Factor 4 893 place in motion (vs. static) 665 easy to utilize 751 old, falling apart 586 easy to utilize 583 old, personal, human -56Subject j Factor 1 Factor 2 -864 go here for function only 905 open (vs. intimate) -841 modern 816 transport. place Factor 3 -877 nonresidential 646 part of my past 901 busy Factor 4 844 easy to utilize -509 part of my past 596 trees and gardens Factor 5 939 good place 895 like Subject #8 Factor 1 Factor 2 835 easy to utilize 925 high ability to get involved here 764 good 692 like Factor 3 -924." empty spaces" 845 high depth of history and culture Factor 4 -936 open space -738 quiet, contemplative -661 park setting Factor 5 907 place I want to get out of -542 park setting -521 good F. -57Subject #9 Factor 1 922 age and history here Factor 2 -889 no surplus space- tightly designed 888 unique character -857 modern, sterile, uniform -796 place is only a means of getting somewhere 828 aware here of what's going on -820 planned env. else 758 well-designed space 525 smells important here Factor 3 Factor 4 943 quiet 842 inhabitants can change and control env. -854 high level of activity 665 quiet 545 freedom 540 smells important Factor -944 5 confusing place Factor 6 -931 repetitive clientele -715 imtimate space Factor 7 776 space which can be used entirely Subject #10 Factor 1 -871 transient clientele 858 institutional-social beyond original function Factor 2 862 business area 832 good urban use mix 752 unfriendly urban -858 too formal and planned 804 relaxing, pleasurable *1' -58Factor 3 Factor 4 925 sense of place, human scale 962 easy to uti-ize 857 good 849 like Subject #11 Factor 1 Factor 2 964 feeling of wanting something (nothing) 812 place of work 566 chaotic, hysterical 937 good times here -937 emptiness here 849 like -824 hollow 698 feeling of being part of a larger group -590 grey,depressing 557 rich in detail Factor 3 Factor 4 865 new, shiny surface 913 noisy -732 history and memory -698 timeless, unchanging 732 no sense of community or belonging -651 easy to utilize 518 chaotic, Factor 5 879 I'm my usual self here hysterical Factor 6 -754 man-made 544 like -502 rich in detail 9 -59Rotated Subject #12 Factor 1 Factor 2 970 feel good here 845 restful 970 like 817 I think I can have an effect on this place 774 feel part of community here -798 noisy 692 privacy 665 an oasis -557 transportation 749 good -641 boring 545 place I'd like to see stay the same Factor 3 Factor 4 -803 tall place 861 housing place 692 place I use very often 783 place to make love in -744 socially mixed 574 easy to utilize Factor 5 -892 retail facilities 521 transportation Factor 6 -896 physically differentiated -602 interesting arch. Factor 7 868 near mountains 684 country-greenery -649 cosmopolitan 570 an oasis Subject #13 Factor 1 Factor 2 921 routes used every day 822 old --921 place I stayed in a in a short time -756 crowded r -60774 worked here -762 only go here for special occasions Factor 3 946 tidy, conservative 893 older, richer people frequent here # Subject 743 places influential on me 584 pleasant rural 513 good Factor 4 833 don't like -712 unemotional 14 Factor 1 Factor 2 931 like 923 greater security 894 interesting 773 privacy 882 diverse 501 easy to utilize 837 pleasant 816 beautiful Factor 3 Factor 4 854 place with certain areas forbidden to public 878 clean, bright -640 bad place for meeting girls Factor 5 Factor 6 926 natural area 836 good -510 faster pace of life 678 more freedom 626 air-conditioned - .61Subject # 15 Factor 1 983 growing, expanding -983 decaying Factor 2 -859 concrete -758 inhospitable 691 like -768 old 645 greenery Factor 3 Factor 4 821 easy to utilize 851 cosmopolitan -739 place a "satellite" (vs. a center) Factor -847 5 sprawling -718 quiet Factor 6 876 like 688 closed place Subject #16 Factor 1 Factor 2 848 comfortable for lounging, relaxing 915 "Down-to-earth" 774 comfortable -764 efficient, purposeful place "real" place 913 working-class place -715 younger people here -590 newly built -708 threatening Factor 3 Factor 4 867 overly bright 900 easy to utilize 774 bustling -673 deteriorating 753 active place 700 can't trust people here -538 inspires reverence Factor 5 - 817 sensual place Factor 6 846 like V / -62-775 otheir cultures here 524 newly built 1 846 good -575 cantt trust people here Subject #17 Factor 1 908 visual continuity Factor 2 918 dark 908 easy to utilize 698 individual activity 899 good 899 smaller-scale 877 enclosure 853 like 736 natural place 609 complexity 609 soft texture Factor 3 892 expression of indiv. identity 642 personal space 500 individual activity Factor 5 841 autonomous environment 752 complexity 752 soft textured -718 specific activities here 608 place of rest 576 natural place Factor 4 720 polluted air 716 a network (vs. specific space ) -634 natural space I U U I -63Rotated Subject #18 Factor 1 Factor 2 817 easy to utilize 949 transportation 916 commercial 684 clean 676 quiet Factor 3 -967 Factor 4 844 sidevialks- like partly outdoors -967 good 753 urban environment 713 relaxation (vs. work) Factor 5 762 go here by choice 670 place to live Subject # 19 Factor 1 883 comfortable -883 many people here 786 easy to utilize 785 in the country Factor 2 930 good place to meet someone 737 modern -654 closed-in place Factor 3 Factor 4 865 narrow winding streets 948 like 517 plush 920 good 500 near water -I -64Factor 5, 833 Dirty -7 3 6 well organized 543 bad smells -571 modern Subject #20 Factor 1 Factor 2 843 liberal culture 840 busy 790 like 766 concrete 741 beautiful 700 poor design 722 good -508 poor design 503 like to vacation here Factor 3 906 lots of "action" here. 724 good eating places here 632 easy to utilize ~2 -65Appendix D Subjects' Factors (in order of percentage variance accounted for) Factors Subject Number 1 1) aesthetic, 2) friendliness, 3) residential, 4) good, 5) freedom, 6) familiarity, 7) public access. 5 1) friendliness, 2) utility, 3) individuality, 4) urban-rural 6 1) friendliness, 2) space, 3) utility, 4) time 7 1) functional, 2) activity level, 3) residential, 4) utility, 5) aesthetic 8 1) utility, 2) friendliness, 3) space, 4) noise, 5) aesthetic 9 1) aesthetic,2) organization, 3) noise, 4) freedom, 6) friendliness 10 1) friendliness, 2) urban-rural, 3) aesthetic, 4) utility 11 1) friendliness, 2) organization, 3)time,4)noise, 5) freedom, 6) natural 12 1) friendliness, 2) privacy, 3) utility, 4) residential, 5) space, 5) public access, 6) aesthetic 17 1) aesthetic, 2) light, 3) individuality, 4) urban-rural, 5) space LG 2 1) comsumeristic, 2) friendliness, 3) work, 4) size, access, 6) utility, 7) 5) public young 3 1) activity level, 2) friendliness, 3) urban-rural, 4) time 5) friendliness, 6) size 4 1) friendliness, 2) utility, 3) privacy, 4) ornate, zation, 6) consumeristic, 7) urban-rural 13 1) time, 2) aesthetic, 5) organi- 3) conservative, 4) friendliness -66subject number 14 factors 1) aesthetic, 2) privacy, 3) formal, 4) neatness, 5) urbanrural, 6) friendliness 15 1) time, 2) friendliness, 3) utility, 4) urban-rural, 5) space, 6) aesthetic 16 1) friendliness, 2) working class, 3) activity level, 4) utility, 5) sensual, 6) good. 18 1) utility, 2) activity type, 3) aesthetic, 4) urban-rural, 5) friendliness 19 1) friendliness, 2) space, 3) ization 20 1) friendliness, 2) urban-rural, 3) utility ornate, 4) aesthetic, 5) organ-