House style preference and meanings across taste cultures
Transcription
House style preference and meanings across taste cultures
Landscape and Urban Planning 44 (1999) 33±42 House style preference and meanings across taste cultures Jack L. Nasar*, Junmo Kang City and Regional Planning, The Ohio State University, 289 Brown Hall, 190 W. 17th Ave., Columbus, OH 43210, USA Received 27 July 1998; received in revised form 14 September 1998; accepted 12 October 1998 Abstract Theorists have speculated that taste cultures differ in their design preferences. To test this, we interviewed 150 adults (30 representing each of the ®ve taste cultures) in central Ohio. We obtained their responses to photographs of house exteriors representing 15 different styles. Did the taste cultures differ as predicted by the theorists? Not really. The results showed strong similarities in the responses across the groups. However, the similarities decreased as the educational/occupational distance between the groups increased. In ®ndings echoing many other studies, we found strong similarities with highest preferences for the Tudor style, and highest friendliness score for the Farm style. Even for style, preference is not a matter of taste. Widely different groups show commonalities. # 1999 Elsevier Science B.V. All rights reserved. Keywords: Aesthetics; House preference; House meanings; Social class 1. Introduction Architects have long thought that the style of a building conveys social meanings and affects emotional experience. Empirical evidence supports these speculations. One review found that residents use their house exterior to de®ne identity and convey personality traits such as friendliness, privacy and independence, social status, aesthetic sense, life style, ideas and values to others (Despres, 1989a, b). Kinzy (Langdon, 1982) found common preferences among Buffalo suburbanites in response to black and white elevations of nine house styles. They liked Tudor most, Farm next, then Ranch, Mediterranean, Early American, Colonial, Contemporary, and Modern the *Corresponding author. Tel.: +1-614-292-1457; fax: +1-614292-7106; e-mail: [email protected] 0169-2046/99/$19.00 # 1999 Elsevier Science B.V. All rights reserved. PII S0169-2046(98)00109-1 least. Tuttle (1983) compared the preferences of homeowners and developers. He found that they had similar preferences. Again, the public liked Tudor the most and the `high' style ± International ± the least. The preference for Tudor was followed by Queen Ann, Mediterranean, Contemporary, Saltbox, Georgian Colonial, Greek Revival, International and Spanish Pueblo last. In an explicit test of responses to a variety of `high' and `popular' styles, Devlin and Nasar (1989) con®rmed that adult professionals liked `popular' styles and disliked `high' design styles. Finally, in a two-city study, Nasar (1989) found that styles conveyed common meanings across adult respondents from Columbus, Ohio, and Los Angeles. As in many other studies, they gave the most favorable rating to the Tudor style. The inferences from the exterior serve an additional purpose. Exterior form gives observers cues about 34 J.L. Nasar, J. Kang / Landscape and Urban Planning 44 (1999) 33±42 how to behave and what to expect. Consider the ®nding by Taylor et al. (1976) that the presence of resident-generated visual features in a neighborhood related to social expectations about people who live in those neighborhoods. Verschure et al. (1977) found that observers made judgments of the occupant's personality traits (such as trustworthiness) and socio-economic status (income and education) from the house exterior. Sadalla et al. (1987) found that observers' inferences from house exteriors about a homeowner's personality agreed with the owner's actual personality. Cherulnik and Wildrerman (1986) found that students accurately identi®ed the socio-economic status of neighborhoods built in the nineteenth century. For example, they accurately identi®ed the previous professional areas as areas where professionals lived. They judged the owners of the upper-middle class houses as neat, intelligent, energetic and ambitious, and the owners of the other houses as timid, gullible and lazy. They described the ®rst group of houses as tasteful, friendly, beautiful and clean and the second group as dangerous and noisy. In sum, the research consistently shows that attributes of house exteriors convey meanings to others. Do these symbols hold across different groups? According to Lynes (1954) and Gans (1974), the American society has different taste cultures. Lynes (1954) divided the public into four cultures ± High- brow, Upper Middlebrow, Lower Middlebrow, and Lowbrow. He felt that the Upper Middlebrows aspire to what they think Highbrows have; Highbrows look to Lowbrows for new ideas; and Lower Middlebrows just do not care. Gans (1974) saw America as consisting of many ``taste cultures, each with its own art, . . ., which differ mainly in that they express different aesthetic standards.'' He listed ®ve taste cultures: High, UpperMiddle, Lower-Middle, Low and Quasi-Folk, and gives clear de®nitions of each. He believes, as we do, that ``buildings should be designed for people who will use them . . . A beautiful building is . . . one that satis®es the aesthetic standards of its users.'' Table 1 shows the Gans (1974) taste cultures in terms of their education and occupation. According to (Gans, 1974, p. 126) ``people not only tend to choose the taste culture which is congruent with their education, but they cannot be expected to choose content which is incongruent with their education.'' Although other factors may in¯uence taste, Gans rules them out. He says that incomes are not reliable as upwardly mobile people ``will move quickly into a neighborhood that expresses their new status but may never adopt the taste culture associated with that status. This . . . re¯ects, at least in part, their lack of education which prevents culture mobility (p. 115). He rules out religion, ethnicity, religious origin, radical background ``as disappearing rapidly in American society'' (p. 70). Table 1 Five taste cultures according to Gans Taste culture Education (and major) Occupations High college graduate, private or state public university or some graduate/professional school (art, architecture, landscape architecture, industrial design, or other design). creator-oriented artists, designers, design educators; and user oriented, administrators and managers Upper Middle college graduate, private or state public university (non-design, non-art) manager, administrator, technical/professional speciality Lower-Middle high school graduate if over 45, community college graduate or some college at state university or small college if under age 45 administrative support, public school teacher, lowest-level white-collar Low some high school if over age 45, high school grad or pre-professional school if under 45 skill or semi-skilled factory or service worker (blue collar), operator, semi-skilled white collar Quasi-Folk Low less than 8th grade unskilled blue collar, service unemployed J.L. Nasar, J. Kang / Landscape and Urban Planning 44 (1999) 33±42 According to Gans, each group reads different magazines, watches different television programs and likes different kinds of designs. These different patterns of behaviors ®nd some support in market research studies, such as the massive VALS inventory (McConnell, 1990). Concerning design, he writes that High Culture ``likes expressionism, impressionism, abstraction and conceptual art'' (p. 76); Upper Middle prefers ``substantive, unconcerned with innovation in form, and uninterested in making issues of method'' (p. 82) though, it borrows ``from . . . High culture'' (p. 83); Lower Middle dislikes the abstract art favored by the High and Upper Middle, and High and Upper Middle, looks for ``substance: [such that] form must serve to make substance more intelligible or gratifying'' (p. 85), and prefers vernacular or traditional forms; Low dislikes abstract ideas and follows the Lower Middle culture; the Quasi-Folk Low may favor old and rural styles. The ideas of Lynes and Gans, while compelling, have not been tested empirically. Do the different taste cultures differ in their architectural preferences and if so do they differ as predicted by Gans or Lynes? To test this, we conducted a study in two phases. In the ®rst phase, we developed a representative set of house styles. In the second, we obtained responses to these styles from representatives of each taste culture. Table 2 Style lists derived from three kinds of sources Historic and professional texts (22 styles) Builder survey (24 styles) Plan shop books (13 styles) Cape Cod Cape Cottage Cape Cod Carpenter Gothic Classical (or Colonial Revival) Cape Cod Coltswold Dutch Colonial Farm Federal French French Mansard Gambrel Roof Garrison Colonial Georgian Georgian Colonial Greek Revival Medieval Neo-Victorian New Orleans Post Modern 2. Method 2.1. Selecting house-styles for study We wanted to select a set of styles recognizable by both, the experts and non-experts, popular with the current housing industry and available throughout the continental USA. To do this, we ®rst developed several lists of styles. We derived our list from the Historic American Building Survey (The Old-house Journal, 1979; Poppelliers, 1977), and a variety of architectural books (Foley, 1980; McCoy, 1982; Pickering, 1951; Rifkind, 1980; Scully, 1974; Walker, 1981; Whiffen, 1969a, b; Whiffen and Koeper, 1981). These references listed many styles varying from a low of eleven (The Old-house Journal, 1979) to high of about three hundred (Foley, 1980). Focusing on two-story houses, we also eliminated other house forms, such as Mobile 35 Saltbox Spanish Spanish Colonial Tudor (or Elizabethan) Williamsburg Dutch Colonial English Colonial Federal French Colonial Georgian German Colonial Gothic Revival Greek Revival International Italianate Colonial Contemporary Early American Mansion Farm French French Colonial Gambrel Roof Georgetown Georgian Neo-Victorian Period Prairie Queen Ann Richardonian Romanesque Roman Revival Saltbox Second Empire Shingle Spanish Colonial Stick Tudor Saltbox Tudor homes and split levels. We also eliminated styles that no longer exist or that closely resembled other styles. The ®rst column of Table 2 shows the 22 styles that resulted. We also interviewed four of the ®ve largest builders in Central Ohio to ®nd which styles they provided to their prospective buyers. This group of builders 36 J.L. Nasar, J. Kang / Landscape and Urban Planning 44 (1999) 33±42 accounted for 63% of the total permits for detached single-family houses in the three counties surrounding Columbus, Ohio. The second column of Table 2 lists the 24 styles resulting from the interviews. Finally, we reviewed publications by plan shops. The International Directory of Home Plans (1988) by Garlinghouse Co. listed 312 plans. The Home Planners Inc. Encyclopedia of Home Designs (1988) listed 450 plans. We also referred to Home Planner Inc. Twostory Home (1986), Early American Home Plans (1986), Contemporary Home Plans (1986), English Tudor Homes (1986), Spanish and Western Home Design (1986), Traditional and Contemporary Plans (1986) and Most Popular Homes (1996). The third column of Table 2 shows the 13 styles derived from the plan shop publications. The combined list of styles from the three sources totalled to 42. Because 42 styles would be unwieldy for the study plan, we narrowed that list further. We dropped styles of a particular period or territory, encompassing other styles within them, not occurring in the US or in central Ohio (where we needed to take photos), not occurring in the two-story form, or indistinguishable from styles of other labels. This process left us with 15 styles: Gambrel roof, Garrison Colonial, Federal, French, Farm, Greek Revival, Georgian, Queen Ann, Spanish, Italianate, Second Empire, Tudor, International, Prairie and Post Modern. We thought there might be some perceived overlap in the styles. To develop a set of perceptually relevant categories of style, we ran one additional analysis. First, we photographed three examples of each style to scan into the computers. The photos looked head on to the facade as seen from the street. We edited out trees, bushes and cars obstructing a clear view of the front facade. The ®nal scenes were superimposed on a black background, so they only differed in style. We printed the resulting color images as color photos. We interviewed 30 students recruited from eight colleges at the Ohio State University. Most students in the sample were between 18 and 25 years old (77%), white (80%), undergraduates (63%) who had lived in Ohio for more than three years (80%). We asked each student to sort the houses into piles in terms of their similarity. They could use as many piles as necessary and could make corrections on completion of the task. Then, they ranked the houses in each group from the best to worst example of the category. We calculated the frequency with which the students paired each house with each other house. We then used the resulting similarity scores in a cluster analysis to derive groupings of styles by similarity. The analysis used Ward's minimum variance agglomerative hierarchical cluster procedure. This is a widely used method that results in the fewest errors (Harvey, 1986; Milligan and Cooper, 1988). It used the Hubert and Levin (1976) c-index stopping rule. The analysis yielded ten clusters of houses. Three clusters had three examples of the same style: Farm, Greek Revival, and Spanish. The other clusters had mixes of styles. For example, one cluster had the three International examples and one Prairie style. Another had the three Queen Ann examples and the three Italianate. Another had the three Post Modern examples joined with two of the Tudor examples. For the study comparing responses across taste groups, we selected the style from each cluster that had the highest goodness-ofexample score in the cluster. Fig. 1 shows the ten houses. The sample included Farm, French, Greek Revival, International, Spanish, Georgian, Post Modern (Tudor), Queen Ann, Federal, and Garrison Colonial. 2.2. Respondents for taste cultures We interviewed 150 individuals (30 each of the ®ve taste groups). Interviews took place in a regional shopping mall in Columbus, Ohio. Interviewers had a job lists and educational levels for each group. The criteria followed Gans' criteria described in Table 1. The interviewers approached every passerby and asked about educational level and current occupation. The interviewers interviewed only those respondents who ®t criteria for a taste group. The interviews took place in a private of®ce. As we did not expect to ®nd many High culture (designers) or Quasi-Folk Low culture members at the mall, we obtained those samples elsewhere. For the High Culture group, we randomly sampled designer and architect of®ces from the yellow pages. We phoned and interviewed those who agreed to participate. For the Quasi-Folk Low Culture, interviewers contacted people at homeless shelters, public housing communities, Volunteers of America, and several community churches and centers. If responses differed across the groups, this differential pattern of sampling would present a confounding variable. If we found similar responses, in spite of J.L. Nasar, J. Kang / Landscape and Urban Planning 44 (1999) 33±42 37 Fig. 1. Ten houses used in the study of taste groups. From top down, the left column shows Farm, Garrison Colonial, Federal, International, French; and the right column shows Georgian, Greek Revival, Queen Ann, Spanish, Post Modern. 38 J.L. Nasar, J. Kang / Landscape and Urban Planning 44 (1999) 33±42 Fig. 1. (Continued) the added difference in sampling procedures, this would strengthen the likelihood of shared responses. Table 3 shows the characteristics of respondents in each group. Members of the High Culture sample held degrees in architecture, landscape architecture, interior design or art. They worked in those ®elds as well. Members of the upper-middle Taste sample had similar levels of educational achievement, but they had not majored in a design ®eld. Their occupations included manager, administrator, engineer, scientist, home economist, occupational therapist, respiratory therapist, registered nurse, psychologist, and musician. Members of the Lower Middle sample had completed community college, technical school or some college (or if over 45, a high school diploma). Their occupations included teacher, bookkeeper, market research analyst, pharmacy assistant, salesperson, computer operator, house inspector, secretary, receptionist, recreation worker, waiter/waitress, bartender, dietician, bank teller, bank of®cer, electrician, and police of®cer. Members of the Low Taste sample held a high school diploma or had attended a community college or technical school. Jobs included bus driver, machine tool operator, construction machinery operator, cashier, waitress, operator, shipping/receiving clerk, truck driver, construction worker, welder, salesperson, cook, surveyor, teacher's aid. Finally, members of the QuasiFolk Low sample had not completed more than the 8th grade. Most were unemployed, but a few held parttime jobs such as cashier, janitor, waiter, cook, garbage collector, house cleaner, and construction worker. 2.3. Interview procedure Interviewers showed respondents each of the ten houses in 400 600 color photographs. Research suggests several important dimensions of housing meaning. These include perceived status, warmth and preference. Though each meaning may differ across `taste cultures', the theories of Gans (1974) and Lynes (1954) speci®cally talk about differences related to perceived status, and preference. On the one hand, the theories argue that some groups aspire to what they believe other groups have. In this sense, these groups may evaluate the perceived status of housing exteriors. On the other hand, the theories argue that the groups have different tastes or preferences. Rather than ask people directly about their preferences and perceived status, we used a question format that put them into a more realistic decision format. We adapted the instructions from those used by Nasar (1989) to measure status, preference, and friendliness. For status, we asked respondents to assume that residents of the ten houses work together. We asked them to rank the houses from `the one most likely to have the person who takes charge and leads the group to the one least likely to have the leader'. For preference, we asked respondents to imagine that they had just won the `Dream House Lottery', and had to choose the house to receive free. They were asked to place the houses in order from `the one you would most like to receive to the one you would least like to receive'. For friendliness, we asked respondents to imagine having a ¯at tire in a street with the ten houses. We asked them to place the houses in order from `the one you would J.L. Nasar, J. Kang / Landscape and Urban Planning 44 (1999) 33±42 39 Table 3 Respondent characteristics High Upper-Middle Lower-Middle Low Quasi-Folk Total Education Less than 8th grade Some high school High school graduate Community college graduate Some college College grad. Masters or Ph.D. 0.0% 0.0 0.0 0.0 0.0 56.7 43.4 0.0% 0.0 0.0 0.0 0.0 86.7 13.3 0.0% 0.0 0.0 23.3 56.7 20.0 0.0 0.0% 16.7 56.7 10.0 16.7 0.0 0.0 100.0% 0.0 0.0 0.0 0.0 0.0 0.0 20.0% 3.3 11.3 6.7 14.7 32.7 11.3 Income Less than $10 000 $10 000±24 999 $25 000±34 999 $35 000±54 999 $55 000 or more 0.0% 36.7 26.7 20.0 16.7 0.0% 13.8 17.2 37.9 31.0 10.3% 31.0 31.0 20.7 6.7 16.7% 40.0 20.0 20.0 3.3 96.7% 3.3 0.0 0.0 0.0 25.0% 25.0 18.9 19.6 11.5 Home ownership Own Rent Other 34.7% 48.0 17.3 56.7% 40.0 3.3 23.3% 63.3 13.3 23.3% 73.3 3.3 3.3% 33.3 63.3 34.7 48.0 17.3 Race White African-American Asian Other 83.3% 3.3 10.0 3.3 90.0% 10.0 0.0 0.0 60.0% 40.0 0.0 0.0 80.0% 20.0 0.0 0.0 43.3% 50.0 0.0 6.6 71.3% 24.7 2.0 2.0 Age 18±25 28±34 35±44 45 and older 3.3% 40.0 30.0 26.7 10.0% 53.3 16.7 20.0 36.7% 33.3 20.0 10.0 56.7% 13.3 16.7 13.3 23.3% 26.7 33.3 16.6 26.0% 33.3 23.3 17.4 Marital status Married Single Divorced/widowed 60.0% 30.0 10.0 48.3% 44.8 6.9 20.0% 63.3 16.7 23.3% 66.7 10.0 16.7% 46.7 36.7 33.6% 50.3 16.1 Years in Columbus Less than 1 year 1±3 years 4±5 years 6±10 years More than 10 years 3.3% 13.3 3.3 33.3 46.7 33.3% 3.3 6.7 16.7 40.0 36.7% 23.3 3.3 3.3 33.3 23.3% 20.0 3.3 10.0 43.3 10.3% 13.8 17.2 20.7 37.9 21.5% 14.8 6.7 16.8 40.3 Gender Male Female 50.0% 50 46.7% 53.3 50.0% 50.0 50.0% 50.0 53.3% 46.7 50.0% 50.0 40 J.L. Nasar, J. Kang / Landscape and Urban Planning 44 (1999) 33±42 most likely go to for help to the one you would least likely go to for help'. 3. Results Did responses from each taste culture differ from one another as predicted by Gans? The results showed a mixed support for Gans' theory. We ran several kinds of analyses. To examine differences between groups, we looked at paired-comparisons across each pair of groups in their rankings of the styles on each scale. To examine similarities between the groups, we looked at correlations of the scores across the groups. The paired comparisons across each pair of group rankings of the styles revealed no signi®cant differences across any pair of groups for desirability or friendliness. For perceived status, only one group differed from others. The High taste culture differed signi®cantly from each other group (Upper-Middle: t2.08, p<0.05; Lower-Middle: tÿ1.79, p<0.01; Low: tÿ1.76, p<0.01; Quasi-Folk: tÿ1.99, p<0.05). Table 4 shows the results of the Kendal Tau B rank correlation analyses. As you can see, it shows high levels of agreement among the groups. First, consider the status scores. The results show signi®cant correlations between each group's rankings of the houses. Yet, the results also offer support for the prediction of Gans (1974). Although the group ratings have signi®cant correlations, the correlations peak for neighboring groups in education and occupation (those along the diagonal), and they decrease as the education/occupation level difference increase. For example, High has its highest correlation with Upper Middle (r0.96), a lower correlation with Lower Middle (r0.78), a lower correlation with Low (r0.76), and its lowest correlation with Quasi-Folk Low (r0.60). Upper middle has its highest correlation with High, then Lower Middle, Low and then Quasi-Folk Low. Thus, the mean correlation on the diagonal is 0.85. One level below the diagonal, the mean drops to 0.76. One level down, it drops again to 0.70; and the bottom cell correlation drops to 0.60. Gans' prediction of a strong connection between the High and Quasi-Folk culture did not emerge. Next, consider the desirability scores. A similar pattern emerges. With the exception of one comparison (High and Quasi-Folk Low), the results show signi®cant correlations in each group's ranking of the Table 4 Kendall Tau B correlations between taste cultures Desirability Upper Middle Lower Middle Low Quasi-Folk Low High 0.56 b 0.49 a 0.42 a 0.33 Upper Middle 0.72 b 0.69 b 0.51 a Lower Middle 0.58 a 0.72 b Low 0.64 b High 0.96 b 0.78 b 0.76 b 0.60 a Upper Middle 0.82 b 0.76 b 0.64 a Lower Middle 0.85 b 0.73 b Low 0.76 b High 0.67 b 0.38 0.47 0.42 Upper Middle 0.58 a 0.81 b 0.72 b Lower Middle 0.73 b 0.78 b Low 0.78 b Leadership Upper Middle Lower Middle Low Quasi-Folk Low Friendliness Upper Middle Lower Middle Low Quasi-Folk Low a b p<0.05 p<0.01 J.L. Nasar, J. Kang / Landscape and Urban Planning 44 (1999) 33±42 houses. Again, the correlations between groups of similar levels of education and income (along the diagonal) tend to exceed those of the groups with larger differences in education and income. For desirability, the results show two exceptions: the correlations between Upper Middle and Low (r0.69) and between Lower Middle and Low (r0.72) are higher than some of the scores on the main diagonal. Otherwise, as we move further from the diagonal, the correlations tend to decrease. For example, the High taste group has its highest correlation with the Upper Middle. Its correlations drop as we move to LowerMiddle, Low, and Quasi-Folk. The Upper Middle had higher correlations with the Lower-Middle than with the Low or Quasi-Folk. This agrees with Gans' prediction of a strong connection between High and Quasi-Folk. It contradicts his prediction, the Low Middle had a higher correlation with the Quasi-Folk than with the Low group. Finally, consider the friendliness score. The results show signi®cant correlations for seven of the ten comparisons. Unlike the scores for status and desirability, the correlations do not show a pattern related to income and education. The correlations on the diagonal are not higher than the off-diagonal scores. 4. Discussion Gans and Lynes predicted systematic differences in responses across the taste cultures. The present ®ndings offer weak support for their position. The High Culture group had different ratings for perceived status than did the other groups. Otherwise, the groups displayed similar patterns of response. In partial agreement with the model, as differences in education and occupation level increased, the relationship between perceived status scores decreased. This happened to a lesser extent for desirability scores, and to an even lesser extent for friendliness scores. Signi®cant group differences emerged for only two of the ten styles, and these differences were between the High groups and the Low groups. The ®ndings also do not support the theory that High culture draws ideas from the Low or Quasi-Folk Low culture. A large body of research has pointed to differences between architects and others in their responses to architectural stimuli (cf. Nasar, 1994). In agreement 41 with these ®ndings, the study found that the designers differed from the others in their judgments of perceived status. However, the differences did not carry over to their preference or friendliness ratings. Where previous differences have centered on architects or architecture students compared to others, our designer group included architects and other design professionals. Perhaps the other designers had more similar scores to the public than did architects. Research on environmental aesthetics and meaning have documented strong commonalities in preferences for and the connotative meanings conveyed by architecture (cf. Nasar, 1997; Stamps, 1995, 1996). The present ®ndings con®rm the presence of strong commonalities across groups. For example, all groups converged on the Post Modern as most desirable, the International as highest in status, and the Farm as the most friendly. Where do the commonalities come from? They may result from shared experience among the American public. Repeated exposure to house styles associated with various meanings might lead observers to associate those meanings with similar styles. Thus, mass media exposure to house styles associated with various contexts may account for the similarities. Consider, for example, the US White House (Colonial), regularly displayed as a backdrop in TV reports on the President. TV also shows house exteriors associated with various people in shows. The choices of style may both echo and shape popular meanings. In other cultures, having different housing styles and media exposure, one would expect to ®nd different patterns of meaning, but within a culture, commonalities may well still emerge. For urban design, the commonalities in meaning suggest that design controls for style may well shape the meanings conveyed by developments. Informed by research, such controls can improve the appearance of the surroundings for large numbers of people. References Cherulnik, P.D., Wildrerman, S.K., 1986. Symbols of status in urban neighborhoods: Contemporary perceptions of nineteenthcentury Boston. Environment and Behavior 18, 604±622. Contemporary Home Plans, 1986. Home Planners, Inc., Farmington Hills, MI. Despres, C., 1989a. The meaning of home: Literature review and directions for future research and theoretical development. 42 J.L. Nasar, J. Kang / Landscape and Urban Planning 44 (1999) 33±42 Paper presented at the International Housing Symposium, Sweden. Despres, C., 1989b. On style: integrating architectural theory and environmental design research. Paper presented at the Environmental Design Research Association 20, Black Mountain, NC. Devlin, K., Nasar, J.L., 1989. The beauty and the beast: some preliminary comparison of `high' versus `popular' residential architecture and public vs. architecture judgments of same. Journal of Environmental Psychology 9, 333±344. Early American Home Plans, 1986. Home Planners, Inc., Farmington Hills, MI. Encyclopedia of Home Designs, 1988. Home Planners, Inc., Farmington Hills, MI. English Tudor Homes, 1986. Home Planners, Inc., Farmington Hills, MI. Foley, M.M., 1980. The American House. Harper and Row, New York. Gans, H.J., 1974. Popular Culture and High Culture: An Analysis and Evaluation of Taste. Basic Books, Inc., New York. Harvey, R.J., 1986. Quantitative approaches to job classification: a review and critique. Personnel Psychology 39, 267±289. Hubert, L.J., Levin, J.R., 1976. A general statistical framework for assessing categorical clustering in free call. Psychological Bulletin 83, 1072±1080. International Directory of Home Plans, 1988. Garlinghouse Co., Kansas. Langdon, P., 1982. Suburbanites rating house styles. New York Times, April 22. Lynes, R., 1954. The Taste-makers. Harpers and Brothers, New York. McConnell, G.E., 1990. A VALS overview of U.S. shopping patterns and attitudes. SRI International and Business Intelligence Program, D90-1442, Menlo Park, CA. McCoy, E., 1982. Guide to U.S. architecture 1940±1980. Arts and Architecture Press, Santa Monica, CA. Milligan, G.W., Cooper, M.C., 1988. Methodological review: clustering methods. Applied Psychological Measurements 11, 329±354. Most Popular Homes, 1996. Home Planners Inc., Farmington Hills, MI. Nasar, J.L., 1989. Symbolic meanings of house styles. Environment and Behavior 21, 235±257. Nasar, J.L., 1994. Urban design aesthetics: the evaluative qualities of building exteriors. Environment and Behavior 26, 377±401. Nasar, J.L., 1997. The Evaluative Image of the Vity. Sage, Thousand Oaks, CA. Old-house Journal, 1979. Field guide to old-house styles. Oldhouse Journal, Brooklyn, NY. Pickering, E., 1951. The Homes of America. Thomas Y. Cromwell Co., New York. Poppelliers, J.C., 1977. What Style is It? Preservation Press, Washington, DC. Rifkind, C., 1980. A Field Guide to American Architecture. New American Library, New York. Sadalla, E.K., Vershure, B., Burroughs, J., 1987. Identity symbolism in housing. Environment and Behavior 19, 559±587. Scully, V., 1974. The Shingle Style Today. George Brazillier, Inc., New York. Spanish and Western Home Design, 1986. Home Planners, Inc.. Farmington Hills, MI. Stamps, A.E., 1995. In: Nasar, J.L., Grannis, P., Hanyu, K. (Eds.), Beauty is in. . .? Twenty-sixth Annual Conference of the Environmental Design Research Association, proceedings. EDRA, Oklahoma City, OK, pp. 48±53. Stamps, A.E., 1996. People and places: variance components of environmental preferences. Perceptual and Motor Skills 82, 323±334. Taylor, R.B., Brower, S., Stough, R., 1976. User-generated visual features as signs in the urban residential environment. In: Suedfeld, P., Russell, J.A. (Eds). The behavioral basis of design, Book 1: Session Summaries and Papers/EDRA6 Dowden, Hutchinson and Ross, Inc., Stroudsburg, PA, pp. 94±101. Traditional and Contemporary Plans, 1986. Home Planners, Inc., Farmington Hills, MI. Tuttle, D.P.O., 1983. Suburban Fantasies. Unpublished master's thesis, University of Wisconsin, Milwaukee. Two-story Home, 1986. Home Planners, Inc., Farmington Hills, MI. Verschure, B., Magel, S., Sadalla, E.K., 1977. House form and social identity. In: Suedfeld, P., Russell, J.A., Ward, L.M., Szigeti, F., David, G. (Eds.), The Behavioral Basis of Design, Book 2: Session Summaries and Papers/EDRA 7. Dowden, Hutchinson, and Ross, Inc., Stroudsburg, PA, pp. 238±278. Walker, L., 1981. American Shelter: An Illustrated Encyclopedia of the American Home. Overlook Press, Woodstock, NY. Whiffen, M., 1969a. American Architecture Since 1780: Guide to Style. M.I.T. Press, Cambridge, MA. Whiffen, M., 1969b. American Architecture Since 1607±1976. M.I.T. Press, Cambridge, MA. Whiffen, M.G., Koeper, F., 1981. American Architecture 1607± 1976. M.I.T. Press, Cambridge, MA.