Environment and Behavior

Environment and Behavior http://eab.sagepub.com/

Effects of Area, Height, Elongation, and Color on Perceived Spaciousness Arthur E. Stamps III Environment and Behavior 2011 43: 252 originally published online 8 April 2010 DOI: 10.1177/0013916509354696 The online version of this article can be found at: http://eab.sagepub.com/content/43/2/252

Published by: http://www.sagepublications.com

On behalf of: Environmental Design Research Association

Additional services and information for Environment and Behavior can be found at: Email Alerts: http://eab.sagepub.com/cgi/alerts Subscriptions: http://eab.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://eab.sagepub.com/content/43/2/252.refs.html

Downloaded from eab.sagepub.com at UNIV CALIFORNIA DAVIS on August 30, 2011

Article

Effects of Area, Height, Elongation, and Color on Perceived Spaciousness

Environment and Behavior 43(2) 252–273 © 2011 SAGE Publications Reprints and permission: http://www. sagepub.com/journalsPermissions.nav DOI: 10.1177/0013916509354696 http://eab.sagepub.com

Arthur E. Stamps III1

Abstract This article reports findings from three experiments, covering 46 environments and 66 participants, on how strongly four properties of the physical environment influence perceived spaciousness.The properties were horizontal area, boundary height, elongation, and color. Ten original findings were reported. Overall, horizontal area had the strongest effect on perceived spaciousness (r = .60; more floor area increases perceived spaciousness), followed by height (r = –.22; lower boundaries increase perceived spaciousness). The effect of color on perceived spaciousness, when amount of light is controlled, was much smaller (r = .14). Findings for elongation were different for concave and convex spaces (r’s of –.22 and +.26). Quantitative syntheses of the current work with previous work are presented, as is numerical guidance for cost-effective future work. Keywords spaciousness, smart growth, sustainable design Space! the formal frontier. These are some studies from the research enterprise. Its five part mission: to create new worlds, to seek out new views and new spatial relations, to boldly write what no one has writ before, to report new findings on the question of what makes a space seem . . . well . . . er. . . more spacious? There are both theoretical and practical reasons for considering this question. The theoretical reason was described in previous articles

1

Institute of Environmental Quality, San Francisco, CA

Corresponding Author: Arthur E. Stamps III, Institute of Environmental Quality, 290 Rutledge Street, San Francisco, CA 94110 Email: [email protected]


253

Stamps

(Stamps, 2007, 2010): animals, including people, feel threatened if they do not have sufficient space. Environments that do not provide sufficient space are ambient stressors and thus should be avoided if possible or, if unavoidable, mitigated as much as possible. The practical reason for discovering what properties of the environment influence perceptions of spaciousness is that the creation of resource-conserving cities requires increases in density. The importance of high density was noted by, among others, the visionary city planner R.L. Meier in the 1970’s (Meier, 1975), and has since become a major design goal for policies such as sustainable design, countering global warming, and transit villages (Calthorpe & Fulton, 2001; International City Management Association, 2003, pp. 11-20; Talen, 2003; Williams, Burton, & Jenks, 2000). The combination of the need at the global level for increased density and the need at the individual level for adequate space thus generates the question of how, within a given volume of space, can it be made to appear larger than it really is? Hence the studies reported in this article.

Previous Findings There is a substantial amount of research on how strongly various environmental properties influence impressions of spaciousness. Emphasis in this article is placed on research that reports the strength of relationships—for example, efficacy—in terms of correlations or standardized mean differences. The reasons for this criterion are given in the section on statistical protocols below. The relevant literature on spaciousness, covering 11 studies, 214 scenes, and 337 participants was recently reviewed (Stamps, 2007). The following environmental properties were investigated: (a) Effects of horizontal area on perceived spaciousness were reported by Gärling (1970a, 1970b), Benedikt and Burnham (1985), Inui and Miyata (1973), Franz, von der Heyde, and Bülthoff (2003), and Franz and Wiener (2005), with the overall finding being that the larger the horizontal area within a boundary, the more spacious the space seemed to be. (b) Effects on perceived spaciousness due to amount of light were reported by Martyniuk, Flynn, Spencer, and Hendrick (1973), Kirschbaum and Tonello (1997), and Inui and Miyata (1973),with the overall finding that brighter spaces seemed more spacious. (c) Relationships between shape and perceived spaciousness were reported in two studies. Sadalla and Oxley (1984) constructed rooms of with walls of gray plywood panels and had different degrees of elongation ranging from a square room (1:1 ratio of width to length) up to a corridor (1:9 ratio). The shapes of these rooms were convex. The correlation between estimated size of the room and elongation, based on data reported in Tables 1 and 2 in the referenced article, was r = –.27 (on n = 7 rooms for which data were


254

Environment and Behavior 43(2)

given), indicating that long, narrow convex spaces would appear to be less spacious than fat, short spaces of the same area. Ishikawa, Okabe, Sadahiro, and Kakumoto (1998) investigated design alternatives to a Japanese street. The alternatives were inclusion of setbacks that varied in proportions. Some were long and shallow; others were short but deep. In this case, the shapes of the streets were concave. The concept was intriguing, but since there was not sufficient data to calculate an effect size, any conclusions regarding this particular point would have to be withheld pending replication that did report effect sizes. (d) There was one study for the effect of occlusion on perceived spaciousness, done by Imamoglu (1973), on the amount of furniture in rooms. The effect on spaciousness was r = –.54 (n = 3): the more furniture, the less spacious the room appeared. (e) There was one study on the permeability of the boundary. Franz et al. (2003) reported data indicating that perceived spaciousness and percentage of walls in 16 museum rooms correlated at r = .12. (f) The effect of boundary roughness on perceived spaciousness was reported in a pair of studies by Stamps and Krishnan (2006), with the finding that rougher boundaries made spaces seem larger (r’s of .10 on n = 16 rooms and r = .28 on another 12 rooms). Seven additional studies, covering 106 scenes and 217 participants, have been done subsequent to that review. Two of the studies, reported in Stamps (2007), reported effect sizes of floor area, occlusion (expressed as interior partitions), and light (expressed in cd/m2), for art galleries shown as 24 static images and as 8 virtual reality models. Results indicated that floor area correlated with spaciousness at r = .26 in the static medium and r = .42 in the dynamic medium; occlusion (the presence or absence of interior partitions) correlated with perceived spaciousness at r = .23 in the static medium and r = .56 in the static medium; amount of light correlated at r = .20 in the static medium and r = .17 in the dynamic, and, overall, responses obtained from static simulations correlated at r = .79 with responses obtained with the dynamic medium. The same hypotheses were statistically tenable with both media. Another pair of studies, reported in Stamps (2009) patched the work reported in Ishikawa et al. (1998) by reporting effect sizes for effects of recesses in streets with concave shapes. The venue was the deliciously photogenic traditional Japanese machinami style as expressed in the Gion district of Kyoto (Durston & Mizuno, 2002; Plutschow, 1979). Factors were area of setbacks and elongation of setbacks. Media were static color images and virtual reality models. In the static media, area of setback correlated at r = .45 and elongation of setback correlated with perceived area at r = .28. In the dynamic medium, the respective correlations were r = .56 and .17. The correlation of perceived spaciousness between the six virtual reality models and


255

Stamps

Table 1. Quantitative Summary of a priori Database on Perceived Spaciousness Factor

Venue

n

r

Source

Streets 12 .96 Gärling (1970a) Horizontal area Streets 12 .78 Gärling (1970b) Rooms 15 .63 Inui and Miyata (1973) Art galleries 16 .84 Franz, von der Heyde, and Bülthoff (2003) Art galleries static 24 .26 Stamps (2007) Art galleries dynamic 8 .42 Stamps (2007) Streets static 18 .45 Stamps (2009) Streets dynamic 12 .56 Stamps (2009) Rooms 12 .56 Stamps (2010) Synthesis 129 .65 .05 ci = [.54, .73] Light Rooms 6 .27 Martyniuk, Flynn, Spencer, and Hendrick (1973) Rooms 13 .94 Inui and Miyata (1973) Rooms 12 .66 Stamps (2010) Landscapes 18 .18 Stamps (2010) Synthesis 49 .64 .05 ci = [.45, .77] 7 -.27 Sadalla and Oxley (1984) Elongation—Convex Rooms Synthesis 7 -.27 .05 ci = [-.80, .49] Street recesses static 18 .28 Stamps (2009) Elongation— Concave Street recesses dynamic 12 .17 Stamps (2009) Synthesis 30 .24 .05 ci = [-.09, .52] Art galleries static 24 .23 Stamps (2007) Occlusion Art galleries dynamic 8 .56 Stamps (2007) Synthesis 32 .30 .05 ci = [-.01, .56] Boundary Rooms 16 .12 Franz et al. (2003) permeability Rooms 12 .55 Stamps (2010) Landscapes 8 .62 Stamps (2010)n Landscapes 18 .70 Stamps (2010) Synthesis 54 .51 .05 ci = [.03, .67] Boundary Rooms 16 .10 Stamps and Krishnan (2006) roughness Rooms 12 .28 Stamps and Krishnan (2006) Synthesis 28 .17 .05 ci = [-.17, .48] Boundary depth Landscapes 8 .03 Stamps (2010) Landscapes 18 .10 Stamps (2010) Synthesis 26 .08 .05 ci = [-.27, .42] Art galleries 11a Static/dynamic .79 Stamps (2007) media Streets 9a .87 Stamps (2009) Synthesis 20 .82 .05 ci = [.63, .92] a. Harmonic means from two different numbers of stimuli.


256


Table 2. Repeated Measures Analysis of Variance, Experiment 1 Source Participants (Within) Area Height Setback ratio Residual Total

SS 316.78 (250.92) 64.93 90.02 68.69 467.68 1008.10

df

MS

F

p

24 13.19 (11) 22.81 1 64.93 37.75 4e-09 1 90.02 52.33 5e-12 1 68.69 39.93 1e-9 272 1.72 299

% Variance

6.2% 8.7% 6.6%

the 28 corresponding static images was .87 (on n = 6). Again, the same hypotheses were statistically tenable in both media, replicating the finding obtained for the art galleries regarding the equivalence of results over static and dynamic media. The next three studies are reported in Stamps (2010). A study of 12 rooms indicated that perceived spaciousness correlated at r = .56 with floor area, r = .55 with percentage of walls and roof that are glass, and r = .66 with luminosity. Another study, this time of 8 landscapes, indicated that perceived spaciousness correlated at r = .03 with depth of boundary and r = .62 with the percentage of view not covered by solid objects. These findings were replicated in another study of 18 landscapes, for which the correlations with perceived spaciousness were r = .70 for percentage of view not covered with solid objects, .18 for light, and .10 for boundary depth. Thus, the a priori empirical database on how the environment influences perceived spaciousness covered 18 studies, 320 scenes, 554 participants, and 8 environmental properties. Estimates of the efficacies of the eight environmental properties with respect to perceived spaciousness are listed in Table 1. Overall, the property with the largest effect on perceived spaciousness was horizontal area (r = .65, .05 ci = [.54, .73]), followed by light (.64, [.45, .77), boundary permeability (.51, [.03, .67), interior occlusion (.30, [–.01, .56]), elongation for convex shapes (–.27, [–.80, .49]), elongation for concave shapes (r = .24, [–.09, .52]), boundary roughness (.17. [–.17, .48), and boundary depth (.08, [–.27, .42]). Also, responses of spaciousness were highly correlated between the static and dynamic media (r = .82. .05 ci = [.63, .92]).

Selection of Variables Based on the data described above, four environmental properties were selected for inquiry. The first variable was horizontal area. This was chosen to increase the


257

Stamps

diversity of environments for the area/spaciousness relationship. The second was height, because we were unable to locate prior reports of effect sizes for the influence of height on spaciousness. The third factor was elongation, because the relevant studies generated conflicting findings. It seemed possible that the discrepancy due to elongation in one experiment was a function of recesses in a concave environment (streets with alcoves), whereas the others consisted of elongation of a convex space (rectangular rooms). Accordingly, it was decided to attempt replication with both convex and concave shapes. The last variable was color. The amount of light (luminosity) had already been shown to have an effect on spaciousness, so the next step seemed to be variation in colors with equal luminosity. The effects of occlusion, boundary permeability, boundary roughness, and boundary depth could be controlled by holding them constant for all environments within an experiment. Both static and dynamic media produced the same results, so either would be adequate for the present work. We chose to use the dynamic virtual reality medium.

Presuppositions This article is based on extensive previous work for topics of (a) simulation validity, (b) equivalence of different scaling models, and (c) presentation protocol options such as stimulus order and viewing conditions. The following assumptions are made: (a) estimates of affective responses obtained on-site correlate highly with affective responses obtained from static color images, (b) different scaling models produce virtually identical results, and (c) results are highly reproducible over different presentation orders and venues. The claim for simulation validity is based on 40 years of research, covering 4,200 participants and 1,215 environments, indicating that (1) responses to static color images correlate highly and reproducibly with responses obtained in the field, (2) choice of medium (sketch, black and white photograph, computer simulation, etc.) has an effect on responses but this effect can be controlled by using the same medium for all stimuli in an experiment, and (3) it is probably prudent to validate each simulation medium before using it (Stamps, 2000, pp. 100-113). The claim for validity of static color images versus on-site evaluations was based on 185 environments (Stamps, 2000, p. 103) and was subsequently replicated with 470 environments (Palmer & Hoffman, 2001, p. 155). The claim for the equivalence of different scaling protocols is based on a review of eight articles with 1,150 stimuli indicating that the choice of scaling protocol made very little difference in the obtained results (Stamps, 2000, pp. 100-101). The claim that presentation protocols have had very minor effects on findings is supported by studies replicating findings obtained within the present laboratory (Stamps, 1992), between the present and another laboratory (Stamps & Nasar, 1997), independent


258


replication with 1,148 participants (Feimer, 1984), and between the present laboratory and results obtained from a review covering 19,000 participants from 23 countries and more than 3,200 environments (Stamps, 1999).

Statistical Protocols Analyses of variance, covariance, or regression for repeated measures data were done following Cohen and Cohen (1993, Chapter 11). Emphasis on contrasts and reporting results in terms of effect sizes such as standardized mean differences (d) or correlations (r) was done following Rosenthal and Rosnow (1991). Mathematically, d and r are equivalent and can readily calculated from each other (see Rosenthal & Rosnow (1991) for the equations) but sometimes the results are easier to understand in one of the measures. Sample sizes were calculated using power analysis following Cohen (1988). Summarizing findings over multiple studies was done with meta-analyses following Hedges and Olkin (1985). A free plug-in for doing meta-analysis in Excel is given in Bax, Yu, Ikeda, Tsuruta, and Moons (2006, 2008). These considerations constitute what has been called the effect size paradigm for statistical quality control in basic research. Descriptions of how to execute this paradigm and examples taken from the environment and behavior literature are given in Stamps (2002).

Experiment 1: Height, Area, and Elongation Stimuli and Experimental Design The venue for this experiment was the traditional Japanese street used in the previous study of recess setbacks but with the added dimension of height of buildings. Buildings were created with heights of 1, 2, or 3 stories, or, in terms of meters to eaves, 3, 6.7, and 10 m. Twelve sites were selected from the previous study using the criteria that (a) the full range of variables would be replicated, and (b) the number of scenes would be kept low because of the costs of using dynamic simulations. Twelve streets were created, using a factorial experimental design of setback area (2) by elongation of setback (2) by height (3), for a total of 12 streets. The site plans and screen shots of four scenes are shown in Figure 1.

Participants and Sample Size The required number of participants was calculated with power analysis. Relevant previous findings were that horizontal area and perceived spaciousness correlated at r = .73 and elongation correlated with perceived spaciousness at r = –.27. The smaller value was used in this study. A


259

Stamps

Figure 1. Site plans, screen shots, and plots of mean spaciousness for experiment 1

correlation of –.27 translates into about 7% of variance. For a = .05, three intended tests (area, elongation, height), a target effect size of 13%, power = .80, and a repeated measures experimental design with 12 stimuli, the minimum number of participants was 15. In the event, 25 participants were recruited by a professional survey research firm from the adult population of


260


a major city in the United States. There were 12 males and 13 females. Politically there were 8 liberals, 9 moderates, and 8 conservatives. The mean and standard deviation of age were 44.7 and 16.3 years. Occupations ranged from hair stylist to college professor.

Task Virtual reality models were shown on a laptop computer using a custom computer program. The screen measured 337 mm by 208 mm. Luminosity was 150 lux. Participants sat approximately 400 mm from the screen in a room with an ambient light level of 150 lux. The computer program had two parts. The first part was a demonstration that showed the participants how to use the controls. After completing the demonstration, participants viewed the main part. In this part, the first screen stated what judgment was requested (“Please rate the following pictures on the criterion of how not spacious (1) or spacious (8) they appear.”), along with two images showing the extreme conditions in the experimental design. Then each stimulus was shown with a row of buttons on the bottom. The buttons were numbered from 1 to 8. When a button was pressed, an “OK” button appeared. When the “OK” button was pressed, the next stimulus was shown. Participants could change their minds and press other numbered buttons until they pressed the “OK” button. Participants were allowed to take as much time as they needed.

Results and Discussion The analysis of variance is listed in Table 2. All three environmental properties had substantial effects on impressions of spaciousness. Height had the strongest effect (8.7% of variance, p = 5e-12), followed by horizontal area (6.2%, p = 4e-9), and setback ratio (6.6%, p = 1e-9). More detailed findings are shown in Table 3. Table 3 lists the means for each level of each factor, the standardized mean contrasts (d) and correlations (r) between levels of factors. The values of d and r are listed because they are necessary to synthesize the findings of this study with the other studies in the literature. Thus, the efficacy of area of street, over the range of 194 to 258 m2, was r = .31, with larger areas appearing more spacious (finding #1, or F1). The efficacy of height of building, over the range of 3 to 10 m, on impressions of spaciousness was r = .49, with lower buildings making the street seem more spacious (F2). For elongation, indicated by the elongation of the setback, the efficacy was r = –.32 (F3), with long, shallow setbacks making the street seem more spacious. For quick comparisons, plots of the contrasts are also shown at the bottom of Figure 1.


261

Stamps

Table 3. Contrasts for Perceived Spaciousness, Experiment 1 Factor

Level

Stimuli

Area (m2) Height (m) Setback ratio

258 194 3.0 6.7 6.7 10.0 3.0 10.0 1:4 1:1

DEFJKL ABCGHI GADJ HBEK HBEK ICFL GADJ ICFL GHIDEF ABCJKL

M

d

r

F(1,264)

4.56 .65 .31 37.72 3.72 5.04 .95 .43 45.34 3.81 3.81 .18 .09 1.72 3.53 5.04 1.13 .49 64.76 3.57 4.57 .67 .32 33.76 3.71

p 3e-9 1e-10 0.19 3e-14 2e-8

Experiment 2: Height, Area, and Elongation Stimuli and Experimental Design In this study, the venue was changed from outdoor to indoor. The environments were very simple, plain, convex spaces with a wood floor, gray walls with vertical joints 1.22 m on center, and a white ceiling. Three properties of the environment were created: horizontal area, elongation, and height. The levels of area were 12, 16, and 20 m2. The levels of elongation were width-to-length ratios of 1:1, 1:2, and 1:9, or, in verbal terms, a square, a rectangle, and a corridor. Heights were 2.44 and 3.66 m. These sizes were chosen because, in local units, they could be made with 4 × 8 foot or 4 × 12 foot plywood panels. The experimental design was a factorial of area (3) by elongation (3) by height (2), for a total of 18 rooms. Figure 2 shows the floor plans and screen shots of four rooms.

Participants and Sample Size The same target effect size was used in studies 1 and 2, but, with more stimuli, fewer participants were needed. For a = .05, three tests, power at .80, and 18 stimuli, the minimum number of participants fell to 6. Twenty-four participants were recruited by the professional survey research firm. The participant sample was balanced for sex and political affiliation. Ages ranged from 22 to 72 years with a mean of 45.1 and standard deviation of 13.5 years. Occupations ranged from unemployed to business owner.

Task The same task was used in studies 1 and 2. Downloaded from eab.sagepub.com at UNIV CALIFORNIA DAVIS on August 30, 2011

262


Figure 2. Site plans, screen shots, and plots of mean spaciousness for experiment 2.

Note: Rooms A, B, C, D, E, F, G, H, and I were 3.04 m high. Rooms J, K, L, M, N, O, P, Q, and R were 2.43 m high.

Results and Discussion The analysis of variance is listed in Table 4. The factor with the largest effect on perceived spaciousness was elongation (10.3% of variance, p = 3e-17). The effect of area was less (4.3%, p = 6e-9), and the effect of height was undetectable (0% of variance, p = .39). In terms of efficacies (Table 5), the correlation of elongation with perceived spaciousness was r = .40, with all the effect being attributable to the long, narrow corridors. Long, narrow


263

Stamps

Table 4. Repeated Measures Analysis of Variance, Experiment 2 Source Participants (Within) Area (m2) Height (m) Elongation Residual Total

SS

df

431.05 (1004.30) 68.05 1.44 150.33 784.46 1435.39

MS

F

% Variance

p

23 408 1 68.05 35.26 6e-9 1 1.44 0.75 .39 1 150.33 77.89 3e-17 405 1.93 431

4.6% 0.0% 10.3%

Table 5. Contrasts for Perceived Spaciousness, Experiment 2 Factor

Level

Stimuli

Area (m2) Elongation Height (m)

12 16 16 20 12 20 1:1 1:2 1:2 1:9 1:1 1:9 2.44 3.66

ABCJKL DEFMNO DEFMNO GHIPQR ABCJKL GHIPQR ADGJMP BEHKNQ BEHKNQ CFILOR ADGJMP CFILOR ABCDEFGHI JKLMNOPQR

M

d

r

F(1, 391)

3.06 3.52 .33 .16 7.77 3.52 4.03 .37 .18 9.77 3.06 4.03 .70 .33 34.98 3.92 3.95 0.0 0.0 0.0 3.95 .90 .41 58.48 2.69 3.92 .88 .40 55.92 2.69 3.48 3.60 .08 .04 0.74

p .005 .002 7e-9 1.0 1e-13 5e-13 .39

spaces were perceived as being much less spacious than square or rectangular spaces of the same area. There was no detectable difference in spaciousness between the square and rectangular rooms (r = .0, p = 1.0).

Experiment 3: Height, Area, Elongation, and Color Stimuli and Experimental Design This study also used rooms as stimuli. Three criteria were used to design these rooms. First, the volume was held constant at 125 m3. Thus, this experiment could address the question of how, given a program calling for high density housing, could that building’s volume be subdivided to make each unit seem as spacious as Downloaded from eab.sagepub.com at UNIV CALIFORNIA DAVIS on August 30, 2011

264


possible? Should the floor area be increased and the ceiling lowered, or vice versa? Should the rooms be compact or elongated? Would changing the wall color compensate for a lack of square footage or ceiling height? A second criterion was that the ratio among levels of factors should be such that the largest level would be twice the size of the lowest. With four levels for each factor, the required ratio between levels was 3 2 , which might explain some of the otherwise incomprehensible dimensions. Third, the overall sizes of both the smallest and largest rooms had to be reasonable. The result consisted of the floor plans shown in Figure 3. Levels of horizontal area were 49, 38.9, 30.9, and 24.5 m2. Heights were 2.5, 3.15, 3.96, and 5.0 m. Elongation ratio’s were 1:1, 1:1.26, 1:1.57, and 1:2. Because luminosity has been shown to influence perceived spaciousness, it had to be controlled. This was done using colors created under the CIELAB color space with constant luminosity, with constant reflectivity of the walls, and a constant level of source lighting (200 cd/m2 of floor area). For details on color spaces, see Wyszecki and Styles (2000) or Pascale (2003). Essentially, this implies that the overall amount of light (e.g., the luminosity) was controlled, so the factor of color in this experiment was variation in hue and saturation. CIELAB coordinates for each color, as given in Photoshop, were 90, 0, –40 for blue, 90, –40, 0 for green, 90,0, 40 for yellow, and 90, 40, 0 for pink. A person was rendered in each room to help participants evaluate scale, and each room contained pictures done by the author. The content of the pictures varied widely so descriptions of them would be difficult, but, considering who the painter was, they are, of course, all works of Art. The experimental design was a Graeco-Latin square (Cochran & Cox, 1957, p. 146) of horizontal area (4), elongation (4), height (4), and color (4), with a total of 16 rooms. Screen shots of four rooms are also shown in Figure 3.

Participants and Sample Size Since experiments 1 and 2 were done before experiment 3, the estimate of the target effect size could be based on the data listed in Table 1 plus the findings for experiment 1. For horizontal area, the overall correlation with perceived spaciousness was r = .63 (≈39% of variance). For height, the correlation was r = –.49 (24% of variance). For elongation in convex spaces, the correlation was r = –.27 (7%). No prior effect size was available for color, so we used a default value of Cohen’s medium (13% of variance). The minimum of these effect sizes was 7% of variance, so, with a target of 7%, a = .05, power = .89, four tests, and 16 stimuli, the minimum number of participants was 14. Seventeen participants were recruited by the professional survey research firm. There were nine males and eight females. The mean age was 39 years with a standard deviation of 14 years. There were 6 political liberals, 5 moderates, 5 conservatives, and one person who did not state political affiliation. Occupations ranged from writer to commercial real estate. Downloaded from eab.sagepub.com at UNIV CALIFORNIA DAVIS on August 30, 2011

265

Stamps

Figure 3. Site plans and screen shots for experiment 3

Task The same task was used in all three studies.

Results and Discussion The analysis of variance is listed in Table 6. Horizontal area again had the strongest effect on perceived spaciousness (12% of variance, p = 2e-4). Downloaded from eab.sagepub.com at UNIV CALIFORNIA DAVIS on August 30, 2011

266


Figure 4. Plots of mean spaciousness for experiment 3

Effects of elongation and height were an order of magnitude less (2% and 1% of variance, p’s = .002 and .04). Color had an effect two orders of magnitude less than area (0.1% of variance, p = .61). Contrasts are listed in Table 7 and are shown in Figure 4. Thus, the effect of horizontal area, over the range of 24.5 to 49 m2, was F7: r = .54, with larger horizontal area leading to an increase in perceived spaciousness. For height, over the range of 2.5 to 5 m, was F8: r = –.26, with the rooms with lower ceilings being judged as more spacious than the rooms with higher ceilings. For elongation, over the range of 1:1 to 1:2, the effect was F9: r = .14, with no detectable difference between these levels of elongation. Finally, for color, no detectable differences were found for effects of color on impressions of spaciousness. The largest contrast was between yellow and pink, with F10: r = .11.

Discussion Overall, the data reported in this article increased the literature on how the environment influences perceived spaciousness to 21 experiments, 366 environments, and 620 participants. The database for effects of horizontal area, height, elongation, and color are listed in Table 8. Data for each factor will be


267

Stamps

Table 6. Repeated Measures Analysis of Variance, Experiment 3 Source Participants (Within) Area Elongation Height Color Residual Total

SS 263.00 586.75 104.51 18.76 7.92 3.30 492.26 849.75

df

MS

F

p

16 255 1 104.51 57.74 8e-14 1 18.76 10.36 .002 1 7.92 4.37 .04 3 1.10 0.60 .61 249 1.81 271

% Variance

12% 2% 1% 0.1%

discussed in turn. Inferences regarding the identification of new contributions to the literature, the effects of the new data on the previously existing data, and estimating the amount of work needed for future work are based on the effect size paradigm for statistical quality control in basic research as described in Stamps (1996, 1997a, 1997b, 2002).

Horizontal Area New findings for the effect of horizontal area on perceived spaciousness were reported as F1, F4, and F7. In addition to the three new effect sizes, the current work also contributed to the collective body of knowledge by expanding the relevant sample size (n = 129 before the work and n = 175 after the work) and the range of venues over which the effect size was reported (new venues of streets with buildings of different heights; new proportions of rooms). Prior to the work reported in this article, the .05 ci for this effect was [.54, .73]. After the current work, this ci shrank to [.51, .68]. Clearly, this was not much of a change. The reason is that the effect of horizontal area on perceived spaciousness was already pretty firmly established in previous work, so additional work is most unlikely to make a contribution on this particular point. An estimate of how much additional work would make a change in the collective body of knowledge is provided in the last column of Table 8. This column is labeled “nover”. We use nover as an index of how solid the collective evidence is on a particular point. High positive values of nover indicate very solid evidence that would be hard to change with future work. More technically, nover is the amount of additional data that would be required to change the significance of the collective finding. If nover is positive, it would be the amount of work, reporting a finding of r = 0, that would make the collective estimate indistinguishable from random noise (e.g., p > .05). Higher values


268


Table 7. Contrasts for Perceived Spaciousness, Experiment 3 Factor Area (m2) Height (m) Elongation ratio Color

Level

Stimuli

49.0 38.9 38.9 30.9 30.9 24.5 49.0 24.5 2.50 3.15 3.15 3.96 3.96 5.00 2.50 5.00 1:100 1:1.26 1:1.26 1:1.587 1:1.587 1:2.00 1:1.00 1:2.00 Blue Yellow Yellow Pink Pink Green Blue Green

AEIM BFJN BFJN CGKO CGKO DHLP AEIM DHLP AFKP BELO BELO CHIN CHIN DGJM AFKP DGJM ABCD EFGH EFGH IJKL IJKL MNOP ABCD MNOP AHJO BGIP BFIP CFLM CFLM DEKN AHJO DEKN

M

d

r

F(1, 240)

4.57 .63 .30 13.77 3.73 3.73 .33 .16 3.81 3.29 3.29 .30 .15 3.09 2.90 4.57 1.27 .54 55.07 2.90 4.03 .23 .12 1.87 3.72 5.72 .22 .11 1.70 3.42 3.42 .08 .04 0.21 3.32 4.03 .53 .26 9.76 3.32 4.09 .31 .15 3.32 3.68 3.68 .49 .24 8.20 3.03 3.03 .51 .24 8.20 3.70 3.03 .29 .14 2.86 3.70 3.59 .06 .03 0.10 3.51 3.51 3.81 .22 .11 1.70 3.81 .17 .08 0.95 3.59 3.59 .00 .00 1.00 3.58

p 2e-4 .05 .08 2e-12 .17 .20 .64 .002 .07 .004 .003 .09 .75 .20 .32 .32

mean the effect is already established and further work may not be the best use of one’s research resources. The effect of horizontal area on perceived spaciousness would appear to fall into this category, joining claims such as the validity of static color simulations, the interchangeability of alternate scaling methods, and the effect of experimental viewing venues for evaluating affective responses to environments.


269

Stamps

Table 8. Relationships Among Past, Present, and Future Work on Spaciousness Factor Horizontal area Height Elongation— Concave Elongation— Convex Color

Source

n

r

Sn

rˆ

.05 ci

nover

Previous worka 129 .65 .54, .73 F1 Experiment 1 12 .31 F4 Experiment 2 18 .35 F7 Experiment 3 16 .54 Summary 175 .60 .51, .68 5,203 Previous work 0 0 -1.00, 1.00 F2 Experiment 1 12 -.49 F5 Experiment 2 18 .04 F8 Experiment 3 16 -.26 Summary 46 -.22 -.46, .04 -14 Previous work 30 .24 -.09, .52 F3 Experiment 1 12 .32 Summary 42 .26 -.001, .49 -2 Previous work 7 -.27 -.78, .28 F6 Experiment 2 18 -.40 F9 Experiment 3 16 .14 Summary 41 -.18 -.44, .11 -.52 Previous work 0 0 -1.00, 1.00 F10 Experiment 3 16 .14 16 14 -.30, .53 -126

a. Please see Table 1 for details on previous work.

Height For height, we were unable to find previously reported effect sizes, so the .05 ci before the current work was [–1.0, 1.0]. After the three new findings (F2, F5, F8), the sample size was increased from 0 to 46 environments and the .05 ci was upgraded to [–.46, .04]. For the effect of boundary height on perceived spaciousness, the ci does not currently include zero, or, in terms of p levels, p > .05. In this case, nover will be negative, and, technically, is the amount of data, reporting the current estimated effect size, that would make the collective finding significant (p < .05). A negative nover that is close to 0 means that only a small amount of work would be needed to establish a relationship beyond random chance. For the effect of height on perceived spaciousness, nover is –14, indicating that an experiment with nstim = 14 could make a useful contribution to the literature. The relationship of nover to future research can be seen by comparing how much work and how much benefit would be involved in choosing whether to work next on horizontal area or on boundary height. For horizontal area, a study of 5,203 stimuli would be needed; for boundary height, that number drops to 12 stimuli. If one’s resources are


270


limited, it would seem that future work on boundary height would be more rewarding than future work on horizontal area.

Elongation The relationship of elongation to perceived spaciousness was split into two variations because, although the same measurement was made (length divided by width), the prior findings were divergent. For the street scenes, elongation was measured for one part of a concave space. For rooms, elongation was measured for the whole, concave shape. New findings (F3 for the streets; F6 and F9 for rooms) supported the previous work with streets with long, shallow alcoves would be more spacious than streets with short, deep alcoves and rooms with compact shapes would be more spacious than long, narrow spaces such as corridors. However, with both nover’s being negative, it would be premature to make the call on these points. Establishing the alcove effect would require less work than establishing the room effect (nover’s = –2 and –52 respectively, for anyone contemplating such future work.

Color For color, the range of stimuli was extended to go beyond the effect of amount of light. The relevant a priori finding was that amount of light has nearly the same effect on perceived spaciousness as does horizontal floor area (r’s of .65 and .64). But color, at least as perceived by people, also has two other properties. Readers may be accustomed to thinking of color in terms of hue, value, and saturation; for analytical work, the CIELAB system is generally more useful. The work reported in this article (F10) used rigorous controls in both the technical specification of color and in the creation of the environments so that the effect of overall amount of light was eliminated. The result: sans amount of light, the effect of color on perceived spaciousness was small (r = .16) and will take a study with an additional 126 environments to distinguish this effect from chance. In conclusion, therefore, we suggest that since sustainable design often implies higher densities and insufficient space can be a strong ambient stressor, there are good reasons to refine our understanding of how perceived spaciousness can be mitigated through environmental design. To paraphrase numerous previous authors, more work is needed. Engage!

Summary This article reports findings from 3 experiments, covering 46 environments and 66 participants, on how strongly four properties of the physical environment


271

Stamps

influence perceived enclosure. The properties were horizontal area, boundary height, elongation, and color. Ten original findings were reported. Overall, horizontal area had the strongest effect on perceived spaciousness (r = .60; more floor area increases perceived spaciousness), followed by height (r = –.22; lower boundaries increase perceived spaciousness). The effect of color on perceived spaciousness, when amount of light is controlled was much smaller (r = .14). Findings for elongation were different for concave and convex spaces (r’s of –.22 and +.26). Quantitative syntheses of the current work with previous work are presented, as is numerical guidance for cost-effective future work. Declaration of Conflicting Interests The author(s) declared no potential conflicts of interests with respect to the authorship and/or publication of this article.

Funding The author(s) received no financial support for the research and/or authorship of this article.

References American Psychological Association. (2001). Publication manual of the American Psychological Association (5th ed.). Washington, DC: Author. Bax, L., Yu, L. M., Ikeda, N., Tsuruta, H., & Moons, K. G. M. (2006). Development and validation of MIX: Comprehensive free software for meta-analysis of causal research data. BMC Medical Research Methodology, 6, 50. Bax, L., Yu, L. M., Ikeda, N., Tsuruta, H., & Moons, K. G. M. (2008). MIX: Comprehensive free software for meta-analysis of causal research data (Version 1.7). Retrieved November 23, 2009, from http://mix-for-meta-analysis.info Benedikt, M. L., & Burnham, C. A. (1985). Perceiving architectural space: From optic arrays to isovists. In W. H. Warren & R. E. Shaw (Eds.), Persistence and change (pp. 103-114). Hillsdale, NJ: Lawrence Erlbaum. Calthorpe, P., & Fulton, W. (2001).The regional city. Washington, DC: Island Press. Cochran, W. G., & Cox, G. M. (1957). Experimental designs. New York: John Wiley. Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum. Cohen, J., & Cohen, P. (1993). Applied regression/correlation analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum. Durston, D., & Mizuno, K. (2002). Kyoto: Seven paths to the heart of the city. Tokyo: Kodansha International. Feimer, N. R. (1984). Environmental perception: The effects of media, evaluative context, and observer sample. Journal of Environmental Psychology, 4, 61-80.


272


Franz, G., von der Heyde, M., & Bülthoff, H. (2003). An empirical approach to the experience of architectural space in VR. Retrieved December 23, 2004, from http://www.kyb.tuebingen.mpg.de/publications/pdfs/pdf2232.pdf Franz, G., & Wiener, J. M. (2005). Exploring isovist-based correlates of spatial behavior and experience. Lecture Notes in Artificial Intelligence, 3343, 42-57. Gärling, T. (1970a). Studies in visual perception of architectural spaces and rooms III: A relation between judged depth and size of space. Scandinavian Journal of Psychology, 11, 124-131. Gärling, T. (1970b). Studies in visual perception of architectural spaces and rooms IV: The relation of judged depth to judged size of space under different viewing conditions. Scandinavian Journal of Psychology, 11, 133-145. Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis. Orlando, FL: Academic Press. Imamoglu, V. (1973). The effect of furniture on the subjective evaluation of spaciousness and estimation of size of rooms. In R. Küller (Ed.), Architectural psychology (pp. 314-352). Stroudsburg, PA: Dowden, Hutchinson & Ross. International City Management Association. (2003). Getting to smart growth II: 100 more policies for implementation. Washington, DC: International City Management Association. Inui, M., & Miyata, T. (1973). Spaciousness in interiors. Lighting Research and Technology, 5, 103-111. Ishikawa, T., Okabe, A., Sadahiro, Y., & Kakumoto, S. (1998). An experimental analysis of the perceptions of the area using 3-D stereo dynamic graphics. Environment and Behavior, 30, 216-234. Kirschbaum, C. F., & Tonello, G. (1997). Visual appearance of office lighting. Right Light, 1, 143-148. Martyniuk, O., Flynn, J. E., Spencer, T. J., & Hendrick, C. (1973). Effect of environmental lighting on impression and behavior. In R. Küller (Ed.), Architectural psychology (pp. 51-63). Stroudsburg, PA: Dowden, Hutchinson & Ross. Meier, R. L. (1975). Planning for an urban world: The design of resource-conserving cities. Cambridge, MA: MIT Press. Palmer, J. F., & Hoffman, R. E. (2001). Rating reliability and representation validity in scenic landscape assessments. Landscape and Urban Planning, 54, 149-161. Pascale, D. (2003). A review of color spaces from xyY to R’G’B’. Montreal: The BabelColor Company. Plutschow, H. E. (1979). Introducing Kyoto. Tokyo: Kodansha International. Rosenthal, R., & Rosnow, R. L. (1991). Essentials of behavioral research: Methods and data analysis. New York: McGraw-Hill. Sadalla, E. K., & Oxley, D. (1984). The perception of room size: The rectangularity illusion. Environment and Behavior, 16, 291-306.


273

Stamps

Stamps, A. E. (1992). Bootstrap investigation of respondent sample size for environmental preference. Perceptual and Motor Skills, 75, 220-222. Stamps, A. E. (1996). Effect sizes as a lingua franca of environmental aesthetics. In J. L. Nasar & B. Brown (Eds.), Public and private spaces: EDRA 27 (pp. 151-162). Salt Lake City, UT: Environmental Design Research Association. Stamps, A. E. (1997a). Advances in peer review research: An introduction. Science and Engineering Ethics, 3, 3-10. Stamps, A. E. (1997b). Using a dialectical scientific brief in peer review. Science and Engineering Ethics, 3, 85-98. Stamps, A. E. (1999). Demographic effects in environmental preferences: A metaanalysis. Journal of Planning Literature, 14, 155-175. Stamps, A. E. (2000). Psychology and the aesthetics of the built environment. Norwell, MA: Kluwer Academic. Stamps, A. E. (2002). Meta analysis. In R. Bechtel & A. Churchman (Eds.), The handbook of environmental psychology (pp. 222-232). New York: John Wiley. Stamps, A. E. (2007). Evaluating spaciousness in static and dynamic media. Design Studies, 28, 535-557. Stamps, A. E. (2009). On shape and spaciousness. Environment and Behavior, 41, 526-548. Stamps, A. E. (2010). Effects of permeability on perceived enclosure and spaciousness. Environment and Behavior, 42, 864-886. Stamps, A. E., & Krishnan, V. V. (2006). Spaciousness and boundary roughness. Environment and Behavior, 38, 841-872. Stamps, A. E., & Nasar, J. L. (1997). Design review and public preferences: Effects of geographical location, public consensus, sensation seeking, and architectural styles. Journal of Environmental Psychology, 17, 11-32. Talen, E. (2003). Measuring urbanism: Issues in smart growth research. Journal of Urban Design, 8, 195-315. Williams, K., Burton, E., & Jenks, M. (2000). Achieving urban form. London: E & F Spon. Wyszecki, G., & Styles, W. S. (2000). Color science: Concepts and methods, quantitative data and formulae (2nd ed.). New York: John Wiley.

Bio Arthur E. Stamps III extracted his PhD from U.C. Berkeley in 1980. The work focused on statistics (done in the psychology department), philosophy (in the Graduate School of Business), and futures research (Department of City Planning). He has published more than 90 articles, does a substantial amount of peer review on submissions relating to environment and behavior, and serves on the editorial boards of several major journals. Current research focuses on cognition and affect. He works at the Institute of Environmental Quality in San Francisco.


Environment and Behavior

Environment and Behavior

Suggest Documents