You are on page 1of 26

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/258132464

Effects of Multiple Boundaries on Perceived Spaciousness and Enclosure

Article in Environment and Behavior · October 2013


DOI: 10.1177/0013916512446808

CITATIONS READS

35 587

1 author:

Arthur Stamps

110 PUBLICATIONS 3,601 CITATIONS

SEE PROFILE

All content following this page was uploaded by Arthur Stamps on 12 February 2016.

The user has requested enhancement of the downloaded file.


446808
46808StampsEnvironment and Behavior
© 2012 SAGE Publications

Reprints and permissions:


EAB45710.1177/00139165124

sagepub.com/journalsPermissions.nav

Article
Environment and Behavior

Effects of Multiple
45(7) 851­–875
© The Author(s) 2012
Reprints and permissions:
Boundaries on sagepub.com/journalsPermissions.nav
DOI: 10.1177/0013916512446808
Perceived eab.sagepub.com

Spaciousness
and Enclosure

Arthur E. Stamps III1

Abstract
This article reports seven new, original findings, based on 4 experiments, 56
environmental scenes, and 71 participants, on how the factors of area over
which one could walk (boundary height, boundary porosity, and boundary
proximity) influence perceived spaciousness or enclosure. Perceived spa-
ciousness was most strongly related by the area over which one could walk.
Enclosure was most strongly related to boundary height. Proximate bound-
aries had stronger effects on perceived enclosure than did distal boundaries.
Results were highly reproducible over vastly different environmental venues,
indicating that the research protocols worked very well.

Keywords
permeability theory, enclosure, spaciousness, locomotive permeability,
perceptual permeability

The primary question addressed in this article is “How do properties of mul-


tiple boundaries influence perceived spaciousness or perceived enclosure?”
The motivation for investigating this question is theoretical. The theory in

1
Institute of Environmental Quality, San Francisco, CA, USA

Corresponding Author:
Arthur E. Stamps III, Institute of Environmental Quality, 290 Rutledge Street, San Francisco,
CA 94110, USA.
Email: artstamps@comcast.net
852 Environment and Behavior 45(7)

question is the permeability theory, which, in a nutshell, proposes the


following:

1. The most important function an environment can provide for its


inhabitants is safety.
2. Among other things, safety depends on how far and how easily one
can sense or move through an environment.
3. Ability to sense or move through something is called permeability,
from which comes the name of this theory.
4. Therefore, data are needed regarding how the physical properties
of environments mitigate either perceptual or locomotive perme-
ability.

The research on permeability is based on the concept that judgments of


enclosure and spaciousness are proxies for judgments of potential safety or
danger. For example, being enclosed with respect to vision implies that
enemies could be close but not visible. That would be dangerous, so spaces
surrounded by regions not permeable by vision should be judged as more
enclosed than spaces more open to vision. Horizontal area indicates range of
possible locomotion, which means more ability to get away if necessary, and
so, by the enclosure/safety relationship, more horizontal area should be per-
ceived as being more spacious and less enclosed.
More expansive presentations of permeability theory can be found in pre-
vious articles (Stamps, 2010a, 2011). As a first approximation, environments
can mitigate permeability in four ways: (a) neither type of permeability is
possible, (b) both types of permeability are possible, (c) visual permeability
is possible but not locomotive permeability, and (d) locomotive permeability
is possible but not visual permeability. This classification generates four
zones of permeability (both, visual but not locomotive, locomotive but not
visual, and neither), with iconic examples being the plains of Kansas, a win-
dow, fog, and a brick wall.
A substantial amount of data has been reported on how properties of envi-
ronments can influence judgments of enclosure or spaciousness. The litera-
ture on enclosure is reviewed in Stamps (2002). For spaciousness, reviews
are given in Duval, Charles, and Veitch (2002) and Stamps (2009). Specific
properties of the physical environment for which effect sizes have been
reported with respect to perceived enclosure include (a) distance, (b) gaps,
(c) height, (d) horizontal area, (e) light, (f) location with respect to observer,
(g) material, (h) permeability, (i) shape, and (j) solid walls. Environmental
properties for which effect sizes on perceived spaciousness have been
Stamps 853

reported include (a) boundary roughness, (b) height, (c) horizontal area,
(d) light, (e) occlusion, and (f) shape. However, one limitation of those stud-
ies is that the environments previously under inquiry all had a single bound-
ary, leading to the question of what happens to perceived spaciousness or
enclosure when there is more than one boundary?
There is also research on the connection between enclosure and safety.
Herzog and Miller (1998) had respondents rate 18 alleys and 18 nature scenes
for preference, mystery, danger, openness, and curvature, and reported that
rated danger correlated at r = −.45 with rated openness. Herzog and Chernick
(2000) investigated perceived tranquility and perceived danger in 48 urban
and natural settings. A correlation of −.72 was found between rated openness
and rated danger. Another study investigated perceived enclosure and per-
ceived threat in Greek villages (Stamps, 2005a). Three villages with different
densities were selected from a planning book (Doxiadis, 1968). The villages
were Pappas (80 inhabitants per hectare), Kapinski (110 inhabitants per hect-
are), and Larissa (150 inhabitants per hectare). Seven locations and angles of
view were chosen at random within each city, and renderings were made of
the architectural masses visible at those locations and angles. Altogether,
there were 21 stimuli. A total of 63 participants rated each stimulus on a scale
of 1 = extremely open to 8 = to extremely enclosed. Ratings were also pro-
vided on a scale of 1 = extremely safe to 8 = extremely threatening. Perceived
enclosure and perceived safety correlated at r = −.82. Four more relevant
studies were reported in Stamps (2005b). Data for all four studies were pro-
vided by 19 participants. The venue for the first study consisted of rooms.
The rooms measured 5.75 m wide × 5.75 m deep × 3.5 m high. The walls had
horizontal gaps. The experimental design was total gap height (0.5, 1.0, 1.5
m) by gap location (top and bottom, middle and top, bottom and middle), for
a factorial of gap height (3) × gap location (3) and a total of 9 stimuli. In this
experiment, judged enclosure and judged threat correlated at r = .67. The
venue for the next study consisted of plazas. The plazas measured 49 m × 49
m and had boundaries of buildings at three heights: two stories, four stories,
and six stories. In terms of meters, the boundary heights were 7.8, 13.4, and
19.8 m. In this study, enclosure and threat correlated at r = .55. The venue
for the third study was a circular colonnade with a diameter of 20 m. The
columns were designed to cross a factor of visual permeability (20%, 40%,
60%, or 80%) with gaps that would or would not permit locomotion by a
person (150, 900 mm). In this study, judged enclosure and judged safety were
not related (r = −.06). The venue for the fourth study was a 6 × 6 m garden.
The boundaries were of two heights: below or above eye level (1.25 and 2.25
m). The materials of the boundaries were either fences or plants. The
854 Environment and Behavior 45(7)

densities of the boundaries were open or dense. The experimental design was
height (2) × material (2) × density (2) for a total of 8 stimuli. In this experi-
ment, the relationship between judged enclosure and judged safety was r =
−.12. Finally, Alkhresheh (2007) investigated the variables of height and
width of commercial streets. The venue consisted of synthetic streetscapes
based on buildings in Gainesville, Florida. Images in the original document
(Alkhresheh, 2007) showed office or residential occupancy over first floor
commercial occupancy. Widths of streets ranged from 6 to 40 m. Building
heights ranged from 6 to 238 m or 2 to 80 stories. Altogether, there were
42 streets, but the experimental design was not balanced, and so clear conclu-
sions could not be made for all 42 streets. However, a subset of 9 streets was
found that did have a balanced experimental design of building height (6, 9,
and 12 m) by street width (6, 9, and 12 m). Eight-three participants provided
data for three responses: enclosure, safety, and comfort. For the 9 stimuli in
the balanced experimental design, rated enclosure was more strongly related
to street width than building height (rs of −.88 vs. .32). Safety was also more
strongly related to street width than building height (rs of .89 vs. −.12).
Enclosure and safety were negatively correlated (r = −.94), and judgments
of perceived safety and perceived comfort were virtually identical (r = .96).
Overall, the literature on enclosure and threat suggests that these two vari-
ables have been highly correlated in the venues of streets and rooms but not
gardens.

A Priori Hypotheses
Based on the material described above, the following questions were inves-
tigated:

Hypothesis 1: Which is more strongly related to perceived spacious-


ness: locomotive or visual permeability? The reasoning is that spa-
ciousness is more strongly related to locomotive permeability than
to visual permeability, and so the corresponding hypothesis was
that the amount of area through which one could walk (walking
area) would account for more variance than would boundary height
(height of boundary).
Hypothesis 2: Does boundary height influence perceived enclosure?
The reasoning is that boundary height reduces visual permeability,
and so more height should lead to more perceived enclosure.
Hypothesis 3: Does boundary permeability increase perceived enclo-
sure? Again, boundary permeability is visual permeability, and so
Stamps 855

the more porous the boundary, the less enclosed a space should
seem.
Hypothesis 4: Does boundary proximity influence perceived enclo-
sure? In this article, boundary proximity was expressed as whether
a boundary was nearer to or farther from an observer. The nearer
boundaries were labeled “proximate,” and the farther boundar-
ies were labeled “distal.” If perceived spaciousness and enclosure
are proxies for perceived safety, and safety depends on distance to
possible threat, then the proximate boundary should have stronger
effects on perceived spaciousness or perceived enclosure. This con-
cept might be called the “primacy of the proximate.” A correspond-
ing hypothesis for boundary height is that the effect on enclosure
of changing the proximate boundary height should be larger than
changing the distal boundary height, or, in other words, it matters if
the transition between boundary heights is proximate-to-distal ver-
sus distal-to-proximate.
Hypothesis 5: The concept of the primacy of the proximate can also be
applied to the property of boundary permeability. In this case, the
hypothesis becomes the effect of boundary permeability is greater if
the transition between boundary permeability is from proximate-to-
distal rather than distal-to-proximate (see Figure 6 for clarification
of this concept).

General Method
The experiments in this article used simple experimental protocols because
extensive previous research indicated that more elaborate protocols were not
necessary for factors such as scaling model, representation of affective
responses, presentation venue, simulation medium (Feimer, 1984), or meth-
ods of scaling (Oswald & Wu, 2010; Stamps, 2000). Static color images
were used because a review of 84 findings obtained from 967 environments
and 6,323 participants indicated that responses obtained on-site correlated at
r = .86 with static color images and r = .83 with dynamic presentations
(Stamps, 2010b). Responses were mostly visual preference, but there were
also data for 15 other behavioral measures as well as perceived spaciousness.
A review of demographic differences in environmental perception, covering
more than 19,000 participants from 21 countries, 3,821 scenes, and 12 demo-
graphic factors, indicated a very high consensus (r > .82) across all demo-
graphic groups except for children less than 12 years old versus adults,
political groups such as special interest groups versus nonspecial interest
856 Environment and Behavior 45(7)

groups, and designers versus nondesigners for avant garde projects


(Stamps, 1999).
Results are reported as repeated-measures analyses of variance (Cohen &
Cohen, 1993), contrasts (Rosenthal & Rosnow, 1985), and correlations
(Rosenthal & Rosnow, 1991). The analyses of variance are listed in the online
appendix. Contrast analysis consists of creating variables that precisely
express focused hypotheses and then calculating the effect size for each
hypothesis. Plots of the contrasts are given in the online appendix. Correlations
were reported to enable use of power calculations for determining the required
number of participants (Cohen, 1988) and also to enable the use of meta-
analysis in subsequent statistical syntheses of the present work (Hedges &
Olkin, 1985).

Experiment 1
Hypotheses

This experiment was designed to test H1 (spaciousness is more strongly


influenced by locomotive permeability than by visual permeability) and H2
(higher boundaries imply less perceived spaciousness).

Method
Stimuli and Experimental Design. To test for effects of multiple boundar-
ies, it is necessary to use a venue with multiple boundaries. One such venue
is a Japanese garden. Design elements in Japanese gardens include borrowed
scenery (Itoh, 1981), creating solid or porous boundaries (Itoh, 1981), step-
ping stones (Itoh, 1981), white sand (Itoh, 1981), and rocks (Keane, 1996).
The gardens in this experiment were based on Zen courtyards in Ryoanji,
Kyoto (Keane, 1996). Fences were solid bamboo as shown in Ohashi (1997,
plate 44). Stepping stones were created as shown in Keane (1996). The cen-
tral part of the gardens had rocks placed in white sand, and stepping stones
arranged to form a path of travel. The central part of the gardens had stone
curbs to delineate the boundary for locomotion. Beyond the curb were two
boundaries. The proximate boundary was a bamboo fence. The distal bound-
ary consisted of trees. There were three tree heights: 3, 9, and 12 m. There
were two gardens. Each garden was shown from two locations. One location
showed the garden as seen along its long axis. This produced the “axial”
view. The other location showed the garden as seen at an oblique angle. This
produced the “oblique” view. The area of the central part of the garden dif-
fered at each location. The area of the amount of the central part of the garden
Stamps 857

Figure 1. Site plans, experimental design, and stimuli for Experiment 1.

was used to create the variable of walking area. Details on how walking area
was calculated are given in the online appendix. Two sites were created for
each of the three heights and both types of view. The resulting experimental
design was distal boundary height (6, 9, 12 m) by walking area (115, 152, 156,
165 m2), for a total of 12 stimuli. Figure 1 shows the site plans (Site 1 and Site 2)
and the 12 stimuli.
858 Environment and Behavior 45(7)

Participants and Sample Size. The required number of participants


was calculated using power analysis (Cohen, 1988). The target effect
size was obtained from a previous study of perceived spaciousness in a
site with the same overall dimensions as the rock gardens (Stamps,
2009). In that study, amount of walking area accounted for 7.2% of vari-
ance in perceived spaciousness. For experiment-wise α = .05, dfh (the
number of planned tests) = 2, power = 0.80, and 12 stimuli, the minimum
required number of participants was 13. A total of 18 participants was
recruited by a professional survey research firm from the adult popula-
tion of a major city in the United States. The participant sample was
balanced for gender and political affiliation. Mean age was 44 years
(SD = 12).
Task. Images were shown on a laptop computer using a custom computer
program. The screen measured 337 mm × 208 mm. The screen resolution was
1,280 × 800 pixels. Participants sat approximately 400 mm from the screen in
a room with an ambient light level of 150 cd/m2. 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 show-
ing 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. Data were recorded when a button was pressed and accepted by
pressing an “OK” button.

Results and Discussion


A repeated-measures analysis of variance indicated that perceived spacious-
ness was more strongly related to walking area than to boundary height
(3.3% vs. 0.1%). The contrasts between the levels of distal boundary height
were not significant. For the contrast between 6 and 9 m, the means were
M6 = 4.75 and M9 = 4.67 with F(1, 187) = 0.13 and α = .71. For the levels of
9 and 12 m, the means were M9 = 4.67 and M12 = 4.48 with F(1, 187) = 0.62
and α = .43. For the factor of walking area, the means and statistics were
M115 = 4.17, M152 = 4.85, F(1,187) = 0.70, α = .5e−5; M152 = 4.85, M156 = 4.22,
F(1,187) = 5.66, α = .02; and M156 = 4.22, M163 = 5.29, F(1,187) = 16.48,
α = 7e−5. In terms of correlations with perceived spaciousness over n = 12
stimuli, the results were r = −.21 for distal boundary height and r = .66 for
walking area. The difference between r = −.21 and r = .66 was significant
(Z = 2.13, α = .02).
Stamps 859

Main points. Thus, these results suggest that increasing the height of trees
behind a solid fence will not make the area inside that fence appear to be
more spacious. In this garden, there was no borrowing of scenery. However,
the amount of area through which one could walk did have a detectable effect
on perceived spaciousness. Walking area had a stronger effect on perceived
spaciousness than did distal boundary height.

Experiment 2
Hypotheses

In the preceding experiment, the height of a porous distal boundary beyond


a solid proximate boundary had no detectable effect on perceived spacious-
ness. This experiment reversed the locations of the boundaries so that the
solid boundary was behind the porous boundary. In terms of design, the dif-
ference was plants behind fences as compared with fences behind plants. The
same hypotheses were addressed in Experiments 1 and 2 (locomotive perme-
ability and visual permeability). The same venue (Japanese rock gardens)
was used in Experiments 1 and 2.

Method
Stimuli and Experimental Design. Another set of 12 rock gardens was
created, but this time the distal boundaries were solid fences and the proxi-
mate boundaries were plants. There were two sites. Site 3 had the same walk-
ing area as did Site 1 in Experiment 1. Site 4 had the same walking area as did
Site 2 in Experiment 1. The experimental design was the same in both experi-
ments: distal boundary height (3) by walking area (4). The site plans and
stimuli are shown in Figure 2.
Participants and Sample Size. Results from Experiment 1 suggested a
target effect size of 3.3% of variance. For α = .05, dfh = 2, nstim (nstim
refers to the number of stimuli) = 12, and power = 0.80, 26 participants
were recommended. A total of 24 participants were recruited by a profes-
sional survey research firm from the adult population of a major city in the
United States. The participant sample was balanced for gender and political
affiliation. Mean age was 49 years (SD = 17).

Results and Discussion


The repeated-measures analysis of variance indicated, as predicted and as
found in Experiment 1, perceived spaciousness was more strongly related to
860 Environment and Behavior 45(7)

Figure 2. Site plans, experimental design, and stimuli for Experiment 2.

walking area than to boundary height (9.6% vs. 2.2%). For the factor of
walking area, the means and statistics were M115 = 4.53, M152 = 4.83,
F(1, 253) = 1.99, α = .10; M152 = 4.83, M156 = 5.18, F(1, 253) = 2.57, α = .07;
Stamps 861

and M156 = 5.18, M163 = 6.04, F(1, 253) =15.85, α = 4e−5. However, contrast
analysis indicated that the height of the distal boundary did influence per-
ceived spaciousness. The effect was not detectable for the difference between
6 and 9 m fences, M6 = 5.46, M9 = 5.20, F(1, 253) = 1.93, α = .17, but there
was a detectable effect between distal boundary heights of 9 and 12 m,
M9 =5.20, M12 = 4.78, F(1, 253) = 4.95, α = .03. In terms of correlations with
perceived spaciousness over n = 12 stimuli, the results were r = −.43 for
distal boundary height and r = .88 for walking area. The difference between
r = −.43 and r = .88 was significant (Z = 3.89, α = 1e−4).
Main points. Thus, the data from Experiment 2 replicated the effect found
in Experiment 1 regarding how perceived spaciousness is related to loco-
motive permeability. Walking area had a stronger effect on perceived spa-
ciousness than did distal boundary height. For height of distal boundary,
there was a difference between the results of the two experiments. When a
proximate boundary was solid, there was no detectable effect on spacious-
ness because of increasing the height of a porous distal boundary. However,
when the relative locations of solid and porous boundaries were reversed,
raising the height of the distal boundary did influence spaciousness. In
design terms, if the fence is closer, the height of the distant plants did not
matter, but if the plants were closer, then the height of the more distant
fence did matter. It would seem that if one wanted to borrow scenery, it
might be best to start with porous boundaries. In statistical terms, the cor-
relation of distal height and judged spaciousness was r = −.21 for a solid
proximate boundary and a porous distal boundary. However, when the
porous boundary was proximate and the distal boundary solid, the correla-
tion with spaciousness was much stronger at r = −.43. With ns of 12, the
difference between these two correlations did not reach the conventional
.05 value (Z = 0.53, α = .70), suggesting that a useful next step would be to
execute additional experiments in which the factors of boundary location,
boundary porosity, and boundary height were fully crossed. More specifi-
cally, the findings from Experiments 1 and 2 suggest a new hypothesis as
follows:

Hypothesis 6: Does the visual permeability of a near (proximate)


boundary change the effects on spaciousness or enclosure of a distal
boundary? The reasoning is that because the proximate boundary is
visually permeable, the distal boundary should be easier to see, so
the hypothesis is that the more permeable the proximate boundary,
the stronger the effect of a distal boundary should be.
862 Environment and Behavior 45(7)

Experiment 3
In this experiment, the factors of boundary location, boundary porosity, and
boundary height were fully crossed. The venue was an interior room using
design features from traditional Islamic architecture (Ettinghausen, Grabar,
& Jenkins-Madina, 2001; Fletcher, 1963; Hillenbrand, 1994). Like the
Japanese gardens, traditional Islamic architecture often exhibits multiple
boundaries with differing degrees of permeability (Hillenbrand, 1994).
Iconic examples include the Dome of the Rock (Ettinghausen et al., 2001),
the Alhambra (Hillenbrand, 1994), and the Taj Mahal (Fletcher, 1963). Some
of the semipermeable boundaries, such as colonnades, can be attributed to
technical issues of limited roof span or the need to ventilate large, open inte-
rior spaces. However, other semipermeable boundaries were created to
enhance security. Window screens, for example, were designed to protect
inhabitants from the vision of strangers (Ettinghausen et al., 2001), and often
were composed of elaborate, delicate geometrical forms (Fletcher, 1963).
Another example of a permeable boundary is a separate square enclosure
within a mosque that is composed of open lattice work. This is called a maq-
sura (Hillenbrand, 1994). The presence of this design element appears to
have been motivated by the facts that two of the first four caliphs were assas-
sinated during prayer, but it was still necessary for the caliphs to be visible
to their people. In other words, a design feature was needed that would per-
mit visual permeability but not locomotive permeability. It would seem that
the traditional Islamic architects were very early adopters of the permeability
theory.

Hypotheses
Hypotheses tested included H2 (boundary height) H3 (boundary porosity),
H4 (the interaction of proximity and boundary height), H5 (the interaction of
proximity and boundary permeability), and H6 (the effect of a distal bound-
ary is stronger if the proximate boundary is porous rather than solid). H6 was
based on the data from the first two experiments, so successful tests of H6
indicate replication of previous findings.

Method
Stimuli and Experimental Design. The rooms in this experiment were cre-
ated within a 12 m wide × 12 m long × 8 m high envelope. The walls were
made of white marble and embossed with a frieze derived from a traditional
Moroccan window screen. The floor was a turquoise blue carpet. Golden
Stamps 863

Figure 3. Floor plan and types of boundaries in Experiment 3.

marble was used to construct pony walls of different heights, porosity, and
location. In each room, there were two boundaries: a proximate boundary and
a distal boundary. Each boundary had two properties. The properties were
height (1.5, 3.0 m) and porosity (the Moroccan screen, the screen with a solid
back). Thus, there were four types of individual boundaries (low porous, low
solid, high porous, and high solid). The floor plan and types of boundaries are
shown in Figure 3. Each room had two boundaries, producing a third prop-
erty, proximity. This property had two levels: proximate and distal. The
experimental design consisted of the four combinations of proximate by dis-
tal location, and within each were nested the four combinations of height and
porosity, for a total of 16 stimuli. The experimental design and stimuli are
shown in Figure 4.
864
Figure 4. Experimental design and stimuli for Experiment 3.
Stamps 865

Generation of contrasts. The factor of boundary height was expressed as the


contrast for perceived enclosure between the stimuli in which both boundar-
ies were high against the stimuli in which both boundaries were low. For the
stimuli as labeled in Figure 4, that is the contrast in perceived enclosure
between stimuli KLOP and stimuli ABEF. For boundary porosity, the con-
trast was between stimuli in which both boundaries were solid and rooms in
which both boundaries were porous. The two groups of stimuli for porosity
were AICK and FHNP. For the factor of proximity of height, the contrast was
between stimuli in which the proximate boundary was low, and the distal
boundary was high against the stimuli in which the proximate boundary was
low and the distal boundary was high. The respective groups of stimuli were
IJMN and CDGH. For the factor of proximity of porosity, the contrast was
between stimuli in which the proximate boundary was solid, and the distal
boundary was porous against the stimuli in which the proximate boundary
was porous and the distal boundary was solid. The groups were EGMO and
BDJL. The coding for the last factor (H6: the effect of proximate porosity on
the effect of distal height) was more complicated. The effect of distal height
was the difference in perceived enclosure for stimuli in which the distal
boundary was high against the perceived enclosure when the distal boundary
was low. The effect of proximate porosity on the effect of distal height then
becomes a difference between proximate porosity and differences in distal
heights. Inspection of Figure 4 and a little algebra produces the following
coding for H6: GHOPABIJ to EFMNCDKL.
Participants and Sample Size. Experiment 3 used different factors than
were used in Experiments 1 and 2, so there were no prior estimates of a target
effect size. Lacking prior data, Cohen’s (1988) value for a medium effect size
was used. That value is 13% of variance. For α = .05, dfh = 5, power = 0.80,
and 16 stimuli, the number of required participants was 7. The participant
sample consisted of 29 university undergraduate engineering students. There
were 26 males and 3 females. Political affiliations were 11 liberal, 14 moder-
ate, 2 conservative, and 2 who did not state political affiliation.
Task. The stimuli were shown in a PowerPoint presentation during a class.
The instructions were

Here are 16 images of rooms. We would please ask you to look at the
rooms and rate each one on your impression of how enclosed it
appears, on a scale of open (1) to enclosed (8). In order to help you
calibrate your responses, the first two images show the ranges of
variation in the physical properties of the images.
866 Environment and Behavior 45(7)

The next screen then showed two images that exhibited the range of the
factors in the study. After that, each screen showed a room, and participants
rated it.

Results and Discussion


Analysis of variance indicated that boundary height had the strongest effect on
perceived enclosure (31.5%). The factor with the next strongest effect on per-
ceived enclosure was the interaction between height and proximity at 4.74%. It
made a difference if the transitions were from high to low or low to high.
Boundary permeability accounted for a small but detectable amount of variance
(1.9%). The interaction of proximity and boundary permeability was far too
small to detect. However, the porosity of the proximate boundary did change
the effect of the height of the distal boundary (0.4% of variance, α = .02).
For boundary height, the mean values for enclosure were MKLOP = 6.29,
MABEF = 2.89, F(1, 420) = 335, and α = 5e−56. For boundary permeability,
the data were MFHNP = 5.34, MACIK = 4.51, F(1, 420) = 199.96, and α = 5e−6.
For the interaction of proximity and height, the data were MIJMN = 6.01,
MCDGH = 4.69, F(1, 420) = 50.79, and α = 2e−12. For the interaction of prox-
imity and porosity, the data were MEFMO = 4.99, MBDJL = 5.05, F(1, 420) =
0.01, and α = 1.00. Finally, for the factor of proximate porosity on distal
height, the data were MGHOPABIJ = 4.83, MEFMNCDKL = 5.11, F(1, 420) = 4.22,
and α = .02.
In terms of correlations with perceived enclosure, the results were as fol-
lows: For H2 (boundary height), the correlation was r = .84. For H3 (bound-
ary permeability), the correlation was r = −.20. For H4 (interaction of
proximity and boundary height), the correlation was r = .33. For H5 (interac-
tion of proximity and boundary permeability), the correlation was r = .01, and
for H6 (replication of Experiments 1 and 2), the correlation was r = .09. All
correlations reported in this paragraph were calculated over the mean
response values for n = 16 stimuli.

Experiment 4
Hypotheses

This experiment used the same factors as were used in Experiment 3 (H2:
boundary height, H3: boundary permeability, H4: interaction of proximity
and height, H5: interaction of proximity and boundary permeability, and
H6: interaction of permeability of proximate boundary by distal height).
Stamps 867

However, the venue was changed to ascertain whether the findings from
Experiment 3 could be replicated in a different type of environment. The
environments in Experiment 4 were Persian gardens (Khansari, Moghtader,
& Yavari, 2004; Porter & Thévenart, 2003). In particular, the gardens were
modeled after the “four garden” or chahar-bagh scheme (Porter & Thévenart,
2003). In this scheme, the garden is situated within walls, typically created
with colonnades. The center of the garden contains a water pool or fountain,
and four open canals flow off the central pool along both axes, thus dividing
the garden into four parts. An example of this type of courtyard from
Samarquand can be seen in Hillenbrand (1994). The walls for that courtyard
consist of two stories of pointed-arch colonnades with entry points designed
as much larger pointed arches that are open to the exterior. A large variety of
plants were used in these gardens, often in multiple rows that varied in height
and porosity. For example, one garden in Shiraz had low, porous plants
located next to the water and larger trees located near the courtyard walls
(Khansari et al., 2004), whereas another garden, also in Shiraz, had very low
dense hedges next to the water and trees behind (Khansari et al., 2004). Still
another garden, this time in Kashan, had the trees near the water and the low
plants behind the trees (Khansari et al., 2004). Because the four-garden
scheme exhibited multiple boundaries that varied in height, porosity, and
proximity, this scheme was chosen as the venue for Experiment 4.

Method
Stimuli and Experimental Design. The overall envelope of the gardens
was the same as were the dimensions of the room in Experiment 3 (12 × 12 ×
8 m). However, the proximate and distal boundaries were created with plants
rather than marble screens. The plants were created to be either porous or
dense, with heights of either 1.5 or 3.0 m. The plants were located at the same
distances as were the pony walls in Experiment 3. Figure 5 shows the site
plan for the gardens and also sections. The sections illustrate the concepts of
the main effect of height and of interaction between proximity and boundary
height. If the boundaries have the same heights (both low or both high), then
the contrast in height is the difference in perceived enclosure between those
stimuli (conditions both low or both high in Figure 5). The interaction of
proximity with the effect of height is the difference between stimuli that have
low proximate and high distal boundaries (condition Prox low and Distal
high in Figure 5) against the stimuli that have high proximate and low distal
boundaries (condition Prox high and Distal low in Figure 5). Similar reason-
ing applies to the interactions of proximity and porosity.
868 Environment and Behavior 45(7)

Figure 5. Site plan and sections for Experiment 4.

The experimental design consisted of the four combinations of proximate


by distal location, and within each were nested the four combinations of
height and porosity, for a total of 16 stimuli. The same contrasts were also
calculated in Experiments 3 and 4. Figure 6 shows the experimental design
and stimuli for Experiment 4.
Figure 6. Experimental design and stimuli for Experiment 4.

869
870 Environment and Behavior 45(7)

Participants and Sample Size. The target effect size was obtained from
the results of Experiment 3. In this experiment, amounts of variance
accounted for ranged from 31.5% to less than 0.1%. Given this diversity, a
de minimis effect size of 2% was chosen as the target. The hypothesis
degrees of freedom were set to 1 because any experiment that would detect
a small effect would also detect larger effects. With α = .05, dfh = 1, power = 0.80,
and nstim = 16, 25 participants were recommended. As 24 participants were
already available when Experiment 2 was done, these participants also
provided data for Experiment 4. Experiments 2 and 4 were done at the
same time.

Results and Discussion


Analysis of variance indicated that boundary height and boundary permea-
bility each accounted for 9% of response variance. There was a detectable
effect for the interaction of proximity and permeability (2.2%). The interac-
tion of proximity and boundary permeability was far too small to detect
(<0.1%). However, the result from Experiments 1 and 2 regarding the effect
of proximate porosity on the effect of distal height was again replicated
(0.3% of variance, α = .05).
For boundary height, the mean values for enclosure were MKLOP = 5.86,
MABEF = 4.29, F(1, 345) = 72.45, and α = 2e−16. For boundary permeability,
the data were MFHNP = 5.93, MACIK = 4.37, F(1, 345) = 71.50, and α = 4e−16.
For the interaction of proximity and height, the data were MIJMN = 5.44,
MCDGH = 5.35, F(1, 345) = 0.20, and α = .80. For the interaction of proximity
and porosity, the data were MEFMO = 5.70, MBDJL = 4.92, F(1, 345) = 17.87, and
α = 1e−5. Finally, for the factor of proximate porosity on distal height, the data
were MGHOPABIJ = 5.12, MEFMNCDKL = 5.35, F(1, 345) = 3.00, and α = .05.
In terms of correlations with perceived enclosure, the results were as fol-
lows: For H2 (boundary height), the correlation was r = .59. For H3 (bound-
ary permeability), the correlation was r = −.58. For H4 (interaction of
proximity and boundary height), the correlation was r = .03. For H5 (interac-
tion of proximity and boundary permeability), the correlation was r = .29, and
for H6 (replication of Experiments 1 and 2), the correlation was r = .12 All
correlations reported in this paragraph were calculated over n = 16 stimuli. In
addition, since the same factors and levels were expressed in Experiments 3
and 4, it was possible to calculate how well the results from Experiment 3
generalized to the venue of Experiment 4. The correlation of perceived
Stamps 871

Table 1. Summary of Results.


Hypothesis Source r r^
Hypothesis 1: Spaciousness is related to Experiment 1 .66
locomotive area Experiment 2 .88 .79
Hypothesis 1: Spaciousness is related to boundary Experiment 1 −.21
height Experiment 2 −.43 −.32
Hypothesis 2: Enclosure is related to boundary Experiment 3 .84
height Experiment 4 .59 .74
Hypothesis 3: Enclosure is related to boundary Experiment 3 −.20
permeability Experiment 3 −.58 −.41
Hypothesis 4: Going from high proximate Experiment 3 .33
boundaries to low distal boundaries is more Experiment 4 .03 .18
enclosing than going from low to high
Hypothesis 5: Going from solid proximate Experiment 3 .01
boundaries to porous distal boundaries is more Experiment 4 .29 .15
enclosing than going from porous to solid
Hypothesis 6: Porosity of proximate boundaries Experiment 3 .09
increases the effect of distal height Experiment 4 .12 .11
Results obtained from balanced, formal Experiments 3, 4 .73
experiments designed and conducted within
the effect size paradigm are reproducible and
generalize over different environments

enclosure over the 16 stimuli in Experiment 3 and the 16 stimuli in Experiment


4 was r = .73, t(14) = 4.07, α = 5e−4.

Synthesis of Findings
Table 1 lists the main findings in this article. As noted above, the findings
from multiple experiments were statistically combined using meta-analysis.
The specific meta-analytic method used here is the combination of correla-
tions from multiple experiments into a collective correlation (r^ ). These col-
lective correlations are the effect sizes of the current results for each of the a
priori predictions and two a posteriori claims.
The first prediction was that spaciousness would be more strongly related
to horizontal area within a boundary than to boundary features such as height.
872 Environment and Behavior 45(7)

The reasoning is that perceived spaciousness is based on perceived range of


locomotive permeability, whereas boundaries are related more to visual per-
meability. The respective collective correlations were r^ = .79 for spacious-
ness and horizontal area and r^ = −.32 for spaciousness and boundary height.
The difference between these two correlations is highly significant (Z = 4.55,
α = 3e−5). Accordingly, the present data solidly support a prediction made
from permeability theory.
The second hypothesis is that boundary height should influence perceived
enclosure because the higher the boundary, the less the visual permeability,
and less visual permeability implies more danger potential, and because
enclosure is a proxy for danger, higher boundaries should make places seem
more enclosed. The applicable collective correlation was r^ = .74. This means
the data supported another prediction from permeability theory: Higher
boundaries make spaces appear to be more enclosed.
The third prediction is that permeable boundaries should make places
seem less enclosed. The reasoning is that the more permeable a boundary, the
more of an environment will be visible. Increased visibility means less dan-
ger, and because perceived enclosure is a proxy for danger, the more perme-
able a boundary, the less enclosed a space should seem. The relevant datum
is r^ = −.41. Again, the data were as predicted by the permeability theory.
The fourth prediction was that a nearer boundary should have a stronger
effect on perceived enclosure than a farther boundary. This is the principle of
the “primacy of the proximate.” The reasoning is that the closer a threat can
be without being detected, the more dangerous (and hence enclosed) a space
should seem. More specifically, the prediction was for boundary height: The
heights of a proximate boundary have a stronger influence on perceived
enclosure than do the heights of distal boundaries. The effect size for this
prediction was r^ = .18. This result supports the prediction: Spaces arranged
from high boundaries to low boundaries will seem more enclosed than spaces
arranged from low to high boundaries.
The fifth prediction also tested the prediction of the primacy of the proxi-
mate, but this time for the property of boundary permeability. The collective
effect size was r^ = .15. This was statistically the same result as was obtained
for the primacy of proximate height (r^ = .18). This result was also as pre-
dicted by permeability theory. However, a caveat is in order. Inspection of
the four correlations for the principle of the primacy of the proximate indi-
cates that height had more influence in Experiment 3 but permeability had
more influence in Experiment 4. This distinction suggests that further
research will be needed to clarify the conditions within which the principle of
the primacy of the proximate holds.
Stamps 873

The sixth prediction was made after Experiments 1 and 2 but before
Experiments 3 and 4. From Experiments 1 and 2, it seemed that if a proxi-
mate boundary was porous, the effect of the height of a distal boundary
should be greater than if the proximate boundary was solid. Data from
Experiments 3 and 4 were in conformance with the prediction at r^ = .11. Data
from Experiments 3 and 4 were also virtually identical (rs of .09 and .12),
indicating successful replication over all four experiments.
The final result from this article was that judgments of perceived enclo-
sure, when evaluated using the same formal factors but in different venues,
correlated at r^ = .73. This result indicates that not only were the predictions
of permeability theory correct but also that the findings successfully general-
ized over different environments.

Possible Limitations
The work done in this article used three kinds of environments (Japanese
rock gardens, Persian rooms, Persian courtyards), and so there are many
other kinds of environments in which the relationships between safety,
enclosure, and environmental properties could be explored. As mentioned at
the beginning of this article, sometimes the relationship between safety and
enclosure has been strong and sometimes not. Additional relevant data have
been provided in Stamps (2000). Clearly, there is much work to be done
before definitive claims can be made about the ranges over which the con-
nection between safety and enclosure obtains.

Summary
This article presents seven new, original findings on how multiple boundar-
ies influence perceived spaciousness or enclosure. Perceived spaciousness
was strongly related to how much locomotion was possible within boundar-
ies, and, to a lesser extent, to boundary height. Perceived enclosure was
strongly related to height of boundary and, to a lesser extent, to boundary
porosity. Boundary proximity also influenced perceived enclosure, with
proximate boundaries having stronger effects than distal boundaries. Finally,
results were highly reproducible over very different environmental venues.

Declaration of Conflicting Interests


The author(s) declared no potential conflicts of interest with respect to the research,
authorship, and/or publication of this article.
874 Environment and Behavior 45(7)

Funding

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

References
Alkhresheh, M. M. (2007). Enclosure as a function of height-to-width ratio and scale:
Its influence on user’s sense of comfort and safety in urban street space (Doctoral
dissertation). University of Florida, Gainesville, FL.
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 behav-
ioral sciences. Hillsdale, NJ: Lawrence Erlbaum.
Doxiadis, C. A. (1968). Ekistics: An introduction to the science of human settlements.
Oxford, UK: Oxford University Press.
Duval, C. L., Charles, K. E., & Veitch, J. A. (2002). Open-plan density and environ-
mental satisfaction. Retrieved from http://irc.cnrc.gc.ca/ircpubs
Ettinghausen, R., Grabar, O., & Jenkins-Madina, M. (2001). Islamic art and architec-
ture 650-1250. New Haven, CT: Yale University Press.
Feimer, N. R. (1984). Environmental perception: The effects of media, evaluative
context, and observer sample. Journal of Environmental Psychology, 4, 61-80.
Fletcher, B. (1963). A history of architecture on the comparative method (17th ed.).
New York, NY: Charles Scribners’ Sons.
Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis. Orlando, FL:
Academic Press.
Herzog, T. R., & Chernick, K. K. (2000). Tranquility and danger in urban and natural
settings. Journal of Environmental Psychology, 20, 29-39.
Herzog, T. R., & Miller, E. J. (1998). The role of mystery in perceived danger and
environmental preference. Environment and Behavior, 30(4), 429-449.
Hillenbrand, R. (1994). Islamic architecture: Form, function and meaning. New York,
NY: Columbia University Press.
Itoh, T. (1981). Space and illusion in the Japanese garden. New York, NY: Weatherhill/
Tankosha.
Keane, M. P. (1996). Japanese garden design. Boston, MA: Tuttle.
Khansari, M., Moghtader, M., & Yavari, G. (2004). The Persian garden: Echoes of
paradise. Washington, DC: Mage.
Ohashi, H. (1997). The Japanese garden: Islands of serenity. Tokyo, Japan: Graphic-sha.
Oswald, A. J., & Wu, S. (2010). Objective confirmation of subjective measures of
human well-being: Evidence from the U.S.A. Science, 327, 576-579.
Stamps 875

Porter, Y., & Thévenart, A. (2003). Palaces and gardens of Persia. Paris, France:
Flammarion.
Rosenthal, R., & Rosnow, R. L. (1985). Contrast analysis: Focused comparisons in
the analysis of variance. Cambridge, UK: Cambridge University Press.
Rosenthal, R., & Rosnow, R. L. (1991). Essentials of behavioral research: Methods
and data analysis. New York, NY: McGraw-Hill.
Stamps, A. E. (1999). Demographic effects in environmental preferences: A meta-
analysis. 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, May 22-26). Research on enclosure. Paper presented at EDRA
33, Philadelphia, PA.
Stamps, A. E. (2005a). Enclosure and safety in urbanscapes. Environment & Behav-
ior, 37, 102-133.
Stamps, A. E. (2005b). Visual permeability, locomotive permeability, safety and
enclosure. Environment & Behavior, 37, 587-619.
Stamps, A. E. (2009). On shape and spaciousness. Environment & Behavior, 41, 526-548.
Stamps, A. E. (2010a). Effects of permeability on perceived enclosure and spacious-
ness. Environment & Behavior, 42, 864-886.
Stamps, A. E. (2010b). Use of static and dynamic media to simulate environments: A
meta-analysis. Perceptual & Motor Skills, 111, 1-12.
Stamps, A. E. (2011). Effects of area, height, elongation and color on perceived spa-
ciousness. Environment & Behavior, 43, 252-273.

Author Biography
Arthur E. Stamps III holds a Ph.D. in Architecture from UC Berkeley and is a
licensed architect. His interests are design review, environmental perception and
preference, and peer review. He works at the Institute of Environmental Quality, a
501 (c) (3) non-profit research organization in San Francisco.

View publication stats

You might also like