Predicting Window Size Preferences:
Commonness or Cognitive Evaluation?
An Honors Thesis (ID 499)
by
Amy J. Westergren
Ball State University
Muncie, IN
May, 1988
Expected date of graduation: May, 1988
-
Predicting Window Size Preferences
1
.~
\f/;~
. '
Predicting Window Size Preferences:
Commonness or Cognitive Evaluation?
Amy J. Westergren
,-
Ball State University
Running head: PREDICTING WINDOW SIZE PREFERENCES
-,
Predicting Window Size Preferences
2
Abstract
Are window size preferences based on experience or based on cognitive
evaluation of many factors about windows? The literature provides some
indirect support for both arguments. The purpose of the present experiment
was to determine if one of these explanations is better than the other.
Approximately 150 undergraduates were run in each of four between-group
experimental conditions (N=600). Each condition represented a combination
of two of four possible questionnaires. The four questionnnaires assessed:
(1) subjective personal preference (2) commonness (3) other people's
preferences and (4) reasons underlying wanting or not wanting windows.
Correlations were performed on the responses to the four questionnaires.
-
Personal preferences were reliably correlated with the other three types of
assessments. Analyses of variance were used to look at the differences
between the first and second instruments, thus providing an assessment of
context effects. Obtained context effects were very small, concerning just
a few defferences in preferences between individual spaces on
questionnaires. Theoretical implications and environmental applications
are discussed.
Predicting Window Size Preferences
--
3
Predicting Window Size Preferences:
Commonness or Cognitive Evaluation?
Windows and lighting have been the focus of many studies. Some have
concentrated on the relationship between lighting level and performance of
specific tasks (Blackwell, 1959; Weston, 1949; Simonson and Brozek, 1952;
Canter and Stringer, 1975). Other researchers have investigated human
responses to daylighting/windows such as comfort, satisfaction, and
feelings of pleasantness (Bennett, 1974; Boyce, 1973). Keighley (1973a)
found that satisfaction with windows decreases as window height
I
-
decreases. Optimum window height is lowest when the external scene is a
distant view (Keighly, 1973b). Cuttle (1983) found that large windows are
preferred and that workers see it as important to be able to work by a
window when they are in an office. Studies have shown daylight and/or
sunlight is preferred over artificial light in settings such as office,
hospital, and home (Cuttle, 1983; Heerwagen and Heerwagen, 1986;
Hopkinson, 1967; Keighley, 1973a, 1973b; Verderber, 1986, Wooton and
Barkow, 1983).
What about the effects of windowless environments? Information on
this subject is limited and the results obtained for studies on windowless
classrooms are very contradictory. Demos (1965,1967) found no
significant differences in achievement, health, or personality between
-
subjects in windowless classrooms and those whose classrooms contained
windows. In fact, students in windowless classroom actually held favorable
attitudes about their "windowlessness". Absenteeism was not found to be
significant in a windowless elementary school (Larson, 1965). Karmel
Predicting Window Size Preferences
-
4
(1965) found that children attending windowless schools drew pictures of
schools with windows more often than other children. When questioned
directly, children of windowless classrooms also expressed a desire for
windows (Larson, 1965).
There are two schools of thought which offer explanations regarding the
basis of window preferences. One school believes that people simply prefer
what is common or most typical. The other school posits that features of
the object under scrutiny, in this case windows, are carefully evaluated and
weighted resulting in a preference.
Much support has been given to the idea that preferences reflect
commonness. Mere exposure to an object/event increases positive feelings
-
I
about the object/event (Harrison, 1977; Hill, 1978; Matlin and Stang, 1978).
The organism will, therefore, develop a preference for the "common" object.
Matlin (1971) found that old stimuli were preferred to new stimuli.
Moreland and Zajonc (1977, 1979) have shown recognition of these old (or
common) stimuli need not exist for the exposure effect to occur. Zajonc and
Markus (1982) suggest that preferences can be, under some circumstances,
relatively independent of cognitive evaluation. Interestingly, they found
that when preferences were based on both exposure effects and familiarity,
subjects often attributed their preferences to stimulus features.
The validity of Zajonc's view that preferences can be independent of
cognitive evaluation is counter to the other school of thought which
assumes that cognitive factors necessarily precede preferences. For
-
instance, provision of light, ventilation, and view are just a few of the
basic features of windows that can give rise to a preference for windows in
a space. Markus and Gray (1973) found that a number of factors determined
satisfaction with windows in residential environments (for example,
Predicting Window Size Preferences
-
5
brightness, spaciousness, amount of greenery and "nature" visible, and the
amount and kind of activity occurring). This study and others like it suggest
that many factors go into preference judgments. Of course, the assumption
behind this model is that if the function of objects (windows) is fully
explicated and importance-weighted, preferences can be empirically
predicted.
An experiment was performed to address the issue of whether cognitive
analysis or typicality underlies preference judgments for windows.
Method
Subjects
Subjects were 610 undergraduate students, 212 males and 398 females,
participating in the psychological sciences research participant pool at Ball
State University. The mean age was 19.5 years and the median age of the
subjects was 19.
Question naires
Four questionnaires were constructed to assess subjective reactions
regarding window size for a variety of common settings (Butler, Biner, and
Jones, 1988). The settings were the kitchen, library, family room, bed
room, large lecture hall, garage, dormroom, small classroom, public
bathroom, living room, dining room, bathroom at home, computer workroom,
office, gym, laundry room, and restaurant. The first questionnaire (QA)
-
consisted of 17 questions (one question per setting) asssessing personal
window size preferences (e.g., When I am doing thing in the kitchen I
prefer:). The second questionnaire (QB) consisted of 17 questions regarding
the perception of others' window preferences (e.g., When other people
Predicting Window Size Preferences
--
6
are doing things in the kitchen, they prefer:). The third questionnaire (QC)
consisted of 17 questions asking subjects to subjectively assess the most
common window size for each space (e.g., For a kitchen, which of the
following is most common?) For these questionnaires (QA, QB, QC)
responses were made on a four-point scale (1 =no windows, 2=a small
window, 3=a medium-sized window, 4=a large window). The final
questionnaire (QD) was a 170-question instrument composed of ten
questions for each of the 17 different settings. Specifically, subjects were
asked to consider and judge the importance of ten reasons for wanting or
not wanting windows for each of the spaces. The ten reasons were taken
from earlier window preference research done by Butler and Biner (in press).
I
-
These reasons were proposed as relevant factors underlying window
proferences. They were: air quality, emergency use, mood and social
interaction, to see outside, to make space seem more larger or spacious, to
affect other people's impressions from the outside, sunlight, privacy, source
of nuisance or distracton, and limitation of flexibility of furniture
arrangement. Subjects responded on a five-point scale (A= "Not at all
important" to E="Very important").
Procedure
Four experimental conditions were created by pairing different
combinations of the four instruments. Subjects assigned to condition 1
received QA and then QB. Subjects in condition 2 received QB and then QA.
Subjects in condition 3 received QC and then QA. Finally, condition 4
-
subjects received QD followed by QA. These four experimental conditions
represented a between subjects factor. Subjects were run in large groups in
a university lecture hall. It took subjects between 10-45 minutes to
complete the experiment. Subjects recorded their responses on
Predicting Window Size Preferences
-
7
computer-scorable answer sheets. On each answer sheet subjects were
also asked to fill in their age, sex, and a code number for the condition.
This code in no way identified them with their answers, but was necessary
for the analysis of the data. Responding to these instructions and to the
questions on each questionnaire required careful attention. Subjects who
did not read or listen to instructions, therefore, made easily identifiable
mistakes and could be eliminated from the study. Approximately 150
subjects were run in each experimental condition.
Results
For the first questionnaire in each condition (e.g., A of condition A8),
means were calculated for each space. Pearson-moment correlations
-
between questionnaires were then calculated using these means. All four
questionnaires were found to be highly correlated across spaces. Personal
subjective preference (OA) and perception of others' preferences (08) were
found to be the most highly correlated questionnaires [£(15) = .99, Q. < .001].
Related to the hypothesis, the correlation between personal subjective
preference (OA) and reasons for wanting or not wanting windows (00) was
slightly, but not significantly higher [£(15)
= .95, Q. <
.001] than the
correlation between personal subjective preference (OA) and commonness
(OC), [£(15) = .94, Q. < .001]. The correlation between commonness (OC) and
perception of others' preferences (08) was slightly, but not significantly
lower [£(15)
= .94, Q. <
.001] than the correlation between reasons (00) and
others' preferences (08) [[(15)
-
= .97, Q. < .001].
Interestingly, the lowest
correlation was between OC (commonness) and OD (reasons for wanting or
not wanting windows) [£(15)
= .89, Q. <
.001]. The only significant difference
among these correlations was between the OA to 08 correlation and the OC
to 00 correlation, Z
= 3.07, P < .01.
Predicting Window Size Preferences
8
In order to assess the effects that judging others' preferences,
commonness, and reasons have on personal subjective preference, a four
[context (AB, BA, CA, DA)--between subjects] X 17 (space--within subjects)
analysis of variance was performed on the peronsal subjective preferences.
No significant main effect was found for context [E(3,567)
= 2.23, Q < .083].
That is, completing the other questionnaires did not generally affect
personal window size preference. There was a significant difference of
preferences among individual spaces on the questionnaires [E(16, 9072)
=
462.82, Q < .0001]. These personal preference differences among spaces
accounted for a large proportion of the variance, Eta 2
= .40. A significant
interaction was found between context and spaces [E(48, 9072)
= 2.10, 12 <
.001], but it accounted for less than one percent of the variance. In other
words, the context (completing other questionnaires) apparently had a small
effect on some spaces but not in others (see Figure 1).
Insert Figure 1 about here
In order to assess the effect order of questionnaire has on personal
subjective preference, a second analysis of variance was run comparing
condition AB to condition BA. This 2 [order (AB vs. BA)--between subjects]
X 2 [questionnaire (A vs. B)--within subjects] X 17 (spaces--within
-
subjects) ANOVA, as in the previous ANOVA, revealed that individual spaces
on the questionnaires were highly significant sources of variance [E(16,
4528)
= 304.48, Q < .001].
These differences accounted for a large
Predicting Window Size Preferences
9
-
percentage of the variance, Eta2 =.40. An interaction was found between
space (1-17), questionnaire (A or B), and order (AB or BA), [E(16, 4528)
=
2.45, Q < .001]. This interaction indicates that certain spaces on certain
questionnaires in certain orders varied significantly.
As can be seen in Figure 2, the pattern of judgments for individual
spaces on questionnaire A (personal subjective preference) remained
relatively stable regardless of order (AB, BA). Subjects judgments of
others' preferences varied 3.04 times more than subjects judgments of
their own subjective preferences, sSB/ssA
= 3.04.
When subjects judged B
in condition 1 (AB), the effect is most prevalent.
Insert Figure 2 and 3 about here
Discussion
The present data suggests that neither commonness nor cognitive
evaluation alone accounts for window size preference judgments. Zajonc
and Markus (1982) believe that preference judgments can often be a mere
reflection of commonness. In other words, people simply prefer what is
common. Our data reveals a strong relationship between commonness and
preference judgments suggesting that Zajonc and Markus's belief that
-
preferences are based on commonness is correct. However, we also found
evidence which shows that cognitive evaluation and preferences are also
highly related. This later evidence suggests that preferences are also based
on cognitive evaluation. Because both theories are supported, both must
Predicting Window Size Preferences
10
-
be that rejected as bases of how preference judgments are made. That is,
our data show that neither commonness nor reasons is the sole determinor
of preference judgments for window size. Both schools of thought
concerning preference jugments (commonness and reasons) are probably
incomplete.
A model which seems to make more sense than these two theories is an
interactionist model in which commonness and reasons are independently
acting and sometimes reacting to produce preferences. In other words, the
interactionist model is supported by data which show that commonness and
reasons are not as strongly related than are other relationships investigated
in this study (e.g., commonness and preferences). This suggests that
reasons and commonness may be independent factors working together to
produce preference judgments.
For a number of theoretical reasons it was believed that the context (AB,
BA, CA, DA) in which subjects answered the questionnaires would influence
subjects' personal preference judgments. Helson (1964) has argued in his
well-known adaptation-level theory that many human responses can be
influenced by the context in which stimuli are presented. That is, what
subjects do just before performing a behavior, making a judgment, and the
like can influence subsequent performance. For example, apparent
brightness of lights depends upon wheter you were just in a dark basement
or out in the bright sunshine.
A second reason for expecting context effects, particularly in this study,
-
is the phenomenon of social desirability bias. It has been used by social
psychologists to explain the instability of subjects judgments on
questionnaires. "On most topics where society's norms dictate or even
suggest that most people prefer one kind of behavior over another, we can
Predicting Window Size Preferences
-
11
expect an overreporting of the desirable behaviors and and underreporting of
the undesirable ones" (Wrightsman, 1977, p. 61). Then in the present study,
preference judgments could have been influenced by the desire to look good,
to conform or to stand out from the group. Similarly, it is possible that a
response consistency bias is operating. Specifically, it could be
hypothesized that subjects' personal preferences may be similar to the
commonness judgments they make just prior to giving their own personal
preference judgments because subjects somehow make an effort to appear
consistent in their ratings with what they believe to be normal or typical.
This effect, of course, would not be expected to occur when personal
preferences are assessed first. The results then would be a contextual
-
f
effect on personal judgments of best window size. n they judge commonness
prior to giving their own preferences.
Surprisingly, only small context effects on questionnaire A over all four
conditions (AB,BA,CA,DA) were demonstrated in the present study. A
slightly larger context effect was evident only when subjects judged other's
preferences (OB). This effect was particularly noted when subjects judged
other's preferences after judging their own. There are several possiblities
as to why little context effects were found. First, it is possible that people
really know what they prefer and these preference judgments are so reliable
that changing the context has little influence on them. Therefore, the
adaptation-level suggested by Helson (1964) nor response consistency may
not be evident in situations where subjects have strong opinions or
knowledge about a topic. Second, our design which encouraged truthfulness
by its anonynmity and by its lack of demand characteristics decreased the
likelihood that context effects would occur. That is, social desirability
bias may not occur in situations where the topic is "neutral" and lacks the
Predicting Window Size Preferences
12
demand characteristics often seen in instruments, for instance, asking
questions about sexual practices or charitable behavior.
Our findings also reveal the usefulness of personal subjective judgments
for window size preferences. Pointing to the previous mentioned problem of
context effects, researchers in the past have been reluctant to use personal
preference judgments.
Researchers have suggested that subject reports
about their attitudes are far from being identical to their actual behaviors
(Duval & Wicklund, 1972). Researchers have suggested that self-report
information has a greater tendency to be biased by extraneous factors than
are report about other's behavior, attitudes, etc. For example, Baird (1976)
has demonstrated that by asking subjects about others' behaviors, much of
the subjectivity and its accompanying biases, found when people report
their own behavior, can be alleviated. In other words, more valid
assessment of actual human behavior can be attained by asking people what
others do then by asking them what they themselves do. Our data, on the
other hand, indicates the contrary. Subjects judgments of others'
preferences were less reliable than their own subjective personal
preferences. In fact, the variance in the judgments of others' preferences
was three times greater than that of the personal subjective preferences.
This effect was most evident when subjects judged others' preferences
after their own (Condition AB).
In the present study, context not only had an effect on personal
preferences, but also on judgment of others' preferences. Why? Social
-
psychologists might explain these differences as due to response
consistency bias. This argument can not fully explain our data however. We
found a consistency effect in only one context (BA). If social conformity or
consistency is the reason behind the similarity between B and A for BA, why
Predicting Window Size Preferences
13
-
was it not symmetrical? If people desire to be similar to others in
Condition BA should not they also desire to be similar in Condition AB? In
the present dat, this was simply not the case.
This context effect may have occurred because of a memory bias. When
subjects make personal subjective preferences followed by judgment of
others' preferences (Condition AB) they may have difficulty judging others'
preferences because they are trying to exclude both their personal
preferences and experiences from the questions. Subjects may tend to
include themselves when asked about others' preferences in B of Condition
BA. The distinction to exclude oneself is clearer in Condition AB than in
Condition BA. Subjects find jugments harder to make because without the
ability to use their own experiences, subjects lack knowledge about others'
habits, preferences, etc.
Previous research on windowless classrooms has been ambivalent. Our
study shows that there are significant differences in window size
preferences depending on how large the classroom is. It was found that
larger windows are desired in small classrooms while smaller ones are
preferred in large lecture halls. How do windows actually effect behaviors
in these spaces? What activities performed in these settings require
windows or the lack of windows? There may be activities in which the
presence or absence of windows would be advantageous. Such questions
aqre in need of further research.
One major criticism of our study is its inability to be translated into IES
-
standards. Subjects are asked whether they would prefer no windows, a
small window, a medium-sized window, or a large window in specific
settings. Since we did not anchor these choices with actual size
measurements, we have no way of knowing what a small window meant
Predicting Window Size Preferences
--
14
to individual subjects. More disturbing is that the cognitive definitions of
our present anchors likely chance as a function of the space in question. For
instance, a window with the dimensions of "X" by "V" might be a "large
window for the bathroom but a relatively small one for a living room.
Future research should include anchors which would be easily translated by
IES. For example, specific percentages of wall space taken up by windows
could be used signify the size of the window.
A second criticism stems from the population used in the present study.
Many would argue that the external validity of this study is low because the
sample was comprised of college students. Such criticisms are indeed
noteworthy, but should not detract from the results. This study has
I
-
demonstrated that pencil and paper instruments can produce sensical
preference data and do so relatively free of context effects. Furthermore,
this method would appear to be easily adaptable for use on other populations
such as the elderly to obtain preference judgments in situations which are
common to them.
-
Predicting Window Size Preferences
15
-
Author Notes
I would like to thank Dr. Paul M. Biner and Dr. Darrell L. Butler of Ball
State University for their wisdom and patience in helping me prepare this
manuscript.
-
Predicting Window Size Preferences
16
References
Baird, J. C., Noma, E., Nagy, J. N. and Quinn, J. (1976). Predicted and observed
activity patterns in campus settings. Perceptual and Motor Skills, !3.,
615-624.
Bennett, C. (1974). Let's shed a little light. In D.H. Carson (ed.),
Man-environment interactons: Environmental evaluations and
applications. (pp. 15-28). Washington, D. C.: Environment Design Research
Association.
Blackwell, H. R. (1959). The development and use of a quantitative method
for specification of interior illumination levels on the basis of performance data. Illumination Engineering, ~, 317-353.
Boyce, P. R. (1973). Age, Illuminance, visual performance, and preference.
Lighting Research Technology,~, 125-144.
Butler, D. L., Biner, P. M., and Jones, S. K. (1988). Common spaces and
activities of college students evaluated using a recall method. Unpublished manuscript.
Butler, D. L., and Biner, P. M. (in press). Effects of setting on window preferences and factors associated with those preferences. Environment and
Behavior.
Canter, D. and Stringer, P. (1975). Environmental interaction: Psychological
approaches to our physical surroundings. New York: University Press.
Cuttle, K. (1983). People and windows in workplaces. Paper presented at the
Conference on People and Physical Environment Research, Wellington, New
Zealand.
-
Demos, G. D. (1965). Controlled physical classroom environments and their
effects upon elementary school children. Riverside County, CA, Palm
Springs School District.
Demos, G. D., Davis, S., and Zuwaylif, F. F. (1967). Controlled physical
environments. Building Research, ~, 60-62.
Predicting Window Size Preferences
17
-
Duvall, S., and Wicklund, R. A. (1972). A theory of objective self-awareness.
New York: Academic Press.
Harrison, A. A. (1977). Mere exposure. Advances in Experimental Social
psychology, 10..
Heerwagen, J. H. and Heerwagen, D. R. (1986). Lighting and psychobiological
comfort. Lightjng Design and Application, .1.6(4}, 47-51.
Helson, H. (1964). Adaptation-level theory. New York: Harper and Row.
Hill, W. F. (1978). Effect of mere exposure on preference in non-human
mammals. Psychological Bulletin,
1177-1198.
as,
Hopkinson, R. G. {1967}. The psychophysics of sunlighting, "Sunlight in
Buildings." Proceedings of the CIE, 13-19, {Bouwcentrum International,
Rotterdam}.
,-
Karmel, L. J. {1965}. Effects of windowless classroom environments on high
school students. perceptual and Motor Skills, 20, 277-278.
Keighley, E. C. {1973a}. Visual requirements and reduced fenestration in
office buildings: A study of window shape. Building Science, ,6, 311320.
Keighley, E. C. {1973b}. Visual requirements and reduced fenestration in
offices: A study of multiple apertures and window area. Building
Science. 8. 321-331 .
Larson, C. T. {Ed.} {1965}. The effect of windowless classrooms on elementary school children. The university of Michigan: Department of
Arch itecture.
-
Markus, T. A. and Gray, A. {October, 1973}. Windows in low rise, high density
housing: The psychological significance of sunshine, daylight, view, and
visual privacy. Proceedings of CIE Conference.
Matlin, M. W. {1971}. Response competition, recognition, and affect. Journal
of Personality and Social Psychology, 19, 295-300.
Predicting Window Size Preferences
18
-
Matlin, M. W. and Stang, D. J. (1978). The pollyana rinciple. Cambridge:
Schenkman.
Moreland, R. L. and Zajonc, R. B. (1977). Is stimulus recognition a necessary
condition for the occurence of exposure effects. Journal of Personality
and Social Psychology. ~ 191-199.
Moreland, R. L. and Zajonc, R. B. (1979). Exposure effects may not depend on
stimulus recognition. Journal of Personality and Social Psychology,
1085-1089.
az,
Simonson, E. and Brozek, J. (1952). Flicker fusion frequency: Backgrounds
and applications. Physiological Reviews, 32, 349-378.
Verderber, S. (1986). Dimensions of person-window transactions in the
hospital environment. Environment and Behavior,,la, 450-466.
Weston, H. C. (1949). Sight. light. and efficiency. London: Lewis.
Wooton, E. and Barkow, B. (1983). An investigation of the effects of windows
and lighting in offices. Proceedings of the 1983 Daylight Conference,
Phoenix, Arizona.
Wrightsman, L. S. (1977). Social psychology (2nd ed.). Monterery, CA: Brooks
and Cole Publishing Company.
Zajonc, R. B. and Markus, H. (1982). Affective and cognitive factors in
preferences. Journal of Consumer Research, ,9, 123-131.
-
t-
)
GRAPH OF CONDITION BY QUESTION
4 -.----------
2
LEGEND
CONDITION AS
CONDITION BA
CONDITION CA
, . . . ''' . . . . 11 . . . . . . . . .
1-,
01
Figure 1
•
()2
•
0'3
•
•
•
•
•
•
•
•
•
•
•
•
•
~
04
05
06
07
08
09
010
Ql1
012
013
Q14
015
016
017
CONDITION DA
-)
----
----------
GRAPH OF CONDITION BY QUESTION FOR QUESTIONNAIRE A
40-.------------------~-------------------------------------------------,
30
\
\
\
~
20
1\ I \
~
/
V
\I
\J
V
A
~
f
\ j
LEGEND
CONDITION AB
10 "\--.---.---.
•
•
•
•
•
•
•
•
•
•
•
.~ CONDITION BA
OA1 OA2 OA3 OA4 OA5 OA6 OA7 OAB OA9 OA10 OA11 OAI2 OA13 OA14 QA15 OA16 OA17
Figure 2
)
--_._---- .._-
\.---J
------_._------------
GRAPH OF CONDITION BY QUESTION FOR QUESTIONNAIRE B
40.--------------------------------------------------------------,
~\
J ~
30 ,\
\ j
/~
J ~
~
j
\
\\
20
~
\V/
\J
,~\
~ \\
\
~
~
I
f
\
1
t:-ll
-J
LEGEND
CONDITION AS
10 '\---.---r--..-- I
•
I
•
I
•
•
--.----.
•
•
.~ CONDITION SA
OBI ()8;~ 0[3] OB4 085 086 087 088 089 0810 0811 0812 0813 0814 0815 0816 0817
Figure 3
----------_._------------"