Building and Environment 70 (2013) 171e188

Contents lists available at ScienceDirect

Building and Environment

journal homepage: www.elsevier.com/locate/buildenv

A study on student perceptions of higher education classrooms:

Impact of classroom attributes on student satisfaction and
performance
Zheng Yang a, Burcin Becerik-Gerber b, *, Laura Mino c
a
Sonny Astani Department of Civil & Environmental Engineering, University of Southern California, 3620 S. Vermont Avenue, KAP 217, Los Angeles,
CA 90089-2531, United States
b
Sonny Astani Department of Civil & Environmental Engineering, University of Southern California, 3620 S. Vermont Avenue, KAP 224C, Los Angeles,
CA 90089-2531, United States
c
Arup Group Limited, 12777 West Jefferson Blvd Suite 100, Los Angeles, CA 90066, United States

a r t i c l e i n f o a b s t r a c t

Article history: Exploring how classroom attributes affect student satisfaction and performance in higher education
Received 22 May 2013 classrooms continues to be a critical initiative among educators and researchers. Although specific
Received in revised form classroom attributes and their impacts on student satisfaction and performance have been investigated
29 August 2013
independently, a holistic investigation of many of these attributes and their individual and cumulative
Accepted 30 August 2013
impacts on student perceptions of their learning environments is missing. This paper takes a statistical
approach to assess ambient, spatial, and technological attributes that can be found in higher education
Keywords:
classrooms through an online survey conducted in six classrooms in a university. The paper provides
Learning environments
Student perception
insight for future evaluation of higher education learning environments by linking two Likert scales: one
Classroom attributes rating student satisfaction with classroom attributes and the other rating the impact of these attributes
Student satisfaction on student performance, and by analyzing the relationships between reported perceptions and student
Student performance evaluations of different conditions. The results revealed that student perceptions rely heavily on spatial
Higher education classrooms attributes, specifically visibility and furniture, and ambient attributes, specifically air quality and tem-
perature, which are highly impacted by the design, management and maintenance of classrooms. The
paper also investigated the impacts of non-classroom factors, including gender, seating location, cu-
mulative GPA, college year and expected course grade, on student perceptions of learning environments.
Results showed that perceptions of visibility, acoustics and furniture were more sensitive to non-
classroom factors, followed by temperature, air quality, artificial lighting, room layout and software.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction attendance, but after minimum requirements are achieved, the

With the advent of sick building syndrome, numerous concerns
have been raised over the impact of classroom attributes on student
performance. Psychologists, educators, architects, and engineers
have studied ideal learning environments for age groups ranging
from preschool to higher education. While these studies all have
attempted to characterize the optimal conditions, a consensus has
yet to be reached. Adverse ambient conditions such as extreme
temperatures, inadequate lighting, and poor air quality undoubt-
edly have negative impacts on student performance, retention, and
* Corresponding author. Tel.: þ1 213 740 4383.
E-mail addresses: zhengyan@usc.edu (Z. Yang), becerik@usc.edu (B. Becerik-
Gerber), laura.mino@arup.com (L. Mino).
essential impacts and relations of these attributes are still unclear
[1e4]. The spatial design of classrooms, including furniture, visi-
bility, and layout, has also been a frequent topic of investigation
[5,6]. While rows of desks facing one direction is the default
classroom layout, open floor plans, group seating, and comfortable
furniture have been implemented with contradictory results [1,7].
Over the past decade, technology has become a central component
of many classrooms in the form of audio and visual equipment,
participation tools, personal computers, and Internet access. While
classroom technology has generally been viewed as a positive
addition, in some cases technology has been found to have no
impact or even a negative impact on the students’ learning expe-
riences [8,9]. Still others contend that the ideal learning environ-
ment is context-dependent and thus can never be permanently
defined. Studies have found personal factors such as gender and age

0360-1323/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.buildenv.2013.08.030
172 Z. Yang et al. / Building and Environment 70 (2013) 171e188
to play a significant role in users’ assessments of various spatial and
technological attributes [10,11]. Steve et al. have argued that the
ideal learning environment is always changing as innovations in
technologies, teaching strategies, and design philosophies require
continuous measurement and reassessment [1].
This paper presents the findings of a comprehensive and sys-
tematic analysis of student perceptions of different classroom at-
tributes, determinants of each attribute perception, and the
impacts of non-classroom factors on perception. First, the authors
investigate individual and cumulative impacts of ambient, spatial,
and technological classroom attributes that exist in higher educa-
tion classrooms. These three categories of classroom attributes
were selected to reflect the range of attributes frequently consid-
ered in the assessment of learning environments. A total of ten
attributes were included in the investigation. These are tempera-
ture, air quality, and lighting, both natural and artificial, acoustics,
visibility, classroom layout, furniture, hardware such as projector,
computer, clicker, smart board, camera, etc. and software including
software installed on classroom computers einstructor and student
computers- and the Internet. These attributes also have the po-
tential to be improved by more informed planning and investment
in classroom design, management and maintenance. The authors
investigate how students perceived these attributes individually as
well as how the perceptions of different attributes correlate with
each other. Student responses to a two-stage online survey con-
ducted in six classrooms at a university were used to analyze how
reported perceptions of various classroom attributes are related to
students’ evaluations of different conditions. Finally, as perception
has been established as personal and contextual, the responses
were assessed for their variance according to the non-classroom
factors such as gender, college year, expected course grade, cu-
mulative GPA, and students’ seating locations. The research pre-
sented in this paper aims to answer the following research
questions:
1. How do students perceive ambient, spatial, and technological
classroom attributes, and what are the interdependencies
among perceptions of different attributes?
2. What are the relationships between reported perceptions and
corresponding descriptive conditions, and how should a
perception concerning a specific attribute be interpreted?
3. How do student perceptions of classroom attributes vary with
non-classroom (personal and contextual) factors such as
gender, college year, seating location, expected course grade
and cumulative GPA?
2. Classroom attributes influencing student perceptions of
learning environments
Classrooms should be configured to provide the best learning
environments possible to promote student learning [12]. Trickett
and Moos [13], and Walberg and Anderson [14] conducted one of
the first studies about learning environments in late 60’s and early
70’s. Since then, several studies have built on their work and
extended it to different educational settings [15e18], and a large
number of instruments have been developed to assess students’
perceptions of various aspects of learning environments [19e21].
In general, students’ perceptions can be divided into three cate-
gories: perception of the psychosocial environment such as
belongingness and connection with classmates [22e27]; percep-
tion of the psychological environment such as motivation, self-
efficacy and achievement [27e31]; and perception of the phys-
ical environment such as classroom size, lighting and technology
[32e37]. This classification scheme is consistent with the one
originated from Moos’ work (i.e., relationship dimensions, per-
sonal development dimensions, maintenance and system change
dimensions) [38].
Each of the above perception types could add to the global un-
derstanding of the impact of learning environments on student
satisfaction and performance [36]. This paper focuses on the
physical learning environments – one of the most salient features of
a classroom setting [39,40]. Well-organized classroom environ-
ments can facilitate student learning and increase students’ eval-
uations of the instructor and the course [33,41,42]. Moreover,
physical learning environments can be improved through class-
room design, maintenance and management [36]. While several
studies have analyzed the impact of physical learning environ-
ments on student satisfaction and performance in K-12 education
settings, there are relatively few attempts that study physical
learning environments and their impact on student satisfaction and
performance in higher education [33], and the findings from pri-
mary and secondary education classrooms may not be applicable in
higher education settings [36,43].
Physical learning environments should be evaluated by
studying both the physical attributes and the students’ percep-
tions of those attributes. Based on the literature, physical attri-
butes could be classified into three categories. The first category is
the ambient environment, including attributes such as tempera-
ture, acoustics, lighting, daylight and air quality [44]. The second
category consists of attributes related to the spatial environment,
such as classroom layout [45], classroom furniture [5], visibility
and accessibility of sightline [46]. The third category encompasses
technology-related attributes including appropriateness of func-
tions of high-tech hardware, ease of software use [47], and speed
of net transfers.
These three types of physical attributes are correlated [36,48]
and closely related to the learning outcomes and student
behavior [32,49], which in turn determine student satisfaction and
performance. For example, some of the prior work has focused on
learning environment attributes that influence student satisfaction
[50e53]. Aldridge and Rowley found that high-quality education
environments, better learning opportunities and positive percep-
tions strongly improved the student performance [54]. Hill and
Epps suggested that attributes with satisfactory conditions, such as
lighting, temperature, and space management, increased student
satisfaction with learning environments [33]. There is no perfect
classroom environment to satisfy all types of academic activities
[35], sometimes not much can be done to change a specific attri-
bute [36], and how much a physical environment matters depends
on a student’s perceptions [55]. Accordingly, student perceptions
should be treated as important determinants by efforts that
focus on improving learning environments in higher education
settings [56].
2.1. Ambient, spatial and technological attributes
The ambient attributes studied in this paper are temperature, air
quality, acoustics and lighting. Surveys and quantitative tests found
temperature to be the most influential ambient attribute in deter-
mining student perceptions of classrooms [57,58]. Various studies
have shown that students easily accept slightly cool thermal con-
ditions [59] but prefer slightly warm environments [60], and
temperatures above 74F influence student performance in math
and reading [61]. Air quality is closely tied to temperature and
ventilation as it defines the concentration of indoor pollutants,
especially that of CO2 [62,63]. Poor air quality has been correlated
with low efficiency, high absence rate, unsatisfactory performance
and failure [64,65]. Regarding classroom acoustics, student per-
ceptions are affected by internal environmental sounds such as
 Z. Yang et al. / Building and Environment 70 (2013) 171e188
HVAC fans or student conversations [66], as well as by external
sources such as the sound of traffic [67]. Poor classroom acoustics
can contribute to a negative learning environment for students
[68], as excessive noise causes distraction and annoyance [69]. Any
object that produces background noise should be eliminated [32]
and other acoustical reverberations must be carefully considered
and evaluated [70]. The perception of lighting is governed by light
levels, spatial distribution of light, glare and color rendering in a
space. There are varying opinions on whether natural or artificial
light provides optimal classroom conditions. Natural light is known
to have positive psychological and biological effects on students.
Natural light can improve student feeling, behavior, and concen-
tration [58,71] but can also result in undesirable glare and solar
heating effects [72]. In most cases, where natural light is available,
artificial light is still required to maintain appropriate levels at all
times of day, and abundant windows can also cause distraction
[73]. Computers and writing surfaces such as white boards can
reflect unwanted glare in classrooms [74]. Poor lighting can result
in headaches, eyestrain, and fatigue. Lighting colors and levels have
also been shown to impact perceptions of thermal conditions [75],
and different colors may elicit positive feelings from students and
promote learning [76]. Finally, personal factors like gender and age
have debatable influence on preferences for lighting color and
illumination [77e79].
The spatial attributes studied in this paper are room layout,
furniture and visibility. Layout is defined as the arrangement and
boundaries of a space, while furniture involves the comfort, ergo-
nomics, and functionality of chairs and work stations. Visibility in a
classroom is determined by distance and line of sight between
students and an instructor or visual aids such as a projector or black
board. Compared to ambient attributes, spatial attributes are less
standardized by industry codes, and fewer studies have focused on
the impact of spatial design on student performance. However,
some consensuses have been achieved. For example, placing the
lectern at a short side of the room enhances visibility [80], as a clear
line of sight must be ensured for every seat [32]. The size and shape
of the classroom are also of great importance to provide space for
basic classroom interactions and activities [12,73]. Spacing between
seats and rows should be ample enough to allow for movement and
accessibility [74,81]. Sometimes this space could be organized by
adding shelves or cubbies for privacy [82,83]. Additionally, path-
ways between work stations and trafficked areas should be sepa-
rated clearly [83,84]. The placement of the instructor’s desk can
also be manipulated, for example, placed in front of or in the middle
of the classroom [85,86]. Typically, lecture classes benefit from
rows of seats, which keep students more focused and attentive to
the instructor. With this arrangement, students face away from
visual distractions [32,74,81], and the students who sit in front
rows or central rows are more involved in classroom activities and
the lecture [87], as the reduced distance guarantees student-
instructor eye contact [88] and clear visibility of blackboards,
whiteboards or projector screens [89]. For discussion-type classes,
group clusters or horseshoe arrangements are supportive of stu-
dents’ interactions [87], but a clear distinction between individual
and group areas should be made [32]. Finally, ergonomic and
adjustable furniture is considered optimal to fit students of
different sizes [90]. Ergonomically comfortable furniture also keeps
students focused and attentive in classroom environments [91,92].
Permanently attached furniture should be minimized to increase
the flexibility of space usage [93,94].
The technological attributes studied in this paper are hardware
(e.g., projector, computer, clicker, smart board, camera, etc.) and
software (e.g., software installed on instructor and student com-
puters and the Internet). In recent years, the growing use of
equipment including computers, smart boards, DVD players, and
173
projectors has motivated researchers to assess the impact of tech-
nology on learning outcomes and students’ evaluation of their
learning environments. Assessing technology as an educational
tool, Schmid et al. found that low and medium uses of technology to
be more effective in classrooms than high use [9]. Shuell and Faber
found that technology was most effective when it was used as a
“cognitive tool” rather than a “presentation tool” [95]. Finally,
Lowerison et al. found no significant relationship between com-
puter use and global course evaluations, indicating the potential
ineffectiveness of this classroom technology [8]. The same study,
however, also found that students value computer use outside the
classroom, for activities such as career training and social interac-
tion. AV (Audio Video) technology, such as projectors, speakers,
document readers, and digital writing surfaces, are intended to
improve students’ acoustical and visual perceptions. Kennedy et al.
demonstrated that increased use of visual aids improved students’
listening ease. Often, the use of teaching technology results in
students taking fewer notes in class, as the lectures are faster-paced
and digitally available [96]. However, all classroom technologies
should be operated proficiently [35] to ensure that no time is
wasted and technical issues do not interrupt students’ learning
[73]. In addition, students with higher self-reported GPAs were
found to have lower expectations for the use of projectors as
teaching aids [97]. These findings reveal that while technology can
be an effective teaching tool, its use in classrooms does not guar-
antee improved student satisfaction or performance, and may even
result in the disengagement of students from the content taught in
classrooms [98].
3. Survey methodology
Survey-based investigations are preferred in the assessment of
learning environments compared to direct observations as ob-
servers may miss important data, and students are the only ones
who can provide their perceptions of learning environments [99].
As instruments for learning environment research, surveys are well
developed and widely applicable to different lines of inquiry
involving classroom attributes. The learning environment in-
ventory for student assessment of classrooms [100,101] and the
college and university classroom environment inventory for eval-
uating higher education classrooms [102,103] are two of the most
commonly used surveys. This study also uses a survey-based
approach to investigate the impact of a wide range of physical at-
tributes on student satisfaction and performance in higher educa-
tion classrooms.
3.1. Survey design
To gather data on students’ evaluations of their classroom en-
vironments, an online self-administrated survey (Appendix I) was
developed and distributed to students enrolled in courses in six
preselected classrooms of a university. Students were provided a
link to the survey in advance and were instructed to take the survey
within 15 min after the class ended. The research team had gone
through several iterations and both education professionals and
students tested the validity of survey questions in an attempt to
minimize bias and misinterpretation of the survey questions. A
campus IRB approval was obtained for this study.
The first part of the survey asked students to provide personal,
non-classroom information, such as their gender, college year,
seating locations in the classroom, their cumulative GPAs, and ex-
pected course grades. Students’ responses were collected to vali-
date the data and evaluate non-classroom factors that potentially
impact student perceptions of their learning environments
(research question 3).
174 Z. Yang et al. / Building and Environment 70 (2013) 171e188
Table 1
A double five-point Likert scale for satisfaction and impact votes.
Rate your satisfaction with the Rate the degree to which you believe
following attributes in this the following attributes in this
classroom classroom impact your performance
Very dissatisfied Very satisfied No impact Large impact
1 2 3 4 5 1 2 3 4 5
The second part of the survey was a double Likert scale for rating
student satisfaction (the satisfaction vote) and students’ rating of the
impact of ten classroom factors on their performance (the impact
vote) (Table 1). A double scale is used to increase the reading con-
sistency and ease of understanding [104]. Gathering the satisfaction
vote and impact vote separately and then integrating the results make
the rating process more convenient and precise than asking stu-
dents to directly describe their perceptions of the classroom attri-
butes (research question 1). Based on the literature review, the
following ten attributes were included in this paper: temperature,
air quality, artificial lighting, daylight (only for classrooms with
windows, students in the classroom without any daylight could
answer “NA” to the questions related to daylight), acoustics (the
audio contact with instructor and the ability to hear the presenter,
etc.), visibility (ability to see blackboard, whiteboard, projector,
visual aids, etc.), furniture in classrooms, classroom layout, hard-
ware (projector, computer, clicker, smart board, etc.), and software
(software installed on classroom computers, and the Internet).
The third part of the survey asked students to select which pre-
defined conditions in the ambient, spatial, and technological cate-
gories were present in their classrooms (research question 2). For
example, students could select whether the classrooms were cold,
humid, or too bright. This portion of the survey aimed to investigate
the relationships between perceptions of classroom attributes and
corresponding descriptive conditions and to explore the possible
determinants of certain perceptions.
3.2. Survey administration
Six classrooms were selected to reflect an accurate sampling of
existing and renovated classrooms as well as a range of higher
education classroom types with various levels of technological
capabilities. These classrooms were selected in collaboration with
the university’s facilities management and technology enhanced
learning groups. The selected classrooms included two distance
education network (DEN) classrooms with high-tech capabilities to
record and show lectures online (Fig. 1), two large auditorium
Fig. 1. Distance education network (DEN) classroom 1 (left) DEN classroom 2 (right).
classrooms with seating capacities of approximately 350 students
and 150 students (Fig. 2) and two typical discussion classrooms
with different orientations and layouts (Fig. 3). The detailed spec-
ifications of the six classrooms are described in Table 2.
A total of 29 courses were taught in the selected classrooms
during the two periods of the survey. These courses were offered in
wide range of disciplines including engineering, business, science,
and the humanities. The first survey period spanned four weeks
from January 18 to February 12, 2010. The second survey period also
lasted four weeks from November 8, 2010 to December 3, 2010. The
first period coincided with the beginning of the spring semester,
while the second period was at the end of the fall semester,
concluding prior to finals. Conducting the survey during two
separate semesters allowed for increased diversity in the surveyed
students. The survey was accessible via smart phones and com-
puters with the intention that students would record their class-
room experience immediately after their classes ended. The start
time and end time of each response were recorded; most of the
collected responses (98% of respondents, 659 responses) were
within 15 min after the class end time (the remaining 15 re-
spondents submitted their response within 3 h after the class end
time). The results from the two-stage survey were analyzed first to
test their consistency. The distributions of responses to all ques-
tions in both periods were not significantly different (p > 0.05 for
two tailed t-test). Therefore, aggregated results were used in the
analysis.
3.3. Respondent characteristics
All students in each of the six classrooms were eligible to
participate in the survey. A total of 674 responses were received
from the two-stage survey: 255 responses (out of 450 distributed
questionnaires) during the first survey period and 585 responses
(out of 662 distributed questionnaires) during the second period.
The higher participation rate during the second survey period is a
consequence of improved means of communication with students
by having direct access to student email addresses. Finally 627 valid
responses were used for analysis after removing incomplete,
duplicate or invalid responses. With a 90% confidence level, the
margin of error was 3.1%. Of the valid responses, 36% were fresh-
men, 22% were sophomores, 14% were juniors, 12% were seniors,
12% were graduate master’s students, and 3% were graduate PhD
students. The higher percentages of undergraduate underclassmen
can be attributed to the fact that the large auditoriums, which seat
the most students, typically hosted large freshmen and sophomore
courses. The gender ratio of respondent students was 57% female
and 43% male.
 Z. Yang et al. / Building and Environment 70 (2013) 171e188
Fig. 2. Auditorium classroom 1 with 350 seats (left) and auditorium classroom 2 with 150 seats (right).
4. Methodology for data analysis
4.1. Perception scores
A perception score was calculated for each classroom attribute
based on a student’s satisfaction vote and impact vote for each
classroom attribute. Due to the fact that dissatisfaction leads to
negative perception while satisfaction is associated with positive
perception, each satisfaction vote was subtracted by 3 to be sym-
metric from the origin. The product of the transformed satisfaction
vote and corresponding impact vote then yielded a score, called the
perception score, between 10 and 10. Integrating the satisfaction
vote with the impact vote for assessment is crucial as learning is the
primary function of classroom environments. Ambient, spatial, and
technological attributes that result in optimal student satisfaction
do not necessarily result in optimal student performance. For
example, a certain light level proven to be the most satisfying
lighting in experiments could be reported by a student as bright for
studying, and a slightly cooler classroom that may be unsatisfactory
could cause the student to be more focused.
The perception score for one classroom attribute might affect
the assessment of another. Accordingly, the interdependency
among the ten classroom attributes must also be evaluated,
instead of treating each attribute as an independent variable. For
example, lighting and air quality can both impact perceptions of
temperature. For this purpose, the perception scores were further
analyzed using a factor analysis (FA) for identifying the dimension
of a respondent’s selections. Specifically, principal component
analysis (PCA) was used for determining the interdependency of
classroom attributes and measuring their underlying clustering.
Prior to performing the PCA, two prerequisites were satisfied. First,
the Kaiser-Meyer-Olkin (KMO) value was examined; the KMO is
used for measuring sample adequacy with a threshold of 0.5 (>0.6
Fig. 3. Discussion classroom 1 (left), discussion classroom 2 (right).
175
mediocre, >0.7 middling, >0.8 meritorious, >0.9 marvelous for
the factor analysis). Second, the p value of Bartlett’s test of sphe-
ricity was calculated to test whether the correlation matrix is an
identity matrix. If the significance value is above 0.05, the factor
scores could not produce an identity matrix, making the PCA
inapplicable [105].
4.2. Regression interpretation for perception scores
After obtaining the perception score for each classroom attri-
bute and identifying their mutual dependencies, the relationships
between the perception scores of classroom attributes and corre-
sponding descriptive conditions were investigated, and the
possible determinants of each attribute perception were explored.
The students were asked to select the presence of conditions to
describe current status of each classroom attribute, such as a “stuffy
air” condition for the “air quality” attribute, from which the relation
of each condition to final perception score was measured. A multi-
linear regression was run for weighing the relations and test for
statistical significance. The larger the calculated coefficient is, the
closer the relationship between the descriptive condition and the
corresponding perception. If a certain condition had no significant
influence in a multi-linear regression formula, its coefficient was
given the value of 0. A positive coefficient indicated a positive
relation and vice versa.
There are two methods for analyzing the influences of condi-
tions and their mutual relations. One is the adjusted determination
coefficient (Adjusted R Square), used to represent the percentage of
the dependent variable (perception score) that can be explained by
the independent variables (descriptive conditions). The other
method is tolerance for assessing the multicollinearity between
variables. The smaller the tolerance value is, the stronger the
multicollinearity appears.
 176
Table 2
Specifications of the six classrooms.

Classroom Layout Dimension Capacity Lighting Window HVAC Ventilation Furniture Technology
(width*depth
*height)

DEN 1 Seats face the front; rigid rows of 25ft*40ft*12ft Medium Artificial No window Central HVAC, Mechanical ventilation Low flexibility AV wall; DVD player; projector;
table arrangement; depth to width occupancy lighting with controllable with reported air push-button control panel; auxiliary
aspect ratio doesn’t exceed 1:1.5; (25e45) thermostat exchange issues audio/video input jacks; PC

Z. Yang et al. / Building and Environment 70 (2013) 171e188

has carpeted area computer; clicker base station
DEN 2 Seats face the front; rigid rows of 32ft*30ft*12ft Low Artificial Inoperable Central HVAC, Mechanical and natural Low flexibility AV wall; touch screen control
table arrangement; depth to width occupancy lighting windows with controllable ventilation panel; DVD player; projector;
ratio exceeds 1:1.5; has carpeted (15e30) and daylight thermostat auxiliary audio/video input jacks;
area PC computer interface; clicker
base station
Auditorium 1 Ramped, seats face the front; fixed 50ft*80ft*20ft Very large Artificial No window Central HVAC, Mechanical ventilation No flexibility; seats AV wall; tethered microphone;
seat arrangement; fold-up writing occupancy lighting no thermostat with some air flow issues bolted to the ground touch screen control panel; DVD
surface; depth to width ratio exceeds (300e350) player; auxiliary audio/video
1:1.5; two out-swinging doors; no input jacks; PC computer
carpeted area interface; assistive listening
system; document camera
Auditorium 2 Ramped, seats face the front; fixed 55ft*35ft*18ft Large Artificial Small Central HVAC, Mechanical ventilation, No flexibility; seats AV wall; DVD player; auxiliary
seat arrangement; fold-up writing occupancy lighting skylight recently upgraded recently updated air and bolted to the ground audio/video input jacks; PC
surface; depth to width ratio doesn’t (130e150) and daylight temperature control comfort system computer interface; push button
exceed 1:1.5; two out-swinging doors; controlled room
no carpeted area
Discussion Seats face different directions; group 32ft*35ft*12ft Medium Artificial Inoperable Central HVAC, with Mechanical and natural High flexibility; AV wall box based media; DVD
classroom 1 seating arrangement; depth to width occupancy lighting windows controllable ventilation flip-top tables; easy player; auxiliary audio/video
ratio exceeds 1:1.5; has carpeted area; (20e40) and daylight thermostat for layout changes; input jacks; PC computer
flexible layout chairs on wheels interface; with reported
configuration issues
Discussion Seats face the front; group seating 45ft*25ft*12ft Medium Artificial Inoperable Central HVAC, Mechanical and natural High flexibility; AV wall box based media; DVD
classroom 2 arrangement; occupancy lighting windows with controllable ventilation, with some flip-top tables; easy player; auxiliary audio/video
depth to width ratio doesn’t exceed (40e60) and daylight thermostat air flow issues for layout changes input jacks; PC computer
1:1.5; has carpeted area; flexible interface; special request audio
layout enforcement; assistive listening
system
 Z. Yang et al. / Building and Environment 70 (2013) 171e188
Table 3
Internal consistency (alpha reliability) for Perception Score and Regression
Interpretation.
Alpha reliability
DEN Auditorium Discussion
classrooms classrooms classrooms
Perception score 0.72 0.65 0.67
Regression 0.63 0.70 0.62
Interpretation
In order to further test the internal consistency of the survey,
Cronbach Alpha reliability [106] (the most commonly used mea-
sure of internal consistency) is calculated for both the perception
score and regression interpretation. The results are shown in
Table 3, from which it can be seen that all of the Alpha reliability
values are above 0.5, demonstrating the survey is internally
consistent for different classrooms [107].
4.3. Non-classroom factor analysis
In addition to the perception score, and the correlation and
regression analyses, the non-classroom factors were considered as
fingerprints. ANOVA (Analysis of Variance) was used to test for
significant differences in mean values of samples from more than
two populations raised by the class intervals of the target non-
classroom factor. When the calculated probability PðH0 Þ is greater
than 0.05, the hypothesis that the target factor has no impact
should be accepted. If PðH0 Þ  0:05, it can be concluded that the
non-classroom factor has a significant impact. When H0 is rejected,
a multiple comparison procedure (Tukey’s Procedure or T Method)
was then carried out to analyze which of the means differ from one
another. With the help of the Studentized range distribution, a
collection of confidence intervals with simultaneous confidence
level 95% for all pairwise comparisons can be obtained. The interval
that doesn’t include 0 yields the conclusion that the corresponding
means differ significantly from each other.
5. Survey analysis
5.1. Perception scores
Ten classroom attributes were included in the analysis. These
attributes are temperature, air quality, artificial lighting, daylight,
acoustics, visibility, room layout, furniture, hardware, and software.
The data set has an acceptable internal consistency (a measure
Fig. 4. Mean satisfaction votes and impact votes and perception scores for each classroom attribute (solid blue line: vote threshold). (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.)
177
based on the correlation to test whether the same general con-
structs could produce same results), with a Cronbach’s a value of
0.824 for the satisfaction vote and 0.749 for the impact vote. The
mean satisfaction vote and impact vote for each of the ten class-
Overall room attributes are presented in Fig. 4.
It can be seen in Fig. 4 that both the satisfaction votes and
0.62 impact votes for eight of the ten attributes are above the thresholds
0.60 (satisfaction vote ¼ 0 and impact vote ¼ 3). The students consid-
ered air quality, visibility, furniture, hardware and artificial lighting
very satisfactory (satisfaction vote>¼ 1). The students were least
satisfied with daylight and acoustics in the classrooms (satisfaction
vote 0.5). Meanwhile, they rated visibility, and hardware as
having very large impact on their performances (impact vote4).
Students believed temperature, artificial lighting, acoustics, furni-
ture and software impacted their performance (3 < impact
vote<4). Perception scores for most of the attributes were high
(perception score>2), indicating that students put a premium on
their current learning environments (Fig. 5). Although the students
were not satisfied with daylight, they indicated this attribute does
not impact their performance substantially. However, this was not
the case for acoustics. The students were not satisfied with
acoustics and indicated that acoustics have a somewhat high
impact on their performance.
Regarding the factor analysis, the KMO value was 0.698, above
the acceptable threshold (>0.6 mediocre, >0.7 middling, >0.8
meritorious, >0.9 marvelous for the factor analysis) indicating the
survey responses were sufficient for the factor analysis. The asso-
ciated p ¼ value was 0.001, demonstrating the correlation matrix is
not an identity matrix, therefore, the Bartlett’s test is passed and
the PCA is applicable for clustering the classroom factors. The factor
cluster and items loading over 0.1 are shown in Table 4.
The PCA with Varimax rotation was conducted using the SPSS 20
software. Cluster 1 explains 31.25% of the variance in the overall
perception. This clustering includes the attributes of visibility,
furniture, and room layout, all of which are related to spatial
characteristics and can therefore be named as “spatial attributes.”
Cluster 2 explains 28.98% of the variance. It contains the four at-
tributes related to the ambient environment, which are air quality,
temperature, artificial lighting, acoustics and daylight. They are
named as “ambient attributes.” Cluster 3 accounts for 11.61% of the
variance. It encompasses hardware and software, which are
referred to here as “technological attributes.” The clustering results
of spatial, ambient and technological factors are consistent with the
categories from the literature review, and indicate that overall
student perceptions are highly reliant on the spatial-category and
ambient-category attributes, which are greatly impacted by the
design, management and maintenance of classrooms.
178 Z. Yang et al. / Building and Environment 70 (2013) 171e188
Fig. 5. Mean perception scores for each classroom attribute (lateral axis: perception score threshold).
5.2. Regression interpretation results for perception score
5.2.1. Spatial attributes
For evaluating the student perceptions of visibility in the
classrooms, the students were asked to select the conditions
affecting their abilities to see. A linear regression was used to
analyze the relationship between the conditions and corresponding
perception scores. For all of the coefficients, the five conditions
account for 79.2% of the student visibility perceptions. As shown in
Fig. 6, the most positively significant conditions are “none” and the
perception of “slope of classroom”, indicating the presence of both
conditions could increase the perception score, while the most
negatively significant conditions are “sight block by other students”
and “far from the front of the classroom,” their existence would
lower the perception score. The least significant relation is the
condition of “sight block by equipment”. Based on the tolerance
results, all conditions except “none” have tolerance values around
0.8, close to 1.00, indicating they are independent from each other
but co-linear with the condition of “none”. Low tolerance value
means the target conditions is strongly correlated to other condi-
tions and its coefficient may not to be significantly different from 0.
When analyzing the student perceptions of furniture, 57.4% of
the perception score is explained by the three descriptive condi-
tions (Fig. 7). The “level of comfort with furniture” has a higher
significant relation to furniture perception than any of the other
conditions. Specifically, very weak dependencies are found be-
tween furniture perception and “number of chairs and desks,”
which implies that students are more impacted by the functionality
and comfort of furniture than by the amount of furniture.
Room layout is another spatial attribute studied in this paper.
The linear regression results show 51.2% of the perception score can
be attributed to those three conditions (Fig. 8). The most influential
condition is the “layout for interaction and collaboration with
others,” and “sufficiency of workspace for course tasks” has the
Table 4
Principal component analysis for ten classroom factors with loading values.
Classroom attributes Cluster 1 Cluster 2
Visibility 0.813
Furniture 0.758
Room layout 0.523
Air quality 0.804
Temperature 0.786
Artificial lighting 0.547
Acoustics 0.382
Daylight 0.215
Hardware
Software
second most significant relationship to room layout perception.
Room layout perception is inversely correlated with the “suffi-
ciency of space for moving around classroom”.
5.2.2. Ambient attributes
For the temperature perception analysis, the linear regression
results are presented in Fig. 9. From the adjusted R square, a ma-
jority of the perception score (76.3%) was determined by the pre-
defined conditions, within which “heat from sun” was the most
significant condition influencing the temperature perception, due
in part to local weather conditions in Southern California.
“Noticeably different temperature than other classrooms or hall-
ways” also greatly determined the student perceptions of class-
room temperature. As “heat from classroom equipment” shows a
small relationship, the major contributors to negative temperature
perception appear to be cold air from outside. The results also
showed that heat always has a positive relationship to perception,
while cold air is inversely related to perception. The students prefer
warmer learning environments and experience discomfort when
the building HVAC system is running excessively in classrooms.
For evaluating the student perceptions of air quality in the
classrooms, as can be seen in Fig. 10, 86.4% of the perception score
was determined by the five conditions. “Stuffy air” has the most
significant relationship to negative perception with air quality
(highest absolute value in coefficient), ranking above “dirty air,”
“odorous air”, and “humid air,” all of which are inversely related to
the perception score. In contrast, “dry air” and “drafty air” have
positive coefficients, indicating students prefer dry and well-
ventilated learning environments. Finally, the tolerance values of
all coefficients are above 0.75 and close to 1, indicating these con-
ditions are independent from each other.
The determinants of perception scores for artificial lighting and
daylight are analyzed separately. For artificial lighting, the regres-
sion results are presented in Fig. 11. The eight conditions determine
82.4% of the perception score. Specifically, “undesirable color” and
“too much glare” are found to have the most significant relation-
ships to artificial lighting perception, followed by the condition of
“too bright.” This result indicates that students are sensitive to and
Cluster 3 negatively affected by excessive brightness, glare and undesirable
color. By calculating multicollinearity, all tolerance values are
slightly below 1, demonstrating all conditions are independent and
could not be linearly expressed by others.
Daylight analysis is done only for the classrooms with windows,
and 76.3% of the perception score can be interpreted by the pro-
vided conditions, especially the positive condition of “adequate
illumination” and the negative conditions of “too dark” and
0.547 “insufficient sunlight/windows” (Fig. 12). The results show internal
0.383
consistency: several conditions highly related to daylight
 Z. Yang et al. / Building and Environment 70 (2013) 171e188
Fig. 6. Coefficients of conditions for interpreting visibility perception and multicollinearity among descriptive conditions (Adjusted R Square ¼ 0.792).
perception suggest that adequate sunlight and brightness are
essential to students, although “too bright” and “too much sun-
light” cause a slight negative perception.
Concerning the perception of acoustics, the students were asked
to select the environmental sounds that affect their abilities to hear
the presenter in the classroom. The causal relationship is analyzed
and shown in Fig. 13. From the results, the adjusted R square
indicated that 82.8% of the perception score could be determined
by the predefined environmental sounds. “None” is the most
positively significant condition affecting the acoustics perception,
suggesting that students prefer quiet learning environments. As
sounds from air conditioning vents were proven to be the second
most influential condition, noise from HVAC system actually causes
issues. Finally, students are more sensitive to external noise from
people speaking outside the classroom than they are to the sound
of speech among their classmates in the same room.
6. Non-classroom factor analysis
Non-classroom factors can provide important contextual infor-
mation about student perceptions. For example, previous studies
have found females to be generally less satisfied with their thermal
environments than males [10,108]. This section investigates the
differences caused by non-classroom factors in the perception
scores. Students were asked to provide the following non-
classroom factors about themselves: gender, college year, seating
location, cumulative GPA (QeQ plot was applied to analyze the
normality of the GPA distribution and the significance value is
larger than 0.5, indicating GPA data is thus approximately a normal
distribution with mean of 3.3), and their prediction of the grade
they expected to earn for current course. The results of ANOVA
provide insights into the contextual impacts on perception, and
Fig. 7. Coefficients of conditions for interpreting furniture perception and multi-
collinearity among condition conditions (Adjusted R Square ¼ 0.574).
179
references for classroom design, maintenance and management
practices.
It is necessary to confirm that all of the non-classroom factors
are independent from each other. The following hypothesis is
tested: there is no bivariate correlation among non-classroom
factors. Four factors - gender, college year, seating location and
GPA ratio - are taken into consideration, and the correlation sig-
nificance values are shown in Table 5. The GPA ratio is an index for
expressing normalized expectations for a course outcome. It is
calculated by dividing the expected grade by the cumulative GPA.
GPA ratio is chosen as a metric rather than the expected grade or
cumulative GPA, as it better reflects normalized expectations from a
student. The factor expected grade alone is too subjective, and cu-
mulative GPA alone depends on a student’s previous academic
performance and may be unrelated to the student perceptions of
their current classroom environment. Basically, five class intervals
for GPA ratio were set: very high expectation (2), high expectation
(1.33e2), normal expectation (0.75e1.33), low expectation
(0.75 ¼ 0.5), and very low expectation (0.5).
Based on the results (Table 4), all significance values are above
0.05 (commonly used significance level in statistics), indicating the
hypothesis should be accepted. There is no correlation between any
bivariate non-classroom factors, demonstrating the in-
dependencies and laying the foundation for the following factor
variance analysis. The variations of each non-classroom factor are
listed in Table 6.
For assessing the impact of variations of non-classroom factors
on student perceptions, the encoded results are processed using
ANOVA and presented in Table 7. When the significance value
(P-value) for a specific variation of a non-classroom factor is less
than 0.05, that variation is interpreted as causing a significant dif-
ference in the corresponding perception.
Fig. 8. Coefficients of conditions for interpreting room layout perception and multi-
collinearity among descriptive conditions (Adjusted R Square ¼ 0.512).
180 Z. Yang et al. / Building and Environment 70 (2013) 171e188

Fig. 9. Coefficients of conditions for interpreting temperature perception and multicollinearity among descriptive conditions (Adjusted R Square ¼ 0.763).

Fig. 10. Coefficients of conditions for interpreting air quality perception and multicollinearity among descriptive conditions (Adjusted R Square ¼ 0.864).

Fig. 11. Coefficients of conditions for interpreting artificial lighting perception and multicollinearity among descriptive conditions (Adjusted R Square ¼ 0.824).

Fig. 12. Coefficients of conditions for interpreting daylight perception and multicollinearity among descriptive conditions (Adjusted R Square ¼ 0.763).
 Z. Yang et al. / Building and Environment 70 (2013) 171e188
Fig. 13. Coefficients of conditions for interpreting acoustics perception and multicollinearity among descriptive conditions (Adjusted R Square ¼ 0.828).
For gender, temperature, acoustics, visibility, and software are all
significant (with P-values below 0.05), indicating those perceptions
vary significantly with gender (Table 6). Specifically, female stu-
dents generally evaluate temperature and visibility more positively
but acoustics and software more negatively compared to male
students (Table 8) ([ is relatively high score; Y is relatively lower
score, e means there is no significant difference).
Regarding the impact of college year on the perception scores,
the students have no disparate requirements for hardware, soft-
ware, temperature, acoustics and daylight. In contrast, student
perceptions vary by college year on the following attributes: visi-
bility, air quality, artificial lighting, furniture, and room layout.
According to the results from Tukey’s Procedure (Table 9), sopho-
more students rated the air quality, artificial lighting and furniture
lower than students from other years, while freshman students and
senior students gave relatively high evaluations to their learning
environments. Comparatively, graduate students, had high de-
mands for air quality and room layout but were lenient regarding
artificial lighting. College year thus appears to be highly related to
student perceptions of some of the classroom attributes. How to
satisfy students’ requirements in different years is worthy of further
investigation.
Seating location also has a marginally significant impact on the
perception of acoustics, furniture and visibility, most likely due to
uneven distributions of sound, furniture arrangement, and sight-
line barriers throughout the classrooms. Specifically, sitting in the
middle of the classroom yielded the highest perception scores for
visibility and acoustics while sitting at the right-back of the class-
room was associated with the poorest perceptions, which can be
attributed to the distance from instructor or the proximity of these
seats to entrances and hallways. Concerning the perception of
furniture, students sitting in front always provided higher scores
(Table 10). However, seating location did not significantly affect
student perceptions of temperature, air quality, artificial lighting,
daylight, room layout, hardware or software.
The GPA ratio also helps describe the variance in student per-
ceptions of their learning environments. Except for the visibility
attribute, GPA ratio was significantly associated with perceptions of
all other attributes, indicating that students with different course
Table 5
Correlation significance between non-classroom factors.
Factors Gender College GPA ratio Seating location
Gender NA 0.208 0.14 0.21
College 0.208 NA 0.304 0.473
GPA ratio 0.14 0.304 NA 0.127
Seating location 0.21 0.473 0.127 NA
181
expectations and historical performances evaluate learning envi-
ronments discriminately. From the results of Tukey’s Procedure
shown in Table 11, a student expecting lower course performance
than their overall GPA is more likely to give high perception scores
for classroom attributes of temperature, artificial lighting, acoustics
and room layout, while another student expecting to receive a
higher course grade than his or her overall GPA usually places
higher demands on the classroom and therefore be more sensitive
to spatial and ambient attributes.
7. Discussion and conclusions
A comprehensive statistical analysis was conducted to investi-
gate student perceptions of different classroom attributes, de-
terminants of each attribute perception, and the impact of non-
classroom factors on perception. The study found that student
perceptions of their learning environments highly relied on spatial
attributes (such as room layout and furniture) and ambient attri-
butes (such as temperature and air quality), and the reported per-
ceptions were roughly determined by the presence of
corresponding descriptive conditions. Additionally, non-classroom
factors provided important contextual fingerprints for student
perceptions of classroom attributes especially for visibility, acous-
tics and furniture. These findings illustrate the potential value of
effort to improve design, management and maintenance for higher
education classrooms, while also providing guidance about bene-
ficial changes to implement.
The paper investigated individual and cumulative impacts of
ambient, spatial, and technological classroom attributes that exist
in higher education classrooms. Except for the daylight, air quality,
acoustics and room layout, the perception scores for all other at-
tributes were above the threshold (perception score ¼ 0), indi-
cating that students put a premium on their current learning
environments. Students gave low satisfaction and impact votes to
artificial lighting; they were dissatisfied with artificial lighting but
they indicated that artificial lighting did not impact their perfor-
mance. Like artificial lighting, students were dissatisfied with
acoustics. However, students specified that acoustics had high
Table 6
Differences in non-classroom factors.
Contextual factor Differences
Gender Male Female
College year Freshman Sophomore Senior Junior Graduate
Seating location Left front Right front Middle Left back Right back
GPA ratio Very low Low Normal High Very high
182 Z. Yang et al. / Building and Environment 70 (2013) 171e188

Table 7 Table 11
ANOVA analysis results for the impact of non-classroom factors on perceptions. Comparisons of perceptions scores in terms of GPA ratio.

Gender College year Seating location GPA ratio Attribute Very low Low Normal High Very high

Temperature 0 0.547 0.058 0.013 Temperature [ e e Y e

Air quality 0.248 0.017 0.209 0.022 Air quality e e e e Y
Artificial lighting 0.281 0.026 0.077 0.026 Artificial lighting e [ e e Y
Daylight 0.39 0.134 0.117 0.165 Acoustics [ e e Y e
Acoustics 0.025 0.213 0.012 0.027 Furniture e e e Y Y
Furniture 0.696 0 0.003 0.03 Room layout e [ Y e e
Room layout 0.117 0.017 0.058 0.024 Software e e [ e Y
Visibility 0.031 0.015 0.001 0.087
Hardware 0.074 0.738 0.175 0.005
Software 0.042 0.825 0.202 0.003
satisfaction votes from students and did not vary greatly among
classrooms. However, this paper only investigated the general
Table 8
classroom attributes, which could be divided into more detailed
Comparisons of perceptions scores in terms of gender. attributes.
This paper analyzed the relationships between reported
Attribute Male Female
perception scores and the presence of descriptive conditions to find
Temperature Y [ the determinants of each attribute perception. Some of the
Acoustics [ Y
adjusted R square values were much less than 1 (especially for
Visibility Y e
Software [ e room layout and furniture attributes), indicating there are still
unknown conditions that contribute to the perception scores. Thus,
the possible inclusion of other conditions could be explored. Future
research could include all potential conditions as possible causes of
impact on their performance, indicating that improving acoustics
attribute perception, which may make causality analysis more
might have the highest priority to improve student perceptions.
effective and reliable. With regard to conditions themselves, stu-
Finally, they gave high satisfaction votes but low impact votes to air
dents should not only select the presence of conditions but also rate
quality and room layout attributes. Further exploration into satis-
their satisfaction with each condition as an independent variable.
faction and impact votes is needed to investigate the correlations
Sensitivity analysis could then be conducted to examine how
that are uncovered in this paper.
perception varies with changes in satisfaction. Another limitation
An analysis of interdependencies among perceptions of
of this study is that the causality between the perception scores of
different attributes was performed using the PCA. The clustering
classroom attributes and the corresponding descriptive conditions
results were consistent with the categories from the literature re-
were predefined as linear. In future research, non-linear regressions
view, consisting of spatial attributes, ambient attributes and tech-
such as polynomial regression and machine learning approaches,
nological attributes. As the perceptions of some attributes were
like artificial neural networks, could be applied to capture the
correlated to each other, the adjustment to one attribute influenced
inherently probabilistic relations and even predict perception
the perception of other attributes. Therefore, all the attributes
scores.
within the same category should be considered as a whole, and
Lastly, the paper provided an analysis of variance to demon-
system optimization techniques could be applied to address indi-
strate that student perceptions of some classroom attributes were
vidual attributes. The results also revealed that spatial attributes
associated with non-classroom factors as hypothesized. The find-
should preferentially be improved in classroom design, manage-
ings also indicated that student perceptions were related to seating
ment and maintenance because they accounted for 31.25% of the
locations. Additionally, variations in course expectations and his-
variance in the overall perception. Ambient attributes were also
torical performances could lead to discriminant evaluations of
important (28.98% of the variance), and their status should be
learning environments. Better classroom climate and collaborative
continuously monitored and analyzed. In particular, temperature
atmospheres may be viable ways to eliminate the negative in-
was proven to be the most influential attribute, consistent with
fluences. A future direction for research would be to extend the
previous research findings. Technological attributes received high
ANOVA analysis to a multifactor analysis of variance for investi-
gating such multifactor situations. For example, the GPA ratio could
be integrated with college year to comprehensively analyze their
Table 9 cross-impacts on student perceptions. Also, there are several non-
Comparisons of perceptions scores in terms of college year.
classroom factors that were not examined in the current study
Attribute Freshman Sophomore Junior Senior Graduate that might also influence student perceptions. Such factors such as
Air quality [ Y e e Y classroom size, student majors and course time/durations could
Artificial lighting e Y e [ [ also be involved in the variance analysis.
Furniture [ Y e e e
Visibility e e Y [ e
Room layout e e e [ Y Acknowledgments

The authors would like to thank Otto Khera of Technology

Table 10 Enhanced Learning for his continuous support and help in selecting
Comparisons of perceptions scores in terms of seating location. the surveyed classrooms and preparing the survey. We would also
like to thank Otto Khera and Minkyu Kim for their constructive
Attribute Left front Right front Middle Left back Right back
feedback on the paper. Lastly, we would like to thank the in-
Acoustics e e [ e Y structors who helped us distribute the survey and the students,
Visibility e e [ e Y
Furniture [ [ e e e
who took the time to answer the survey. Without them, this study
would not be possible.
 Z. Yang et al. / Building and Environment 70 (2013) 171e188 183

Appendix I
184 Z. Yang et al. / Building and Environment 70 (2013) 171e188
Z. Yang et al. / Building and Environment 70 (2013) 171e188 185
186 Z. Yang et al. / Building and Environment 70 (2013) 171e188
 Z. Yang et al. / Building and Environment 70 (2013) 171e188 187

References [25] Dorman JP, Fraser BJ. Psychosocial environment and affective outcomes in
technology-rich classrooms: testing a causal model. Soc Psychol Educ
2009;12(1):77e99.
[1] Higgins S, Hall E, Wall K, Woolner P, McCaughey C. The impact of school
[26] Velayutham S, Aldridge JM. Influence of psychosocial classroom environ-
environments: a literature review. The Centre for Learning and Teaching,
ment on students’ motivation and self-regulation in science learning: a
School of Education, Communication and Language Science. University of
structural equation modeling approach. Res Sci Educ 2013:1e21.
Newcastle; Accessed online on 2005 10:04e08.
[27] Fraser BJ. Classroom learning environments: retrospect, context and pro-
[2] Wargocki P, Wyon D. Research report on effects of HVAC on student per-
spectIn Second international handbook of science education. Springer; 2012.
formance. ASHRAE J 2006;48(10):22e40.
p. 1191e239.
[3] Wargocki P, Wyon DP. The effects of moderately raised classroom temper-
[28] Chionh YH, Fraser BJ. Classroom environment, achievement, attitudes and
atures and classroom ventilation rate on the performance of schoolwork by
self-esteem in geography and mathematics in Singapore. Int Res
children (RP-1257). HVAC&R Res 2007;13(2):193e220.
Geographical Environ Educ 2009;18(1):29e44.
[4] Corgnati SP, Filippi M, Viazzo S. Perception of the thermal environment in
[29] Tseng YC. A distributed emergency navigation algorithm for wireless sensor
high school and university classrooms: subjective preferences and thermal
networks. IEEE Computing; 2011.
comfort. Build Environ 2007;42(2):951e9.
[30] Lee JC, Yin H, Zhang Z. Exploring the influence of the classroom environment
[5] Brooks DC. Space and consequences: the impact of different formal learning
on students’ motivation and self-regulated learning in Hong Kong. Asia-pa-
spaces on instructor and student behavior. J Learn Spaces 2012;1(2).
cific Educ Res 2010;18(2).
[6] Montgomery T. Space matters experiences of managing static formal
[31] Umphrey LR, Wickersham JA, Sherblom JC. Student perceptions of the in-
learning spaces. Active Learn Higher Educ 2008;9(2):122e38.
structor’s relational characteristics, the classroom communication experi-
[7] Amedeo D, Dyck JA. Activity-enhancing arenas of designs: a case study of the
ence, and the interaction involvement in face-to-face versus video
classroom layout. J Architectural Plan Res 2003;20(4):323e43.
conference instruction. Commun Res Rep 2008;25(2):102e14.
[8] Lowerison G, Sclater J, Schmid RF, Abrami PC. Student perceived effective-
[32] Guardino C, Fullerton E. Changing behaviors by changing the classroom
ness of computer technology use in post-secondary classrooms. Comput
environment. Teaching Exceptional Child 2010;42(6):8e13.
Educ 2006;47(4):465e89.
[33] Hill MC, Epps KK. Does physical classroom environment affect student per-
[9] Schmid RF, Bernard RM, Borokhovski E, Tamim R, Abrami PC, Wade CA,
formance, student satisfaction, and student evaluation of teaching in the
et al. Technology’s effect on achievement in higher education: a stage I
college environment? Acad Educ Leadersh 2009;14(1):P15e9.
meta-analysis of classroom applications. J Comput Higher Educ
[34] Espey M. Does space matter? Classroom design and team-based learning.
2009;21(2):95e109.
Appl Econ Perspect Policy 2008;30(4):764e75.
[10] Choi J, Aziz A, Loftness V. Investigation on the impacts of different genders
[35] Lei SA. Classroom physical design influencing student learning and evalua-
and ages on satisfaction with thermal environments in office buildings. Build
tions of college instructors: a review of literature. Education 2010;131(1):
Environ 2010;45(6):1529e35.
128e34.
[11] Mahdavi A, Unzeitig U. Occupancy implications of spatial, indoor-
[36] Fulton RD. A conceptual model for understanding the physical attributes of
environmental, and organizational features of office spaces. Build Environ
learning environments. New Dir Adult Contin Educ 1991;1991(50):13e22.
2005;40(1):113e23.
[37] Simonsen B, Fairbanks S, Briesch A, Myers D, Sugai G. Evidence-based
[12] Roskos K, Neuman SB. The classroom environment. Reading Teach
practices in classroom management: considerations for research to practice.
2011;65(2):110e4.
Educ Treat Child 2008;31(3):351e80.
[13] Trickett EJ, Moos RH. Social environment of junior high and high school
[38] Moos RH. The social climate scales: an overview. Consulting Psychologists
classrooms. J Educ Psychol 1973;65(1):93.
Press; 1980.
[14] Walberg HJ, Anderson GJ. Classroom climate and individual learning. J Educ
[39] Zandvliet DB, Fraser BJ. Physical and psychosocial environments associated
Psychol 1968;59(6p1):414.
with networked classrooms. Learn Environments Res 2005;8(1):1e17.
[15] Fraser BJ. Two decades of classroom environment research; 1991.
[40] Langerman PD. You can be a successful teacher of adults. National Associa-
[16] Fraser BJ. Research on classroom and school climateIn Handbook of research
tion for Public Continuing and Adult Education; 1974.
on science teaching and learning. New York: Macmillan; 1994. p. 493e541.
[41] Barth MM. Deciphering student evaluations of teaching: a factor analysis
[17] Fraser BJ. Science learning environments: assessment, effects and determi-
approach. J Educ Business 2008;84(1):40e6.
nantsIn International handbook of science education Dordrecht. The
[42] Merritt DJ. Bias, the brain, and student evaluations of teaching. John’s L Rev
Netherlands: Kluwer; 1998. p. 527e64.
2008;82:235.
[18] Goh Swee Chiew, Khine Myint Swe, editors. Studies in educational learning
[43] Fraser BJ. Classroom environment. Routledge; 2011.
environments: an international perspective. Singapore: World Scientific Pub
[44] Felix E, Brown M. The case for a learning space performance rating system.
Co Inc; 2002.
J Learn Spaces 2011;1(1).
[19] Fraser BJ. Classroom environment instruments: development, validity and
[45] Fu F. Comparison of students’ satisfaction and dissatisfaction factors in
applications. Learn Environments Res 1998;1(1):7e34.
different classroom types in higher education. Hybrid Learning: Springer;
[20] Fraser BJ. Learning environments research: yesterday, today and tomorrowIn
2010. p. 415e26.
Studies in educational learning environments: an international perspective;
[46] Tanner C Kenneth. The influence of school architecture on academic
2002. p. 1e25.
achievement. J Educ Adm 2000;38(4):309e30.
[21] Fraser BJ, Fisher Darrell L, McRobbie Campbell J. Development, validation and
[47] Yang Z, Liu Q. Research and development of web-based virtual online
use of personal and class forms of a new classroom environment instrument.
classroom. Comput Educ 2007;48(2):171e84.
In: Annual meeting of the American educational research association, New
[48] Building energy efficiency and sustainability. Johnson Controls Inc; 2010.
York 1996.
[49] Guardino C. Modifying the classroom environment to reduce disruptive
[22] Cho H, Gay G, Davidson B, Ingraffea A. Social networks, communication
behavior and increase academic engagement in classrooms with students
styles, and learning performance in a CSCL community. Comput Educ
who have a hearing loss; 2008. Doctoral Dissertation). Available from Pro-
2007;49(2):309e29.
Quest Dissertations and Theses database. (No. AAT3304010).
[23] Rueda R, Loera G, Fujii K, Moulton K, Ragusa G, Oh YJ, et al. Learning and
[50] Butt BZ. A study examining the students satisfaction in higher education.
motivational characteristics of urban high school students. Los Angeles, CA:
Procedia-social Behav Sci 2010;2(2):5446e50.
Mental Health America of Los Angeles; 2008.
[51] Temple P. Learning spaces in higher education: an under-researched topic.
[24] Frisby BN, Martin MM. Instructorestudent and studentestudent rapport in
London Rev Educ 2008;6(3):229e41.
the classroom. Commun Educ 2010;59(2):146e64.
188 Z. Yang et al. / Building and Environment 70 (2013) 171e188
[52] So H, Brush TA. Student perceptions of collaborative learning, social presence
and satisfaction in a blended learning environment: relationships and critical
factors. Comput Educ 2008;51(1):318e36.
[53] Mai L. A comparative study between UK and US: The student satisfaction
in higher education and its influential factors. J Marketing Manage
2005;21(7e8):859e78.
[54] Aldridge S, Rowley J. Measuring customer satisfaction in higher education.
Qual Assur Educ 1998;6(4):197e204.
[55] Vosko RS. Shaping spaces for lifelong learning. Lifelong Learn 1984;8(2):4.
7, 28.
[56] Siegel P. Where innovation matters, IT matters. Educause Rev 2003;38:96e7.
[57] Liu S, Yoshino H, Mochida A. A measurement study on the indoor climate of a
college classroom. Int J Ventilation 2011;10(3):251e61.
[58] Earthman GI. Prioritization of 31 criteria for school building adequacy. Bal-
timore, MD: American Civil Liberties Union Foundation of Maryland; 2004.
[59] Hwang R, Lin T, Kuo N. Field experiments on thermal comfort in campus
classrooms in Taiwan. Energy Build 2006;38(1):53e62.
[60] Jung GJ, Song SK, Ahn YC, Oh GS, Im YB. Experimental research on thermal
comfort in the university classroom of regular semesters in Korea. J Mech Sci
Technol 2011;25(2):503e12.
[61] Lackney JA. Thirty-three educational design principles for schools & com-
munity learning centers. ERIC Clearinghouse; 2000.
[62] Lee S, Chang M. Indoor and outdoor air quality investigation at schools in
Hong Kong. Chemosphere 2000;41(1):109e13.
[63] ASHRAE. ASHRAE handbook: Fundamentals. Atlanta, GA: ASHRAE; 2009.
[64] Shaughnessy R, Haverinen-Shaughnessy U, Nevalainen A, Moschandreas D.
A preliminary study on the association between ventilation rates in class-
rooms and student performance. Indoor Air 2006;16(6):465e8.
[65] Shendell DG, Prill R, Fisk WJ, Apte MG, Blake D, Faulkner D. Associations
between classroom CO2 concentrations and student attendance in Wash-
ington and Idaho. Indoor Air 2004;14(5):333e41.
[66] Dockrell JE, Shield BM. Acoustical barriers in classrooms: the impact of noise
on performance in the classroom. Br Educ Res J 2006;32(3):509e25.
[67] British Research Establishment. National noise incidence study 2000/2001
(England and Wales). Updated: 2002. Available at: www.defra.gov.uk/
envirnment/noise/nis0001.
[68] Shield BM, Dockrell JE. The effects of noise on children at school: a review.
Build Acoust 2003;10(2):97e116.
[69] Dockrell JE, Shield B. Children’s perceptions of their acoustic environment at
school and at home. J Acoust Soc Am 2004;115:2964.
[70] Lang D. Essential criteria for an ideal learning environment. Center for Envi-
ronment, Education and Design Studies, University of Washington; 1996.
[71] Rayneri LJ, Gerber BL, Wiley LP. The relationship between classroom environ-
ment and the learning style preferences of gifted middle school students and the
impact on levels of performance. Gifted Child Quarterly 2006;50(2):104e18.
[72] Winterbottom M, Wilkins A. Lighting and discomfort in the classroom.
J Environ Psychol 2009;29(1):63e75.
[73] Veltri S, Banning JH, Davies TG. The community college classroom environ-
ment: student perceptions. Coll Student J 2006;40(3):517e27.
[74] Wong NH, Jan WLS. Total building performance evaluation of academic
institution in Singapore. Build Environ 2003;38(1):161e76.
[75] Maki Y, Shukuya M. Visual and thermal comfort and its relations to exergy
consumption in a classroom with daylighting. Int J Exergy 2012;11(4):481e92.
[76] Papadatos SP. Color them motivateddcolor’s psychological effects on stu-
dents. NASSP Bull 1973;57(370):92e4.
[77] Sleegers P, Moolenaar N, Galetzka M, Pruyn A, Sarroukh B, Zande B. Lighting
affects students’ concentration positively: findings from three Dutch studies.
Lighting Res Technol 2012;45(2):159e75.
[78] Mott MS, Robinson DH, Walden A, Burnette J, Rutherford AS. Illuminating the
effects of dynamic lighting on student learning. SAGE Open 2012;2(2).
[79] Samani SA, Samani SA. The impact of indoor lighting on students’ learning
performance in learning environments: a knowledge internalization
perspective. Int J Business Soc Sci 2012;3(24):P127e36.
[80] Al-Haboubi M, Ishteeaque E. Designing new classroom buildings. J Archit Eng
2000;6(4):129e32.
[81] Classroom and technology design and construction minimum requirements.
Pennsylvania State University; 2009.
[82] Evans GW, Lovell B. Design modification in an open-plan school. J Educ
Psychol 1979;71(1):41.
[83] Weinstein CS. The physical environment of the school: a review of the
research. Rev Educ Res 1979;49(4):577e610.
[84] Visser J. Aspects of physical provision for pupils with emotional and
behavioural difficulties. Support Learn 2001;16(2):64e8.
[85] Proshansky E, Wolfe M. The physical setting and open education. Sch Rev
1974;82(4):557e74.
[86] Zifferblatt SM. Architecture and human behavior: toward increased under-
standing of a functional relationship. Educational Technology; 1972.
[87] Marx A, Fuhrer U, Hartig T. Effects of classroom seating arrangements on
children’s question-asking. Learn Environments Res 1999;2(3):249e63.
[88] Niemeyer DC. Hard facts on smart classroom design: Ideas, guidelines, and
layouts. Rowman & Littlefield; 2003.
[89] Navid Shahzad. Planning a classroom. Updated: 2003. Available at: http://
www.dailytimes.com.pk/default.asp?page¼story_22-2-2003_pg3_5; 2013.
[90] Panagiotopoulou G, Christoulas K, Papanckolaou A, Mandroukas K. Class-
room furniture dimensions and anthropometric measures in primary school.
Appl Ergon 2004;35(2):121e8.
[91] Bullard J, editor. Creating environments for learning. Pearson; 2009.
[92] Knight G, Noyes J. Children’s behaviour and the design of school furniture.
Ergonomics 1999;42(5):747e60.
[93] Lippman PC. The L-shaped classroom: a pattern for promoting learning.
DesignShare; 2004.
[94] University of Melbourne. Archive: Teaching and research services for staff:
The University of Melbourne. Updated: 2011. Available at: trs.unimelb.edu.
au/archive/history.html.
[95] Shuell TJ, Farber SL. Students’ perceptions of technology use in college
courses. J Educ Comput Res 2001;24(2):119e38.
[96] Kennedy SM, Hodgson M, Edgett LD, Lamb N, Rempel R. Subjective assess-
ment of listening environments in university classrooms: perceptions of
students. J Acoust Soc Am 2006;119:299.
[97] Tang TL, Austin MJ. Students’ perceptions of teaching technologies, appli-
cation of technologies, and academic performance. Comput Educ 2009;53(4):
1241e55.
[98] Holbert KE, Karady GG. Strategies, challenges and prospects for active learning
in the computer-based classroom. Educ IEEE Trans 2009;52(1):31e8.
[99] Brophy JE, Good TL. Teacher behavior and student achievement. Institute for
Research on Teaching, Michigan State University; 1984.
[100] Fraser BJ. Assessment of learning environments: manual for learning envi-
ronment inventory (LEI) and my class inventory (MCI). Third Version; 1982.
[101] Serena P, Anna B. Italian nursing students’ perception of their clinical
learning environment as measured with the CLEI tool. Nurse Educ Today
2009;29(8):886e90.
[102] Fraser BJ, Aldridge JM, Soerjaningsih W. Instructor-student interpersonal
interaction and student outcomes at the university level in Indonesia. Open
Educ J 2010;3:32e44.
[103] Fraser BJ, Treagust DF. Validity and use of an instrument for assessing
classroom psychosocial environment in higher education. Higher Educ
1986;15(1/2):37e57.
[104] Levermore GJ. Building energy management systems: applications to low
energy HVAC and natural ventilation control. E & FN Spon; 2000.
[105] George D. SPSS for windows step by step: a simple study guide and refer-
ence, 17.0 Update, 10/e. India: Pearson Education; 2003.
[106] Cronbach LJ. Coefficient alpha and the internal structure of tests. Psycho-
metrika 1951;16(3):297e334.
[107] Numally JC. Psychometric theory. NY: McGraw-Hill; 1978.
[108] Karjalainen S. Gender differences in thermal comfort and use of thermostats
in everyday thermal environments. Build.Environ 2007;42(4):1594e603.