Journal of Building Engineering 20 (2018) 51–59

Contents lists available at ScienceDirect

Journal of Building Engineering

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

Towards quantifying human experience in the built environment: A T

crowdsourcing based experiment to identify influential architectural design
features
⁎
Semiha Ergan , Zhuoya Shi, Xinran Yu
Department of Civil and Urban Engineering, Tandon School of Engineering, New York University, 15 MetroTech Center, Brooklyn, NY 11201, United States

A R T I C LE I N FO A B S T R A C T

Keywords: One of the main challenges in the quantification of the influence of architectural design features on human
Architecture design experience is to define the set of architectural design features that people notice immediately in a space as well
Crowdsourcing as to define the type of influence these design features can have on people. Through a crowdsourced experiment,
Human experience this study provides evidences on the architectural design features that people notice immediately in a space,
Neuroscience
preferences of people on the spaces configured with these features, and the influence level of these features on
Built environment
overall experience in spaces. Statistical analysis on around 400 subjects’ data show that certain features such as
the openness of space, presence of windows and daylighting, flexibility in isolation/socialization, level of arti-
ficial lighting, density of spaces, and color of surfaces are easy to notice by people and are also powerful to
change the human experience. The findings provide an ordering of the identified design features based on their
noticeability and influence levels for practitioners to consider in their design decisions. Findings also establish
the basis towards objectively quantifying the impact of architectural design features on human experience in
spaces.

1. Introduction human experience is the lack of knowledge on capability of individual

architectural design features on influencing human experience and how
Buildings impact people in various dimensions, including pro- differential is that influence across a group of architectural design
ductivity and well-being. Ongoing research on buildings and human features. By knowing the features that people are more responsive to
interactions has shown that healthier buildings result in 1–5% reduc- and understanding the influence these features have on human, it is
tion in healthcare costs, reduced stress, satisfaction and engagement, possible to aim for designs that enhance the experience people get in
lower absenteeism rates, and higher productivity [43,55]. Physical architectural spaces.
workspace design is also defined as a critical factor for decisions to stay Various efforts in the current design practice aim to improve design
in a job position and job performance [1]. With people spending more such as green building initiatives in response to multiple environmental
than 90% of their time indoors [31], it is essential to identify and concerns (e.g., carbon footprint, greenhouse gases), WELL building
quantify the relationship between the interior built environment and standard in response to the need for healthier buildings for occupants,
human experience, which is closely tied to enriching productivity and and evidence-based design (EBD) for increasing performance of build-
efficiency. Human experience in spaces is defined as the impact of the ings for occupants [48,73]. Particularly, WELL and EBD aim to provide
space on people’s mood, level of comfort, and interactions with sur- healthier and high performing buildings. WELL highlights seven fea-
roundings [12,74] and change with architectural design features (e.g., tures (i.e., air quality, water quality, nourishment, lighting, fitness,
symmetry of objects, connectivity of spaces, openness of spaces). Ar- comfort, and mind) to achieve wellness of people working or living in
chitectural design features are the elements that constitute a space and buildings. Under each feature, the WELL building standard provides
gives it its unique characteristic where people interact with [66]. Lit- requirements to be satisfied for a certification. WELL building standard
erature is full of discussions on various types of architectural design in overall aims for having sustainable measures for health of occupants,
features and their expected impact on experience. However, one of the and is not formulated for configuring architectural design features to
challenges in quantifying the impact of interior built environment on enhance human experience. On the other hand, EBD is a performance-

⁎
Corresponding author.
E-mail addresses: [email protected] (S. Ergan), [email protected] (Z. Shi), [email protected] (X. Yu).

https://doi.org/10.1016/j.jobe.2018.07.004
Received 16 May 2018; Received in revised form 4 July 2018; Accepted 4 July 2018
Available online 06 July 2018
2352-7102/ © 2018 Elsevier Ltd. All rights reserved.
S. Ergan et al. Journal of Building Engineering 20 (2018) 51–59

based design practice that seeks evidences from research, expert opi- effect of architectural design features, researchers mainly highlighted
nions, or occupant feedbacks to base design decisions. It is mainly ap- natural daylight levels, presence/absence of windows, size of windows
plied at healthcare facilities for patient satisfaction, healthcare provi- if present, natural views, and presence/absence of restorative images
ders’ productivity and safety, and improving healthcare performance in (e.g., a visual stimuli depicting natural scenes) as architectural design
general [63]. Typically, post-occupancy evaluations and building-use features that impact restorativeness and recovery periods (e.g.,
surveys are administered to collect self-reports of people about their [27,28,38,70,65,11,13,46,6,63,24]) (Table 1). Windows with a view of
satisfaction in the spaces. However, there is no systematic method to nature and views of natural /indoor plants have restorative influences,
detect which architectural design features are more influential in the and are related to positively-toned moods, activity levels, and sustained
satisfaction levels reported. attention (e.g., [35,69,64,4,25]). Literature also indicates the un-
This paper provides the findings of a crowdsourcing experiment pleasant outcomes (i.e., fear, dissatisfaction, and stress) caused by high-
designed and expert elicitations conducted with professional architects rise buildings which are often built with big windows if not all glass
to (a) capture the architectural design features that people are more exterior facades [77]. However, in this study, the authors concentrated
responsive to, and (b) the preferred configurations of these features in on the impact of interior design features on human experience and drew
spaces, and (c) blind responses of people on experience for a given conclusions regarding them.
configuration of a design feature. The experiment was designed by Impact of design on stress and anxiety level of individuals is also
building of off the architectural design features utilized by architects widely known. A rigorous literature review on linkage between the
and studied in the literature to give user a particular experience (e.g., stress and anxiety levels and architectural design revealed various de-
restorativeness, motivation to work) in a space. The experiment pre- sign features (Table 1). Discussion in the literature shows that when in
sented a dual image set for each architectural design feature in a rea- absence or in wrong configuration, these features could result in higher
listic space setting without revealing the design feature being eval- stress indicators in people. Synthesis of studies shows that crowding,
uated, and corresponding bipolar scales that aim to help participants to color of surfaces, level of artificial lighting, and presence/absence of
express their experience if they were to be in these spaces. A semantic visual cues (e.g., interior/exterior landmarks) were identified as part of
bipolar scale represents a rating scale that is designed to measure the environmental stressors in the built environment as shown in Table 1
connotative meaning of different architectural features. For example, [11,15,29,52,60,63,64]. Poor lighting levels proved to trigger stress in
the architectural design feature of ‘presence/absence of daylight’ in a people while navigating in spaces (e.g., [60]), change circadian
space can be rated on bipolar scales of pleasant – unpleasant, relaxed- rhythms, and result in disruption in work cycles and longer recovery
tense, energized-tired. Expert elicitations was performed with eighteen periods in hospital settings [3,56]. Different colors said to affect human
architects from the profession with diverse backgrounds and average 20 experience differently depending on the wavelength of the colors in
years of experience to define architectural design features that influ- spaces [19,61]. Occupants of office spaces colored in red and blue
ence human experience. The next section provides the results of an found to result in higher stress and anxiety levels and higher depression
extensive literature review at the interaction of neuroscience and ar- levels, respectively due to the stimulating effect of red and depressing
chitecture with a sound point of departure for the design of the ex- effect of blue [36,61]. Positive outcomes were achieved for students
periment. The outcomes of this literature review is the synthesis of working in spaces with warmer colors (e.g., pink) (hence colors with
architectural design features and corresponding human experience they longer wavelength) than colder colors (e.g., gray) (hence colors with
are related to. shorter wavelength). In addition to these design features, landmarks in
interior spaces or urban settings have been used as visual cues for
2. Background research wayfinding, spatial recognition, and sense of familiarity with environ-
ments [21,33,39,47,63]. Lack of such landmark objects triggers stress
This study builds on and extends the research studies conducted at and anxiety [5,60].
the intersection of neuroscience and architecture. The point of de- The perceived density of interior spaces (measured with respect to
parture for this study is organized as a) influence of architectural design square footage and aspect ratio), the height of ceilings, openness of
features on human experience, and b) semantic bipolar scales used at spaces, and in/flexibility in isolation vs. socialization in spaces were
the intersection of neuroscience and architecture. mainly discussed in the literature as main factors that can play a role in
the sense of crowding in a designed space (Table 1) [10,13,67,75].
2.1. Influence of architectural design features on human experience Studies showed that high density and short depth of spaces result in
distress and eventually social withdrawal [14]. Complementary re-
Architectural design features are the elements in a space that give its search reported that larger horizontal areas are perceived as more
unique characteristics [66]. Those design elements are believed to spacious [59]. Another factor that was discussed in the literature was
change the experience a user has in a space, and the degree of impact the height of ceilings, where rooms with higher ceilings being perceived
varies with each feature and its configuration. Various architectural spacious than lower ceilings – hence less sense of crowding and stress
design features that relate to human experience are discussed in the [17,67]. Aspect ratio in rooms changes the perceived spaciousness
literature. Particular interest is on the cognitive influence of those ar- where corridor like layouts appeared less spacious as compared to
chitectural features, and how people feel when they are in a space that square shaped spaces of the same size [54]. Enclosed rooms are asso-
is formed by those specific features. To synthesize the findings from ciated with higher stress levels and decisions to exit such spaces
literature in this domain, we categorized the architectural design fea- [16,67]. Finally, flexibility in getting privacy in spaces whenever
tures that are discussed in the literature for influencing human ex- needed but also providing the flexibility to enable socialization among
perience based on the ‘type of influence’ they have on people. The re- users of spaces was mentioned in the literature as a factor for reducing
sults are provided in Table 1. stress. Such studies were especially abundant in healthcare and office
As shown in Table 1, types of influence of architectural design facilities (e.g., [23,58,62,13]).
features on human experience can be categorized in four groups as The aesthetics and pleasure is the third main user experience we
‘restorativeness’, ‘stress and anxiety’, ‘aesthetics and pleasure’, and identified from former research. Studies on aesthetics in architecture
‘motivation’. Various studies at the intersection of neuroscience and and ways to design aesthetically appealing spaces are abundant
architecture looked at the impact of design on human restoration. A [26,37,44,57,71,8]. Features that are mainly discussed in the literature
restorative environment is the one that promotes restoration [22,7]. for changing the perception of aesthetics in spaces include symmetry of
The literature on this aspect is mainly on healthcare facilities, re- objects in interior/exterior spaces, spatial alignment in these objects,
habilitation and correction centers, and schools. For restorativeness shape/layout of spaces, and contour of objects [2,34,40,41]. Symmetry

52
S. Ergan et al. Journal of Building Engineering 20 (2018) 51–59

Table 1
A synthesis of previous research studies on influence of architectural design features on human experience.
Human experience Architectural design features Ref.

Restorativeness Presence /absence of natural daylight E.g., [6,11,13,35,38,63,69,70,27,28].

Presence/absence of windows E.g., [35,38,64,70,46,24,27,28].
Size of windows (if present) E.g., [70,25]
Presence of restorative nature images/Exposure to nature in spaces E.g., [11,13,35,38,46,63–65,69,70,4].
Stress and anxiety Crowding Density of spaces E.g., [10,13,14,59,67,75].
Height of ceilings E.g., [17,67].
Flexibility in isolation and socialization in spaces E.g., [23,58,62,13].
Openness of spaces E.g., [15,54,67,16].
Color of surfaces E.g., [61,19,36].
Level of artificial lighting E.g., [52,15,60,3,56,63].
Presence/absence of visual cues (e.g., interior/exterior landmarks) E.g., [47,63,5,60,4,21,39,33].
Aesthetics and pleasure Symmetry of objects E.g., [34,41].
Existence of spatial alignment E.g., [40]
Shape /layout of spaces E.g., [53,76,63,23].
Contours of objects in spaces E.g., [2,45,51,78]
Motivation Texture/material of surfaces E.g., [29].
Ease of access to spaces with a facility E.g., [32,48].
Sequence of connectivity of spaces/ Movement and usage patterns E.g., [32].

of objects is defined as the notion of how exactly similar parts mirror and Negative Affect Schedule (PANAS), where participants are asked to
each other along a central axis [34] and was defined as an essential rate their emotional experience towards a specific product or design on
determinant of aesthetics. The relationship between the sensation of a five-point Likert scale using a predefined set of words [72]. An ex-
aesthetic and symmetry was investigated earlier with EEG and EMG tended version of PANAS is called PANAS-X, which includes more
experiments and the results show that symmetrical objects increase the words (60 words instead of 20) to explain and rate emotions. A se-
sense of aesthetics and provide a sense of comparison among disparate mantic bipolar scale represents a rating scale that is designed to mea-
elements, allowing receivers to better navigate through an environment sure the connotative meaning of different architectural features. For
[41]. The layout of the room is proved to be one of the main char- example, the presence/absence of daylight in a space can be rated as
acteristics that affects people’s satisfaction for both office environment (pleasant/unpleasant). A bipolar scale prompts a respondent to balance
and healthcare facilities [53,63,76]. Literature on the contour of objects two opposite attributes, determining the relative proportion of these
and human perception showed that curvilinear objects were more likely opposite attributes with respect to the space experience a person has.
to be preferred [2,51,68], particularly for route recognition [78], Extensive literature review at the intersection of neuroscience and
warmth [2], and relaxation [45]. architecture showed that there are various research studies that eval-
Finally, the influence of design on ‘motivation’ to work and pro- uated the effect of individual design characteristics (e.g., window size,
ductivity of workers is well-established as well [23]. Features that are space size) on human experience using bipolar scales. Results show that
mainly studied in literature are identified as the presence of interior restorativeness was measured using bipolar scales of relaxed/tense,
color coding, texture/material of surfaces, ease of access to spaces with energized/tired, pleasant/unpleasant, lively/bored, and delighted/dis-
a facility, and connectivity of spaces [23,29,49]. Color and texture are appointed [18,20,4,46,5,6,9]. Stress and anxiety in space settings was
recognized as powerful design features for changing user experience measured using bipolar scales of free/encumbered, roomy/cramped,
and enabling navigation in interior spaces [29]. Warm colors are re- pleasant/unpleasant, calm/restless, delighted/disappointed, free/en-
cognized as landmark during the spatial navigation process because it closed, relaxed/tensed, happy/unhappy, clear/confused, and lively/
increases participants’ awareness [29]. General colors used in a logical bored [14,15,17,18,29,30,42,52,60,63]. Aesthetics were associated
way can also serve as signs in space navigation process. Space config- with the bipolar scales of focused/distracted, ordered/chaotic, calm/
uration in terms of ease of access to spaces and sequence of connected restless, pleasant/unpleasant, comfortable/uncomfortable, relaxed/
spaces are correlated with the level of productivity [32]. Sequence is tense, excited/bored, and engaged/aroused [10,17,30,40]. Finally,
related more on getting the map of an area and using this information motivation in architectural spaces were associated with bipolar scales
for increasing efficiency in work spaces. of excited/bored, delighted/disappointed, calm/restless, and lively/
As discussed so far, the synthesis of previous research studies re- bored (e.g., [63,60]).
sulted in four major categories of human experience in architectural These previous studies constitute a point of departure in defining
design spaces and 18 features to examine. The gap still exists in the the semantic bipolar scales that can be used for measuring the influence
literature to understand how configurations of these architectural de- of each architectural design feature on human experience. The a subset
sign features impact a certain experience (e.g., stress, approach moti- of identified scales (through merging of the synonymous scales) were
vation) and the level of influence each design feature has on people. used in the crowdsourcing experiment to identify a subset of statisti-
This research builds on these identified 18 design features (with com- cally significant semantic differential scales that people refer-to explain
bined findings from expert architect elicitations), defines their influence their experience in spaces.
levels on people, their frequency of noticeability by people; and the
preferred configurations in spaces for positive experiences.
3. Research method and scope

2.2. Semantic bipolar scales used at the intersection of neuroscience and The objective of this study is to define the architectural design
architecture features that people are more responsive to (how much each feature
influences people, and their noticeability levels) along with the bipolar
A common method used in research to study human experience scales that are differential in capturing human experience, and identify
regarding a specific design is to obtain self-reports from participants to the preferred configurations of these features (through collecting blind
define and rate their feelings. This method is also normally referred to responses of people about different configurations without revealing
as a measure of effect. A widely adopted emotion scale is the Positive the underlying difference). The steps of the research method are

53
S. Ergan et al. Journal of Building Engineering 20 (2018) 51–59

from literature. Only additional feature that was identified was the
acoustics and ambient noise, which could not be represented in dual
image sets and was excluded from the study. Experts also did not
mention symmetry of architectural features and visual cues/landmarks
in their discussions. However, we incorporated these features that were
identified from literature to the study to study a comprehensive set of
features.

3.2. Generation of dual image sets for architectural design features

For each identified feature (e.g., symmetry of architectural com-

ponents, size of windows) a pair of images were generated. Each image
represents one configuration of that feature (e.g., symmetrical space vs.
asymmetrical space, regular space vs. irregular space, large windows vs.
small windows) on the far end of the corresponding bipolar scale
(Fig. 2). Dual images only had one feature changed and the rest of the
settings in that space was kept constant. The images were created to be
realistic and meaningful as much as possible to allow participants to
fully get a realistic experience in the space shown on the images.

3.3. Design of the crowdsourcing experiment and the pilot study

A total number of 17 architectural design features was represented

in this experiment as given in Table 1, except the sequence of con-
nectivity of spaces/ movement and usage patterns, as it was not pos-
sible to represent this feature in a single dual image set as discovered
with initial pilot study. This feature was excluded from the study. The
generated dual image sets were incorporated into an online crowd-
sourcing platform that was shared with public. The survey had around
80 questions with a core module and a separate module for each ar-
chitecture feature with five questions each. The core module consists of
demographic questions, and an Ishihara color blindness test plate 8 (to
eliminate participants with color blindness). Each module other than
the core asks participants to pick the space they prefer given in a dual
image set, rate the spaces provided in each dual set using the semantic
bipolar scales presented (multiple scales per dual image set), and write
Fig. 1. An overview of the steps involved in the research. descriptive words that describe their feeling for each image. The dual
images represented the positive and negative ends of corresponding
overviewed in Fig. 1. The literature review on defining architectural features evaluated based on the suggested meanings in literature. For
design features and corresponding user experiences mentioned in the example, for size of windows, a high window to wall ratio (60%) re-
literature on neuroscience for architecture was the first step of the re- presented the positive end of the scale, a low window to wall ratio
search and discussed in details in the previous section. Next step was (20%) represented the negative end of the scale. For contour of objects,
the confirmation of these findings from literature with 18 expert ar- round representations were used for the positive end versus sharp
chitects from leading architectural firms. Resulting set of architectural edged representations were used for the negative end. Participants were
design features and scales were then used to create realistic re- not given any information about the feature and their configurations
presentation of each architectural design feature in dual image sets (one being evaluated in each dual image set. The designed study was eval-
corresponding to the positive and the other one corresponding to the uated by an expert environmental psychologist and his feedback was
negative end of the bipolar scale) using building information models. incorporated to the experiment before launching it to the general
These dual image sets were used to design a crowdsourcing experiment. public. On the crowdsourcing platform, only participants from the
A pilot study with an environmental psychologist and a group of par- United States and Canada were allowed to participate in the study to
ticipants was conducted to revise the crowdsourcing experiment and eliminate cultural influences on responses.
the questions. The final step was launching of the crowdsourcing ex-
periment on a widely-used crowdsourcing platform and analyzing the 3.4. Data analysis
collected data. Each step after the literature review is overviewed
below. The results of crowdsourcing experiment with 356 participants was
analyzed in both descriptive and statistical ways. The responses were
analyzed statistically to identify the configurations that people prefer
3.1. Conducting expert elicitation for defining the list of architectural design for architectural design features, and their influence level on people.
features to use in the experiment Several statistical tests were conducted using a statistical analysis
software to evaluate the statistical significance of the results at p = 0.05
Eighteen expert architects, with an average 20 years of experience and a 95% level of confidence. The analysis involved computing means
in design, from three leading architectural firms participated in the for every architectural design feature and analyzing the data statisti-
expert elicitation phase where interviews were held with them to cally to test the statistical significance of the difference between every
identify the features they modify to change user experience in spaces. pair of images (i.e., every feature and its two bipolar configurations).
An overview of the participants is provided in Table 2. The analysis also involved testing the statistical significance of the
The results from expert elicitation overlapped with the findings difference between means for every pair of bipolar scales used for each

54
S. Ergan et al. Journal of Building Engineering 20 (2018) 51–59

Table 2
An overview of participants in the expert elicitation.
Company ID # of participants Design experience Years of experience

1 5 Healthcare; Office and workspace; Interior 15; 11; 10; 3; 24 Average: 13 years
2 4 Office and workspace; Academic; Hotel; Police facilities; Crime labs; Detention centers; Court 25; 20; 30; 24 Average: 25 years
houses; Science/research facilities
3 9 Office and workspace; Residential; Retail; Hotels; Academic; Science/research facilities; Airports; 20; 21; 10; 10; 30; 15; 30; 35; 12 Average:
Interior; Healthcare; Court houses 20 years

Feature: Symmetry of architectural components

S ymmetri cal A symmetri cal

Feature: Existence of spatial alignment in interior
spaces

A l i gned N ot A l i gned
Fig. 2. Two example architectural design features represented on two ends of bipolar scales.

architectural design feature and eliminating the insignificant bipolar is obvious that this sample was derived from a large population and
scales that cannot express the feelings of people for that feature. covered every aspect of age, gender, and education levels.
Variables collected were tested for normality to decide whether
parametric tests could be used to analyze them. Repeated measures 4.2. Descriptive analysis of preferred configurations for each architectural
ANOVA test with Mauchly's test of sphericity was used due to the design feature
nature of the sample, as it detects any overall differences between re-
lated means. The goal is to test whether there is a statistically sig- The results of the crowdsourced data analysis show that preferences
nificant difference between the configurations of the dependent vari- of people on architectural design features have a visible difference for
able (i.e., architectural design feature). The null hypothesis (H0) is that the provided configurations, as shown in Fig. 4. Responses showed that
mean rating is the same for both configurations of a given architectural people preferred the configuration of the architectural design features
design feature (e.g., symmetrical and asymmetrical). The alternative that corresponded to the positive end of the bipolar scales. In other
hypothesis is that mean rating is significantly different for those con- words, as shown in Fig. 4, people preferred natural daylight, spaces
figurations. Partial eta-squared (hp2) was used to measure the effect size with large windows, liked spaces that had exposure to nature, preferred
of each architectural feature, thus enable putting the features in an high ceilings, spaces with flexibility for isolation and socialization,
order according to their strength of influence and noticeability. A spaces that are open instead of enclosed, color coding on surfaces, have
suggested scale for hp2 = 0.02 is considered ‘small’ effect size, 0.13 enough level of luminance instead of darker spaces, spaces with visual
‘medium’ effect size, and 0.3 ‘large’ effect size [50]. Partial eta-squared cues and landmarks for geo-referencing, spaces where objects were
covers how much variance in a dependent variable (i.e., architectural symmetrical and spatially aligned, regular layouts instead of irregular
feature) is explained by the independent variables (i.e., user experience layouts in rooms, rounded surfaces instead of sharp edged ones, var-
translated into bipolar scales). iations in texture/material and spaces that has ease of access with
visible access points (e.g., a visible entrance).
4. Research findings The preferences of people on configurations of design features were
all parallel to what is discussed in literature for giving a positive and
4.1. Demographics negative user experience. So, the results were complementary to the
hypotheses tested in earlier research studies. Only for the “density of
Data was collected from an initial number of 356 participants. 26 of space” feature, 69% of the participants preferred the dense space,
them could not pass the Ishihara color blindness test, and their results which is on the contrary to the initial hypothesis. This could be partially
were eliminated from the dataset. 297 people completed the experi- attributed to the dual image set used where a socialization area (i.e., a
ment (80% of the total participants). The distribution of participants’ cafeteria) was depicted instead of a private space, where functionality
with respect to age, gender, and education level is provided in Fig. 3. It of the space required denser settings.

55
S. Ergan et al. Journal of Building Engineering 20 (2018) 51–59

Age Gender Education

Fig. 3. Demographics of participants.

4.3. Statistical analysis ratings received from participants for positive configurations of features
are statistically significantly different (with p-values < 0.05) from the
Table 3 summarizes the statistical analysis on preferences on con- mean ratings for the negative configurations of architectural features.
figurations of architectural design features. The results show that pre- According to the results of the t-test, people felt significantly different
ferences of people for each architecture feature are statistically sig- when they imagined themselves living in the spaces configured to give
nificantly different between the two configurations presented, with positive and negative user experiences. The only exceptions observed
majority preferring the configurations that align with positive end of were for the bipolar scales “pleasant-unpleasant” used for obtaining
the bipolar scales. self-reports of participants on spatial alignment, density of space, and
The null hypothesis was that the proportion of participants' pre- texture/material; “free- encumbered” used for measuring experience on
ferences for the positive configuration of an architecture design feature “height of ceilings”; and “calm-stressed” used for measuring experience
was the same as the negative configuration. The results show that on “color of surfaces” and “ease of access”. The p-value for the ratings
proportion of selection between the two configurations of a given de- on user experiences is larger than the threshold 0.05, meaning that the
sign feature is statistically significantly different (with p-value < 0.05). user experiences are not statistically significantly different between the
For the rest of the design features, people preferred the spaces that pre-recognized positive and negative architecture features. Hence, it
were configured to give a positive experience (e.g., rooms that are open was concluded that these bipolar scales were not suitable to get user
instead of enclosed) as provided in the second column in Table 3 and experience ratings for the corresponding design features and should not
the responses of people were statistically significantly different for be used to capture user experience in spaces.
preference on each dual image set as shown with p-values in the middle
column. In general, the results are also complementary of the suggested
4.4. Comparison on influential power of architectural design features
configuration of each design feature in the literature to give a positive
experience (e.g., a space with exposure to nature is better than a space
It was apparent from the statistical analysis that all of the studied
with no exposure to nature).
architectural design features were statistically significantly different
To test the significance level of responses on the bi-polar scales, t-
between the two configurations of the features except for the “height of
test was also conducted to assess the means of user ratings between the
ceilings”. However, the effect size of each feature, measured by partial
two configurations of a given architectural design feature. The null
eta-squared, was different. Table 4 ranks the influential power of each
hypothesis was that the means of the rating of participants on how they
design feature based on the magnitude of how people perceive those
perceive each configuration of a design feature was the same, which
features. The table also provides the responses of people in their own
meant that people felt indifferent about the configurations provided in
words on how each space made them feel.
dual image sets. As shown in Table 3, the results show that the mean
It is apparent from Table 4 that a space that has exposure to nature,

Fig. 4. Distribution of responses on preferred configurations of architectural design features.

56
S. Ergan et al. Journal of Building Engineering 20 (2018) 51–59

Table 3
Summary of statistical analysis on responses.
Feature Preferred configuration % of preferred configuration Proportion test p-value Bipolar scales t-test p-value

−10
Symmetry of components Symmetrical 68.7% 40.7 1.7e ordered −12.95 2.2e−16
focused −9.94 2.2e−16
pleasant −6.95 9.1e−12
Spatial-Alignment Aligned 58.2% 7.5 0.005 calm −7.53 2.0e−13
pleasant 0.27 0.78
ordered 11.59 2.2e−16
Contours of objects Round 65.3% 26.1 3.1e−07 relaxed 9.11 2.2e−16
engaged 6.57 1.1e−10
pleasant −5.67 2.2e−08
Level of luminance Bright 87.3% 152.2 2.2e−16 relaxed −9.71 2.2e−16
happy 9.64 2.2e−16
pleasant 9.86 2.2e−16
Presence of windows/ Daylighting With window/ with daylighting 85.7% 137.8 2.2e−16 relaxed −4.67 4.4e−06
energized −4.86 1.9e−06
pleasant −4.62 5.7e−06
Size of windows High window to wall ratio 89.1% 162.0 2.2e−16 relaxed 5.59 3.5e−08
energized 6.95 1.0e−11
pleasant −6.10 1.9e−09
Shape of interior spaces Regular shape and aspect ratio 75.9% 70.5 2.2e−16 relaxed −6.98 8.6e−12
calm 7.02 6.7e−12
excited 2.59 0.009
Height of ceilings High ceiling 62.4% 33.3 7.7e−09 free 9.2 2.2 e−16
roomy 9.4 2.2e−16
pleasant −7.9 1.6e−14
Density of spaces Dense 69.2% 37.3 9.7e−10 free 9.72 2.2e−16
roomy −14.50 2.2e−16
pleasant 0.36 0.71
Color of surfaces Colorful 64.8% 21.9 2.7e−06 calm 0.26 0.79
delighted 8.09 4.1e−15
clear 7.79 3.7e−14
Flexibility in isolation /socialization Flexible (not rigid) 77.5% 76.0 2.0e−16 free 23.63 2.2e−16
calm 2.42 0.015
pleasant −11.57 2.2e−16
Openness of space Open space with partial enclosure 82.1% 91.3 2.2e−16 delighted 16.63 2.2e−16
free 22.81 2.2e−16
pleasant −15.04 2.2e−16
Restorative image/Exposure to nature Exposure to nature 93.6% 189.3 2.2e−16 lively 25.85 2.2e−16
delighted 20.13 2.2e−16
relaxed 14.12 2.2e−16
Texture/material Modern/variation 73.9% 55.6 8.6e−14 delighted 3.77 0.0002
excited 3.54 0.0004
pleasant −1.85 0.06
Ease of access Visible 94.2% 188.4 2.2e−16 calm −1.45 0.14
pleasant −19.01 2.2e−16
lively −23.89 2.2e−16

ease of access, open (instead of enclosed), has windows and natural It is apparent that majority of the features that are easy to notice by
daylight, and a high flexibility in isolation when needed and enable people in a space are also powerful to change the experience as shown
socialization when preferred is a space that has high influence on with the partial eta-squared values. The findings provide an ordering of
people. These features are the top features based on their partial eta- the identified design features based on their noticeability and influence
squared values. Beyond these, lighting level, texture/material of sur- levels for practitioners to consider in their design decisions.
faces and symmetry are highly influential as well. These design features
can be highlighted in the first place to change user experiences in a 5. Conclusions and future work
space as compared to the rest of the features.
When the frequency of words and the type of words are analyzed, it A crowdsourcing experiment was designed using the outcomes of
is apparent that the words that people used to express their experience the expert elicitation and literature review to identify the influential
well align with the feature being evaluated blindly. People could be and noticeable design features utilized to give different user experi-
able to express their feelings in words that are almost identical to the ences in spaces. The results show that a space that has exposure to
feature being evaluated (e.g., symmetrical, bright, open). The results nature, ease of access, open (instead of enclosed), has windows and
also show that certain features such as openness of space, presence of natural daylight, and a high flexibility in isolation when needed and
windows and daylighting, flexibility in isolation/socialization, level of enable socialization when preferred is a space that has high influence
artificial lighting, density of spaces, and color of surfaces are easy to on people. These features were defined as the top features based on
notice by people when used to give a different user experience. These their partial eta-squared values in the statistical analysis. The results
features are provided in italic text in Table 4. These features are easy to also show that certain features such as openness of space, presence of
notice and express with wordings that match the feature being analyzed windows and daylighting, flexibility in isolation/socialization, level of
as reflected on the frequency of referral of these wordings by partici- artificial lighting, density of spaces, and color of surfaces are easy to
pants. For two features, spatial alignment and exposure to nature, lack notice by people when used to give a different user experience. It was
of these features are easy to notice and express feelings as compared to apparent that majority of the features that were easy to notice by people
existence of them, as highlighted with underlined wordings in Table 4. in a space were also powerful to change the experience. The findings

57
S. Ergan et al. Journal of Building Engineering 20 (2018) 51–59

Table 4
Influential power of each design feature on human experience.
Architectural design feature Partial eta – squared Effect size Frequently used words (for the positive configuration) Frequency of words used

Restorative nature image/exposure to nature .677 Large Interesting/inviting/happy vs. boring/plain/dull/bland 80

108
Ease of access to spaces .603 Large Welcoming/inviting vs. 78
Boring/dull 97
Openness of spaces .533 Large Open/bright/spacious vs. 128
Cramped/closed/tight 125
Presence of windows /Natural daylight .528 Large Light/open/window vs. 122
Closed/windowless 110
Flexibility in isolation vs. socialization .426 Large Open/free/exposed vs. 142
Private/closed/cramped 123
Level of artificial lighting (luminance) .345 Large Bright/light vs. 110
Dark/dim 104
Texture/material .344 Large Modern/open/bright vs. 84
Dark/old 85
Symmetry of architectural components .275 Large Symmetrical vs. 68
Asymmetrical/uneven/unbalanced 89
Size of windows .208 Large Bright/open/light vs. 97
Dark/closed/dim 48
Density of interior spaces .195 Large Roomy/open/empty/lonely 112
Cramped/crowded/social 119
Spatial alignment in interior spaces .175 Large Aligned/straight/ordered vs. 92
Curved/curvy 142
Contours of objects .130 Medium Round/circle vs. 30
Orderly/square 67
Shape /interior layout .128 Medium Balanced/ordered/normal vs. 76
Odd/strange/unbalanced 67
Color of surfaces .117 Medium Colorful/happy/bright vs. 124
Boring/bland/calm 115
Height of ceilings .008 Small Roomy/opened/relaxed vs. 36
Cramped/closed 70

Bold: Top five influential design features based on partial eta-squared values; Italic text: Noticeable design features in spaces based on frequency of words used to
express experience with respect to meaning of the feature; Underlined text: Noticeable when design feature is configured to give negative experience.

provide an ordering of the identified design features based on their [6] M. Boubekri, B. Hull, L. Boyer, Impact of window size and sunlight penetration on
noticeability and influence levels for practitioners to consider in their office workers’ mood and satisfaction: a novel way of assessing sunlight, Environ.
Behav. 23 (4) (1991) 474–493.
design decisions. [7] C. Chang, P. Chen, W.E. Hammitt, L. Machnik, Psychophysiological responses and
The findings also provide valuable input for researchers. The subset restorative values of wilderness environments, Environments (2007) 479–484.
of features that are found to be influential on people will be used for [8] K. Dale, G. Burrell, An-aesthetics and architecture, in: A. Carr, P. Hancock (Eds.),
Art and Aesthetics at Work, 1st ed., Palgrave Macmillan, London, 2003.
design of experiments aiming to quantify the impact of these features [9] D. Dara-abrams, Learning and navigating built environments: how spatial cognition
on human physiological and task performances. The future work in- and behavior relate to environmental form students: spatial knowledge of a uni-
cludes incorporating the findings of this study to quantify human ex- versity campus, SFP/TR8 Spat. Cogn. (2014) 1–8.
[10] E.M. De Croon, J.K. Sluiter, P.P.F.M. Kuijer, M.H.W. Frings-Dresen, The effect of
perience in virtual tasks (reflecting configurations of architectural de-
office concepts on worker health and performance: a systematic review of the lit-
sign features) using body area sensor networks (reflecting data on erature, Ergonomics 48 (2) (2005) 119–134.
human physiological conditions such as brain signals, heart rate [11] A.S. Devlin, A.B. Arneill, Healthcare environments and patient outcomes: a review
of the literature, Environ. Behav. 35 (5) (2003) 665–694.
variability, facial expressions, and skin conductance).
[12] J.P. Eberhard, Brain Landscape: The Coexistence of Neuroscience and Architecture,
Oxford University Press, New York, U.S., 2009.
Acknowledgement [13] G.W. Evans, The built environment and mental health, J. Urban Health.: Bull. N. Y.
Acad. Med. 80 (4) (2003) 536–555.
[14] G.W. Evans, S.J. Lepore, A. Schroeder, The role of interior design elements in
This material is based upon work supported by the Defense human responses to crowding, J. Personal. Soc. Psychol. 70 (1) (1996) 41–46.
Advanced Research Projects Agency (DARPA) under Grant [15] J. Farbstein, M. Farling, Understanding cognitive processes in correctional settings,
D15AP00098. The views, opinions, and/or findings expressed are those in: Proceedings of the Neuroscience and Correctional Facility Design Workshop,
New Orleans, 7–8 October 2006.
of the author(s) and should not be interpreted as representing the of- [16] L.B. Fich, P. Jönsson, P.H. Kirkegaard, M. Wallergård, A.H. Garde, Å. Hansen, Can
ficial views or policies of the Department of Defense or the U.S. architectural design alter the physiological reaction to psychosocial stress? A virtual
Government. We also would like to thank graduate student Ahmed TSST experiment, Physiol. Behav. 135 (8) (2014) (91–77).
[17] G. Franz, J.M. Wiener, Exploring isovist-based correlates of spatial behavior and
Radwan in helping with data collection for this research. experience, Lect. Notes Artif. Intell. 3343 (2005) 42–57.
[18] G. Franz, M. Von Der Heyde, H.H. Bülthoff, An empirical approach to the experi-
References ence of architectural space in virtual reality-exploring relations between features
and affective appraisals of rectangular indoor spaces, Autom. Constr. 14 (2) (2005)
165–172, https://doi.org/10.1016/j.autcon.2004.07.009.
[1] ASID, Design Impacts Lives: Pre/Post-Occupancy Analysis, American Society of [19] K. Frasca-Beaulieu, Interior design for ambulatory care facilities: how to reduce
Interior Designers, ASID HQ Office Research, 2017 (White paper). stress and anxiety in patients and family, J. Ambul. Care Manag. 22 (1) (1999)
[2] M. Bar, M. Neta, Human prefer curved visual objects, Psychol. Sci. 17 (8) (2006) 67–73.
645–648. [20] M. Geiser, P. Walla, Objective measures of emotion during virtual walk-throughs
[3] K.M. Beauchemin, P. Hays, Sunny hospital rooms expedite recovery from severe urban environments, Appl. Sci. 1 (2011) 1–11.
and refractory depressions, J. Affect. Disord. 40 (1–2) (1996) 49–51. [21] R.G. Golledge, Place recognition and wayfinding: making sense of space, Geoforum
[4] R. Berto, The role of nature in coping with psycho-physiological stress: a literature 23 (2) (1992) 199–214.
review on restorativeness, Behav. Sci. 4 (4) (2014) 394–409. [22] T. Hartig, Restorative environments, Encycl. Appl. Psychol. 3 (2004) 57–66.
[5] C.J. Beukeboom, D. Langeveld, K. Tanja-Dijkstra, Stress-reducing effects of real and [23] B. Haynes, The impact of office layout on productivity, J. Facil. Manag. 6 (3) (2008)
artificial nature in a hospital waiting room, J. Altern. Complement. Med. 18 (4) 189–201.
(2012) 329–333. [24] J. Heerwagen, Design, productivity, and well-being: what are the links? in:

58
S. Ergan et al. Journal of Building Engineering 20 (2018) 51–59

Proceedings of the AIA Conference on Highly Effective Facilities, Cincinnati, Ohio, [52] M. Rashid, C. Zimring, A review of the empirical literature on the relationships
12–14 March 1998. between indoor environment and stress in health care and office settings: problems
[25] J. Heerwagen, G. Orians, The psychological aspects of windows and window design, and prospects of sharing evidence, Environ. Behav. 40 (2) (2008) 151–190.
EDRA 21 (1990) 269–280. [53] J. Reiling, R.G. Hughes, M.R. Murphy, The impact of facility design on patient
[26] P. Hekkert, Design aesthetics: principles of pleasure in product design, Psychol. Sci. safety, in: R.G. Hughes (Ed.), Patient Safety and Quality: An Evidence-Based
48 (2006) 157–172. Handbook for Nurses, Agency for Healthcare Research and Quality (US), Rockville,
[27] L. Heschong, Daylight and Retail Sales. California Energy Commission, Public MD, 2008.
Interest Energy Research (PIER) Program, Technical Report P500-03-082-A-5, [54] E.K. Sadalla, D. Oxley, The perception of room size: the rectangularity illusion,
October 2003. Environ. Behav. 16 (1984) 291–306.
[28] L. Heschong, Windows and Classrooms: A Study of Student Performance and the [55] M.R. Salleh, Life event, stress, and illness, Malays. J. Med. Sci. 15 (4) (2008) 9–18.
Indoor Environment. California Energy Commission, Public Interest Energy [56] M. Schweitzer, L. Gilpin, S. Frampton, Healing spaces: elements of environmental
Research (PIER) Program, Technical Report P500-03-082-A-7, October 2003. design that make an impact on health, J. Altern. Complement. Med. 10 (1) (2004) S-
[29] M.L. Hidayetoglu, K. Yildirim, A. Akalin, The effects of color and light on indoor 71–S-83.
wayfinding and the evaluation of the perceived environment, J. Environ. Psychol. [57] R. Scruton, The Aesthetics of Architecture, Princeton University Press, Princeton,
32 (1) (2012) 50–58. NJ, 2013.
[30] M.L. Hidayetoglu, K. Yildirim, K. Cagatay, The effects of training and spatial ex- [58] S.A. Shumaker, S.M. Czajkowski (Eds.), Social Support and Cardiovascular Disease,
perience on the perception of the interior of buildings with a high level of com- Plenum, New York, 1994.
plexity, Sci. Res. Essays 5 (5) (2010) 428–439. [59] A.E. Stamps, Effects of area, height, elongation, and color on perceived spacious-
[31] P. Höppe, Different aspects of assessing indoor and outdoor thermal comfort, ness, Environ. Behav. 43 (2) (2010) 252–273.
Energy Build. 34 (6) (2002) 661–665. [60] E.M. Sternberg, M.A. Wilson, Neuroscience and architecture: seeking common
[32] B.D. Ilozor, P.E.D. Love, G. Treloar, The impact of work settings on organizational ground, Cell 127 (2) (2006) 239–242.
performance measures in built facilities, Facilities 20 (1/2) (2002) 61–67. [61] N.J. Stone, A.J. English, Task type, posters, and workspace color on mood, sa-
[33] G. Janzen, C. Jansen, A neural wayfinding mechanism adjusts for ambiguous tisfaction, and performance, J. Environ. Psychol. 18 (2) (1998) 175–185.
landmark information, NeuroImage 52 (1) (2010) 364–370. [62] R.S. Ulrich, Evidence based environmental design for improving medical outcomes,
[34] J. Khalaj, O. Pedgley, Comparison of semantic intent and realization in product in: Proceedings of the Healing By Design: Building for Health Care in the 21st
design: a study on high-end furniture impressions, Int. J. Des. 8 (3) (2014) 79–96. Century, McGill University Health Centre, Montreal, CA, 20–21 September 2000.
[35] R. Küller, C. Lindsten, Health and behavior of children in classrooms with and [63] R. Ulrich, C. Zimring, X. Zhu, Healthcare leadership: a review of the research lit-
without windows, J. Environ. Psychol. 12 (4) (1992) 305–317. erature on evidence-based healthcare design, Health Environ. Res. Des. J. 1 (3)
[36] N. Kwallek, C.M. Lewis, A.S. Robbins, Effects of office interior color on workers’ (2008).
mood and productivity, Percept. Mot. Skills 66 (1988) 123–128. [64] R.S. Ulrich, View through a window may influence recovery from surgery, Science
[37] J. Lang, Symbolic aesthetics in architecture: toward a research agenda, in: 224 (4647) (1984) 420–421.
J.L. Nasar (Ed.), Environmental Aesthetics, 1st ed., Cambridge University Press, UK, [65] A. Van den Berg, T. Hartig, H. Staats, Preference for nature in urbanized societies:
1988, pp. 11–26. stress, restoration, and the pursuit of sustainability, J. Soc. Issues 63 (1) (2007)
[38] B. Lawson, The Language of Space, Architectural Press, Oxford, 2001. 79–96.
[39] J. Lingwood, M. Blades, E.K. Farran, Y. Courbois, D. Matthews, The development of [66] J.P. Van Leeuwen, H. Wagter, Architectural design-by-features, in: R. Junge (Ed.),
wayfinding abilities in children: learning routes with and without landmarks, J. CAAD Futures, Springer, Dordrecht, 1997.
Environ. Psychol. 41 (2015) 74–80. [67] O. Vartanian, G. Navarrete, A. Chatterjee, L.B. Fich, J.L. Gonzalez-Mora, H. Leder,
[40] A. Lustig, M. Lustig, The role of visual attention in architectural design, in: M. Skov, Architectural design and the brain: effects of ceiling height and perceived
Proceedings of the ANFA Conference, Salk Institute for Biological Studies, La Jolla, enclosure on beauty judgments and approach-avoidance decisions, J. Environ.
CA, 18–20 September 2014, pp. 60–61. Psychol. 41 (2015) 10–18.
[41] A.D.J. Makin, M.M. Wilton, A. Pecchinenda, M. Bertamini, Symmetry perception [68] O. Vartanian, G. Navarrete, A. Chatterjee, L.B. Fich, H. Leder, C. Modroño,
and affective responses: a combined EEG/EMG study, Neuropsychologia 50 (14) M. Nadal, N. Rostrup, M. Skov, Impact of contour on aesthetic judgments and ap-
(2012) 3250–3261. proach-avoidance decisions in architecture, PNAS 110 (Suppl 2) (2013)
[42] J. Meyers-Levy, R. Zhu, The influence of ceiling height: the effect of priming on the S10446–S10453.
type of processing that people use, J. Consum. Res. 34 (2) (2007) 174–186. [69] J.A. Veitch, A.D. Galasiu, The Physiological and Psychological Effects of Windows,
[43] B. Morton, J. Ramos, The drive toward healthier buildings 2014: the market drivers Daylight, and View at Home Review and Research Agenda no. IRC-RR-325, NRC
and impact of building design and construction on occupant health, well-being and Institute for Research in Construction, Ottawa, Canada, 2012.
productivity, Smart Mark. Rep. Dodge Data Anal. (2014). [70] S. Verderber, Dimensions of person-window transactions in the hospital environ-
[44] J.L. Nasar, Urban design aesthetics the evaluative qualities of building exteriors, ment, Environ. Behav. 18 (1986) 450–466.
Environ. Behav. 26 (3) (1994) 377–401. [71] I. Vilnai-Yavetz, A. Rafaeli, C.S. Yaacov, Instrumentality, aesthetics, and symbolism
[45] M.K. Nejad, Curvilinearity in Architecture: Emotional Effect of Curvilinear Forms in of office design, Environ. Behav. 37 (4) (2005) 533–551.
Interior Design (Doctoral dissertation), Texas A&M University, College Station, [72] D. Watson, L.A. Clark, A. Tellegen, Development and validation of brief measures of
Texas, United States, 2003. positive and negative affect: the PANAS scales, J. Personal. Soc. Psychol. 54 (6)
[46] A. Ozdemir, The effect of window views' openness and naturalness on the percep- (1988) 1063–1070.
tion of rooms' spaciousness and brightness: a visual preference study, Sci. Res. [73] WELL, The WELL Building Standard, International Well Building Institute, New
Essays 5 (16) (2010) 2275–2287. York, NY, United States, 2014.
[47] R. Parsons, L.G. Tassinary, R.S. Ulrich, M.R. Hebl, M. Grossman-Alexander, The [74] J.C. Western, A. Buttimer, D. Seamon, The human experience of space and place,
view from the road: implications for stress recovery and immunization, J. Environ. Econ. Geogr. 57 (3) (1981) 275–278.
Psychol. 18 (2) (1998) 113–140. [75] S. Winchip, M. Inman, P.C. Dunn, Stress due to crowding in multifamily dwelling
[48] D. Pati, A framework for evaluating evidence in evidence-based design, Health interior spaces, Home Econ. Res. J. 18 (1989) 179–188.
Environ. Res. Des. J. 4 (3) (2011) 50–71. [76] T. Zborowsky, M. Kreitzer, People, place, and process: the role of place in creating
[49] D.M. Pedersen, T.L. Topham, Perception of enclosure effects of architectural sur- optimal healing environments, Creat. Nurs. 15 (4) (2009) 186–190.
faces in a large scale interior space, Percept. Mot. Skills 70 (1) (1990) 299–304. [77] R. Gifford, The consequences of living in high-rise buildings, Archit. Sci. Rev. 50 (1)
[50] C.A. Pierce, R.A. Block, H. Aguinis, Cautionary note on reporting eta-squared values (2007) 2–17.
from multifactor anova designs, Educ. Psychol. Meas. 64 (6) (2004) 916–924. [78] P. Jasen-Osmann, M. Heil, The process of spatial knowledge acquisition in a square
[51] P.C. Quinn, C.R. Brown, M.L. Streppa, Perceptual organization of complex visual and a circular virtual environment, Adv. Cogn. Psychol. 3 (3) (2007) 389–397.
configurations by young infants, Infant Behav. Dev. 20 (1) (1997) 35–46.

59