*2.3. Ethical Aspects*

The Human Research Ethics Committee (HREC) at the Delft University of Technology approved the application to conduct this study on the 31st of January 2022. A consent form was included at the beginning of the questionnaire, stating all data will be treated anonymously. This form also mentioned that students could skip any part of the questionnaire if they felt uncomfortable answering it.

#### *2.4. Data Management and Analysis*

The data were exported from the Qualtrics XM platform to SPSS version 26.0 software (SPSS Inc, Chicago, IL, USA) for data analysis. Descriptive statistics were performed to calculate the frequencies, percentages, maximum, minimum and standard deviation (SD), and mean of the variables related to demographics, emotional state, IEQ comfort perception, psychosocial perception, IEQ preferences and the importance of IEQ-related items to study better, and psychosocial preferences.

TwoStep cluster analysis is a segmentation method that enables the creation of profiles of clusters based on any form of data, including categorical data [32]. This method was also used in previous studies within the domain of IEQ to determine profiles of clusters [12,24–28]. Accordingly, TwoStep cluster analysis was performed and validated twice and separately to create two distinct cluster models. The first TwoStep cluster analysis was performed to cluster the students based on their IEQ preferences, while the second one clustered them based on their psychosocial preferences. The input variables for the IEQ preferences model comprised eight variables: ventilation and fresh air, temperature, view to the outside, sounds from the outside, sounds from the inside, smells, artificial light, and daylight. The input variables for the psychosocial preferences model comprised nine variables: storage, cleanliness, amenities, chair type, presence and company of others, size of the room, bonding or identifying with the place, ability to adapt or control the place, and privacy. The settings of the TwoStep cluster analysis were based on selecting loglikelihood, determination of the number of clusters automatically, and Akaike's information criterion (AIC). Once the cluster model was generated, four validation steps were conducted : (1) silhouette measure of the cluster model is larger than 0.2 (fair and above); (2) Chisquare tests were performed to examine the relationship between the input variables of the cluster analysis and the final cluster model, with *p*-value less than 0.05 considered as statistically significant; (3) the predictor importance scores of the input variables were larger than 0.02; and (4) the dataset was randomly split half (50%) to re-run the final solution model on each half to ensure that both solutions were similar to the final solution.

After the TwoStep cluster analysis, descriptive analysis was conducted to calculate the frequencies, percentages, and SD for different variables of each cluster (e.g., health, IEQ perception, IEQ preferences). To compare differences between the clusters, Chi-square and ANOVA tests were used (for nominal and continuous variables, respectively). Each student belongs to two clusters, a cluster of IEQ preferences, and a cluster of psychosocial preferences, resulting in clusters of students with the same IEQ preferences but different psychosocial preferences, and vice versa. Hence, it is important to investigate the overlap between the two models to better understand in detail the profile of students within these two models. The overlap between the two cluster models was identified using cross-tabulation. In addition, frequencies, percentages, and SD for different variables of each profile within the overlap between the two models were calculated. The significant differences between the variables among the different profiles were tested using Chi-square and ANOVA tests. Chi-square calculations with less than 5 in one cell were excluded from the analysis.

#### **3. Results**

#### *3.1. Students Characteristics*

Table 1 presents several characteristics (e.g., age, gender, time spent at home, study place, and lifestyle) of the respondents in 2021 and 2022. Since only the mean time spent at home was significantly different between the two groups; students in 2021 spent more time at home than students in 2022, this study mainly focused on questions related to study places and excluded the questions related to students' homes, such as building-related symptoms. The mean age of the 451 students was 20 years old. The ratio of female to male students was 1.6. Students within this study spent their studying time mostly at their homes (74%), while 26% of them stayed in educational buildings for studying. The students stayed at their homes around 17 h per day during weekdays, and 16 h per day during the weekend.


**Table 1.** Students characteristics in 2021 and 2022.

### *3.2. Students' Preferences of Their Study Places*

Figure 1 presents the mean and SD values of the eight IEQ preferences aspects. Daylight (8.4 ± 1.5) was the most important aspect of the whole study sample. This is followed by both view to the outside (8.2 ± 1.8) and temperature (8.2 ± 1.3). In contrast, smells (6.2 ± 2.3), artificial light (6.2 ± 2.0), and sounds from the outside (6.3 ± 2.2) were the least important IEQ aspects. Figure 2 illustrates the mean and SD values of the nine psychosocial preference aspects. Amenities (8.0 ± 1.5) and cleanliness (7.6 ± 1.7) were the most important psychosocial aspects of the study place. On the other hand, students in this study reported the lowest scores on three psychosocial aspects: presence and company of others (5.3 ± 2.5), bounding or identifying with the place (5.4 ± 2.5), and size of the room (5.5 ± 2.0).

**Figure 1.** IEQ preferences of study places.

#### *3.3. TwoStep Cluster Analysis*

TwoStep cluster analysis was carried out to categorize profiles of students based on their IEQ preferences and separate psychosocial preferences in their study places. This was carried out by using the original variables that consists of eight variables of the IEQ preferences and nine variables for the psychosocial preferences. The results of the TwoStep cluster analysis resulted in two models: the IEQ preferences model, and the psychosocial preferences model. Each of these two models comprised three distinct clusters. The Silhouette coefficient was fair for both models; 0.3 for the IEQ preferences model, and 0.2 for the psychosocial preferences model.

The predictor importance of the eight input variables for the IEQ preferences model, as well as the nine input variables for the psychosocial preferences model, was found to be strong and larger than 0.02. Additionally, after randomly splitting the dataset into two halves, only a few changes were found between the two halves and the final solution (Table 2). Furthermore, all eight IEQ preference variables were found to be statistically significant in relation to the IEQ preferences model (*p* < 0.05). Similarly, the nine psychosocial preference variables were found to be statistically significant in relation to the psychosocial preferences model.


**Table 2.** Predictor importance of the input variables for both models.

#### 3.3.1. IEQ Preferences Model

The clusters of the IEQ preference clusters are described in Table 3 and Appendix B. Table 3 only includes the variables that were statistically different among the clusters within the IEQ preferences model (*p* < 0.05). The IEQ preferences model resulted in three clusters: IEQC1 (concerned with all IEQ aspects), IEQC2 (concerned with daylight and view to the outside), and IEQC3 (concerned with only temperature). These three clusters scored a high importance level for daylight (ranged from 7.0 to 9.0), view to the outside (ranged from 6.7 to 9.0), and temperature (ranged from 7.8 to 8.7).

**Table 3.** Descriptives of IEQ clusters.


#### 3.3.2. Psychosocial Preferences Model

Descriptions of the psychosocial preference clusters are presented in Table 4 and Appendix C. Table 4 only illustrates the variables that were found to be statistically different among the three clusters within the psychosocial preferences model (*p* < 0.05). This model consists of three distinct clusters: PSC1 (Preference for most of psychosocial aspects), PSC2 (preference for presence and company of others), and PSC3 (preference only for amenities and cleanliness). Generally, the students within these clusters reported a high importance for two aspects, which are cleanliness (ranged from 7.1 to 9.0) and amenities (ranged from 7.5 to 8.9).


**Table 4.** Descriptive of psychosocial clusters.

### *3.4. Overlap between the IEQ and the Psychosocial Preferences Model*

The overlap between the IEQ and psychosocial preferences model resulted in nine distinct profiles that are illustrated in Figure 3. Descriptions of these profiles, presented in Table 5, are statistically significantly different between the profiles. A comprehensive description for these nine groups is illustrated in Appendix D.

In general, all nine profiles are concerned with three IEQ preferences, which are daylight (ranged from 6.6 to 9.3), view to the outside (ranged from 6.4 to 9.1), and temperature (ranged from 7.6 to 8.9). Pertaining to the psychosocial preferences, most of the profiles scored high importance levels for two aspects: amenities (ranged from 7.4 to 8.9) and cleanliness (ranged from 6.8 to 9.0). Therefore, the description for each profile is based on highlighting which profile scored the highest and/or lowest importance level for both IEQ and psychosocial preferences among all profiles.

Each name of the nine profiles consists of two parts: the first part is related to IEQ preferences, and the second part is related to psychosocial preferences. The IEQ preferences part consists of one of three names that are; (1) concerned, which means all IEQ preferences are important, (2) visual concerned, which implies that daylight and view to the outside are important, and (3) unconcerned, which indicates that almost all IEQ preferences are not very important except for temperature. The psychosocial preferences part includes one of the four categories; (1) perfectionist, which implies high importance levels for most of the psychosocial aspects, (2) extrovert, which reflects the high importance level for the presence and company of others, (3) introvert, which means that privacy is highly important, (4) non-perfectionist, which indicates that most of the psychosocial aspects are not highly important, except amenities.

**Figure 3.** The nine profiles of students based on the overlap between the IEQ preferences model and psychosocial preferences model.



**Table 5.** *Cont.*

3.4.1. Overlap between IEQC1 with Psychosocial Clusters

The overlap between the IEQC1 and the three psychosocial clusters resulted in three profiles: IEQC1-PSC1: the concerned perfectionist; IEQC1-PSC2: the concerned extrovert; and IEQC1-PSC3: the concerned the non-perfectionist.

• IEQC1-PSC1: the concerned perfectionist.

The concerned perfectionist profile comprises 59 students (14%), of which 29% are male and 71% are female students. These students are the largest group that experienced positive events (37%). In terms of lifestyle, this group has the lowest number of students that consume alcohol (75%). Regarding the IEQ preferences, the concerned perfectionist students rated the highest importance for sounds from the inside (8.2), smells (7.9), sounds from the outside (7.8), and artificial light (7.2). Furthermore, they rated the highest importance (as compared to the other groups) for six IEQ-related items, which are control of room temperature (8.3), control of room ventilation (8.2), control of shading (8.2), control of surrounding sounds (8.1), personal desk ventilation and fresh air (8.1) and presence of plants (7.0). As this profile overlaps with PSC1 who are concerned with all psychosocial preferences (except presence and company of others), it is the most concerned with cleanliness (9.0), amenities (8.9), chair type (8.1), ability to adapt or control the place (7.4), and bonding or identifying with the place (6.8).

• IEQC1-PSC2: the concerned extrovert.

The concerned extrovert profile consists of 63 students (14%), of which the percentages of male and female students are similar as the concerned perfectionist profile (29% and 71%, respectively). This profile is the second highest group that experienced recently positive events (35%). Regarding health, students within this profile are the second highest group that suffered from anxiety (30%). It can be noted that these students are the group to have the least control over all IEQ factors in their most used study place. Pertaining to IEQ preferences, the concerned extrovert students are concerned about all IEQ aspects, especially daylight (9.0) and view to the outside (8.8). As this profile overlaps with PSC2, it is the profile that is most concerned with the presence and company of others in their study places (6.9), while they rated the lowest importance for storage (5.3).

• IEQC1-PSC3: the concerned non-perfectionist

The concerned non-perfectionist profile comprises 35 students (8%), which includes the lowest percentage of male students (17%) and the highest percentage of female students (83%). Nearly half of them (45%) were feeling relaxed when they were completing the questionnaire. It can be noted that this profile rated the highest for negative affect (12.1). With regards to their lifestyle, students within this profile are the highest in terms of alcohol consumption (91%), as well as doing physical activity (94%). In terms of IEQ preferences, the concerned non-perfectionist students rated the highest importance for ventilation and fresh air (9.0), temperature (8.9) and sounds from the inside (8.2). Regarding IEQ-related items, these students rated the highest importance for headphones (7.9). As this profile overlaps with PSC3, which rated the least importance scores for most of the psychosocial preferences, it is the least concerned with bonding or identifying with the place (3.0), ability to adapt or control the place (3.9) and size of the room (4.0).

3.4.2. Overlap between IEQC2 with Psychosocial Clusters

The overlap between the IEQC2 and the three psychosocial clusters resulted in three profiles: IEQC2-PSC1: the visual concerned perfectionist; IEQC2-PSC2: the visual concerned extrovert; and IEQC2-PSC3: the visual concerned non-perfectionist.

• IEQC2-PSC1: the visual concerned perfectionist

The visual concerned perfectionist profile is the second smallest profile size that comprises 34 students (8%), of which 47% are male students and 53% are female students. It is the profile that least experienced recently positive events (23%), while it rated the highest positive affect (18.3). Most of the students (97%) within this profile spent their studying time at their homes. Pertaining to IEQ perception, these students comprise the profile that is least dissatisfied with air smell (35%). With regards to the IEQ preferences, the visual concerned perfectionist students rated the highest importance for daylight (9.3) and view to the outside (9.1). Regarding the IEQ-related items, these students rated the highest importance for lamp on my desk (7.1), and the presence of plants (7.0). With regards to psychosocial preferences, they rated the highest importance for cleanliness (9.0), amenities (8.9) and storage (8.4). In addition, the visual concerned perfectionist students scored a high importance level for privacy (7.9).
