*3.2. Influence of Daylight Illumination on Visual Comfort* 3.2.1. Data Preprocessing Analysis of Covariance

The level of basic electrodermal activity is related to personality characteristics. Individuals who have a higher basic electrodermal activity tend to be more introverted, nervous, and emotionally unstable. In contrast, individuals with a lower basic level are found to be more cheerful and outgoing, and have a more balanced mentality and better psychological adaptation. The time-domain mean of SC of each elderly person under different illumination was plotted on a scatterplot. According to this scatterplot (Figure 9), although the change in the electrodermal activity level of each elderly person was relatively small under different illumination conditions, a large difference in its values between different subjects was observed. This difference may originate from different physiques and cannot be artificially controlled. Covariance analysis was used to explore whether the individual's basic electrodermal activity level would affect the analysis of the SC time-domain characteristic mean under different illumination levels, and to clarify whether the SC time-domain mean values measured by different subjects could be directly compared.

**Figure 9.** SC changes in individuals at different daylight illumination levels.

In this study, the SC time-domain characteristic mean values (hereinafter referred to as the baseline value) of the elderly persons during the baseline period was the covariate. The illumination was the independent variable, and the SC time-domain characteristic mean (hereinafter referred to as the SC mean value) under different illumination was the dependent variable. The statistical results of covariance are shown in the table below. The between-subjects effects test shows the baseline value of covariate *p* (Sig.) = 0.000 < 0.01 (Table 3). Hence, it can be considered that an interactive relationship existed between the SC mean value of the dependent variable and the baseline value of the covariate, indicating that a person's initial electrodermal activity level did affect the electrodermal activity level under different illumination conditions.


**Table 3.** Between-subjects effects test.

a. R<sup>2</sup> = 0.948 (After adjustment R2 = 0.946).

After deducting the effect of covariate baseline on the experiment, *p* (Sig.) had a value of 0.006 < 0.05 (Table 4), showing that the illumination of the independent variable still had a significant effect on the SC mean value of the dependent variable.

**Table 4.** Univariate tests.


SC Time-Domain Mean Value Normalization

To eliminate the influence of each elderly person's initial electrodermal activity level, it was necessary to perform data normalization. The specific operation of normalization involves subtracting the SC mean value of an elderly person's reading under different illumination conditions from the baseline value, and dividing the result by the baseline value. The change rate Δk obtained is the normalized result. The Δk value indicates the state arousal level. A high Δk indicates high state arousal, whereas a low value indicates low state arousal. The equation below shows the normalization process, where Δk denotes the normalized change rate of the SC. *Xemotion* refers to the SC mean value under a certain illumination, and *xcalm* represents the baseline value. The formula is as follows [20]:

$$
\Delta \mathbf{k} = \frac{\overline{\mathbf{X}}\_{emotion} - \overline{\mathbf{x}}\_{calm}}{\overline{\mathbf{x}}\_{calm}}
$$

### 3.2.2. State Arousal

### Correlation Analysis

The variance homogeneity analysis of Δk data revealed a significance *p* = 0.000 < 0.05; thus, it was impossible to use the parametric test. Therefore, the nonparametric Kaplan– Meier (KM) analysis was performed to test the difference between each group of data. The obtained significance was *p* = 0.006 < 0.05, indicating a significant difference. As such, the data could be further analyzed. The illumination and Δk values did not satisfy the normal distribution; thus, Spearman correlation analysis was performed. The results show that the correlation coefficient was 0.108 > 0.01 (Table 5), revealing a significant correlation.

**Table 5.** Spearman correlation coefficient of illumination and Δk.


#### Cumulative Analysis

First, analysis of variance (ANOVA) was performed on Δk under each group of illumination to test whether there was significant difference in state arousal under different illumination conditions. A variance homogeneity test revealed that *p* (Sig.) = 0.460 > 0.05; thus, ANOVA could be continued. The results are shown in the table below. *p* (Sig.) = 0 < 0.01 indicated a significant difference in state arousal under different illumination conditions (Table 6).

**Table 6.** ANOVA analysis results.


The Δk value can indicate the arousal degree. The greater the value, the greater the arousal, whereas the smaller the value, the smaller the arousal degree. By accumulating the Δk value under each group of illuminations, the trend of arousal under different illumination conditions can be determined (Figure 10). For 600–900 lx, the state arousal degree was the largest. Under other illumination values, the state arousal degree was small (Table 7). However, the positive or negative state of each group remained unknown. Thus, further analysis in combination with the results of subjective questionnaire was required.

**Figure 10.** Trend of cumulative Δk value.

**Table 7.** Cumulative Δk value result.


3.2.3. Visual Comfort

Correlation Analysis

The score data of visual comfort questionnaire were analyzed by variance homogeneity, and the significance was *p* = 0.387 > 0.05. Hence, one-way ANOVA was performed, and the significance determined was *p* = 0.000 < 0.05, indicating a significant difference; thus, in-depth analysis could be conducted. The illumination value and comfort score did not meet the normal distribution; therefore, the Spearman correlation analysis was performed. The results show that the correlation coefficient was 0.312 > 0.01, revealing a significant correlation (Table 8).


**Table 8.** Spearman correlation coefficient of illumination and visual comfort score.

#### Descriptive Analysis

The scores of visual comfort questionnaire of each group were counted, and the proportion of "Very Uncomfortable", "Uncomfortable", "Normal", "Comfortable", and "Very Comfortable" in each group was counted. From the questionnaire scores, it was found that (Figure 11):


Based on the above situation, it can be preliminarily concluded that 300 and 400 lx resulted in negative visual perception for most elderly persons, who were unable to read without physical or psychological stress. With the increase in illumination, the visual perception tended to be positive at 500 lx. At 600 and 700 lx, positive feedback was obtained from most elderly persons, and their feelings were biased towards comfort. When the illumination was 800 lx, the opinions began to diverge. Although the number of elderly persons who had normal feelings increased, more elderly persons felt comfort. At 900 lx, the number of elderly persons who considered they had normal feelings were in the majority, at nearly half. At 1000 lx, the number of the elderly persons who considered they had normal feelings was similar to that who were uncomfortable, but a small proportion of elderly persons felt comfortable.

By averaging the visual comfort scores under each group of illumination values, the general trend of visual comfort could be obtained (Table 9). The range of discomfort fell within 0–3 points. From the average value, 300, 400, and 1000 lx can be considered as negative discomfort in the range of less than 3 points. The scores under 900 and 500 lx were close to 3 points, which can be judged as normal. The scores under 600, 700, and 800 lx were in the range of more than 3 points, which can be considered as positive comfort.

**Figure 11.** Percentage of subjective scores at different illumination levels.

**Table 9.** Average value of the visual comfort score.


#### *3.3. Daylight Illumination Threshold Analysis*

The average score of the visual comfort questionnaire of each group was compared with the accumulated Δk value. The comparison diagram is show in Figure 12.

Δk indicates state arousal level, and the visual comfort score indicates the trend of "Comfort" (>3 points), "Normal" (3.1> *n* > 2.9), and "Uncomfortable" (<3 points), which can be further divided into detail according to the level of 1–5 points (Table 10). According to the results of the subjective questionnaire, at 300–400 lx, the subjective visual comfort score was below 3, indicating discomfort. State arousal was also low, making it harder for the elderly persons to engage. When the illumination was increased to 500 lx, the situation improved, and most of the elderly persons were able to complete the reading task more easily. At the same time, the elderly persons' states of arousal rose during this procedure. At 600 lx, the subjective visual comfort score reached a maximum of 4.36 points, which was classified as "Very comfortable." This indicated that the elderly persons could read without feeling stressed, and their arousal level had improved slightly. When the light level was increased to 700 and 800 lx, the subjective comfort score decreased, but both scores were greater than 3 and still fell into the "comfort" category. At these illumination levels, the elderly persons were under greater pressure to complete the reading task than at 600 lx. Their state arousal was higher, which was a positive outcome. At 900 lx, the subjective visual comfort score remained at 3, despite the fact that state arousal remained high. Subjective visual comfort fell below a score of 3 at 1000 lx, and was classified as "uncomfortable", with a considerable decline in state arousal.

**Figure 12.** Comparison diagram of Δk and visual comfort scores.


**Table 10.** Ranking of scores of the visual comfort questionnaire.

### **4. Discussion**

Based on the analysis of the experimental data, a trend can be noticed. When daylight illumination was set at 300–500 lx, subjective visual comfort scores rose in tandem with state arousal, showing that the elderly persons were gradually becoming more active and engaged. At an illumination level of 600–800 lx, the subjective visual comfort score dropped from its maximum point but remained in the "comfort" category, while state arousal levels continued to rise. Based on the elderly persons' subjective assessments, when the illumination was between 600 and 800 lx, the condition of state arousal was positive. It can be seen that physiological and psychological activity improved the elderly person's subjective perception. The levels of both 700 and the 600 lx had a subjective visual comfort score of 4 or more; 700 lx may be preferable since it resulted in a higher level of state arousal.

However, this did not necessarily apply to the 900 and 500 lx levels, which were both rated 3 out of 5 as "normal." According to the proportion of subjective visual comfort scores, more than half of the elderly persons assessed 500 lx as "normal." At 900 lx, several subjects experienced trouble adjusting the uniformity of light in their field of view, and more glare cases were reported. Some subjects also reported preferring the higher illumination of 900 lx, which helped them see more clearly. As a result, high state arousal can be both a positive and a negative outcome. In terms of consistency of evaluation, the 500 lx level is superior to that of 900 lx.

#### **5. Conclusions**

It can be seen from the experimental data that the peak of subjective visual comfort score appeared at 600 lx, and the peak of state arousal level appeared at 700 lx. The reason for this peak gap can be analyzed from the changes of the two indicators. When the daylight illumination was low (300 lx), as the illumination increased, the elderly persons invested more energy, and the visual comfort also improved. However, when the illumination was increased to 800 lx, the elderly persons felt a certain pressure when reading, and the increase in state arousal then offset a part of this pressure. When the illumination was higher than 1000 lx, the elderly persons appeared to have low state arousal, and the visual comfort was also greatly reduced.

Within the context of this experiment, the most comfortable illumination range for the elderly persons to read in daylight was 600–800 lx, and 700 lx was optimal. This level was higher than the illumination of 300 lx recommended for reading under artificial lighting in the Architectural Lighting Design Standard of China. This showed that, due to the dependence of the elderly persons on daylight, their demand for daylight illumination was higher than that under constant artificial lighting. This conclusion was also confirmed by a study in UK, in which individuals tolerated significantly higher levels of daylight illumination than CIBSE's typical artificial lighting recommendations unless there was glare or direct sunlight [21].

The main conclusions can be summarized as follows:


**Figure 13.** Schematic of illumination value and comfort level.

This experiment was chosen to take place in Shenyang, a city in China's harsh cold area, which is located at high latitudes. Long winters, low solar azimuths, and short daylight hours characterize cities in high latitudes. The elderly persons who live here are restricted to indoors and prefer the more daylight is brought indoors because they are psychologically closer to it. However, in cities located at low latitudes with high levels of daylight radiation (e.g., Guangdong), the issues to consider are quite different. On sunny days, low latitudes are characterized by intense daylight radiation, whereas cloudy situations result in insufficient daylight hours. Although it is critical to bring in as much daylight as possible, too much direct light may generate glare and diminish the comfort of the light environment. The elderly persons in various climatic zones may have different adaptations to daylight, and additional research in various regions is needed.

In this research, extraneous light environment indicators such as glare [22] and illumination uniformity [23] were controlled for, although these two indications are also crucial for evaluating the quality of the light environment. At the same time, the primary purpose of the light environment indicators is to inform the architects' lighting design strategy. The physiology, psychology, and behavior of the elderly persons are affected by design considerations such as building orientation, room sizes, window sizes and parameters, frames and position, types of glazing, transmission characteristics of glazing, cleanliness of glazing, and interior room surfaces [24,25]. The comfort of the daylight environment is a systematic "human-behavior-environment" problem, which is better suited to multifactorial research with the use of appropriate algorithms. Before that, however, experiments are needed to clarify the specific relationships between the different factors. The best design strategy based on daylight performance indicators may be discovered using enough experimental sample data and a multi-objective algorithm [26].

#### **6. Limitation**

First, the set illumination range was 300–1000 lx in the experiments. This was based on the recommended value of reading illumination under artificial lighting in the Architectural Lighting Design Standard of China and the pre-judgment of the actual situation in the pre-experiment. However, there were also cases where it exceeded 1000 lx or was less than 300 lx in the experiments. Since elderly persons with normal vision and reading habits were selected in this experiment, only in certain cases did elderly persons with low sensitivity to light believe that illumination higher than 1000 lx was more comfortable, or those with relatively better vision and high tolerance of daylight illumination considered that daylight below 300 lx was acceptable. It can be seen that the acceptance of daylight illumination was also different for the elderly persons with different physiological and psychological states. This research focused on a certain category of elderly persons and was not representative of a broad group of elderly persons. In addition, this experiment set the time from 8:00–10:00 in the morning according to the reading habits of the elderly persons. Table 1 summarizes the activities of the elderly persons in one day, among which watching TV, playing cards, playing chess, eating, and chatting were the activities enjoyed by the elderly in other periods; these activities were also not covered in this research. Finally, the site selected for this article was a rectangular room facing south, which was relatively simple. The daylight features of other rooms were different from those facing the south direction and would behave differently over time. In addition, there were complex geometries, which needed to be specifically analyzed according to the movement tracking of the elderly persons' activities. These questions all need to be further explored in the future.

**Author Contributions:** Conceptualization, Y.F. and R.H.; methodology, Y.F. and R.H.; software, R.H. and Y.W.; validation, Y.W.; formal analysis, Y.W.; investigation, R.H. and Y.W.; resources, Y.F.; data curation, R.H. and Y.W.; writing—original draft preparation, Y.W.; writing—review and editing, Y.F. and W.G.; visualization, Y.W.; supervision, Y.F. and W.G.; project administration, Y.F.; funding acquisition, Y.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This paper is supported and funded by two sponsors: "Xingliao talents plan" of the innovative leading talent project of Liaoning Province (tpjs2019001), Liaoning Provincial Natural Science Foundation Guiding Plan in 2019 (2019-zd-0656).

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki, and approved by the Ethics Committee of School of Architecture and Planning, Shenyang Jianzhu University.

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** Data are not publicly available due to restrictions regarding the privacy of the participants.

**Acknowledgments:** We are also grateful to the Cold Land Healthy City and Comfortable Building Research Center of the Shenyang Jianzhu University for providing the equipment support for this study, as well as Zitong Wang and Muye Wang for their experience and analysis regarding the data statistics.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

