Driving Factors of Shading Adjustment Behavior Based on Social Cognitive Theory and Structural Equation Modeling

Abstract

Research on occupant behavior is a crucial aspect of building energy-saving research. Among them, the shading adjustment behavior in buildings, which occurs frequently during building usage, significantly impacts building energy consumption. Due to the randomness of shading adjustment behavior and the complexity of its motivations, interdisciplinary research is required in this field. In order to better analyze the driving factors of shading adjustment behavior and thus provide a reference for promoting building energy-saving technologies and strategies, this paper employs novel methods for research. We establish a structural equation model based on social cognitive theory, then design a questionnaire and collect data. We utilize structural equation modeling to examine the interrelationships between different dimensions. We ultimately determine the connections between different driving factors of shading adjustment behavior. The results show that whether in the behavior of activating or deactivating shading, environmental factors have a positive correlation with personal factors, and both environmental and personal factors significantly influence shading adjustment behavior. Furthermore, within environmental factors, social environmental factors also significantly affect shading adjustment behavior. Lastly, comparing the behaviors of activating and deactivating shading, the impacts of light and thermal environmental factors on shading adjustment behavior show certain differences.

1. Introduction

Green buildings and building energy conservation are essential for reducing carbon emissions and environmental pollution [1]. Effective energy-saving technologies and rational behaviors from building occupants are both required to achieve optimal energy-saving outcomes [2]. Even identical buildings can show significant differences in energy consumption due to varying occupant behaviors [3]. Simulations have shown that different occupant behaviors can cause substantial variations in building energy use, with non-energy-saving behaviors causing actual energy demands to far exceed design values [4,5,6].

Shading adjustment behavior, a focal point in occupant behavior research, significantly impacts building energy consumption. Rationally designing shading systems and devising appropriate shading adjustment strategies can enhance indoor light and thermal environments [7,8,9,10]. Research has indicated that installing internal window shading can effectively reduce solar radiation heat, mitigate peak electricity demand, and save over 17% of air conditioning load [11]. Krarti utilized whole building energy simulation and envelope system modeling techniques to analyze energy consumption. By employing continuous and stepped settings for dynamic sliding shading devices, HVAC system usage can be reduced by 50% annually in hot climates [12]. Hu et al. conducted a lighting control experiment on windows using shading device energy-saving strategies, which reduced HVAC system energy consumption by 10% and saved 90% of lighting energy consumption [13]. Furthermore, reasonable shading adjustments positively influence building lighting levels, reducing lighting energy consumption [14]. The study of building occupants’ behavior also holds significant importance for research on passive building energy conservation. Ozarisoy is studying the overheating risks and energy efficiency of six passive design strategies during testing and implementation [15]. This study will aid in strategically designing retrofit interventions, effectively reducing cooling energy consumption by considering residents’ thermal comfort, thermal adaptability, and energy use. Therefore, exploring control and adjustment strategies for dynamic shading systems in buildings has become a key approach to achieving energy savings. However, occupants’ adjustment and control of shading are often random, posing challenges for research. Investigating the driving factors behind shading adjustment behavior is crucial in this field.

Many studies have investigated shading adjustment behavior through field monitoring [16]. But these studies are insufficient to uncover the underlying reasons for such behavior. This is because the threshold values of environmental parameters that trigger shading adjustment actions vary significantly. For example, Inkarojrit’s study shows that manual blinds usually do not close when the vertical solar radiation on the window is less than 13 W/m² [17]. In contrast, Mahdavi’s study shows that the closing rate of blinds increases when the vertical solar radiation is between 150 and 200 W/m² [18]. These threshold differences indicate various reasons for office workers’ shading adjustment behaviors, making it challenging to comprehensively analyze the potential connections between different factors solely through field monitoring.

Thus, besides field monitoring, comprehensive questionnaire surveys are also necessary. Some international studies have explored lighting conditions related to shading. For instance, Hirning et al.’s study discusses the discomfort caused by window glare in tropical regions [19]. Moscoso et al. conducted a survey on lighting environment satisfaction [20]. However, there are few similar surveys on the driving factors of shading adjustment behavior in office buildings in regions with hot summers and cold winters.

Therefore, this paper investigates the driving factors of shading adjustment behavior in buildings located in regions with hot summers and cold winters by constructing a structural equation model. This model systematically considers the influence of key factors on shading adjustment behavior and the interactions between these factors. Based on the analysis results, recommendations and countermeasures are proposed, which are of great significance for research on occupant behavior and building energy conservation.

2. Literature Review

Research on shading device adjustment behavior began in 1978, when Rubin and colleagues conducted a field study of over 700 offices equipped with shading blinds to investigate the regularity of shading adjustments. They found significant differences in the shading states of offices with different orientations [21]. In the following decades, the driving factors of shading adjustment behavior have been extensively studied, leading to a growing consensus on the key factors influencing shading behavior, while efforts have been made to improve and optimize predictive models for shading behavior.

Zhang and colleagues found that the adjustment frequency of shading devices is two to three times higher under high-intensity solar radiation conditions than under low-intensity conditions, with the frequency of shading adjustments and average occlusion rate increasing with the solar altitude angle [22]. Boyce and colleagues found a significant correlation between the average shading rate and solar intensity [23]. Reinhart and Inoue’s research found that the incident depth of solar radiation, as a driving factor for predicting shading behavior, provides more accurate predictive results [24,25]. Haldi and colleagues found that combining the geometric position of the sun and radiation intensity is more effective in predicting shading behavior, and noted that using indoor parameters such as work surface illuminance as predictors can reduce prediction errors caused by differences in physical room conditions, enhancing the model’s broad applicability [26]. Besides physical factors, non-physical factors also influence shading adjustment behavior. Inoue and colleagues found that the desire to enjoy the view outside the window reduces the likelihood of using shading devices, while people’s tolerance for visual comfort increases [25]. Inkarojrit and colleagues found that about 12% of people adjust shading devices for personal privacy protection [9]. Reinhart and colleagues found that when indoor illuminance is below 1000 lux, people primarily draw shades for privacy protection [24]. Many mechanisms governing office workers’ behavior are inherently random and form complex relationships with various physiological, psychological, and physical factors. Recent research by Abdelwahab and colleagues also shows that shading adjustment behavior is influenced by cultural factors [27]. Although progress has been made in predicting shading adjustment behavior driven by physical factors (such as daylight, sunlight, heating, and ventilation), the role of non-physical motivations in office worker behavior remains poorly explained [28].

Currently, building shading adjustment behavior is mainly influenced by factors such as weather conditions, solar radiation, illuminance, the depth of sunlight penetration, temperature, the type and operation mode of shading devices, building orientation, internal layout, work area location, and air conditioning systems [29]. By organizing and categorizing these factors according to shading function, the influencing factors of shading adjustment are classified, as shown in

Table 1. Influencing factors of building shade adjustment behavior.

Shading Function	Influencing Factors	Major Influencing Factors
Thermal Comfort	Solar radiation heat, thermal adaptability of individuals	Solar radiation (vertical radiation, transmitted radiation), weather conditions, outdoor temperature
Light Comfort	Solar radiation, individuals’ light preferences, shading type, lighting use	Solar radiation (outdoor illuminance, brightness), glare, weather conditions (sunny, cloudy), shading type
View and Privacy Needs	Building factors, surrounding environmental factors, psychological factors of individuals	Orientation, spatial layout, age, gender, culture

.

3. Methodology

The main purpose of this paper is to investigate the driving factors of shading adjustment behavior. These driving factors refer to various internal or external influences that lead individuals or groups to adjust shading to meet different needs. These influences can be environmental, psychological, sociocultural, biological, and technical and economic factors [30,31]. Together, these factors shape people’s shading adjustment behavior. This approach aligns well with Bandura’s social cognitive theory. By integrating social cognitive theory to propose a model of shading adjustment behavior driving factors, and analyzing it through a structural equation model, we can better depict the relationships and influence paths between variables, ultimately deriving the relationships between various driving factors.

3.1. Roadmap

First, a Person–Behavior–Environment model for shading adjustment behavior is established using triadic reciprocal determinism, observational learning, and self-efficacy based on social cognitive theory. A questionnaire is designed based on the variables required by the model, and reliability and validity tests are conducted on the collected data. Hypotheses are proposed, and structural equation models for activating and deactivating shading behaviors are established by collecting data from corresponding questions to describe each observed variable in the model. Subsequently, the structural equation models are fitted, evaluated, and analyzed for hypothesis testing. Finally, conclusions are drawn. The roadmap is depicted in Figure 1.

3.2. Structural Equation Modeling

Structural equation modeling (SEM) is a widely used complex statistical analysis method in social sciences, behavioral sciences, educational research, economics, and medical research. SEM enables researchers to simultaneously examine relationships among multiple dependent variables and the relationships between these variables and one or more independent variables. Through SEM, researchers can construct and test causal relationships between variables in theoretical models, including direct effects, indirect effects, and total effects. The structural equation model comprises measurement equations and structural equations [32].

The measurement equation formula is as follows:

$$Y={\mathrm{\Lambda }}_y\eta +\epsilon ;\mathrm{X}={\Lambda }_{\mathrm{x}}\xi +\delta ;$$

Among them, $Y$ and $\mathrm{X}$ represent the observed matrices of the endogenous variables and exogenous variables, respectively. $\eta$ and $\xi$ represent the latent variable matrices of endogenous variables and exogenous variables, respectively. ${\mathrm{\Lambda }}_{\mathrm{y}}$ and ${\mathrm{\Lambda }}_{\mathrm{x}}$ are the latent variable matrices of the endogenous variables and exogenous variables, respectively. $\epsilon$ and $\delta$ represent the error terms of the observed values of the endogenous and exogenous variables, respectively.

The structural equation formula is as follows:

$$\eta =\mathrm{B}\eta +\Gamma \xi +\zeta ;$$

Among them, $\eta$ is the latent variable matrix of the endogenous variables. $\mathrm{B}$ is the regression coefficient matrix between latent variables of endogenous variables, indicating the causal relationship within latent variables. $\Gamma$ is the effect matrix of the latent variables of exogenous variables on the latent variables of endogenous variables. $\xi$ is the latent variable matrix of the exogenous variables. $\zeta$ represents the error term of the structural equation.

3.3. Social Cognitive Theory

Social cognitive theory (SCT) is a psychological theory proposed by Albert Bandura. Building on social learning theory, SCT emphasizes the critical role of personal beliefs in behavior and thinking processes. It particularly highlights five core capabilities within the human self-system. SCT views human nature as autonomous with great potential, suggesting that individuals can control their destiny while being influenced by their environment. The theory encompasses four main components: triadic reciprocal determinism, observational learning theory, self-efficacy theory, and self-regulation theory, with triadic reciprocal determinism serving as its core foundation [33,34,35].

(1)

Triadic reciprocal determinism posits that human behavior is shaped not by single internal or external factors, but by the continuous interaction among personal factors, environmental factors, and behavior. It proposes a PBE (Person–Behavior–Environment) interaction model comprising these three factors, as depicted in Figure 2.

In the PBE model, personal factors, environmental factors, and behavior continuously interact. Specifically, personal factors and behavior mutually influence each other: on one hand, an individual’s cognitive, emotional, and biological characteristics influence behavior; on the other hand, an individual’s behavior and its outcomes can feedback to influence personal thoughts and emotional states, potentially altering subsequent behavior choices (Path 1). In the interaction between behavior and environment, environmental factors provide the background and conditions for behavior, influencing its direction and intensity. Simultaneously, an individual’s behavior can also modify the environment to meet personal needs or goals (Path 2). In the interaction between the individual and the environment, while environmental factors can impact an individual’s psychology and behavior, the individual’s perception and interpretation of the environment occur through cognitive processing. Therefore, the extent of environmental impact on the individual largely hinges on cognitive factors and interpretations (Path 3).

These interactions among the three sets of factors do not always occur simultaneously, and their mutual influences vary in strength. In specific situations, one factor may dominate, either constraining or enhancing the effects of other factors. This model underscores the multifactorial determinism and dynamic nature of individual behavior, elucidating how individuals develop and adjust behaviors amidst the interplay of internal and external factors.

(2)

Observational learning, also known as vicarious learning, contrasts with direct experiential learning. Essentially, it involves the observer indirectly acquiring cognitive skills or behavioral patterns demonstrated by a model (the observed), and subsequently exhibiting these behaviors in similar scenarios. Unlike Thorndike’s trial-and-error theory and Skinner’s operant conditioning theory, which posit that human knowledge and skills derive primarily from direct learning, Bandura posits that human learning, experience, and skills are largely acquired through indirect observational learning [36,37]. Bandura asserts that observational learning is a pivotal form of human learning and exemplifies human vicarious learning ability. It entails the observer encoding the observed behavioral structures and environmental events demonstrated by the model into symbolic representations within the brain, thereby internally guiding the observer to replicate similar behaviors in the future. Thus, the capability for symbolic representation, one of the five fundamental human capabilities, plays a crucial role in observational learning. Observational learning unfolds through a sequence of four processes, attention, retention, reproduction, and motivation, culminating in the observer’s indirect acquisition of behaviors demonstrated by the model.

(3)

Self-efficacy involves an individual’s cognitive assessment of their ability to organize and execute actions required to achieve goals [38]. It pertains not to the skills one possesses, but rather to one’s judgment of what can be accomplished with those abilities. Simply put, self-efficacy is an individual’s perception and evaluation of their ability to influence themselves, or it can be understood as an internal belief in their understanding and confidence in their abilities. Bandura asserts that self-efficacy, as a crucial driver within an individual’s self-system, influences their behavioral choices, the effort they invest in achieving goals, their perseverance in facing challenges, and their emotional state (whether confident or anxious) when undertaking tasks. It significantly determines whether individuals undertake specific actions, often independently of whether they possess the requisite skills to execute those actions. There are four primary sources of self-efficacy: successful personal experiences, observed experiences of others, encouragement and affirmation from others, and psychological and physiological states encountered when facing tasks. Self-efficacy, to a considerable extent, serves as the key to unlocking an individual’s potential. When an individual’s self-efficacy aligns with their actual knowledge and skills, their potential for achievement can be maximized [39].

Bandura posits that individuals possess the capacity and mechanisms for self-regulation, a manifestation of intrinsic motivation enabling them to plan, compare expectations with actual outcomes, and adjust behaviors accordingly [40]. This cognitive process empowers individuals to manage their lives, plan for the future, and set goals. Individuals persevere and progress by setting future goals, achieving personal targets, fulfilling self-needs, and conducting self-evaluations. Individuals adjust behaviors according to internal standards, employing self-rewards or self-punishment as mechanisms. Broadly, the self-regulation mechanism reflects personal initiative and utility. Narrowly, self-regulation serves as a self-feedback system encompassing self-set behavioral goals, self-observation of behavior, self-evaluation based on observations, and self-reaction. The core contribution of self-regulation theory lies in its emphasis on the proactive and positive aspects of individual psychology, demonstrating human rational cognitive abilities and illustrating the effective control and adjustment of behaviors [41].

3.4. Questionnaire Design

In this study, a questionnaire survey method was employed to collect raw data, with a clear indication that the collected data were solely intended for analyzing the driving factors of shading adjustment behavior, ensuring strict confidentiality of survey content.

To accurately capture respondents’ perspectives, the questionnaire was structured into three parts. The first part provided detailed instructions to clarify the survey’s purpose. The second part gathered basic demographic information, including age, gender, office type, and types of shading devices used. The third part consisted of questions relating to personal factors, environmental factors, and shading adjustment behavior. And the questionnaire is shown in the Appendix A. The relationships between some survey questions and the observed variables are shown in

Table 2. Construction and items.

Item	Question
	Person (Activation)
PA1	In specific seasons, do you feel that the indoor temperature is too hot?
	I frequently adjust shading facilities to ensure that the indoor environment is visually and physically comfortable for me.
PA2	Frequency of activated shading for protecting privacy.
	I believe rational shading can significantly reduce energy consumption.
	For me, shading is not only about adjusting light but also a means to protect privacy.
PA3	Even if there are no significant changes in external conditions, I habitually adjust shading facilities.
	I think adjusting indoor temperature through shading is more natural and comfortable than using air conditioning.
	Environment (Activation)
EA1	Frequency of activated shading for reducing glare.
	Frequency of activated shading for avoiding sunlight affecting work.
EA2	Frequency of activated shading for controlling indoor temperature.
	Frequency of activated shading for addressing indoor overheating issues.
EA3	My family and friends often adjust shading habits.
	I am willing to accept advice from professionals on the best shading practices.
	Activate Shading Behavior
ASB1	I adjust shading facilities daily to adapt to different environmental conditions.
ASB2	I often adjust shading facilities to reflect my satisfaction with current lighting and temperature.
ASB3	Frequency of using shading in the office.
	Person (Deactivation)
PD1	I am highly satisfied with the natural lighting in my office.
	I am willing to sacrifice some comfort for maintenance.
PD2	Energy conservation is an important factor I consider when using shading facilities.
PD3	I prefer natural lighting, even if it means the indoor space may be warmer at times.
	Environment (Deactivation)
ED1	Frequency of deactivated shading for increasing room brightness.
	Frequency of deactivated shading for enjoying sunlight exposure.
ED2	Frequency of deactivated shading for increasing indoor temperature.
ED3	My shading behavior is influenced by the attitudes and behaviors of people around me.
	Others’ shading adjustments affect me.
	I am concerned that my shading adjustments may affect others.
	Deactivate Shading Behavior
DSB1	I adjust shading facilities daily to adapt to different environmental conditions.
DSB2	I often adjust shading facilities to reflect my satisfaction with current lighting and temperature.
DSB3	Frequency of not using shading in the office.

.

To ensure the validity of the research data, various distribution and retrieval methods were utilized, including paper surveys and online platforms such as WeChat and Weibo.

To streamline the data collection and analysis, a Likert five-point scale was employed to establish a rating system from 1 to 5, assessing respondents’ agreement with various question descriptions.

4. Results and Discussion

4.1. Determination of Driving Factors

Currently, the main driving factors of building shading adjustment behavior include weather conditions, solar radiation, illuminance, depth of sunlight penetration, temperature, types and operation modes of sunshading devices, building orientation, interior layout, location of work areas, air conditioning systems, etc. [29].

Based on social cognitive theory, the influencing factors can be further classified to form a table of driving factors of shading adjustment behavior, as shown in

Table 3. Driving factors of shading adjustment behavior.

Latent Variable	Observed Variables
Environmental Factors	Light Environment
	Thermal Environment
	Social Environment
Personal Factors	Comfort Preference
	Personal Awareness
	Personal Habits
Shading Adjustment Behavior	Adjustment Habits
	Adjustment Tendency
	Adjustment Frequency

.

4.2. Questionnaire Analysis

4.2.1. Reliability Test

A total of 268 questionnaires were distributed for this study, of which 258 were returned. After excluding nineteen invalid questionnaires, 239 valid questionnaires remained, resulting in an effective response rate of 89.2%.

To assess the reliability of the questionnaire data, the Cronbach’s alpha coefficient method was employed using SPSS 25.0 software. The results, detailed in

Table 4. Comprehensive data reliability analysis.

Cronbach’s α Coefficient	Cronbach’s α Coefficient Based on Standardized Items
0.916	0.918

and

Table 5. Reliability analysis of various level indicators.

First-Level Indicators	Secondary-Level Indicators	Cronbach’s α Coefficient after Removing Items	Cronbach’s α Coefficient
Environment	Light environment	0.901	0.905
	Thermal environment	0.904
	Social environment	0.899
Person	Personal habits	0.907	0.908
	Personal awareness	0.875
	Comfort preference	0.901
Behavior	Activating habits	0.907	0.911
	Activating tendency	0.902
	Activating frequency	0.904

, indicate an overall Cronbach’s alpha coefficient of 0.916, demonstrating a high level of consistency in the questionnaire data. Each of the three primary indicators also yielded Cronbach’s alpha coefficients exceeding 0.90, confirming the high reliability of the questionnaire data and the robustness of relationships between secondary and primary indicators. Given the minimal impact of deleting observed variables on the Cronbach’s alpha coefficient, all secondary indicators were retained without alteration. In conclusion, the questionnaire used in this study exhibited high reliability, thereby meeting the data requirements for subsequent research.

4.2.2. Validity Test

Initially, we conducted the Kaiser–Meyer–Olkin (KMO) test and Bartlett’s test of sphericity on the selected data to ensure suitability for the factor analysis. Using SPSS 25.0 software, the factor analysis was performed on all observed variables. The results of the KMO test and Bartlett’s test of sphericity are presented in

Table 6. Results of KMO test and Bartlett’s test of sphericity.

Method	Result
KMO Sampling Adequacy Measure	0.836
Bartlett’s Sphericity Test	Approximate chi-square	28.908
	Degrees of freedom	25
	Significance	0.000

. The KMO value of 0.836 from the test indicated adequacy for the exploratory factor analysis, and Bartlett’s test confirmed statistical significance, meeting criteria for a further analysis.

For the factor analysis, the principal component method was employed to compute the factor loading matrix, followed by obtaining the rotated loading matrix using the maximum variance method. The final rotated loading matrix is displayed in

Table 7. Factor loading.

Indicators	Factor 1	Factor 2
Light Environment	0.147	0.702
Thermal Environment	0.030	0.686
Social Environment	0.186	0.567
Comfort Preference	0.616	0.003
Personal Awareness	0.754	0.018
Personal Habits	0.664	0.045

. Based on these loadings, the indicator variables “Light environment”, “Thermal environment”, and “Social environment” collectively measure the latent variable “Environment”. Similarly, the indicator variables “Comfort preference”, “Personal awareness”, and “Personal habits” measure the latent variable “Person”. Upon intuitive examination, the rationale behind this measurement system is evident, supported by empirical confirmation from the data-driven results of the exploratory factor analysis.

4.3. Model Construction

Unlike many statistical methods, structural equation modeling (SEM) allows for a simultaneous analysis of relationships among multiple variables, including direct effects, indirect effects, and bidirectional effects. Using SEM to analyze the driving factors of shading adjustment behavior is an effective method that provides deep insights into shading behavior.

Due to distinct characteristics and driving factors of activating and deactivating shading behaviors, this study distinguishes between them for the analysis. This approach yields three implications:

Firstly, it captures different motivations and factors. Activating and deactivating shading behaviors may be influenced by different factors, and a separate analysis helps identify unique motivations and conditions.

Secondly, it refines the theoretical model. By differentiating these behaviors, a more refined theoretical model can explore their influencing factors separately, enhancing model explanatory power and predictive accuracy.

Thirdly, it enhances practicality and targeted design measures. Distinguishing between these behaviors meets architectural design needs effectively, and understanding their independent driving factors guides development of precise intervention measures and policies.

Therefore, this paper will establish separate structural equation models for activating and deactivating shading behaviors. Combining social cognitive theory with these models, we propose six hypotheses on the relationship between shading adjustment behavior and personal and environmental factors.

Hypothesis 1.

Environmental factors (activating) have a positive correlation with the activating of shading behavior.

Hypothesis 2.

Personal factors (activating) have a positive correlation with the activating of shading behavior.

Hypothesis 3.

Environmental factors (activating) have a positive correlation with personal factors (activating).

Hypothesis 4.

Environmental factors (deactivating) have a positive correlation with the deactivating of shading behavior.

Hypothesis 5.

Personal factors (deactivating) have a positive correlation with the deactivating of shading behavior.

Hypothesis 6.

Environmental factors (deactivating) have a positive correlation with personal factors (deactivating).

The structural model is shown in Figure 3.

4.4. Model Fit

The process of comparing survey questionnaire data with the constructed structural equation model to determine if there is a good fit between them is called a model fit.

In the process of validity testing, different researchers will choose appropriate analysis methods according to the research purpose. Usually, the least squares method and the iterative method are used for fitting estimation, but this paper uses the maximum likelihood estimation method in AMOS 26.0 software for fitting. Firstly, the data from the survey questionnaire are examined for the reliability and validity of each indicator to ensure that each indicator has high reliability and validity, meeting the requirements of the structural equation model. Then, AMOS 26.0 software is used for the analysis, and the final fitting results are obtained. The structural equation models for activating and deactivating shading behaviors are shown in Figure 4, respectively.

4.5. Model Evaluation

In order to assess the fit of the structural equation model, AMOS 26.0 software typically uses the chi-square value (CMIN) as a criterion for judgment. Due to the significant deviation between the measurement results and the model, it is necessary to correct the estimated values of each parameter in the model to achieve better predictive performance. In addition, this is because there is a certain correlation between model parameters, and different types of variables may also influence each other, resulting in bias in the estimation results. Therefore, in order to more accurately assess the adaptability of the model, in addition to considering the chi-square value, other evaluation indicators such as model fit indices need to be comprehensively considered to more comprehensively and effectively evaluate the rationality of the model. This paper briefly introduces commonly used fit indices. See

Table 8. Fitness assessment.

Category	Relative Chi-Square	Residual Mean Square	Comparative Fit Index	Tucker–Lewis Index	Approximate Error Mean Square Root	Non-Normed Fit Index	Normed Fit Index	Incremental Fit Index
Hypothesized Model	0.195	0.041	0.992	0.917	0.053	0.990	0.954	0.992
Fit Standards	<3	<0.05	>0.90	>0.85	<0.80	>0.90	>0.90	>0.90
Fit Assessment	Fit	Fit	Fit	Fit	Fit	Fit	Fit	Fit

for the evaluation criteria of model fit indices.

In summary, all fit indices of the model meet the standards. Therefore, the structural equation model constructed in this study exhibits a high degree of fit after undergoing hypothesis significance testing, meeting the criteria for an overall fit.

4.6. Analysis of Model Results

In the models of activating and deactivating shading behavior, six hypotheses were constructed using the mutual relationships between latent variables. The path coefficients were calculated using AMOS 26.0 software and analyzed to verify the feasibility of these hypotheses. As shown in

Table 9. Path hypothesis testing of activating shading behavior.

Variable	Path	Variable	Estimate	Standard Error	Critical Value	p-Value	Standardized Estimate
Activate shading behavior	<---	Environment (activation)	0.860	0.089	9.602	***	0.97
Activate shading behavior	<---	Person (activation)	0.740	0.091	9.856	***	0.91
Person (activation)	<---	Environment (activation)	0.883	0.072	12.308	***	0.85
Activating habits	<---	Activate shading behavior	1.000			***	0.83
Activating tendency	<---	Activate shading behavior	0.894	0.046	19.298	***	0.89
Activating frequency	<---	Activate shading behavior	0.675	0.046	14.586	***	0.82
Comfort preference (activation)	<---	Person (activation)	1.000			***	0.85
Personal awareness (activation)	<---	Person (activation)	0.902	0.063	14.431	***	0.88
Personal habits (activation)	<---	Person (activation)	0.903	0.058	15.494	***	0.84
Light environment (activation)	<---	Environment (activation)	1.000			***	0.88
Thermal environment (activation)	<---	Environment (activation)	0.953	0.085	11.270	***	0.80
Social environment (activation)	<---	Environment (activation)	0.890	0.069	12.947	***	0.82

*** indicates the value less than 0.001.

and

Table 10. Path hypothesis testing of deactivating shading behavior.

Variable	Path	Variable	Estimate	Standard Error	Critical Value	p-Value	Standardized Estimate
Deactivate shading behavior	<---	Environment (deactivation)	0.880	0.090	12.843	***	0.89
Deactivate shading behavior	<---	Person (deactivation)	0.780	0.079	11.449	***	0.84
Person (deactivation)	<---	Environment (deactivation)	0.737	0.144	5.106	***	0.79
Deactivating habits	<---	Deactivate shading behavior	1.000			***	0.81
Deactivating tendency	<---	Deactivate shading behavior	1.073	0.057	18.960	***	0.74
Deactivating frequency	<---	Deactivate shading behavior	0.820	0.056	16.586	***	0.86
Comfort preference (deactivation)	<---	Person (deactivation)	1.000			***	0.88
Personal awareness (deactivation)	<---	Person (deactivation)	0.902	0.063	14.431	***	0.79
Personal habits (deactivation)	<---	Person (deactivation)	0.780	0.128	13.466	***	0.84
Light environment (deactivation)	<---	Environment (deactivation)	1.000			***	0.89
Thermal environment (deactivation)	<---	Environment (deactivation)	1.396	0.162	8.632	***	0.93
Social environment (deactivation)	<---	Environment (deactivation)	1.069	0.129	8.284	***	0.76

*** indicates the value less than 0.001.

, the hypotheses of the model were validated after examination.

According to the above data, the critical values of all six hypotheses are greater than 1.96, and the p-values are all less than 0.05, meeting the requirements and reaching a significant level. Therefore, all six hypotheses are established, as shown in

Table 11. Conclusions of hypothesis testing.

Number	Model Hypothesis	Result
Hypothesis 1	Environmental factors (activation) have a positive correlation with activating shading behavior	Supported
Hypothesis 2	Personal factors (activation) have a positive correlation with activating shading behavior	Supported
Hypothesis 3	Environmental factors (activation) have a positive correlation with personal factors (activation)	Supported
Hypothesis 4	Environmental factors (deactivation) have a positive correlation with deactivating shading behavior	Supported
Hypothesis 5	Personal factors (deactivation) have a positive correlation with deactivating shading behavior	Supported
Hypothesis 6	Environmental factors (deactivation) have a positive correlation with personal factors (deactivation)	Supported

.

4.7. Discussion

The successful validation of the above hypotheses, in alignment with related scientific research results, further confirms the correctness of the hypothesized model based on social cognitive theory and structural equation modeling. Additionally, the results employed an innovative method to verify the driving effects of social environmental factors and personal factors on shading adjustment behavior. This discovery has twofold significance: First, in studies related to shading adjustment behavior, future research should fully consider the impact of social environmental and personal factors on behavior. For instance, social environmental factors and personal factors should be included as input parameters in predictive models for shading adjustment behavior. On the other hand, this also provides insights for the study of building occupant behavior. The study of human behavior is an interdisciplinary field encompassing psychology, environmental science, sociocultural studies, biology, and other disciplines. When conducting research, it is important to consider feasibility and realism, integrating numerous significant influencing factors to achieve a more accurate fit to actual conditions.

5. Conclusions

Based on social cognitive theory, we conducted a detailed exploration and analysis of the driving factors behind shading adjustment behavior. We categorized classical driving factors into environmental and personal factors, establishing the PBE model of shading adjustment behavior. This model illustrates the interactive influence of environmental factors, personal factors, and shading adjustment behavior. Given the distinct reasons for initiating (or increasing) and ceasing (or decreasing) shading, we developed separate structural equation models to analyze the driving factors of activating and deactivating shading behavior. From the analysis of factor loadings across each path, several conclusions were drawn:

(1)

There is a significant mutual interaction among shading adjustment behavior, personal factors, and environmental factors:

The triadic interaction determinism from social cognitive theory applies effectively to shading adjustment behavior. This indicates that environmental factors positively correlate with personal factors (such as personal awareness and habits) whether in activating shading behavior or deactivating shading behavior. Both environmental and personal factors exert significant influence on shading adjustment behavior.

(2)

The influence of social environmental factors:

In addition to natural environmental factors, social environmental factors also significantly influence shading adjustment behavior. Therefore, it is important to consider promoting more effective shading behavior patterns through social and cultural intervention measures. The application of social cognitive theory in analyzing the driving factors behind shading adjustment behavior offers new strategies for enhancing the acceptance and adoption of shading technology. This includes designing and promoting shading behavior demonstrations that emphasize imitation, and motivating office workers to adopt positive shading practices by sharing successful cases through social media and other platforms.

(3)

Differential effects of environmental factors:

In activating shading behavior, the influence of lighting environmental factors outweighs that of thermal environmental factors. Conversely, in deactivating shading behavior, thermal environmental factors hold greater significance, although lighting environmental factors also exhibit substantial factor loadings. Therefore, when activating shading behavior, prioritizing daylighting requirements as the primary driving factor is crucial. On the other hand, in deactivating shading behavior, balancing the enhancement in indoor thermal comfort and visual perception should be considered pivotal driving factors. These insights provide a foundation for developing subsequent shading adjustment logic models.

(4)

Differential effects of personal awareness factors:

Research indicates that personal consciousness factors, such as privacy awareness and energy-saving awareness, exert a stronger influence on activating shading behavior compared to deactivating shading behavior. This heightened influence may stem from individuals’ tendency to activate shading primarily due to privacy concerns. Therefore, ensuring privacy protection is essential when addressing shading behavior considerations.

Furthermore, we should acknowledge the limitations of this study and, based on these gaps, suggest the following areas for future investigation:

(1)

This research primarily conducted qualitative studies on the driving factors of shading adjustment behavior. Quantifying these factors in specific contexts and applying them across various studies still necessitates extensive follow-up research for deeper exploration.

(2)

In terms of the impact of the external natural environment on buildings, this study employed a new classification method. The driving factors were categorized into environmental and personal factors, with environmental factors further subdivided into light environment, thermal environment, and social environment factors. Regional climate differences are indirectly reflected by combining thermal environment, light environment, and personal factors, influencing indoor shading adjustment behavior. This approach obscures the unique climatic characteristics of different regions. Given the significant climate variations across regions, research focusing on the distinct climatic and sociocultural features of various areas is essential.

(3)

This study utilized social cognitive theory and structural equation modeling to investigate the driving factors of shading adjustment behavior. The study of building occupant behavior is interdisciplinary, integrating aspects of psychology, environmental science, sociocultural studies, and biology. Exploring the applicability of the research methods used in this study to broader investigations of building occupant behavior will be a focus of our future research.