Children’s Well-Being in the Context of Perceived Inclusion and Digitalization: Evidence from a Survey of Rural Japanese Classrooms

Abstract

Even in highly developed countries such as Japan, urban–rural disparities in inclusion and digitalization persist, offering lessons for other nations confronting similar divides. Diversity and inclusion in school environments appear to be associated with children’s well-being. However, few studies have examined how children perceive inclusion in the classroom or how such perceptions—along with digital device use and interpersonal factors—relate to their subjective well-being (SWB). This study broadens the scope of research by incorporating inquisitiveness and generativity, examining these associations among children in rural Japan. A cross-sectional survey was conducted with 2158 elementary and junior high school students in Kochi Prefecture. Students were classified into five diversity-related categories, and multinomial logistic and median regression models were applied to analyze associations with the core outcomes. Notably, two-thirds of the students were classified into the inclusion category. SWB was positively associated with inclusion and negatively with exclusion, while inquisitiveness and generativity were higher among students in the inclusion and differentiation categories. Both traits were positively associated with adult responsiveness, as well as adherence to digital use rules. The findings suggest that inclusive classroom climates and supportive digital practices foster children’s inquisitiveness, generativity, and SWB, although associations are correlational, not causal.

1. Introduction

The 21st century has increasingly been characterized by uncertainty, as frequent and unpredictable events—such as economic crises, pandemics, and environmental disruptions—continue to reshape social and economic landscapes (Ball, 2012; Dishon & Gilead, 2021; Pietrocola et al., 2025). In response to this instability, diversity and inclusion have gained prominence as essential values across sectors such as business, education, and public policy (Chimakati & Kelemba, 2023; Davis & Miller, 1996; Saxena, 2014; Stamps & Foley, 2023). A growing body of research suggests that environments fostering psychological safety and inclusion are associated with enhanced creativity, problem-solving, and overall well-being (Carmeli et al., 2010; Frazier et al., 2017). Within organizational studies, human resource diversity has been conceptualized in terms of four distinct states: inclusion (high uniqueness and high belonging), differentiation (high uniqueness and low belonging), assimilation (low uniqueness and high belonging), and exclusion (low uniqueness and low belonging) (Shore et al., 2011, 2018). While some scholars argue that homogeneity may enhance team cohesion and decision-making efficiency in small groups, others suggest the potential role of diversity in promoting collaboration and innovation within larger organizations and societies (Distefano & Maznevski, 2000; Mannix & Neale, 2005; Modi et al., 2025). As global uncertainty deepens, understanding how individuals experience inclusion—and how this experience relates to their well-being—has become increasingly relevant for designing democratic and resilient societies. However, most existing research has focused on adults in workplaces and communities, leaving significant gaps in our understanding of how children experience diversity and inclusion in educational settings. Moreover, the rapid expansion of digital technologies (e.g., smartphones, tablets, online platforms) has begun to reshape children’s social environments in and beyond the classroom. While digital engagement offers new opportunities for connection and learning, it also raises questions about its influence on students’ sense of inclusion and psychological well-being. Against this backdrop, this study examines how elementary and junior high school students in rural Japan perceive classroom diversity and inclusion, how their use of digital devices may shape their social interactions, and how these factors relate to their subjective well-being (SWB). In addition to SWB, we consider related psychological outcomes such as inquisitiveness and generativity, along with the moderating role of adult responsiveness, as these interrelated dimensions provide a comprehensive understanding of children’s experiences in diverse and digitalized classrooms. By focusing on children’s perspectives, the study aims to contribute to the emerging discourse on inclusion and digitalization in educational contexts.

While numerous studies have examined diversity in workgroups, the concept of inclusion has recently gained significant attention in research on well-being within school settings (Adams & Meyers, 2020; Downey et al., 2015; Juvonen et al., 2019; Randel, 2025; Versteegen & Adams, 2025). As scholarship on diversity and inclusion has progressed, researchers have increasingly explored how these factors contribute to organizational performance, as well as the mechanisms through which the potential benefits of diversity and inclusion can be realized (Gomez & Bernet, 2019; Stahl & Maznevski, 2021; Trkulja et al., 2024). A growing body of work now emphasizes the importance of cultivating inclusive environments—whether in workplaces or schools—where individuals perceive themselves as valued members of a group (Nishii & Leroy, 2022; Sabharwal, 2014; Shafaei et al., 2024; Woods et al., 2024). To understand the psychological basis of such perceptions, Brewer’s Optimal Distinctiveness Theory (ODT) offers a useful framework (Brewer, 1991, 1993; Way et al., 2022; Zhao & Glynn, 2022). According to ODT, individuals are motivated to find social contexts that satisfy two fundamental and often competing needs: the need for belongingness and the need for uniqueness. Supporting this proposition, Pickett et al. (2002) found that individuals preferentially select groups that allow them to balance these needs, especially in uncertain or dynamic environments. Building on this theoretical foundation, we define inclusion in this study as the extent to which individuals perceive themselves to be recognized as important group members (Shore et al., 2011, 2018). This perception arises from experiences that simultaneously fulfill both the need for belonging and the need for uniqueness (Good et al., 2012; Mor Barak et al., 2022; Randel et al., 2018) and serves as the conceptual foundation for our analysis.

Children born and raised in environments where smartphones and other digital devices are ubiquitous are often referred to as digital natives (Agárdi & Alt, 2024; Bennett et al., 2008; Helsper & Eynon, 2010). With the rapid advancement of technologies such as quantum computing and artificial intelligence (AI), digital innovation continues to accelerate at an unprecedented pace (Coccia, 2024; How & Cheah, 2024; Taylor, 2025). These developments are transforming children’s daily lives and educational experiences and may also shape their future career paths and aspirations (Akour & Alenezi, 2022; Mhlanga, 2023; Southworth et al., 2023). While the inclusivity or exclusivity of social environments—such as classrooms and workplaces—remains a key determinant of children’s well-being, these environments are undergoing significant change. The growing presence of internet technologies and AI suggests they increasingly involve non-human actors (e.g., AI chatbots, recommendation systems, and virtual assistants), alongside traditional human interactions (Bobillier Chaumon, 2021; Georgiou, 2023; Kozyreva et al., 2020). In an empirical study involving 511 Japanese children, Hirose (2024) found that curiosity-driven questioning is positively associated with children’s subjective well-being (SWB) and that the quality of adult responses plays a critical role in nurturing children’s intrinsic curiosity. From this perspective, reciprocal interactions—whereby a child poses a question out of curiosity and receives an engaged response from an adult—are considered crucial for fostering children’s SWB (Åkerman et al., 2024; Eaude, 2009; Jirout et al., 2024; Park & Peterson, 2006). However, the potential influence of alternative sources of response—such as internet searches or AI-generated assistance—on children’s curiosity, well-being, and intergenerational relationships remains largely unexplored (Kang et al., 2021; Mhlanga, 2022; Rubin et al., 2024). Overall, while adult engagement remains vital, it is also important to examine how emerging digital actors such as AI and smart devices shape children’s inquiry and well-being (Banks et al., 2024; Clemente-Suárez et al., 2024; Ullrich et al., 2022).

Erikson (1963) introduced the concept of generativity within the framework of life-course theory, defining it as a concern for and commitment to guiding and nurturing the next generation. Generativity can be expressed through a wide range of values and behaviors, including mentoring, community volunteering, or passing down family traditions and skills to younger generations (McAdams & Logan, 2004; Peterson, 2006; Timilsina et al., 2019; Wiktorowicz et al., 2022). To capture individual differences in generativity, several psychometric scales have been developed to assess its various dimensions (Schoklitsch & Baumann, 2012). Among the most widely used is the Loyola Generativity Scale (LGS), which measures generative concern—that is, the emotional and motivational aspects of generativity—and has been commonly used in research (e.g., Jones & McAdams, 2013; Lawford et al., 2005; McAdams & de St. Aubin, 1992; McAdams et al., 2001; Peterson & Duncan, 1999). Another widely used instrument is the Generative Behavior Checklist (GBC), which evaluates generative behavior by assessing the frequency of relevant actions in the past two months (McAdams et al., 1993; Schoklitsch & Baumann, 2012). Studies using these two scales have consistently found a positive association between generative concern and generative behavior, suggesting a conceptual distinction between motivation and behavior, as well as an empirical link between them (McAdams et al., 1993).

Recognizing the importance of cultural context, Marushima and Arimitsu (2007) developed the Revised Generative Concern Scale (r-GCS) for use in Japan. The r-GCS comprises three subscales—creativity, sustaining, and care offering—designed to reflect culturally relevant dimensions of generativity. More recently, Hirose (2024) employed the r-GCS in a questionnaire survey targeting Japanese children, adapting the scale to account for their developmental stage and sociocultural background. Across these studies, generativity has consistently emerged as a strong predictor of subjective well-being (SWB), even when controlling for prosocial tendencies and basic sociodemographic characteristics such as age and gender (Jones & McAdams, 2013; Lawford et al., 2005; Peterson & Duncan, 1999; Pratt et al., 2001; Rittenour & Colaner, 2012; Schoklitsch & Baumann, 2012; Tabuchi et al., 2015). Findings based on the r-GCS further suggest that generative traits in children—such as kindness toward younger peers and concern for environmental sustainability—provide further evidence as significant predictors of their SWB (Hirose, 2024). Overall, generativity correlates strongly with sociodemographic factors such as age, education, and income and is also associated with social sustainability indicators, including prosociality and SWB.

Maslow’s theory suggests that the fulfillment of psychological needs contributes to life satisfaction (Maslow, 1954), which is widely regarded as a core aspect of well-being (Diener, 2009). Various instruments have been developed to evaluate well-being, including the Subjective Happiness Scale (SHS), Ryff’s Psychological Well-Being Scales, and the Satisfaction with Life Scale (SWLS), each capturing distinct dimensions of subjective and psychological well-being (see, e.g., Diener et al., 1985, 2003; Lyubomirsky & Lepper, 1999; Ryff, 1989). Well-being reflects not only material conditions but also emotional satisfaction, interpersonal relationships, and happiness. These factors are strongly associated with individuals’ overall quality of life (QOL), a broad concept that encompasses subjective well-being (SWB) as one of its key dimensions. Beyond economic conditions, happiness has been studied in relation to cultural norms, demographic factors, and psychological traits. Among these, age, gender, marital status, education, self-regard, and interpersonal ties have consistently been highlighted in the literature (Diener et al., 1998, 1999; Chitchai et al., 2020; Jan & Masood, 2008; Kahneman et al., 1999; Lee et al., 2000; Oishi & Diener, 2001). While substantial research has focused on the determinants of well-being, increasing attention has been paid to its potential associations with beneficial outcomes, including higher engagement, optimism, and creativity (Au et al., 2020; Magnani & Zhu, 2018; Meisenberg & Woodley, 2015). Factors such as age, economic status, social relationships, and personality traits are significant correlates of individuals’ well-being and life satisfaction. Furthermore, recent studies suggest that well-being is influenced by various factors but may also shape individuals’ cognitive patterns and behaviors. This underscores the reciprocal nature of well-being, highlighting its role as both an outcome and a contributing component within the broader context of QOL (Bibi et al., 2015; Hirose, 2024; Hirose & Kotani, 2022; Leung et al., 2011; Welsch, 2006; Zidansek, 2007).

People with an inquisitive mindset are more likely to exhibit curiosity about unfamiliar things or people and often initiate conversations by asking questions (Bardone & Secchi, 2017; Black, 2005; Hirayama & Kusumi, 2004; Watson, 2019). Building on the conceptualization of inquisitiveness as a component of critical thinking (Facione et al., 1995; Hirayama & Kusumi, 2004; Hogan, 2009), subsequent research has increasingly explored how inquisitive individuals approach learning and social interaction across diverse contexts and how these behaviors may facilitate innovative solutions (Bardone & Secchi, 2017; Harris, 2011; Hogan, 2009; Kawashima & Petrini, 2004; Watson, 2019). In a study involving 426 Japanese university students, Hirayama and Kusumi (2004) investigated how critical thinking dispositions influence the reasoning process. Their findings indicate that inquisitiveness is associated with forming conclusions that are not constrained by one’s preexisting beliefs. More recently, Hirose and Kotani (2022) and Hirose (2024) found that inquisitiveness is positively correlated with both generative concern and subjective well-being (SWB), based on surveys conducted with Japanese adults and children, respectively. Taken together, these findings suggest that inquisitiveness may be an important motivational factor associated with exploratory behaviors, dialogue initiation, and engagement with unfamiliar environments. Such behaviors, including initiating conversations, asking thoughtful questions, and seeking new experiences, may contribute to higher levels of SWB (Baldwin & Moses, 1996; Black, 2005; Cluver et al., 2013; Hirose, 2024; Hirose & Kotani, 2022).

Despite increasing research on diversity and its associations with creativity and well-being, it remains underexplored how children in Japan experience inclusive or exclusive environments. Recent studies conducted in rural contexts, including India, South Africa, and Japan, have shown that children’s access to and use of ICT are significantly related to socio-demographic conditions (Aruleba & Jere, 2022; Jamil, 2021; Kardam et al., 2024; Kormos & Wisdom, 2021; Nae, 2024). However, few studies have examined how these disparities intersect with children’s perceptions of inclusion and their well-being, particularly in developed countries like Japan, where urban–rural divides persist. This study examines how the cognitive, non-cognitive, and digital environments that children engage with may influence their subjective well-being (SWB). By investigating the extent to which diversity and inclusion are fostered within classroom and digital settings, this research aims to provide insights into how inclusive environments are associated with higher levels of inquisitiveness, creativity, and overall well-being.

The study focuses primarily on children’s subjective well-being (SWB) as the central outcome. It also incorporates inquisitiveness and generativity as related psychological dimensions that capture children’s curiosity-driven engagement and sense of future orientation. These constructs, along with digital device usage and adult responsiveness, are examined together because they interactively shape how inclusive or exclusive environments are experienced by children in contemporary classrooms. Specifically, the study classifies a sample of approximately 2150 elementary and junior high school students in Kochi, Japan, into five diversity-related categories: inclusion, assimilation, exclusion, differentiation, and an intermediate category. Using multinomial logistic regression and median regression analysis, the study explores how these categories are associated with digital device usage, inquisitiveness, generativity, and SWB. The results will elucidate the distribution of students across these categories and identify key factors that characterize each. This will contribute to a deeper understanding of how diverse classroom and digital environments are associated with children’s development and SWB.

Taken together, this study is guided by the following research questions:

• How are Japanese school children distributed across diversity-related categories—such as inclusion, assimilation, exclusion, differentiation, and intermediate—and what factors are associated with their placement in these categories?
• How is children’s SWB associated with their diversity-related categories, inquisitiveness, generativity, and digital device usage?
• How is children’s generativity associated with their diversity-related categories, inquisitiveness, and digital device usage?
• How is children’s inquisitiveness associated with their diversity-related categories, the responsiveness of adults to their questions, and digital device usage?

2. Materials and Methods

2.1. Study Regions

This study was conducted in Kochi Prefecture, located in the southern part of Shikoku Island, Japan. Facing the Pacific Ocean, Kochi is characterized by its mountainous terrain, with flatlands concentrated along coastal areas such as Kochi City and Nankoku City. As of 2025, the prefecture had an estimated population of 648,313 and a total area of 7103 km², making it one of the least densely populated prefectures among Japan’s 47 administrative divisions. The study centers on “City X”, a municipality situated in the flatland region of central Kochi. As of 2025, City X had an estimated population of 45,669 and covered an area of 125.30 km². Its economy is supported not only by agriculture and livestock farming but also by the growth of manufacturing industries and logistics centers, facilitated by infrastructure improvements such as an airport and expressways. City X represents a regional hub where urban and rural features coexist. Despite these developments, Kochi Prefecture faces significant demographic challenges, particularly population aging and decline. According to the prefectural government, Kochi’s population decreased by 1.56% in 2025 compared to the previous year—the second-highest rate of decline among all prefectures in Japan. This figure highlights the severity of depopulation in the region (Dilley et al., 2022; Murayama et al., 2022). In this context, the present study also aims to shed light on the distinctive features of children’s diversity experiences—such as inclusion, assimilation, exclusion, differentiation, and intermediate positions—and their interactions with emerging actors, including digital devices (e.g., tablets and smartphones) and internet connectivity (e.g., broadband infrastructure) within school classrooms. Focusing on this rural Japanese community, which is not only experiencing significant population decline but also places high value on interpersonal assurance and is known for strong social conformity (Yamagishi, 2011; Yamagishi & Yamagishi, 1994), the study seeks to reveal how these dynamics manifest within a highly homogeneous and tightly knit local society.

2.2. Design

This study employed a cross-sectional, questionnaire-based design to examine associations among children’s experiences of classroom inclusion, inquisitiveness, generativity, and subjective well-being (SWB). The study also incorporated digital device use and perceived consultation with adults as additional variables of interest (Figure 1). A total of 2158 elementary and junior high school students completed self-report questionnaires in classroom settings. To account for the temporal structure implied by the constructs, we treated variables such as SWB, inquisitiveness, and generativity as outcomes in some models, while in others, inquisitiveness and generativity served as predictors of SWB. Independent variables were selected based on their presumed temporal precedence over the outcomes—for example, sociodemographic characteristics, relatively stable personality traits, and retrospective assessments of past experiences. Median regression was employed to accommodate the bounded and non-normally distributed nature of the outcome measures. This study serves as an initial step toward longitudinal or experimental research on the factors potentially associated with children’s well-being in increasingly digital and diverse educational contexts. The research adheres to APA guidelines for reporting in educational studies.

2.3. Materials and Procedure

The participants in this study were elementary and junior high school students residing in a municipality in Kochi Prefecture, Shikoku, Japan. A total of 2158 valid responses were collected from 2469 invited students, yielding a response rate of $87.4\%$. The responses were distributed across four districts as follows: District A (n = 1157; junior high school = 440, elementary school = 717), District B (n = 173; junior high school = 52, elementary school = 121), District C (n = 391; junior high school = 126, elementary school = 265), and District D (n = 437; junior high school = 165, elementary school = 272). The participants had a mean age of $11.51$ years (SD = $2.01$), with an age range from 8 to 15 years. Female participants accounted for $46\%$ of the total sample. Data collection was conducted between December 2024 and January 2025.

After obtaining written consent from both the participants and their parents, we conducted a questionnaire survey in several school facilities within the study area. The questionnaire was written in simple language understandable by elementary school students and was designed for ease of response using touchscreen devices. The questionnaire asked participants about the following items in order: (i) The Subjective Happiness Scale (SHS) as an indicator of subjective well-being (SWB); (ii) Inquisitiveness, including curiosity and the behavioral tendency to ask questions of adults and peers; (iii) The Revised Generative Concern Scale (r-GCS) as an indicator of children’s generativity; (iv) The perceived costs and benefits of adult responses to children’s questions (i.e., positive and negative effects). These items, together with children’s communication with family members, constitute the interpersonal factors used to capture the quality of social interactions in home and school contexts; (v) Sociodemographic factors, including gender, age, and household size. (vi) Rules governing and frequency of tablet and PC usage; and (vii) Categorical perceptions of belongingness and uniqueness, including inclusion, assimilation, differentiation, exclusion, and an intermediate category. The variables collected in this survey were classified into cognitive, non-cognitive, and sociodemographic domains in relation to generativity and SWB, as part of the “Essential Elements of Sustainable Living” framework (Hirose et al., 2023), described in Figure 1.

2.4. Measures

2.4.1. Individuals’ Status in Terms of Diversity-Related Categories

To assess students’ perceived status regarding diversity and inclusion in the classroom, we classified them into five diversity-related categories: inclusion, assimilation, exclusion, differentiation, and intermediate. Figure 2 illustrates the distribution of students across these five categories. We adopt the concept of the $2\times 2$ framework of inclusion proposed by Shore et al. (2011). Shore et al. (2011) introduced the “Inclusion Framework”, arguing that uniqueness and belongingness interact to create a sense of diversity. In our study, we developed questionnaire items to measure uniqueness and belongingness and classified respondents into five categories—“inclusion”, “exclusion”, “assimilation”, “differentiation”, and “intermediate”—based on their scores, which represent a novel aspect of our approach.1 Prior research on diversity and inclusion highlights the potential benefits of simultaneously experiencing belongingness and uniqueness (Friedman et al., 1998; Shore et al., 2011). Although the measure is parsimonious, its validity is supported by previous studies showing that similar single- or two-item scales can reliably capture the core dimensions of inclusion.

The questionnaire included two items: (1) “I am able to express my true self in my school classroom”, and (2) “I feel a sense of belonging in my school classroom.” Participants responded to each item on a seven-point Likert scale, where 1 indicated that the statement never applied to them and 7 indicated that it applied very often or nearly always. The vertical axis was set to represent “self-expression” and the horizontal axis to represent “sense of belonging”, both on a scale ranging from 1 to 7, as shown in Figure 2. Participants who scored between 5 and 7 for “self-expression” and between 1 and 3 for “sense of belonging” were classified into the differentiation category. Participants who scored between 1 and 3 for “self-expression” and between 5 and 7 for “sense of belonging” were classified into the assimilation category. Participants who scored between 5 and 7 for both “self-expression” and “sense of belonging” were classified into the inclusion category. Finally, participants who scored 4 on either “self-expression” or “sense of belonging” were classified into the intermediate category. These cut-off points (1–3, 4, 5–7) were set in line with prior applications of the inclusion framework, although an element of arbitrariness was inherent in such classifications. In addition, given that Japanese respondents often prefer middle options in survey responses (central tendency bias), those who selected 4 on either or both items were assigned to the intermediate category (Rindfuss et al., 2015), thereby reducing ambiguity in the classification of the other four groups. This classification framework enabled us to capture nuanced diversity experiences in classroom environments. This framework has certain limitations. As the study was conducted in a single city within a rural prefecture (Kochi), the findings may reflect context-specific characteristics of children and may not be readily generalizable to other regions or urban settings. In addition, because the data were collected through a one-time self-administered survey, there is some potential for biases related to the timing of the survey, social desirability, or differences in self-perception. Nonetheless, the study offers important insights into the distribution of children across the five diversity-related categories.

2.4.2. Subjective Happiness Scale (SHS)

To assess participants’ perceived subjective well-being (SWB), we employed the Subjective Happiness Scale (SHS) (Lyubomirsky & Lepper, 1999), a widely used instrument in research on SWB. The SHS consists of four items, each rated on a 7-point Likert scale (1 = “Strongly disagree”, 7 = “Strongly agree”). The fourth item was reverse-coded, and the four items were summed to yield a total score ranging from 4 to 28. This total SHS score was used as the indicator of SWB in our regression analyses. In this study, the SHS showed high internal consistency (Cronbach’s $\alpha =0.86$). The full list of items is provided in Appendix A.

2.4.3. Revised Generative Concern Scale (r-GCS)

Various scales have been developed to capture individual differences in generativity across multiple dimensions (Schoklitsch & Baumann, 2012). To assess “generative concern”, we referred to the Loyola Generativity Scale (LGS), a widely recognized instrument in the literature (e.g., Jones & McAdams, 2013; Lawford et al., 2005; McAdams & de St. Aubin, 1992; McAdams et al., 2001; Peterson & Duncan, 1999). The Generative Behavior Checklist (GBC) is another frequently used measure for assessing generative behaviors (McAdams et al., 1993; Schoklitsch & Baumann, 2012), and previous studies have reported a positive correlation between the LGS and GBC (McAdams et al., 1993). To adapt the concept of generativity to the Japanese cultural context, Marushima and Arimitsu (2007) developed the Revised Generative Concern Scale (r-GCS), which comprises three subscales: “creativity”, “sustaining”, and “care offering.” Given cultural differences and the limited life experience of children, some items from the LGS and GBC may be difficult for younger respondents to interpret. Therefore, in this study, we used the “care offering” subscale of the r-GCS, which has been shown to be appropriate and accessible for children (Hirose, 2024). This subscale focuses on concern for others and for the natural environment. Items were rated on a 5-point Likert scale (0 = “not at all true” to 4 = “very true”), and the item scores were summed to yield a total score ranging from 0 to 28. The subscale showed good internal consistency (Cronbach’s $\alpha =0.82$). The full list of items is provided in Appendix A.

2.4.4. Inquisitiveness Subscale

To assess participants’ inquisitiveness, we employed the inquisitiveness subscale developed by Hirayama and Kusumi (2004), which evaluates inquisitiveness as a component of critical thinking. This subscale measures curiosity about unfamiliar situations and the behavioral tendency to ask questions of both adults and peers (Futami et al., 2020; Hirayama & Kusumi, 2004; Nakagawa, 2016). It consists of ten items rated on a 5-point Likert scale (1 = “Strongly disagree” to 5 = “Strongly agree”). Item scores were summed to yield a total score ranging from 10 to 50. In this study, the subscale showed excellent internal consistency (Cronbach’s $\alpha =0.89$). This total score was used as the indicator of inquisitiveness in our regression analyses. The full list of questionnaire items is provided in Appendix A. Furthermore, previous studies have validated this subscale as a reliable measure of both behavioral tendencies and attitudinal orientations across diverse contexts (Hirose, 2024; Hirose & Kotani, 2022; Nakagawa, 2016).

2.4.5. Cost–Benefit Perceptions in Help-Seeking Behavior

To assess children’s perceptions of the expected costs and benefits of consulting adults, we employed the positive and negative effects subscales developed by (Nagai & Arai, 2007). The concept of cost–benefit perceptions in help-seeking behavior has been widely explored in social psychology (Hirose, 2024; Nagai & Arai, 2007; Nagai & Koike, 2018). The positive subscale consists of eight items measuring the perceived likelihood of supportive responses from adults when children seek advice. By contrast, the negative subscale includes six items assessing the perceived risk of dismissive or unsupportive reactions. All items were rated on a 5-point Likert scale (1 = “Strongly disagree”, 5 = “Strongly agree”). The total possible scores range from 8 to 40 for the positive subscale and from 6 to 30 for the negative subscale. The complete list of items is provided in Appendix A. In our sample, the positive and negative subscales showed high internal consistency (Cronbach’s $\alpha =0.91$ and $\alpha =0.90$, respectively). Previous studies have validated both subscales as reliable measures of children’s consultation-related attitudes and behaviors across diverse contexts (Nagai & Arai, 2007; Nagai & Koike, 2018).

3. Analysis

3.1. Multinomial Logit Model

The Multinomial Logit (MNL) model is a widely used discrete choice model suitable for analyzing categorical outcomes without a natural ordering. In this study, we employed the MNL model to examine participants’ classification into one of five diversity-related categories—inclusion, exclusion, differentiation, assimilation, and intermediate—as illustrated in Figure 2. Although these categories are derived from responses to items with ordinal-like scales, the resulting classification does not exhibit a clearly defined ordinal structure. Accordingly, the MNL model was deemed appropriate for this analysis. For each individual n, the probability of choosing category i was modeled within the latent utility framework described by Washington et al. (2020).

To address research question (1), we employed the multinomial logit (MNL) model to examine the factors associated with children’s perceived inclusion. This method is widely used in the social sciences due to its interpretability, flexibility, and suitability for categorical outcome variables. The dependent variable comprised five nominal, unordered categories—inclusion, assimilation, differentiation, exclusion, and intermediate—each representing a distinct perceived status regarding classroom diversity. Given the lack of inherent ordering among these outcomes, the MNL model was suitable for estimating the probabilistic relationships between the outcome and a set of explanatory variables, including age, gender, digital device usage, family communication, and psychological traits, namely generativity (r-GCS) and inquisitiveness.

The MNL model estimates the log-odds of membership in each category relative to a baseline category. In our analysis, we designated “Intermediate” as the reference category. The probability that individual i belongs to category j is given by

$$\mathrm{P}(Y_i=j)={\displaystyle \frac{exp\left(X_i{\beta }_j\right)}{{\sum }_{k=1}^{J}exp\left(X_i{\beta }_k\right)}}\mathrm{for}j=1,\dots ,J,\mathrm{with}{\beta }_J=0,$$

(1)

where $Y_i$ is the outcome variable for individual i, $X_i$ is the vector of explanatory variables, and ${\beta }_j$ is the parameter vector associated with category j. Model parameters were estimated using maximum likelihood estimation. In addition to reporting coefficient estimates, we also presented Average Marginal Effects (AMEs) to facilitate interpretation. AMEs quantify the expected change in the predicted probability of each outcome associated with a one-unit change in a given predictor, providing a more intuitive understanding of the model results.

3.2. Median Regression Model

To address research questions (2), (3), and (4), we performed regression analyses using the Subjective Happiness Scale (SHS), the Revised Generative Concern Scale (r-GCS), and the inquisitiveness subscale as dependent variables, respectively. Given that the distributions of these variables are non-normal, often exhibiting skewness, we applied median regression rather than parametric mean-based methods such as ordinary least squares (OLS) regression. We conducted normality tests using the Shapiro–Wilk statistic for all three dependent variables. The null hypothesis of normality was rejected for SHS ($z=7.273, p<0.001$), r-GCS ($z=8.658, p<0.001$), and inquisitiveness (z = 8.578, p < 0.001). Although the Shapiro–Wilk test is generally recommended for sample sizes ranging from 4 to 2000, our sample of 2158 slightly exceeds this upper limit. To ensure robustness, we supplemented the statistical tests with visual inspection of the distributions using histograms, which also indicated deviations from normality, as shown in Figure 3. These findings indicate that SHS, r-GCS, and inquisitiveness are not normally distributed. Prior research suggests that median regression is more robust than mean-based regression, particularly when the dependent variable is bounded, skewed, or otherwise non-normally distributed, or when extreme values are present (Wooldridge, 2016; Wooldridge et al., 2020). Therefore, we adopt median regression to estimate the determinants of subjective well-being, generativity, and inquisitiveness, using the model specifications presented in Equations (2)–(4).

Before testing the hypotheses using regression models, we examined the correlation matrix to assess potential multicollinearity among the independent variables. Correlation analysis describes the strength and direction of linear relationships between two variables; however, it was not used here as a basis for causal inference. Rather, it served as supplementary descriptive information to enhance transparency and to verify that no problematic correlations exist among the explanatory variables. To avoid redundancy and artificial multicollinearity in the correlation matrix, we excluded the five dummy variables representing the diversity-related categories. Instead, we included a numerically coded categorical variable summarizing diversity status. In the regression models, however, we employed the full set of dummy variables to capture the non-linear effects of each category. As shown in

Table 1. Correlation Matrix of Key Variables.

Variables	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22
1. diversity-related categories	1.000
2. Subjective happiness scale (SHS)	−0.431	1.000
3. Inquisitiveness	−0.354	0.411	1.000
4. r-GCS	−0.299	0.341	0.635	1.000
5. Positive responses	−0.391	0.444	0.493	0.490	1.000
6. Negative responses	0.196	−0.207	−0.140	−0.131	−0.318	1.000
7. Ask. Adults	−0.180	0.263	0.420	0.348	0.297	−0.113	1.000
8. Ask. Friends	−0.199	0.204	0.318	0.273	0.300	−0.142	0.407	1.000
9. Age	0.015	−0.095	−0.162	−0.156	0.020	−0.098	−0.112	0.067	1.000
10. Female	−0.062	0.018	0.010	0.139	−0.007	−0.148	0.076	−0.000	−0.003	1.000
11. Household size	0.039	−0.026	0.013	0.019	−0.021	0.028	−0.012	0.019	−0.067	−0.029	1.000
12. Breakfast/wk	−0.060	0.126	0.149	0.137	0.087	0.014	0.105	0.046	−0.192	−0.071	−0.017	1.000
13. Dinner/wk	−0.083	0.123	0.086	0.085	0.111	−0.048	0.046	0.054	−0.031	−0.013	−0.071	0.448	1.000
14. Shop with Adults	−0.024	0.044	0.132	0.172	0.050	0.024	0.069	0.008	−0.278	0.078	0.032	0.221	0.187	1.000
15. Cram school	−0.022	−0.010	0.031	−0.006	−0.032	0.093	0.002	0.006	0.032	−0.056	−0.064	0.049	0.001	0.074	1.000
16. Lesson other than cram school	−0.065	0.125	0.124	0.116	0.059	0.036	0.076	0.031	−0.219	− 0.105	0.107	0.092	0.032	0.099	0.104	1.000
17. Tablet for learning	−0.137	0.150	0.227	0.180	0.179	−0.055	0.108	0.138	0.010	−0.026	0.019	0.056	0.062	0.075	0.007	0.040	1.000
18. PC for learning	−0.198	0.173	0.235	0.217	0.205	−0.138	0.143	0.179	0.037	0.038	−0.021	0.029	0.055	0.039	−0.012	0.042	0.374	1.000
19. Usage time high day	0.056	−0.083	−0.026	−0.061	−0.035	0.020	−0.047	0.021	0.145	−0.040	−0.030	−0.040	−0.032	−0.036	0.019	−0.079	0.042	0.097	1.000
20. Rule and time usage	−0.092	0.136	0.183	0.190	0.125	0.010	0.131	0.034	−0.122	0.002	−0.001	0.154	0.088	0.052	0.013	0.062	0.035	0.097	−0.079	1.000
21. Appropriate use of digital devices	−0.187	0.251	0.312	0.323	0.284	−0.126	0.177	0.133	−0.032	0.083	−0.010	0.080	0.042	0.070	0.002	0.071	0.144	0.188	−0.048	0.155	1.000

Note. 1. Diversity-Related Categories; 2. Subjective Happiness Scale (SHS); 3. Inquisitiveness; 4. r-GCS = Revised Generative Concern Scale; 5. Positive Responses; 6. Negative Responses; 7. Ask Adults = Asking adults about non-study topics; 8. Ask Friends = Asking friends about non-study topics; 9. Age; 10. Female; 11. Household Size; 12. Breakfast/wk = Frequency of family breakfasts per week; 13. Dinner/wk = Frequency of family dinners per week; 14. Shop with Adults = Frequency of shopping with adults per week; 15. Cram School = Weekly cram school attendance; 16. Lesson other than cram school = Weekly non-cram extracurricular activity attendance; 17. Tablet for Learning = Extent of tablet use for learning purposes; 18. PC for Learning = Extent of PC use for learning purposes; 19. Usage Time High Day = Maximum daily screen time in the past week; 20. Rule and Time Usage = Family rules and time restrictions on digital device use; 21. Appropriate Use of Digital Devices = Parental guidance on appropriate digital device use.

, the highest correlations were observed between inquisitiveness and generativity ($r=0.635, p<0.001$), inquisitiveness and adults’ positive responses ($r=0.493, p<0.001$), and subjective well-being (SWB) and adults’ positive responses ($r=0.444, p<0.001$). Importantly, none of the pairwise correlations exceeded the commonly accepted threshold of $\left|r\right|=0.7$, indicating no substantial concerns regarding multicollinearity. In constructing the regression models, particular attention was paid to the conceptual ordering of independent and dependent variables. Independent variables were selected based on their theoretical stability and presumed temporal priority. For example, adults’ positive responses were measured retrospectively, reflecting past experiences. Generativity (r-GCS) and inquisitiveness were treated as relatively stable cognitive and dispositional traits. In contrast, SWB was modeled as the dependent variable, as it reflects respondents’ current emotional states and may fluctuate more readily over time. This model structure was designed to maintain consistency with our theoretical framework.

$${\mathrm{SWB}}_i={\alpha }_0+{\alpha }_1·{\mathrm{r}\text{-}\mathrm{GCS}}_i+{\alpha }_2·{\mathrm{Inq}}_i+{\alpha }_3·{\mathrm{Diversity}}_i+{\mathit{\alpha }}_4·{\mathbf{x}}_i^{\prime }+{ϵ}_i$$

(2)

In Equation (2), ${\mathrm{SWB}}_i$ denotes the subjective well-being score of individual i. The parameters ${\alpha }_0,{\alpha }_1,{\alpha }_2,{\alpha }_3,{\mathit{\alpha }}_4$ are coefficients estimated from the data, and ${ϵ}_i$ represents the error term. The vector ${\mathbf{x}}_i^{\prime }$ includes control variables such as age, gender, and household size. The coefficients ${\alpha }_1$ and ${\alpha }_2$ are of particular interest, as they capture the associations between SWB and generativity (r-GCS) and inquisitiveness (Inq), respectively.

This specification directly addresses research question (2).

To model the generativity score (r−GCS) of participant i, we specified the following regression equation:

$${\mathrm{r}\text{-}\mathrm{GCS}}_i={\beta }_0+{\beta }_1·{\mathrm{Inq}}_i+{\beta }_2·{\mathrm{Digital}}_i+{\beta }_3·{\mathrm{Response}}_i+{\beta }_4·{\mathrm{Diversity}}_i+{\mathit{\beta }}_5·{\mathbf{x}}_i^{\prime }+{ϵ}_i$$

(3)

In Equation (3), ${\mathrm{r}\text{-}\mathrm{GCS}}_i$ denotes the generativity score of participant i. ${\mathrm{Digital}}_i$ represents the level of digital device usage; ${\mathrm{Response}}_i$ reflects the participant’s perception of adult responses to their inquiries; ${\mathrm{Diversity}}_i$ indicates their assigned diversity-related category; and ${\mathbf{x}}_i^{\prime }$ is a vector of sociodemographic control variables, including gender, age, and household size. The parameters ${\beta }_0,{\beta }_1,{\beta }_2,{\beta }_3,{\beta }_4,{\mathit{\beta }}_5$ were estimated, and ${ϵ}_i$ is the error term. The coefficient ${\beta }_1$ is of particular interest, as it captures the association between inquisitiveness (Inq) and generativity (r−GCS), which directly addresses research question (3).

To model inquisitiveness, we specified the following regression equation:

$${\mathrm{Inq}}_i={\gamma }_0+{\gamma }_1·{\mathrm{Digital}}_i+{\gamma }_2·{\mathrm{Response}}_i+{\gamma }_3·{\mathrm{Diversity}}_i+{\mathit{\gamma }}_4·{\mathbf{x}}_i^{\prime }+{ϵ}_i$$

(4)

In Equation (4), ${\mathrm{Inq}}_i$ denotes the inquisitiveness score of participant i; ${\mathrm{Digital}}_i$ represents the level of digital device usage; ${\mathrm{Response}}_i$ reflects the participant’s perception of adult responses to their inquiries; ${\mathrm{Diversity}}_i$ indicates the participant’s assigned diversity-related category; and ${\mathbf{x}}_i^{\prime }$ is a vector of sociodemographic control variables, including gender, age, and household size.

The coefficients ${\gamma }_0,{\gamma }_1,{\gamma }_2,{\gamma }_3,{\mathit{\gamma }}_4$ were estimated, and ${ϵ}_i$ is the error term. The coefficient ${\gamma }_1$ is of primary interest, as it captures the association between inquisitiveness (Inq) and digital device usage, which directly addresses research question (4).

4. Results

4.1. Descriptive Statistics

Table 2. Variable definitions.

Variables	Descriptions
Age	Age is measured in years and ranges from 8 to 15.
Gender	Gender is coded as a binary variable, taking the value 1 if the participant is female and 0 if male.
Household size	Household size is defined as the total number of people living in the participant’s household.
Weekly cram school attendance	Weekly cram school attendance refers to the number of days per week the participant attends a cram school.
Weekly non-cram activity attendance	Number of days per week the participant attends extracurricular activities other than cram schools.
Extracurricular sports attendance	This variable is coded as 1 if the participant attends a sports club outside of school, and 0 otherwise.
Extracurricular music attendance	This variable is coded as 1 if the participant attends music lessons outside of school, and 0 otherwise.
Extent of PC use for learning	This variable measures how frequently the participant uses a PC for learning purposes, based on a 4-point Likert scale (1 = “never” 4 = “very often”).
Extent of a tablet use for learning	This variable measures how frequently the participant uses a tablet for learning purposes, based on a 4-point Likert scale (1 = “never”, 4 = “very often”).
Expected future computer use for work	This variable captures the participant’s expected future computer use for work, rated on a 4-point Likert scale (1 = “not at all”, 4 = “very much”).
Maximum daily screen time (past week)	This variable measured using a 5-point scale indicating the duration of tablet or PC use on the day of highest usage in the past week: 1 = “less than 1 h”, 2 = “1 to less than 2 h”, 3 = “2 to less than 3 h”, 4 = “3 to less than 5 h”, and 5 = “5 h or more.”
Family rules on digital device use	This variable is coded as 1 if the participant reports that their family has rules regarding both usage rules and time limits of digital devices and 0 otherwise.
Adherence to digital device rules	This variable measures the extent to which the participant follows family rules when using digital devices, based on a 4-point Likert scale (1 = “not at all”, 4 = “very well”).
Ask adults about non-study topics	This variable measures how often the participant asks adults about topics unrelated to schoolwork, using a 5-point Likert scale (1 = “never”, 5 = “very often”).
Ask friends about non-study topics	This variable measures how often the participant asks friends about topics unrelated to schoolwork, using a 5-point Likert scale (1 = “never”, 5 = “very often”).
Talks most with mother at home	This variable is coded as 1 if the participant reports mother as primary conversational partner at home and 0 otherwise.
Talks most with father at home	This variable is coded as 1 if the participant reports father as primary conversational partner at home, and 0 otherwise.
Family breakfasts per week	This variable indicates the number of times the participant had breakfast together with family members in the past week.
Family dinners per week	This variable indicates the number of times the participant had dinner together with family members in the past week.
Weekly family shopping frequency	This variable indicates the number of times the participant went shopping at a supermarket or store with family members in the past week.
Diversity-related categories	Five-category nominal variable classifying participants into one of five diversity-related categories: Inclusion, Assimilation, Differentiation, Exclusion, or Intermediate.
Generativity (r-GCS)	Generativity is measured using the Revised Generative Concern Scale (r-GCS).The total score ranges from 0 to 28.
Inquisitiveness	This scale is a subscale of critical thinking and yields scores ranging from 10 to 50.
Subjective happiness scale (SHS)	Total subjective well-being score, ranging from 4 to 28, measured using a 7-point Likert scale.
Cost–Benefit perceptions in help-seeking behavior	This scale consists of two subscales measuring perceived benefits and costs of consulting with adults. Each item is rated on a 5-point Likert scale (1 = “Strongly disagree”, 5 = “Strongly agree”). The total score ranges from 8 to 40 for the positive subscale and from 6 to 30 for the negative subscale.

and

Table 3. Overview of participants’ demographics and key variables.

	Kochi Prefecture
	Mean	Median	SD 1	Min	Max
Participant Data
Age	$11.51$	11	$2.01$	8	15
Gender(reference group = male)	$0.46$	0	$0.50$	0	1
Household size	$4.49$	4	$1.25$	2	8
Weekly cram school attendance	$1.08$	0	$1.76$	0	7
Weekly non-cram activity attendance	$1.63$	1	$1.99$	0	7
Extracurricular sports attendance (reference group = others)	$0.49$	0	$0.50$	0	1
Extracurricular music attendance (reference group = others)	$0.12$	0	$0.32$	0	1
Family Communication
Ask adults about non-study topics	$3.89$	4	$1.13$	1	5
Ask friends about non-study topics	$4.03$	4	$1.07$	1	5
Talks most with mother at home (reference group = others)	$0.56$	1	$0.50$	0	1
Talks most with father at home (reference group = others)	$0.15$	1	$0.36$	0	1
Family breakfasts (times per week)	$2.99$	2	$2.82$	0	7
Family dinners (times per week)	$4.74$	5	$2.43$	0	7
Weekly family shopping frequency	$3.07$	2	$2.64$	0	8
Diversity-related categories
Inclusion	$0.67$	1	$0.47$	0	1
Assimilation	$0.04$	0	$0.20$	0	1
Differentiation	$0.02$	0	$0.15$	0	1
Exclusion	$0.03$	0	$0.18$	0	1
Intermediate	$0.24$	0	$0.42$	0	1
Digital Usage
Frequency of tablet use for learning	$3.20$	4	$0.99$	1	4
Frequency of PC use for learning	$3.42$	4	$0.80$	1	4
Expected future computer use for work	$2.78$	3	$1.00$	1	4
Maximum daily screen time (past week)	$2.49$	2	$1.32$	1	5
Family rules on digital device use (reference group = others)	$0.30$	0	$0.46$	0	1
Adherence to digital device rules	$3.36$	3	$1.07$	1	5
Psychological Traits
Generativity (r-GCS)	$22.60$	23	$3.80$	7	28
Inquisitiveness	$38.91$	40	$7.57$	10	50
Subjective happiness scale (SHS)	$18.33$	$18.5$	$2.79$	6	28
Positive responses from adults	$31.50$	32	$6.75$	8	40
Negative responses from adults	$10.35$	8	$5.32$	6	30
Subjects	n = 2158

¹ SD stands for standard deviation.

present the variable definitions and summary statistics for the sample. The final sample includes 2158 elementary and junior high school students from Kochi Prefecture, with a mean age of $11.51$ years ($\mathrm{SD}=2.01$), ranging from 8 to 15 years. Approximately $46\%$ of participants were female. Regarding classroom diversity-related categories, the largest proportion of students were classified into the “Inclusion” category ($67\%$), followed by the “Intermediate” category ($24\%$), while the “Assimilation”, “Differentiation”, and “Exclusion” categories each accounted for between $2\%$ and $4\%$ of the sample. The mean household size was $4.49$ persons. On average, students attended cram schools $1.08$ days per week and participated in other extracurricular activities $1.63$ days per week. In terms of digital device usage, the average frequency of tablet use for learning was $3.20$ on a four-point scale, while PC use for learning averaged $3.42$. The mean score for expected future computer use for work was $2.78$ on a four-point scale. In terms of family communication at home, children most often reported talking with their mother ($56\%$) and less frequently with their father ($15\%$). Families shared breakfast and dinner an average of $2.99$ and $4.74$ times per week, respectively. Psychological traits were measured using validated psychometric scales, with mean scores of $22.60$ for generativity, $38.91$ for inquisitiveness, and $18.33$ for subjective well-being (SWB). Overall, the descriptive statistics indicate a well-balanced sample in terms of gender, school level, digital engagement, and psychological traits.

Table 4. Proportion of five categories related to the expression of individuality and sense of belonging.

Areas	Inclusion and Diversity
	Inclusion	Assimilation	Differentiation	Exclusive	Intermediate
Overall (n = 2158)	$66.6\%$	$4.0\%$	$2.4\%$	$3.5\%$	$23.5\%$
Female (n = 989)	$68.9\%$	$4.6\%$	$2.3\%$	$3.8\%$	$20.3\%$
Male (n = 1169)	$64.6\%$	$3.5\%$	$2.4\%$	$3.2\%$	$26.3\%$

Proportions may not total 100% due to rounding.

presents the diversity classifications by gender. A higher proportion of girls ($68.9\%$) than boys ($64.6\%$) were classified into the “Inclusion” category, while proportionally more boys belonged to the “Intermediate” category.

4.2. Multinomial Logistic Regression Analysis

To empirically address research question (1), we employed the Multinomial Logit (MNL) model, specifying students’ perceived diversity status as the dependent variable. The dependent variable comprised five categories: inclusion, assimilation, differentiation, exclusion, and intermediate. This method estimates the probability of belonging to one of these categories. The independent variables included subjective well-being (SWB), generativity (r-GCS), compliance with digital device usage rules, relevant sociodemographic factors, and additional covariates outlined in Equation (1).

Table 5. Average marginal effects predicting membership in classroom diversity-related categories.

Variable	Five Categories Related to the Expression of Inclusion
	Inclusion	Assimilation	Differentiation	Exclusion	Intermediate
Age	$0.021$ ***	$-0.010$ ***	$-0.006$ ***	$-0.002$	$-0.004$
	($0.005$)	($0.003$)	($0.002$)	($0.002$)	($0.005$)
Gender (base group = male)	$0.017$	$0.015$ *	$-0.003$	$0.010$	$-0.039$ **
	($0.019$)	($0.001$)	($0.007$)	($0.008$)	($0.018$)
Extent of PC use for learning	$0.051$ ***	$0.001$	$-0.003$	$-0.011$ **	$-0.038$ ***
	($0.012$)	($0.006$)	($0.004$)	($0.005$)	($0.011$)
Extent of tablet use for learning	$0.012$	$0.001$	$-0.005$	$-0.006$	$-0.003$
	($0.010$)	($0.005$)	($0.003$)	($0.004$)	($0.009$)
Maximum daily screen time (past week)	$-0.016$ **	$-0.002$	$0.001$	$-0.002$	$0.006$ ***
	($0.007$)	($0.003$)	($0.003$)	($0.003$)	($0.006$)
Family rules on digital device use
((base group = others)	$0.008$	$-0.004$	$0.001$	$0.025$ ***	$-0.031$
	($0.022$)	($0.010$)	($0.007$)	($0.009$)	($0.020$)
Adherence to digital device rules	$0.041$ ***	$-0.007$	$-0.001$	$-0.015$ ***	$0.018$
	($0.015$)	($0.006$)	($0.005$)	($0.006$)	($0.012$)
Generativity (r-GCS)	$0.010$ ***	$-0.002$	$-0.001$	$-0.001$	$-0.011$ ***
	($0.003$)	($0.001$)	($0.001$)	($0.001$)	($0.003$)
Inquisitiveness	$0.014$ ***	$-0.001$	$-0.001$	$-0.003$ ***	$-0.011$ ***
	($0.002$)	($0.001$)	($0.001$)	($0.001$)	($0.001$)
Family breakfasts per week	$-0.004$	$0.002$	$0.002$	$0.003$	$-0.002$
	($0.004$)	($0.002$)	($0.002$)	($0.002$)	($0.004$)
Family dinners per week	$0.010$ **	$-0.0005$	$-0.003$ **	$-0.001$	$-0.005$
	($0.004$)	($0.002$)	($0.002$)	($0.002$)	($0.004$)
Weekly cram school attendance	$0.001$	$0.004$	$-0.003$	$0.002$	$-0.004$
	($0.005$)	($0.002$)	($0.002$)	($0.002$)	($0.004$)
Weekly non-cram activity attendance	$0.009$ *	$-0.004$	$-0.0001$	$-0.001$	$-0.004$
	($0.005$)	($0.003$)	($0.002$)	($0.002$)	($0.004$)
Household size	$-0.013$ *	$0.003$	$0.0004$	$-0.0005$	$0.010$
	($0.007$)	($0.003$)	($0.003$)	($0.003$)	($0.007$)
Talks most with father at home
(base group = others)	$0.052$ *	$-0.019$	$-0.004$ *	$-0.011$ *	$-0.055$ **
	($0.029$)	($0.019$)	($0.012$)	($0.013$)	($0.027$)
Talks most with mother at home
(base group = others)	$0.052$ ***	$-0.001$	$0.007$	$-0.008$	$-0.050$ ***
	($0.021$)	($0.010$)	($0.008$)	($0.009$)	($0.019$)

*** significant at $1\%$, ** significant at $5\%$, * significant at $10\%$.

presents the average marginal effects (AMEs, which show how a one-unit change in a predictor affects the probability of belonging to each category) from the model, examining predictors of classification into five diversity-related categories: inclusion, assimilation, differentiation, exclusion, and intermediate (reference group). Several significant patterns emerged, highlighting significant associations with age, digital engagement, psychological traits, and family communication. First, age was positively associated with the probability of being in the inclusion category ($p<0.01$) and negatively associated with assimilation and differentiation, indicating that older students are more likely to feel both unique and included in their classrooms. Although the effects of gender were comparatively modest, female students showed a marginally significant probability of being classified into the assimilation group ($p<0.1$) and were significantly less likely to be classified into the intermediate category ($p<0.05$).

Regarding digital learning tools, frequent PC use for learning was positively associated with inclusion ($p<0.001$) and negatively associated with both intermediate and einclusionxclusion, suggesting that structured digital engagement (i.e., purposeful, learning-oriented ICT use) is associated with a stronger sense of belonging and self-expression. In addition, adherence to digital device usage rules was positively associated with inclusion ($p<0.01$) and negatively with exclusion ($p<0.01$), indicating that rule-based digital engagement (ICT use following family guidelines) is associated with differences in inclusive versus exclusive classroom experiences. In contrast, extended device usage time on high-use days (periods of unusually long screen use) was negatively associated with inclusion ($p<0.05$) and positively with intermediate, implying that unstructured screen time (excessive use without clear learning purposes) is associated with lower levels of classroom inclusion. Among psychological attributes, both generativity and inquisitiveness were positively associated with inclusion and negatively with intermediate ($p<0.001$), reinforcing the idea that proactive and reflective dispositions (tendencies to act intentionally and think about one’s actions) are associated with inclusive experiences. Inquisitiveness was also negatively associated with exclusion ($p<0.01$). In terms of family interactions, more frequent family dinners were modestly associated with inclusion ($p<0.05$). Moreover, reporting one’s mother or father as the primary conversation partner at home was significantly and positively associated with inclusion ($p<0.01$ and $p<0.1$, respectively) and negatively with intermediate. In sum, these findings suggest that a combination of individual characteristics, learning-oriented ICT use with clear rules, and supportive family environments may contribute to fostering inclusive classroom experiences.

4.3. Median Regression Analysis of Subjective Well-Being (SWB)

To empirically address research question (2), we conducted a median regression analysis, a method less sensitive to outliers than ordinary least squares regression using the Subjective Happiness Scale (SHS) as the dependent variable to assess subjective well-being (SWB). Independent variables included diversity-related categories, generativity (r-GCS), inquisitiveness, adherence to digital device usage rules, relevant sociodemographic characteristics, and additional covariates specified in Equation (2). The results are summarized in

Table 6. Median regression coefficients predicting subjective well-being (SWB) across sequential models.

Variable	Regression Coefficients for SWB
	Model 1	Model 2	Model 3	Model 4
Inclusion (reference group = intermediate category)	$3.000$ ***	$0.191$ ***	$0.188$ ***	$0.189$ ***
	($0.078$)	($0.135$)	($0.140$)	($0.153$)
Assimilation (reference group = intermediate category)	$2.000$ ***	$0.922$ ***	$0.703$ **	$0.809$ **
	($0.174$)	($0.286$)	($0.297$)	($0.323$)
Differentiation (reference group = intermediate category)	$1.000$ ***	$-0.078$	$-0.230$	$-0.137$
	($0.219$)	($0.361$)	($0.375$)	($0.407$)
Exclusion (reference group = intermediate category)	$-1.000$ ***	$-1.476$ ***	$-1.691$ ***	$-1.790$ ***
	($0.184$)	($0.303$)	($0.314$)	($0.342$)
Adherence to digital device rules	$1.000$ ***	$0.272$ ***	$0.362$ ***	$0.306$ ***
	($0.050$)	($0.085$)	($0.089$)	($0.096$)
Inquisitiveness		$0.107$ ***	$0.087$ ***	$0.085$ ***
		($0.009$)	($0.010$)	($0.011$)
Generativity (r-GCS)		$0.078$ ***	$0.068$ ***	$0.047$ **
		($0.018$)	($0.019$)	($0.021$)
Ask adults about non-study topics			$0.218$ ***	$0.236$ ***
			($0.057$)	($0.062$)
Ask friends about non-study topics			$0.019$	$0.027$
			($0.06$)	($0.06$)
Age			$-0.059$ **	$-0.042$
			($0.028$)	($0.032$)
Gender (reference group = male)			$-0.163$	$-0.086$
			($0.112$)	($0.123$)
Weekly cram school attendance				$-0.06$ *
				($0.03$)
Weekly non-cram activity attendance				$0.077$ **
				($0.031$)
Family breakfasts per week				$0.020$
				($0.024$)
Family dinners per week				$0.051$ *
				($0.027$)

*** significant at $1\%$, ** significant at $5\%$, * significant at $10\%$.

. Models were constructed sequentially to assess the incremental contributions of diversity-related categories, psychological attributes, interpersonal interactions, and demographic controls. Across all specifications, inclusion status was positively and significantly associated with SWB. Estimated effect sizes (regression coefficients) ranged from $\beta =0.190$ to $\beta =3.000$, all significant at the $1\%$ level, indicating a robust relationship between classroom inclusion and students’ well-being. By contrast, exclusion was consistently and negatively associated with SWB, with effect sizes ranging from $\beta =-1.000$ to $\beta =-1.790$, statistically significant at the $1\%$ or $5\%$ level. Assimilation also showed a positive association, with effect sizes between $\beta =0.703$ and $\beta =2.000$, significant at the $1\%$ or $5\%$ level. Differentiation, however, was not significantly associated with SWB. Notably, adherence to digital device usage rules was positively associated with SWB across all models. Effect sizes ranged from $\beta =0.272$ to $\beta =1.000$, remaining significant at the $1\%$ level in the fully adjusted specification.

Inquisitiveness and generativity were significant predictors of SWB. Specifically, inquisitiveness was positively associated with SWB, with effect sizes ranging from $\beta =0.085$ to $\beta =0.107$, all statistically significant at the $1\%$ level (highly unlikely to have occurred by chance) across different model specifications. Generativity was similarly associated with higher levels of SWB, with effects ranging from $\beta =0.047$ to $\beta =0.078$, statistically significant at the $1\%$ or $5\%$ level. Interpersonal interactions, particularly conversations with adults, were also significantly associated with SWB. Talking with adults about topics beyond schoolwork was positively associated with SWB, ranging from $\beta =0.218$ to $\beta =0.236$, significant at the $1\%$ level. In contrast, peer interactions were not significantly associated with SWB. Although age initially showed a negative association with SWB, this effect became non-significant after adjustment for psychological traits and social factors. Among other variables, frequent participation in extracurricular activities was positively related to SWB ($\beta =0.077$, $p<0.05$), whereas more frequent attendance at cram schools exhibited a weak negative association ($\beta =-0.058$, $p<0.1$). Additionally, sharing dinner more frequently with family members was positively associated with SWB ($\beta =0.051$, $p<0.1$). Overall, these findings suggest that children’s SWB is associated not only with personal dispositions such as inquisitiveness and generativity but also with experiences of classroom inclusion, structured digital engagement, and supportive family communication.

4.4. Median Regression Analysis on Generativity

To empirically address research question (3), we conducted a median regression analysis with generativity as the dependent variable. The independent variables included diversity status, inquisitiveness, adherence to digital usage rules, and relevant sociodemographic controls, as specified in Equation (3). The results are summarized in

Table 7. Median regression coefficients on generativity.

Variable	Regression Coefficients for Generativity
	Model 1	Model 2	Model 3	Model 4
Inclusion (reference group = intermediate category)	$0.600$ ***	$0.506$ ***	$0.547$ ***	$0.522$ ***
	($0.197$)	($0.184$)	($0.187$)	($0.183$)
Assimilation (reference group = intermediate category)	$0.600$	$0.605$	$0.425$	$0.388$
	($0.419$)	($0.390$)	($0.396$)	($0.388$)
Differentiation (reference group = intermediate category)	$0.440$	$0.580$	$0.616$	$0.558$
	($0.528$)	($0.492$)	($0.499$)	($0.488$)
Exclusion (reference group = intermediate category)	$0.920$ **	$1.111$ ***	$0.887$ **	$0.800$ **
	($0.443$)	($0.413$)	($0.418$)	($0.411$)
Adherence to digital device rules	$0.880$ ***	$0.864$ ***	$0.753$ ***	$0.686$ ***
	($0.124$)	($0.115$)	($0.117$)	($0.115$)
Inquisitiveness	$0.320$ ***	$0.296$ ***	$0.285$ ***	$0.288$ ***
	($0.011$)	($0.011$)	($0.012$)	($0.012$)
Ask adults about non-study topics			$0.235$ ***	$0.214$ ***
			($0.074$)	($0.074$)
Ask friends about non-study topics			$0.210$ ***	$0.232$ ***
			($0.075$)	($0.075$)
Age				$-0.047$
				($0.038$)
Gender (reference group = male)				$0.601$ ***
				($0.144$)
Maximum daily screen time (past week)				$-0.067$
				($0.055$)
Family rules on digital device use (reference group = others)				$0.361$ **
				($0.160$)
Weekly family shopping frequency				$0.112$ ***
				($0.028$)

*** significant at 1%, ** significant at 5%.

. Across all model specifications, Inclusion status was positively and significantly associated with generativity. The estimated effect sizes (i.e., regression coefficients) ranged from $\beta =0.506$ to $\beta =0.600$, and all were statistically significant at the $1\%$ level. These findings suggest that students who perceive themselves as both distinct and accepted within the classroom are more likely to exhibit stronger generative traits. Notably, exclusion also showed a consistently positive association with generativity, with estimated effects ranging from $\beta =0.800$ to $\beta =1.111$, statistically significant at the $1\%$ or $5\%$ level. This result implies that even students who feel socially isolated may develop a sense of generative concern—possibly reflecting a compensatory or resilience-building response to exclusion. Inquisitiveness was strongly and positively associated with generativity across all models, with estimated effects ranging from $\beta =0.285$ to $\beta =0.320$, all statistically significant at the $1\%$ level. These findings emphasize inquisitiveness as a key cognitive disposition underlying students’ generative concern for others and future generations.

Similarly, adherence to digital usage rules was consistently associated with higher levels of generativity, with estimated effects ranging from $\beta =0.686$ to $\beta =0.880$, all statistically significant at the $1\%$ level. This suggests that structured digital habits may support the development of more prosocial and future-oriented attitudes. Interpersonal variables also played a significant role. Talking with adults about topics beyond schoolwork was positively and significantly associated with generativity, with effects ranging from $\beta =0.214$ to $\beta =0.235$, all significant at the $1\%$ level. Conversations with friends showed a similar positive association with generativity ($\beta =0.210$ to $0.232$, $p<0.01$). These findings suggest that open communication, both intergenerational and peer-based, may contribute to generative tendencies. Among the sociodemographic controls, gender emerged as a significant predictor: female participants exhibited higher generativity scores ($\beta =0.601$, $p<0.01$). In addition, shared shopping experiences with family were positively associated with generativity ($\beta =0.112$, $p<0.01$), suggesting that family-based interactions may foster empathy, care, and a sense of social responsibility.

4.5. Median Regression Analysis on Inquisitiveness

To empirically address research question (4), we conducted a median regression analysis with inquisitiveness as the dependent variable. The independent variables included diversity-related categories, intra- and intergenerational communication, adherence to digital device usage rules, and relevant sociodemographic characteristics, along with additional covariates specified in Equation (4).

Table 8. Estimated Median Regression Coefficients for Inquisitiveness.

Variable	Regression Coefficients for Inquisitiveness
	Model 1	Model 2	Model 3	Model 4
Inclusion (reference group = intermediate category)	$2.905$ ***	$2.451$ ***	$2.596$ ***	$2.455$ ***
	($0.395$)	($0.381$)	($0.385$)	($0.367$)
Assimilation (reference group = intermediate category)	$0.671$	$0.475$	$0.809$	$0.531$
	($1.186$)	($0.797$)	($0.806$)	($0.768$)
Differentiation (reference group = intermediate category)	$2.801$ ***	$2.778$ ***	$3.278$ ***	$2.959$ ***
	($1.046$)	($1.012$)	($1.012$)	($0.964$)
Exclusion (reference group = intermediate category)	$1.119$	$1.037$	$1.364$	$0.959$
	($0.881$)	($0.846$)	($0.855$)	($0.815$)
Positive response	$0.619$ ***	$0.413$ ***	$0.400$ ***	$0.417$ ***
	($0.026$)	($0.026$)	($0.027$)	($0.025$)
Negative response	$0.044$	$0.049$ *	$0.061$ **	$0.058$ **
	($0.031$)	($0.029$)	($0.030$)	($0.029$)
Age	$-0.714$ ***	$-0.533$ ***	$-0.510$ **	$-0.455$ **
	($0.077$)	($0.075$)	($0.076$)	($0.077$)
Extent of tablet use for learning		$0.752$ ***	$0.541$ ***	$0.515$ ***
		($0.152$)	($0.156$)	($0.149$)
Family rules on digital device use (reference group = others)		$0.953$ ***	$0.793$ ***	$0.675$ **
		($0.328$)	($0.332$)	($0.317$)
Adherence to digital device rules		$1.602$ ***	$1.401$ ***	$1.360$ ***
		($0.236$)	($0.239$)	($0.228$)
Ask adults about non-study topics		$1.373$ ***	$1.429$ ***	$1.375$ ***
		($0.148$)	($0.150$)	($0.143$)
Ask friends about non-study topics		$0.784$ ***	$0.703$ ***	$0.695$ ***
		($0.155$)	($0.157$)	($0.150$)
Expected future computer use for work			$0.842$ ***	$0.787$ ***
			($0.154$)	($0.147$)
Gender (reference group = male)			$0.074$	$0.187$
			($0.303$)	($0.303$)
Household size			$0.135$	$0.114$
			($0.119$)	($0.114$)
Weekly cram school attendance				$0.200$ **
				($0.081$)
Extracurricular sports attendance (reference group = others)				$0.454$
				($0.300$)
Extracurricular music attendance (reference group = others)				$0.423$
				($0.448$)
Weekly family shopping frequency				$0.128$ **
				($0.057$)

*** significant at $1\%$, ** significant at $5\%$, * significant at $10\%$.

presents the results of this analysis. Across all models, students classified in the inclusion group were associated with significantly higher levels of inquisitiveness compared to those in the intermediate category. The estimated effect sizes (i.e., regression coefficients) ranged from $\beta =2.451$ to $\beta =2.905$, all statistically significant at the $1\%$ level. Similarly, differentiation was positively associated with inquisitiveness, suggesting that a strong sense of uniqueness—despite the absence of belonging—may be linked to heightened curiosity. The estimated effect sizes ranged from $\beta =2.778$ to $\beta =3.278$, with all statistically significant at the $1\%$ level. In contrast, neither exclusion nor assimilation was significantly associated with inquisitiveness.

Positive responses from adults were consistently and strongly associated with inquisitiveness across all models. The estimated effect sizes ranged from $\beta =0.400$ to $\beta =0.619$, and all were statistically significant at the $1\%$ level. In contrast, negative responses showed only a modest positive association with inquisitiveness, and the effects were not statistically significant in Model 1. These findings suggest that although unfavorable interactions may sometimes coincide with increased curiosity, their overall contribution is relatively minor. Age was negatively associated with inquisitiveness, suggesting that younger students are more likely to report higher levels of curiosity. The estimated effect sizes ranged from $\beta =-0.455$ to $\beta =-0.714$, with all statistically significant at the $1\%$ or $5\%$ level across all specifications. Digital engagement patterns were also significantly associated with inquisitiveness. Tablet use for learning ($\beta =0.515$ to $\beta =0.752$), the presence of family rules regarding digital device usage ($\beta =0.675$ to $\beta =0.953$), and adherence to those rules ($\beta =1.360$ to $\beta =1.602$) were all positively associated with inquisitiveness, with significance levels ranging from $1\%$ or $5\%$ across Models 2 to 4. In addition, expected future use of computers for work was also a significant predictor of inquisitiveness ($\beta =0.787$ to $\beta =0.842$). Interpersonal communication was likewise positively associated with inquisitiveness. Talking with adults about topics beyond schoolwork ($\beta =1.373$ to $\beta =1.429$) and with peers ($\beta =0.695$ to $\beta =0.784$) were both positively associated with inquisitiveness. These relationships were statistically significant across relevant model specifications. Among the sociodemographic variables, weekly attendance at cram schools ($\beta =0.200$, $p<0.05$) and frequency of family shopping trips ($\beta =0.128$, $p<0.05$) emerged as modest but statistically significant associations of inquisitiveness, though only in Model 4. In contrast, other factors such as gender and household size did not exhibit significant associations.

5. Discussion

5.1. Summary of Findings

We are now in a position to summarize the findings in relation to the four research questions posed at the end of the introduction. To address these questions, we conducted a cross-sectional questionnaire survey with 2158 Japanese elementary and junior high school students. Drawing on established frameworks, we examined how diversity experiences, inquisitiveness, generativity, and digital device usage relate to children’s well-being. We also considered interpersonal factors such as children’s communication with family members and the responsiveness of adults to their questions. These variables were treated as indicators of the quality of social interactions in home and school contexts.

The first research question asked “How are Japanese school children distributed across diversity-related categories—such as inclusion, assimilation, exclusion, differentiation, and intermediate—and what factors are associated with their placement in these categories?” Our results indicated that approximately two-thirds of the sample were classified as “Inclusion”, suggesting that many students perceive their classrooms as environments where both self-expression and a sense of belonging coexist. The remaining students were distributed across the “Assimilation”, “Differentiation”, “Exclusion”, and “Intermediate” categories. Age, gender, patterns and rules of digital device use; family communication; and psychological characteristics—namely inquisitiveness and generativity—were all significantly associated with category classification.

The second research question asked “How is children’s well-being associated with their diversity-related categories, inquisitiveness, generativity, and digital device usage?” Median regression analyses revealed that children in the inclusion group reported significantly higher levels of subjective well-being (SWB), even after adjusting for sociodemographic and contextual variables. Inquisitiveness and generativity were also positively and independently associated with SWB. The third research question asked “How is children’s generativity associated with their diversity-related categories, inquisitiveness, and digital device usage?” Our analysis found positive associations between generativity and inquisitiveness, as well as between generativity and both adherence to digital usage rules and open communication with adults and peers. These findings indicate potential links between generativity and both individual dispositions and structured social and digital environments.

The fourth research question asked:“How is children’s inquisitiveness associated with their diversity-related categories, the responsiveness of adults to their questions, and digital device usage?” Inquisitiveness was more likely to be higher among students in the inclusion and differentiation categories, and it showed positive associations with frequent interpersonal dialogue and the structured use of digital tools for learning. These findings suggest interconnected associations among inclusive classroom environments, individual psychological tendencies, and structured digital and interpersonal experiences, which may also be related to children’s SWB and generativity (creativity, sustaining, and care offering), especially in the context of increasingly diverse and digitally mediated learning environments.

5.2. Broader Reflections

As many sociologists and social psychologists have noted, rural communities have traditionally been characterized by low population mobility and long-standing interpersonal ties (San Martin et al., 2019; Thomson et al., 2018; Yuki & Schug, 2012). In such environments, maintaining social harmony has often relied on a high degree of homogeneity within the community. Yamagishi and Yamagishi (1994) conceptualized this type of stable, cohesive environment as an “assurance society” in which individuals feel secure due to long-term, predictable relationships. In contrast, their studies described a “trust society” as one where individuals must evaluate whether unfamiliar others can be trusted, particularly in contexts marked by frequent social turnover (Yamagishi, 2011; Yamagishi & Yamagishi, 1994). Researchers have argued that contemporary Japanese society is often seen as transitioning from an assurance-based to a trust-based social structure (San Martin et al., 2019; Thomson et al., 2018; Yamagishi, 2011; Yamagishi & Yamagishi, 1994; Yuki & Schug, 2012). In light of this framework, our findings suggest that inquisitiveness—defined as curiosity and the behavioral tendency to ask questions of adults and peers—may be an increasingly important trait, even within communities often regarded as relatively homogeneous, such as rural Japan. At the same time, fostering inclusive environments that embrace cognitive and identity-based diversity remains an ongoing cultural challenge, particularly in rural settings where traditional norms of conformity and harmony continue to prevail.

A possible mechanism behind the observed link between inclusion and inquisitiveness is the role of psychological safety. When children perceive their classrooms as places where their individuality is accepted and they belong, they may feel less concerned about negative evaluation and more willing to ask questions and explore new ideas. This mechanism may be particularly salient in Japan, where cultural norms emphasize reading the atmosphere, maintaining group harmony, and avoiding standing out. In such contexts, inclusive environments can provide important opportunities for children to express individuality without social costs, thereby fostering inquisitiveness (Yamagishi, 2011; Yamagishi & Yamagishi, 1994). At the same time, digitalization further complicates these dynamics: ICT can serve as a supportive tool that enables diverse forms of self-expression, particularly for children who perceive themselves as different from their peers. Structured and rule-based digital use appears to strengthen inclusion and inquisitiveness, whereas unregulated or excessive use may undermine social connections and well-being. Thus, the impact of digital technologies on learning environments is not uniform but depends heavily on the pedagogical and cultural conditions that shape how such tools are used.

In our study, children in the rural Kochi Prefecture were categorized into diversity-related categories based on their self-reported levels of self-expression (vertical axis) and sense of belonging (horizontal axis), as shown in Figure 2. Approximately two-thirds of participants were classified into the inclusion category, indicating that a substantial proportion of children perceive themselves as both accepted and able to express their individuality in classroom settings. The presence of students in the differentiation category, however, suggests an alternative pattern, wherein self-expression occurs without a strong sense of group belonging. This pattern may partly reflect the use of information and communication technologies (ICT), which can provide individualized spaces for identity exploration beyond traditional peer-group dynamics. These observations imply that even in culturally cohesive and low-mobility rural communities, ICT may open new avenues for engagement with broader social networks and diverse perspectives. Future studies should further investigate how digital tools interact with processes of social integration and identity formation in such educational contexts.

As information and communication technologies (ICT) continue to evolve, their impact on children’s well-being depends not only on access but also on the guidance surrounding their use. Our findings suggest that learning-oriented use of personal computers and clearly defined family rules regarding device usage are more likely to show positive associations with children’s inclusion, inquisitiveness, generativity, and subjective well-being (SWB). In contrast, the lack of such structure appears to be associated with higher odds of exclusion. These results highlight the importance of cultivating digital literacy and promoting responsible usage behaviors in both home and educational settings. Given the importance of supportive family environments in guiding children’s digital engagement, one promising approach is Future Design (Saijo, 2020, 2023; Timilsina et al., 2020). This participatory framework promotes intergenerational dialogue, long-term thinking, and consensus-building within families. While it has been further developed in studies addressing sustainability challenges in low- and middle-income countries (Mostafizur et al., 2025; Pandit et al., 2021), it may also be well suited to fostering responsible digital habits at home and creating more inclusive learning environments in developed societies.

Recent international studies have examined how socio-demographic factors influence rural students’ access to and use of ICT (Aruleba & Jere, 2022; Jamil, 2021; Kormos & Wisdom, 2021). For example, Kardam et al. (2024) investigated ICT gadget usage among rural secondary school students in Haryana, India, and found that while smartphones and televisions were widely used, their use varied considerably across schools and was positively correlated with education level but negatively with distance from urban centers. These findings highlight the strong influence of local socio-demographic conditions on ICT use. Similar challenges have also been noted in Japan. Research on the GIGA School Initiative indicates that, despite the nationwide distribution of digital devices, inequalities in actual use remain evident between urban and rural areas (Nae, 2024). Together, these findings resonate with the focus of our study, which underscores the need for systematic and supportive digital practices in rural classrooms (Afzal et al., 2023; Oyanagi, 2024; Twining et al., 2021).

Moreover, our findings highlight that fostering meaningful ICT engagement in rural classrooms requires more than securing physical access to devices; it also depends on the pedagogical and cultural conditions that shape children’s experiences (Butler et al., 2018; Dellagnelo, 2023). For instance, children classified under the “Differentiation” category—those who may not feel fully integrated into their classrooms but nevertheless use digital environments to express their individuality—did not report particularly low levels of well-being. This suggests that even in rural contexts, often regarded as socially homogeneous, ICT may open up new pathways for supporting diverse educational experiences (Acilar & Sæbø, 2023; Selwyn, 2023).

5.3. Limitations and Future Directions

This study has several limitations that warrant consideration in future research. First, as emphasized in the prior literature, the use of panel data—rather than cross-sectional data—would improve the robustness and generalizability of findings (Cole & Maxwell, 2003; Maxwell & Cole, 2007; Maxwell et al., 2011). Future research should prioritize longitudinal or experimental designs to explore causal relationships among cognitive, non-cognitive, and sociodemographic factors more rigorously. Such approaches will be crucial for advancing our understanding of how subjective well-being (SWB) may shape, and be shaped by, preferences for different societal models.

Second, extending this research to diverse cultural and national contexts is essential for assessing the generalizability of the diversity-related categories—namely inclusion, assimilation, exclusion, differentiation, and intermediate. Comparative cross-cultural studies could yield valuable insights into whether these classifications and their psychological correlates are broadly applicable or shaped by sociocultural context. In addition, investigating how processes of globalization intersect with local diversity structures may contribute to a more nuanced understanding of the mechanisms underlying inclusion and identity development across settings.

Third, future research should employ more precise and quantitative methods to capture children’s digital device usage, including its purpose, frequency, duration, and the extent of adherence to family- or school-based guidelines. As smartphones and tablets become increasingly integrated into children’s daily routines, incorporating objective data sources—such as device usage logs or app-based tracking—may enhance the accuracy and granularity of analyses. Despite these limitations, the present study represents an important initial step in exploring how children’s preferences for inclusive environments are associated with digital engagement, inquisitiveness, and generativity. Future investigations should build on these findings to inform educational practices and policies aimed at enhancing subjective well-being (SWB) in increasingly digital and diverse learning contexts.

6. Conclusions

The extent to which diversity and inclusion are fostered within school environments appears to be associated with children’s well-being. Although diversity has increasingly been recognized as essential for promoting creativity and socio-emotional development, relatively few studies have investigated how children experience inclusive or exclusive classroom settings and how such experiences, together with digital device usage, are related to their subjective well-being (SWB).

Drawing on a survey of 2158 elementary and junior high school students in Kochi Prefecture, Japan, this study classified respondents into five diversity-related categories and employed multinomial logistic and median regression models. The results indicate that SWB is positively associated with inclusion and negatively with exclusion, while generativity is positively related to both inclusion and inquisitiveness. Inquisitiveness, in turn, is supported by inclusive and differentiating environments, as well as by responsive adult interactions, although it shows a decline with age. These findings underscore the potential importance of inclusive classroom environments, structured digital device use, and supportive adult relationships in fostering children’s curiosity, care for others, and overall well-being in increasingly digitalized educational contexts.

In addition to these contributions, this study offers several implications for educational practice and policy. Schools could (i) promote inclusion through teacher training in inclusive pedagogies that foster both belongingness and self-expression; (ii) nurture inquisitiveness and generativity by creating psychologically safe classroom environments supported by positive and responsive adult interactions that encourage children to ask questions and share new ideas; (iii) address the rural–urban digital divide not only by ensuring equitable access to devices but also by providing guidance and learning environments that support the use of digital tools tailored to children’s individual characteristics; and (iv) promote rule-based ICT use and strengthen collaboration and communication between families and schools. Such comprehensive and context-sensitive efforts are likely to play an important role in maximizing the potential of ICT to support children’s well-being.