1.1. Background

Sedentary behavior has become pervasive in modern society, posing a significant and independent risk factor for physical and mental health [1]. This behavior is defined as any waking behavior characterized by an energy expenditure of 1.5 metabolic equivalents (METs) or less while sitting, reclining, or lying down. A growing body of evidence recognizes sedentary behavior as an independent health risk factor, distinct from the mere lack of physical activity [2]. Recent global guidelines and meta-analytic evidence have further reinforced sedentary behavior as a standalone behavioral risk factor, irrespective of engagement in moderate-to-vigorous physical activity [3,4].

Previous studies indicate that U.S. adults spend an average of 9.5 hr per day on sedentary behavior, with approximately 47% of this time occurring during leisure and work activities. A substantial proportion of leisure time (82%) is spent on screen-based activities, including watching television, using the Internet, or working on computers [5]. Similarly, in Korea, the 2021 Korea National Health and Nutrition Examination Survey showed high levels of sedentary behavior among adults, with an average of 8.9 sedentary hours per day and 63.7% sitting for at least eight hours daily. Sedentary behavior was more prevalent among adults aged 65 years and older than among those aged 19–64 years, highlighting sedentary lifestyle as a significant public health concern in Korea [6]. These trends are closely associated with societal and environmental shifts, including rapid industrialization, the proliferation of digital devices, and changes in transportation and workplace environments [7,8]. These figures should be seen not only as statistics but also as critical public health warning signals. Importantly, they indicate that sedentary behavior is structurally embedded in daily life rather than reflecting a temporary lifestyle choice, underscoring the need for sustainable behavioral regulation strategies.

Research has consistently demonstrated that increased sedentary behavior is a major risk factor for the development of various chronic diseases. Prolonged and uninterrupted sedentary time triggers a cascade of physiological changes, including insulin resistance, vascular dysfunction, altered carbohydrate metabolism, and chronic low-grade inflammation [2]. It is also associated with a shift from oxidative to glycolytic muscle fibers, reduced cardiopulmonary function, and decreased muscle and bone masses. Furthermore, it contributes to increased body and visceral fat, elevated blood lipid levels, and heightened inflammatory response [3]. Among middle-aged and older adults, sedentary behavior has been associated with cardiovascular diseases, type 2 diabetes, obesity, and certain cancers [7–9]. Emerging longitudinal and meta-analytic evidence further suggests that excessive sedentary behavior independently predicts all-cause mortality, even after controlling physical activity levels [10,11]. Beyond physical health, individuals with higher levels of sedentary behavior often experience social isolation, limitations in daily activities, and weakened physical function, which can lead to mental health problems such as depression, anxiety, and cognitive decline [3,12,13]. A systematic review found that mortality risk increases when sedentary behavior exceeds 7.5–9 hours per day, with a particularly pronounced rise beyond 9.5 hours [11]. Ultimately, prolonged sedentary behavior poses a substantial public health threat by simultaneously undermining physical, psychological, and social well-being and reducing overall quality of life [3,12,13].

Valid assessment instruments are crucial for accurately assessing and managing the risks of sedentary behavior. However, most existing self-report surveys rely on a single question about the “total time spent sitting,” a format that is highly susceptible to underreporting and inaccuracy [13]. Although objective instruments such as accelerometers provide high accuracy, they are often costly and inconvenient for large-scale use [14]. More critically, both subjective and objective measures focus primarily on quantifying sedentary time and fail to capture an individual’s ability to consciously regulate, interrupt, or prevent sedentary behavior—capacities essential for effective management. Furthermore, these instruments frequently lack cross-cultural validity, a significant concern given that behaviors such as Korea’s traditional floor-sitting culture are not typically accounted for in tools developed abroad [15,16]. In the Korean context, where prolonged sitting is especially concentrated among office workers and other sedentary occupational groups and is associated with elevated metabolic risk, a culturally sensitive self-report scale that assesses not only sedentary duration but also the ability to interrupt and regulate sitting may provide a practical basis for identifying high-risk groups and developing tailored behavioral interventions for everyday and workplace settings [17,18]. This conceptual and practical gap highlights the need for a theoretically grounded instrument that assesses sedentary behavior regulation rather than sedentary duration alone.

1.2. Aims of the current study

This study aimed to explore the conceptual meaning and behavioral characteristics of sedentary behavior regulation and develop and validate a multidimensional scale that captures individual differences in regulatory capacity. By focusing on self-regulatory processes rather than on sedentary duration alone, the proposed instrument seeks to address a critical limitation in existing measurement approaches. Ultimately, the Sedentary Behavior Regulation Scale (SBRS) is intended to provide a precise assessment tool to inform the design, tailoring, and evaluation of long-term sedentary behavior prevention and intervention programs across clinical and community settings.

1.3. Conceptual framework

Grounded in the principles of Self-Determination Theory (SDT), this study was designed to develop a new Sedentary Behavior Regulation Scale (SBRS). SDT is a prominent motivation theory that emphasizes the importance of internal psychological resources in personality development and behavioral regulation [19,20]. This theory posits that when basic psychological needs (competence, autonomy, and relatedness) are satisfied, intrinsic motivation is enhanced, thereby facilitating long-term behavioral change and positively influencing outcomes [14,19,21]. Consistent with this framework, sedentary behavior regulation is conceptualized as an autonomously motivated process involving self-awareness of sedentary patterns, active behavioral practices to interrupt or reduce sedentary time, and preventive environmental structuring to minimize prolonged sitting. Accordingly, SBRS is theoretically defined as an individual’s conscious and active process of managing sedentary behavior in daily life. Operationally, this construct is represented by the total score obtained from the developed scale, with higher scores indicating a greater capacity for effective sedentary behavior regulation.

2.1. Research design

This study employed a methodological research design to develop and validate a self-report instrument for assessing sedentary behavior regulation in adults.

2.2. Research participants and sampling

The participants in this study were adults aged 20–59 years residing in South Korea. A proportional stratified sampling method was used to recruit a sample reflective of the national population, based on data from the Korean National Statistical Office. Stratification variables included region, sex, and age. This strategy was adopted to enhance representativeness and reduce sampling bias, consistent with recommendations for scale development studies targeting general adult populations [22]. Within each stratum, eligible participants were randomly selected from the EMBRAIN online research panel using a computerized randomization procedure to ensure proportional allocation while maintaining random selection.

Participants were recruited via EMBRAIN, a representative online research panel in South Korea consisting of a diverse pool of voluntarily registered adults. To ensure a nationally representative sample, the panel is continuously expanded and demographically balanced. To enhance data integrity, EMBRAIN implements stringent quality control measures, including identity verification, duplicate account detection, and the use of attention-check items. Furthermore, we monitored response times and excluded inconsistent or patterned responses to mitigate careless responding. Only participants who met the eligibility criteria and passed all quality control protocols were included in the final analysis.

The inclusion criteria were adults aged 20–59 years who resided in South Korea, were able to read and understand Korean, engaged in sedentary activities as part of their daily work or leisure, and voluntarily consented to participate after receiving an explanation of the study’s purpose. Exclusion criteria included self-reported cognitive impairment, intellectual disability, severe psychiatric or neurological conditions that could interfere with comprehension or self-reporting, and inability to complete an online questionnaire independently. These criteria were specified to clearly define the target population for which the Sedentary Behavior Regulation Scale was intended to be valid and reliable, consistent with COSMIN recommendations regarding population validity [23].

The required sample size for the validation process was determined based on established methodological guidelines for factor analysis. For exploratory factor analysis (EFA), a minimum ratio of 10 participants per item was applied, following the recommendations of DeVellis and Thorpe [22]. This resulted in a minimum sample size of 320 participants with 32 initial items. For confirmatory factor analysis (CFA), a sample size between 200 and 400 was considered appropriate based on the structural equation modeling guidelines [24]. Accordingly, a total target sample size of 600 participants was established to allow for independent samples for EFA and CFA.

After data collection, the full sample was randomly divided into two independent subsamples using simple random allocation, with one subsample assigned to the EFA and the other to the CFA. To assess the comparability of the two subsamples, equivalence in key demographic characteristics relevant to the factor analysis, including age, sex, and region, was examined. No statistically significant differences were observed between the EFA and CFA subsamples, supporting their structural equivalence. Although more advanced procedures such as the SOLOMON method have been proposed to optimize subsample equivalence [25], the combination of proportional stratified sampling, random allocation, and empirical verification of equivalence has been deemed methodologically adequate and consistent with current best practices in scale development research.

2.3. Research procedure

This study followed an eight-step methodological procedure for instrument development and validation based on the framework proposed by DeVellis and Thorpe [22].

2.4. Criterion-related validity

To assess criterion-related validity, a survey questionnaire consisting of three sections was administered: 10 items on general characteristics, 32 items from the newly developed SBRS, and 16 items from the Korean version of the GPAQ [29]. The GPAQ, developed by the World Health Organization (WHO) in 2002, is a widely used tool for assessing physical activity and sedentary behavior [28]. The K-GPAQ has demonstrated acceptable test–retest reliability and criterion-related validity among Korean adults, with prior validation studies reporting moderate to good reliability across physical activity domains (Spearman’s r = .27–.70; κ = .30–.67) and significant, albeit weak, correlations with accelerometer-measured physical activity and sedentary behavior (r = .18–.34) [29]. The GPAQ assesses the frequency (days per week) and duration (minutes per day) of vigorous physical activity, moderate-intensity physical activity, and travel over the past seven days, as well as average daily sedentary time. Physical activity levels were calculated by converting activities to MET-minutes per week in accordance with the official GPAQ analysis guide [28]. Vigorous activity was assigned 8 METs, whereas moderate activity and mobility were assigned 4 METs, with higher scores indicating a higher level of physical activity. In contrast, sedentary behavior was measured as the average daily sitting time, with a higher score indicating a higher level of sedentary behavior.

2.5. Ethical considerations

This study was conducted in accordance with ethical guidelines and received approval from the Institutional Review Board (IRB No: SYU 2025-03-030-001) in March 2025 prior to data collection. Research was conducted through a specialized online survey agency (EMBRAIN) that sent emails to adult panelists (aged 20–59 years) explaining the study’s purpose, procedures, and participation methods. All participants provided informed consent before participating in the survey. The consent statement clearly indicated that participation was voluntary and that participants could withdraw at any time without penalty. Only those who expressed interest in participating were required to fully understand the explanation and then electronically consent to participate in the study by checking the consent box online. The system was designed to ensure that the survey was completed only after consent was obtained. This process served as a substitute for written consent, and all participants freely decided whether to participate. To ensure confidentiality, all collected data were processed anonymously, and personal information was not provided. All participants who completed the survey received a gift card as a reward.

2.6. Data analysis

Data analysis was conducted using SPSS 26.0 and AMOS 23.0 software (IBM Corp., Armonk, NY, USA). Prior to analysis, incomplete responses were removed, and only complete responses were included in EFA, CFA, and reliability analysis. Descriptive statistics were calculated for participants’ general characteristics. Item analysis involved checking the means, standard deviations, skewness, and kurtosis values of the initial items. To assess the assumption of normality required for maximum likelihood estimation, univariate normality was evaluated by inspecting skewness and kurtosis values for all items, which fell within acceptable ranges. Correlation coefficients between the total score and each item were calculated, and items with correlations below .30 were examined.

Construct validity was evaluated using a two-step factor analytic approach. Prior to factor analyses, chi-square tests were conducted to examine the equivalence between the EFA (n = 320) and CFA (n = 280) groups, confirming the homogeneity of the two subsamples. EFA was conducted using Pearson correlation matrices, as the items were measured on a 5-point Likert scale and demonstrated acceptable normality based on skewness and kurtosis values. A maximum likelihood extraction method with Promax rotation was applied. Sampling adequacy was assessed using the Kaiser–Meyer–Olkin (KMO) measure and Bartlett’s test of sphericity prior to factor extraction. Factor retention decisions were informed by multiple criteria, including eigenvalues greater than 1.0 and inspection of the scree plot.

CFA was subsequently performed using maximum likelihood estimation to assess the fit of the structural model. Model fit was evaluated using absolute (χ²/df, RMSEA, and SRMR) and incremental fit indices (CFI and TLI). Convergent validity was confirmed by examining correlations, construct reliability (CR), and average variance extracted (AVE) among the sub-factors. Discriminant validity was evaluated using the Fornell–Larcker criterion [36] and the heterotrait–monotrait (HTMT) ratio. HTMT values were computed based on item-level correlations, with values below the .85 threshold considered indicative of adequate discriminant validity [37].

Finally, criterion-related validity was assessed by calculating correlation coefficients between the SBRS and the Korean version of the GPAQ. Reliability was evaluated using Cronbach’s α and McDonald’s omega (ω) for internal consistency, as well as the intraclass correlation coefficient (ICC) for test–retest reliability.

3.1. Item generation

A review of previous studies on sedentary behavior revealed that existing instruments are primarily limited to specific populations or rely on recording methods to assess sedentary time [28–34]. A key limitation of these instruments is their inability to capture the self-regulatory mechanisms underlying sedentary behavior. To address this gap, consultation was undertaken with a subject matter expert in psychological instrument development. Based on the comprehensive literature review, an initial pool of 55 items was generated. Following a series of rigorous content reviews, 4 items with overlapping content or low relevance to sedentary behavior regulation were removed to prevent participant confusion, resulting in a final set of 51 items.

3.2. Content validity

Following initial item generation, the 51 items underwent rigorous evaluation by a panel of seven experts to establish content validity (S1 Table). Items were excluded if they demonstrated a low item-level content validity index (I-CVI ≤ .80), conceptual redundancy, or insufficient relevance to the study’s objectives (S2 Table). At this stage, the scale-level content validity index calculated using the averaging method (S-CVI/Ave) for the initial item pool was .87, indicating good overall content validity.

After item refinement and reduction to 32 items, the S-CVI/Ave increased to .96. A second round of content validity testing was conducted with a panel of five experts to reexamine the remaining items. All items demonstrated I-CVI values of .80 or higher, leading to their final retention. The modified kappa values ranged from .85 to .99, indicating excellent agreement beyond chance for all items.

3.3. Pilot test

The pilot study evaluated 32 items in 10 adults aged 20–50 years (mean age = 45.2 years). The survey took 5–6 min to complete, and most respondents rated the items as easy to understand.

3.4. General characteristics of the survey participants

There were 600 survey participants, 312 men (52.03%) and 288 women (48.0%). The largest age group comprised participants in their 50s (n = 182; 30.3%). To establish construct validity, the total sample was randomly divided into two independent subsamples for EFA (n = 320) and CFA (n = 280). To verify subsample comparability, main demographic characteristics, including age and sex, were examined between the EFA and CFA groups. No statistically significant differences were found, indicating that the two subsamples were demographically equivalent and suitable for independent factor analytic procedures (Table 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. General Characteristics of the Participants (N = 600).

https://doi.org/10.1371/journal.pone.0346963.t001

3.5. Construct validity

3.5.2. EFA.

An EFA was performed on the remaining 25 items to examine the underlying factor structure. The suitability of the data for factor analysis was confirmed using a KMO value of .93 and a statistically significant Bartlett’s test of sphericity (χ² = 3831.94, df = 300, p < .001). Both eigenvalues greater than 1.0 and scree plot inspection supported a two-factor solution. During the analysis, two items (x8 and x19) were removed because their communality values were below .30. An additional four items (x17, x23, x24, and x25) were removed because their factor loadings were below .40, and three items (x13, x21, and x30) were removed because of cross-loadings, defined as factor loadings of .30 or higher on two factors in the pattern matrix. The content and psychometric properties of the excluded items, including item–total correlation coefficients, communalities, and factor loadings, are presented in S3 Table.

Ultimately, 16 items were selected and loaded onto two distinct factors. The final model’s KMO value was .91, and Bartlett’s test of sphericity remained statistically significant (χ² = 2368.24, df = 120, p < .001). All retained items exhibited factor loadings of .44 or higher, with communality values of .32 or greater. The two factors jointly accounted for 54.31% of the total variance (Table 2). The overall reliability of the final 16-item instrument was excellent (Cronbach’s α = .90). The two sub-factors were named based on the content of their loaded items: Factor 1, “sedentary behavior management and environmental support”; and Factor 2, “active movement in sedentary contexts.”

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Result of Exploratory Factor Analysis (N = 320).

https://doi.org/10.1371/journal.pone.0346963.t002

3.5.3. CFA.

CFA was conducted to examine the structural relationships between the two factors derived from EFA. Model fit was evaluated using established cut-off criteria: χ²/df values below 3.0, CFI and TLI values ≥ .90, SRMR ≤ .08, and RMSEA ≤ .08 were considered indicative of acceptable model fit, while lower χ²/df and RMSEA values and higher CFI and TLI values indicated good fit. The initial CFA model demonstrated suboptimal fit, with a χ²/df value of 3.43, and other indices that did not meet recommended standards (CFI = .88, TLI = .86, SRMR = .084, RMSEA = .093, 90% CI [.083, .104]).

To avoid purely data-driven modifications, an iterative model refinement process was conducted based on modification indices (MI) and theoretical considerations, including conceptual overlap and redundancy among items with highly similar behavioral intent. This process led to the deletion of four items (x28, x31, x15, and x18), resulting in a final model with 12 items (Fig 1). The final revised model demonstrated good fit to the data, with all indices meeting recommended standards: χ²/df = 2.51, CFI = .93, TLI = .92, SRMR = .056, and RMSEA = .074 (90% CI [.058, .089]).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Confirmatory factor analysis of the measurement model.

https://doi.org/10.1371/journal.pone.0346963.g001

All standardized factor loadings (β) were statistically significant, and their 95% bootstrap confidence intervals did not include zero (Table 3). Reliability was confirmed, as both factors showed CR values above .70. However, the AVE was below the recommended .50 benchmark for both factors, indicating insufficient convergent validity. This suggests that further instrumental refinement is required to enhance this aspect. The Fornell–Larcker criterion for discriminant validity was not satisfied, as the square root of AVE for Factor 1 (.67) was lower than the inter-factor correlation (r = .68, p < .001), and the value for Factor 2 (.68) was only equal to the correlation. To further evaluate discriminant validity, the HTMT value was calculated. The HTMT value between Factors 1 and 2 was .69, well below the conservative threshold of .85. These results indicate that although the two factors are moderately correlated, they represent empirically distinct constructs.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Result of Confirmatory Factor Analysis (N = 280).

https://doi.org/10.1371/journal.pone.0346963.t003

3.6. Criterion validity

Criterion validity was verified using the GPAQ. The total SBRS score showed a significant positive correlation with physical activity (r = .19, p < .001) and a significant negative correlation with sedentary time (r = −.33, p < .001) (Table 4). For individual sub-factors, physical activity was positively correlated with Factors 1 (r = .14, p = .011) and 2 (r = .18, p = .001). Conversely, sedentary time showed a significant negative correlation with factors 1 (r = −.28, p < .001) and 2 (r = −.30, p < .001), further supporting the validity of the instrument.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Correlation between SBRS and GPAQ (N = 320).

https://doi.org/10.1371/journal.pone.0346963.t004

3.7. Reliability

Cronbach’s α for the entire instrument was .87, indicating high internal consistency. Factor 1 had a value of .83, and Factor 2 had a value of .82, both of which exceeded the recommended benchmark of .80. The McDonald’s omega (ω) coefficient for the entire instrument was .87, while Factor 1 and Factor 2 showed ω values of .81 and .82, respectively, indicating satisfactory internal consistency under a congeneric measurement model. Furthermore, the ICC for test-retest reliability was .85 for the total instrument, with Factor 1 at .81 and Factor 2 at .82. These high values confirmed the stability of the instrument over time.

3.8. Optimization of the scale

This study finalized the SBRS, a 12-item instrument that assesses an individual’s ability to manage sedentary behaviors. The items were structured on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree), with higher scores indicating more effective sedentary behavior management. The total SBRS score was calculated by summing all 12 items, yielding a possible range of 12–60. The final item composition by sub-factor was as follows: Factor 1 (sedentary behavior management and environmental support) comprised items 1–6, and Factor 2 (active movement in sedentary contexts) comprised items 7–12 (S4 Table). At the current stage of scale development, no empirical cut-off scores have been proposed; thus, SBRS scores are intended to be interpreted relative to sample distributions or for comparative purposes across individuals, groups, or time points.

4.1. Limitations

This study has several limitations. First, data collection relied on an online panel, which may have introduced a sampling bias toward adults familiar with digital environments, thereby limiting the generalizability of the findings to a broader population. Second, the use of self-report questionnaires may have introduced recall or social desirability biases. Future studies should incorporate objective instruments, such as accelerometers or wearable sensors, to provide a more precise assessment of sedentary behavior regulation and cross-validate self-report data. Third, the sample was restricted to Korean adults aged 20–59 years, which limits the generalizability of the findings to older adults, particularly those aged 60 years and older, who may exhibit distinct sedentary behavior patterns, health conditions, and regulatory processes. In addition, cultural specificity should be considered, as sedentary behavior and its regulation may vary across sociocultural contexts. Accordingly, future studies should validate the SBRS across diverse age groups and populations. Fourth, the AVE values for both factors did not reach the recommended threshold of .50, indicating that a substantial proportion of the variance in the observed items may be attributable to measurement error rather than latent constructs. Although the HTMT result suggested that the two factors may be empirically distinguishable, insufficient AVE may limit construct precision and reduce statistical power in subsequent analyses. In particular, associations between SBRS scores and related outcomes may be underestimated, and caution is warranted when interpreting small effect sizes or non-significant findings. Moreover, the failure to satisfy the Fornell–Larcker criterion, together with AVE values below .50, raises the possibility that the two factors may not yet be sufficiently distinct at the construct level, and this issue should be prioritized in future refinement of the scale. Taken together, these limitations indicate that the present study represents an initial, developmental validation aimed at establishing the initial psychometric properties of the SBRS rather than a definitive validation. These findings highlight the need for further refinement of item content and factor structure to enhance variance extraction and measurement precision. Furthermore, items removed because of cross-loadings should be re-evaluated in future studies with closer attention to their theoretical relevance and substantive content, rather than being excluded solely on statistical grounds.

4.2. Implications and recommendations

Future research should address these limitations. First, the generalizability of the SBRS should be confirmed using larger and more diverse samples, including participants from diverse regions and occupational groups. Second, longitudinal research tracking the validity of the SBRS over time is required to clarify the causal relationship between sedentary behavior regulation levels and health indicators (e.g., obesity, metabolic syndrome, and mental health). Third, additional data should be collected to refine and revalidate the current item set, with particular emphasis on revising or replacing items with lower factor loadings to improve convergent validity and AVE. Such refinement is expected to enhance the sensitivity and statistical power of the scale in future research applications.

From an applied perspective, the current version of the SBRS may be most appropriately used as a screening or exploratory tool to capture general tendencies in sedentary behavior regulation rather than as a high-stakes diagnostic instrument. Therefore, future intervention studies should interpret SBRS scores as indicative of the relative regulatory capacity and use them to inform the development of tailored programs for obesity, metabolic syndrome, and mental health. Finally, experimental and population-specific studies—focusing on groups such as office workers, remote workers, and older adults—are needed to further validate the scale and examine the effectiveness of interventions designed to modify sedentary behavior regulation in diverse contexts.