Redevelopment and examination of the psychometric properties of school physical activity environment in Chinese

Abstract

Background

This study adapted the Questionnaire Assessing School Physical Activity Environment (Q-SPACE) into Chinese and evaluated its psychometric properties.

Methods

A total of 253 primary school and 267 secondary school students completed the modified Q-SPACE-Chi twice, with a 2-week interval, between November 2022 and March 2023. A confirmatory factor analysis and the assessment of measurement invariance supported the efficiency of the three-factor Q-SPACE-Chi (i.e., physical environment, social environment, and school policy) across various groups.

Results

The results confirmed the factorial validity of the three-factor modified A-SPACE-Chi, with all items demonstrating significant loadings in the physical environment, special environment, and social policy dimensions. The predictive validity results revealed significant associations between student’s perceived school physical activity environment and the physical activity level. Moreover, the test-retest reliability and internal consistency of the adapted questionnaire were moderately favorable.

Conclusion

This modified Q-SPACE-Chi, with element of school policy therefore presents an additional valuable tool for researchers or practitioners aiming to evaluate how Chinese students perceive the school physical activity environments within their schools while the measurement properties of Q-SPACE-Chi was invariant among primary and secondary students.

Background

Physical activity (PA) provides a range of health benefits in adolescents, such as a reduced risk of chronic diseases, increased self-esteem, improved psychological well-being, as well as the cultivation of resilience, particularly among Chinese adolescents [1, 2]. Furthermore, previous research indicated a positive correlation between PA and academic performance [3]. Despite the multitude of health benefits associated with PA, the declining interest in PA among school students aged 6–18 is concerning [4].

Given the positive impacts of PA on students’ overall development, the decline in PA levels among school students has become a global concern. Studies have indicated a decline in PA from the young to young adult stages, highlighting the need for further comprehensive investigations into the reasons underlying this decline and the formulation of strategies aimed at fostering environments that promote sustained PA [5]. Aira et al. [5] reported a decline in both overall PA and sports participation among Finnish adolescents, specifically between the ages of 15 and 19 years. Similarly, research has revealed the deficiency in PA among adolescents aged 12–16 and children aged 7–11 in Hong Kong in general [4]. The declining trend in PA is of growing concern, with potential long-lasting consequences. Longitudinal studies focused on adolescence and adulthood inactivity have emphasized that adolescents who are actively involved in sports, such as being members of sports clubs, tend to exhibit greater psychological readiness to maintain PA in their adulthood [6]. Conversely, physical inactivity during the formative years, spanning 11–25 years, can contribute to the emergence of cardiovascular risk factors. Notably, approximately 50–75% of obese children and adolescents tend to be obese as adults, consequently facing elevated risks of various diseases including cardiovascular disease, type-2 diabetes, and colon cancer [7]. Therefore, substantial research efforts have been devoted to understanding the engagement of students in PA.

To understand the problem of physical inactivity in children and adolescents, assessing the current approach to promoting PA among students is paramount. Unlike other age demographics such as adults, school students predominantly spend their time at school [8]. Consequently, school and physical education (PE) play pivotal roles in fostering increased PA participation among school students. Notably, in Hong Kong, a typical school day lasts approximately 7–8 h for both primary and secondary students, constituting approximately one-third of a day. This substantial duration indicates the potential influence of the school environment on cultivating a healthy lifestyle [8]. For primary school students in Hong Kong, the PA levels observed during school hours account for more than half of their total PA [9]. Therefore, schools emerge as a critical platform for increasing PA engagement among students.

According to the Social Ecological Model, health behaviors were affected by factors operating across various levels [10]. These include intrapersonal (e.g., biological and psychological factors), interpersonal (e.g., social support and peer pressure), organizational (e.g., school and workplace environments), community (e.g., neighborhood and recreational environment), and public policy (e.g., government policies and healthcare systems) factors. Although participating in PA may appear to be an individual choice, factors at this level only accounted for 40% of the predicted variance in this behavior [11]. Therefore, the focus should shift towards the role of both social and physical environments in PA engagement among primary and secondary school students.

The physical environment, a subsidiary of the encompassing school environment, is considered a crucial factor in research on school environments and student PA. In a mixed-studies systematic review, Morton et al. [12] included 19 quantitative and 16 qualitative studies to investigate the association between physical environment and student PA among adolescents aged 11–18. Quantitative findings revealed that the factors related to the physical environment had a positive impact on PA. These factors included the specific settings for activities, including the type and location as well as the campus area available per student. However, the same positive association was not found for the actual accessibility to features. The qualitative outcomes revealed the presence of sufficient space and the lack of equipment as themes that emerged frequently. By contrast, Foubister et al. [13] examined the association between school policy, the social and physical environment, and changes in adolescent PA and revealed no significant correlation between the physical environment and student PA.

Apart from the physical environment, the social environment encompassing the support of teachers, friends, and parents has emerged as a critical determinant of students’ active participation in PA. Trigueros et al. [14] analyzed the impact of social support perceived by adolescents on their motivation and subsequent engagement in PA. Their study revealed that social support from teachers, friends, and parents exerted a significant influence on the motivation of students towards PA, with the exception of autonomy support from teachers. Another study on the PA of school students highlighted the significant association of teachers’ supportive teaching, students’ perceived competence, and a sense of connection with the community of friends’ participating in the activities with PA among children aged 10–19 years [15]. Overall, these findings emphasized the pivotal role of the social dimension within the school environment in fostering students’ motivation and subsequent engagement in PA.

In addition to the social aspects, the implementation of school policies was identified as another factor related to the school environment that promoted students’ engagement in PA. Woods et al. [16] conducted a systematic review of 25 studies related to school environment policies regarding PA and revealed that implementing effective school policies to counteract physical inactivity and associated health problems among school children is essential. The review outlined nine policy areas that could be employed to enhance PA among school children, with PE, sports, classroom-based PA, and school breaks emerging as particularly effective policy domains. Appropriate measurement of evidence of the effectiveness of implementing the PA policies is thus crucial [16]. A more recent systematic review synthesized findings from 52 studies on the associations between school-based PA policies, school PA practices, and PA behaviors [17]. However, the review reported inconclusive associations due to inadequate information about the quality of measures used [17].

Considering that the complexity and interplay of the various factors within the school environment (physical, social, and policy), the study findings regarding students’ PA may be influenced by the combination of these components. In 2020, Webster et al. [18] proposed a comprehensive framework for school-based PA that encompasses the physical, policy, social aspects of the school environment in elementary schools, middle schools and high schools. These researchers emphasized the need for adopting a collective approach to understanding school environments for enhancing students’ PA. Yet, there was no valid and reliable scale measuring the physical, policy, social aspects of the school environment that impact students’ PA.

Regarding the school PA environment, researchers have employed both observational measures such as Physical Activity Resource Assessment (PARA) and subjective measures [19] One subjective measure is the Questionnaire Assessing School Physical Activity Environment (Q-SPACE), which was first developed by Robertson-Wilson, Lévesque and Holden [20]. The development of this scale involved a literature review and focus groups for generating a preliminary list of items. A pilot test of this list was then conducted with 136 grade 6–8 students. Then, a finalized version (20 items) consisting of two factors [i.e., physical environment (12 items) and social environment (8 items)] was tested on 244 students. The results revealed that the finalized Q-SPACE had acceptable reliability for capturing students’ perceptions of the school PA environment; however, its validity required further investigation, particularly using confirmatory factor analysis (CFA). In 2011, Martin et al. [21] examined its validity through CFA. A shortened version of the Q-SPACE (16 Items) was supported with model fit indices with two subscales (physical and social environments) of eight items each. Its reliability was confirmed by the satisfactory internal consistency (α =. 74–0.80). Four items were ultimately deleted due to their low factor loadings. Examples of these items are ‘The outdoor areas at my school are large enough for students to be physically active’ and ‘The outdoor areas (e.g., playground, field) at my school are in good condition.’ In 2019, the Q-SPACE was translated into Portuguese, and its factorial validity and reliability (internal consistency values 0.6–0.89) for assessing school PA environment were supported among 504 Portuguese students aged 10–15 years [22]. All of the above three studies have highlighted their limitations in the diversity of data and instrument’s usage in different languages and cultural settings. They also recommended future studies to examine the relationship between Q-SPACE and PA participation.

In such a way, the psychometric properties of the Chinese version of the Q-SPACE questionnaire remain relatively unexplored, and the cultural differences and language barriers within the specific context of Hong Kong add further complexity to its application in assessing school PA environment. Given these challenges and the unique characteristics of the Hong Kong setting, developing a modified school PA environment scale for the Hong Kong context is necessary. Moreover, based on research findings that indicate the significant impact of school policy as an aspect of the school environment on students’ PA, this study added a new factor “school policy” into the Chinese version of the Q-SPACE questionnaire.

The purpose of this study was to adapt and redevelop the modified Questionnaire Assessing School Physical Activity Environment in Chinese (Q-SPACE-Chi) for use among primary and secondary school students in Hong Kong. The primary objectives of this study encompassed three key aspects: (a) factorial validity, (b) predictive validity, and (c) the test–retest reliability of the modified Q-SPACE-Chi for primary and secondary school students in Hong Kong. This study hypothesized that the measurement properties of Q-SPACE-Chi was invariant among primary and secondary students. If the Q-SPACE-Chi is found to be valid and reliable through these assessments, it can be used for evaluating the effectiveness of interventions in terms of students’ overall perceptions of the school PA environments or perceptions related to specific school environment dimensions. Moreover, its applicability could be extended to future studies within the Hong Kong and Chinese contexts. The results of this study can be used as a reference for developing school interventions targeted at promoting PA among school students by enhancing the school PA environments.

Methods

Participants

A total of 253 primary school students from three primary schools aged 6–12 and 267 secondary school students from three secondary schools aged 12–18 participated in this study. The selection of these schools was driven by convenience. Next, to increase the representativeness of this study’s participants, they were stratified recruited from (1) the broader geographical regions of Hong Kong, which consists of three main territories (i.e., Hong Kong island, Kowloon, New Territories); (2) genders (271 male and 249 female students); and level of study (primary schools and secondary schools). The sample size was deemed sufficient in relation to the parameter ratio requirement for CFA [23]. After data management was performed to examine outliers, normality and multivariate collinearity, 520 cases remained eligible for further analysis. The participants’ mean age was 12.88 years (SD = 2.24). All the participants attended government-aided schools where the medium of instruction was Chinese. Regarding family background, the majority of the participants’ parents had attained a secondary education level (32.5%), whereas 24.42% had achieved a tertiary or higher education level. Family size distribution revealed that nearly one-third (29%) of participants were the only child at home, whereas 48.1% had 1 sibling and 22.7% had 2 or more siblings. Geographically, only 19.6% of participants resided in Hong Kong Island or Kowloon, whereas the majority (80.4%) lived in the New Territories. For a comprehensive overview of the demographic characteristics of participants, please refer to Table 1.

Table 1.

Characteristics of participants in main study (N = 520)

	M (SD)	Frequency	%	M (SD)	Frequency	%
	Primary students (N = 253)			Secondary students (N = 267)
Age (in years)	10.97 (0.68)			14.7 (1.61)
Gender
Male		124	49		147	55.1
Female		129	51		120	44.9
Education level
*Junior		0	0	103	38.5
*Senior		253	100	164	61.5
Parent’s Education Level
Primary Education		31	12.3	22	8.2
Secondary Education		150	59.3	19	71.2
Tertiary Education		72	28.5	55	20.6
Living area
Hong Kong Island		5	2	6	2.2
Kowloon		11	4.3	80	30
New Territories		237	93.7	181	67.8

Note. Junior = Form 4- Form 6 in primary school or Form 1-Form 3 in secondary school; Senior = Form 4-Form 6 in secondary school

Measures

School environment PA scale

The Q-SPACE-Chi originates from the study of Martin et al. [21] and comprises two primary factors: an 8-item section on the school’s physical PA environment and an 8-item section on the school’s social PA environment. An example question from the questionnaire is ‘Teachers supervise students being physically active at recess or lunch breaks at my school’. Responses to these questions ranged from 1 (strongly agree) to 5 (strongly disagree), with lower scores indicating a more favorable PA environment. The Q-SPACE was firstly developed and found to have adequate internal consistency (α = 0.81–0.86) and test–retest reliability (r =.72–0.78) [20]. Moreover, the validity of the questionnaire was supported by its significant association with school PA levels [20]. Martin et al. [21] further examined its validity using confirmatory factor analysis and the data supported the re-specified 16-item Q-SPACE, with 2 subscales of 8 items each. To adapt the Q-SPACE to the Hong Kong context, the questionnaire underwent the following modifications:

The first step was a rigorous translation procedure (forward translation and back translation), which was performed in accordance with the guidelines proposed by Hambleton and Kanjee [24]. During the forward translation phase, the original instrument [16-item Q-SPACE in English] [21] was meticulously translated into Chinese by language experts associated with the Centre for Translation at a university in Hong Kong. Ambiguities within the items and instructions were identified and subsequently modified. Particularly, one item was changed from.

“The outdoor areas (e.g., playground, field) at my school are in good condition” to “The outdoor areas (e.g., playground) at my school are in good condition” as schools in Hong Kong seldom had their own field in school site. The translator process involved collaborative discussions between the translators and the author, aimed at addressing any specific translation problems that emerged. This iterative process continued until a consensus was achieved on the final version of the translated instrument. Subsequently, in the back-translation stage, the instrument translated in Chinese was retranslated back into the source language (English) by a different group of translators who were not involved in the forward translation process. These translators were blinded to the contents of the original instruments, thus maintaining the integrity of the process.

The finalized version of the instrument underwent content analysis, which was conducted by a panel of six primary and secondary school students, six primary and secondary school teachers, and two researchers focused on school PA promotion. Following this analysis, the panel recommended the addition of a new factor, ‘school policy’ to the instrument [13]. This recommendation was based on both the result of this content analysis and relevant literature in the field of sport policy, such as studies by Foubister et al. [13] and Woods et al. [16]. Following the guidelines of psychometric principles and instrument development by Swan et al., [25] the questionnaire was then expanded to include seven items pertaining to sport policy. Examples of newly added items include, ‘My school considers sport and physical activity important’, and ‘I am allowed to use the outdoor areas at my school after school hours.’ The questionnaire with new items was subjected to content analysis again by the same panel members including primary and secondary school students, primary and secondary school teachers, and researchers in related discipline. After the identified discrepancies in the instrument versions were resolved by reaching a consensus, the 23-item Chinese version of the instrument was finalized.

Physical activity

Participants’ PA levels were measured using the Physical Activity Questionnaire for Older Children-Chinese (PAQ-C) [26]. The PAQ-C requires participants to recall their participation in different PAs as well as their activity during PE classes, lunch breaks, recess, after school, in the evenings, and on weekends. An example of an item in the PAQ-C is, “In the last 7 days, during your PE classes, how often were you very active (playing extensively, running, jumping, or throwing)?” Each item was scored on a 5-point scale, with higher scores indicating higher PA levels. The original 9-item PAQ-C has been demonstrated to have moderate to high test–retest reliability in children aged 9 to 14 years (r =.75 for male and r =.82 for female students) [25].

Procedures

The study was conducted between November 2022 and March 2023. A cover letter outlining the purpose and objectives of the study was sent to the school principals. This was achieved through convenience sampling. Upon receiving agreement from the school principals to participate in the research, site visits were arranged. During these visits, the questionnaires were delivered to participants. The study’s purpose and details were explained to all of the participants at the beginning of the data collection sessions. After providing informed consent from all participants and parents of underaged participants, he or she completed a questionnaire during the site visit. Upon request, the data collection assistants offered assistance to participants, particularly in reading and comprehending the questions. The completion of the questionnaire required an average of 20 min per participant. Additionally, participants were instructed to complete the same questionnaire after a 2-week interval, a 2-week interval was chosen because it provides a balance between minimizing recall bias and allowing for the natural variability in health status [27]. Prior to the study commencement, permission was granted from the university’s institutional review board (reference no.: 2019-2020-0303) for research with human subjects.

Data analysis

Factorial validity of the modified Q-SPACE in Chinese

Before conducting CFA, exploratory factor analysis (EFA) was performed using principal component analysis. The CFA was performed to assess the factorial validity of the modified three-factor Q-SPACE-Chi and its overall model fit. Specifically, five fit indices were selected for assessing the overall model fit, namely (1) normed chi-square (χ2/df), provides a measure of how well the model fits the data relative to the data from a set of measurement items. As chi square is affected by the sample size, normed chi-square (χ2/df) is chosen; (2) root-mean-square error of approximation (RMSEA), assesses how far a hypothesized model is deviated from a perfect model, however, it may exhibit bias towards with larger sample size and degrees of freedom; (3) standardized root mean residual (SRMR), examines the average standardized residuals, which serves as a direct measure of residuals and offers ease of interpretation and robustness to sample size; (4) non-normed fit index (NNFI) and (5) comparative fit index (CFI), use to compare models. Particularly, CFI is one of the mostly used reported fit index as its ability least affected by sample size and clear interpretation. The normed chi-square value is considered acceptable when the value is between 2.0 and 5.0. SRMR and RMSEA values less than 0.80 are considered to indicate an acceptable fit [28]. For the NNFI and CFI, the values should ideally be between 0.00 and 1.00, with a higher value (i.e., 0.9) indicating a better fit [26]. Next, the reliability and validity of the modified Q-SPACE-Chi were tested based on the standardized factor loadings (SFLs), composite reliability (CR), average variance extracted (AVE), and squared multiple correlation (R2). CR serves as an indicator of reliability and provides further validation, enabling scholars to investigate internal consistency. CR values of 0.6 or higher and AVE values of 0.5 or higher are regarded to indicate reliability [29]. A reliable item should also have a squared multiple correlation of 0.50 or higher [30].

Subsequently, the measurement invariance (MI) was examined to assess whether the factor structure of the Q-SPACE-Chi remained consistent among different populations (primary vs. secondary schools). MI is a key methodological feature of comparative psychosocial science that gain prominence. It is prominent because it examines whether students from primary and secondary school students score similarly in all items and whether the items equally reflect the underlying construct in this study. Multiple-group CFA with a series of nested models with increasing parameter restrictions was employed to examine the MI of the Q-SPACE-Chi. In addition to the aforementioned fit indices, the successive, nested models were examined using the chi-square difference test, CFI, and gamma hat index (GH). Due to the potential influence of nonnormality and sample size on the results of the chi-square difference test, the GH and CFI were suggested by Cheung and Rensvold to serve as comparative fit indices for MI [31]. The GH was unaffected by the sample size. Therefore, a nonsignificant chi-square indifference and the difference of GH and CFI values less than 0.01 would indicate the invariance across two samples.

Test–retest reliability of the modified Q-SPACE in Chinese

To assess test–retest reliability, the intraclass correlation coefficient (ICC) was calculated to examine the repeatability of answers across respondents twice in a 2-week interval. Two week interval was selected as it provided sufficient time period so that participants could not recall their first answer and then allowed for true change of the construct to occur [32]. Approximately 2 weeks is generally considered as appropriate in primary and secondary students [33] ICCs ranging from 0 to 1 with values higher than 0.70 are deemed as acceptable [34].

Predictive validity of the modified Q-SPACE in Chinese

For predictive validity analysis, Pearson correlations between school PA environment and PAQ-C were calculated.

Results

Factorial validity of the modified Q-SPACE-Chi

Results of EFA supported the presence of a 3-factor structure of Q-SPACE-Chi, explaining 23.72%, 23.07%, and 11.93% of the variance, respectively. Table 2 presents the results of the CFA. The initial model (Model 1) exhibited a poor fit to the data (χ² (227) = 1456.83, p <.001; χ²/df = 6.42; CFI = 0.824; NNFI = 0.804; SRMR = 0.066; RMSEA = 0.102 [90% CI = 0.097 to 0.107]). However, a misfit was identified after examination of the modification indices, standardized residual, and SFLs. Five items (SP3, P2, SP1, SP2, and S2) were thus sequentially excluded one by one due to their high standardized residuals (exceeding ± 2.00). After the exclusion of these items, the resulting 18-item three-factor model demonstrated the most parsimonious fit to the data (χ² (132) = 507.58, p <.001; χ²/df = 3.48; CFI = 0.925; NNFI = 0.913; SRMR = 0.040; RMSEA = 0.074 [90% CI = 0.067 to 0.081]). All of the items exhibited significant factor loadings within their respective dimensions. The inter-factor correlations and SFLs, squared multiple correlation, composite reliability, and AVE for the measurement model of the Q-SPACE-Chi are presented in Tables 3 and 4, respectively. The standardized SFLs of the items ranged from 0.51 to 0.80, with an average loading of 0.71. The CR values for each factor ranged from 0.62 to 0.85, and the AVE ranged from 0.33 to 0.50. Overall, the CFA results revealed that the 18 items (after model modification) in the Q-SPACE-Chi collectively measured the school PA environment construct across three factors. This construct exhibited satisfactory factorial validity and moderate internal reliability.

Table 2.

Fit indices for the models (Q-SPACE-Chi)

Model	χ²	df	p	χ²/df	RMSEA (90%CI)	CFI	NNFI	SRMR
Model 1	1456.83	227	0.001	6.42	0.102 (0.097-0.107)	0.824	0.804	0.066
Model 2 (excluded items SP3, P2, SP1, SP2, S2)	507.58	132	0.001	3.48	0.074 (0.067 − 0.081)	0.925	0.913	0.040

Note: RMSEA = Root Mean Square Error of Approximation; CI = Confidence Interval; CFI = Comparative Fit Index; NNFI = Non-Normed Fit Index; SRMR = Standardized Root Mean Residual; SP3 = “I can use the school’s indoor areas after school.”; P2 = “My school has good quality sport and physical activity equipment for students to use”; SP1 = “I can borrow the school’s physical activity equipment after school”; S2 = “My school mates often encourage me to participate in sport and physical activity at school”

Table 3.

Inter-factor correlations of (Q-SPACE-Chi)

Factor	I	II	III	IV
I	/
II	88**	/
III	0.82*	0.93**	/

Note: I = Physical Environment; II = Social Environment; III = School Policy

*p <.05; ** p <.01

Table 4.

Completely standardized factor loadings (SFL), standardized error, squared multiple correlation (R²), composite reliability (CR) - MPAQ-C and average variance extracted (AVE)

	Items	SFL	Theta	R 2	CR/AVE
Physical Environment					0.85/0.50
P1	The indoor areas (e.g., gym) at my school are in good condition	0.72	0.48	0.52
P3	My school has enough equipment for student to use	0.74	0.45	0.55
P4	The gym classes at my school occur often enough during the week	0.51	0.74	0.26
P5	I can find out about community physical activity and sport opportunities at my school	0.73	0.46	0.54
P6	We do a variety of activities in gym class at my school	0.79	0.37	0.63
P7	Student can participate in a variety of sports team at my school	0.77	0.41	0.59
P8	My school offers other physical activities or organized sports for student after school	0.71	0.50	0.50
Social Environment					0.82/0.46
S1	We have good coaches at my school	0.65	0.57	0.42
S3	Teachers usually encourage me to be physically active at my school	0.69	0.52	0.48
S4	Teachers supervise students being physically active at recess or lunch breaks at my school	0.59	0.65	0.35
S5	Teachers or others organize different physical activities and sport events for student outside of gym class at my school	0.74	0.46	0.54
S6	Teachers think physical activity is important for student at my school	0.77	0.41	0.59
S7	The indoor and outdoor areas at my school are supervised	0.63	0.61	0.40
S8	Other student make me feel safe when I am physically active	0.65	0.57	0.43
School Policy					0.62/0.33
SP4	My school consider sport and physical activity important	0.80	0.36	0.64
SP5	My school supports teachers to have training on sports and physical activity	0.75	0.42	0.57
SP6	Student are involved in the school decision making about sports and physical activity	0.77	0.40	0.60
SP7	My school assist me in booking physical activities venue outside school areas	0.72	0.49	0.52

The outcomes of the MI analysis of the Q-SPACE-Chi among primary and secondary school students are presented in Table 5. The model fit was satisfactory for both primary school students (χ² (132) = 267.95, p <.001; CFI = 923) and secondary school students (χ² (132) = 470.19, p <.001; CFI = 0.895), confirming the validity of the three-factor model of QSPACE-Chi in both groups. The baseline model (Model i) exhibited a suitable fit the data for both groups (χ² (264) = 738.14, p <.01; CFI = 0.971), indicating that the factor structure of the construct was the same across the groups. In subsequent models, various constraints were applied: equalizing all factor loadings (Model ii, χ² (279) = 763.37, p <.01; CFI = 0.970), equalizing item error variances (Model iii, χ² (297) = 883.87, p <.01; CFI = 0.964), and equalizing factor variances and covariances (Model iv, χ² (303) = 899.51, p <.01; CFI = 0.964) across the groups. Although significant differences were observed in the chi-square difference tests between these models, the alterations in the CFI and GH between models were less than 0.01, indicating that the factor structure, factor loadings, error variance of each item as well as factor variances and covariances were invariant across the groups. The level of factorial invariance for the CFA model of the Q-SPACE-Chi was satisfactory across primary and secondary school students.

Table 5.

Testing for measurement invariance across lower and higher education groups

Model	χ²	df	Model comparison	Δ χ²	Δdf	CFI	ΔCFI	ΔGH
Primary sch.	267.95	132				0.923
Secondary sch.	470.19	132				0.895
i	738.14	264				0.971	< 0.01
ii	763.37	279	Model 1– Model 2	25.23*	15	0.970	< 0.01	< 0.01
iii	883.87	297	Model 2– Model 3	120.5**	18	0.964	< 0.01	< 0.01
iv	899.51	303	Model 3– Model 4	15.64*	6	0.964	< 0.01	< 0.01

Note: χ² = Maximum Likelihood Ratio Chi-Square; CFI = Comparative Fit Index; GH = Gamma Hat Index; Model i: Configural Invariance (unconstraint); Model ii: Constrained factor loadings; Model iii: Constrained factor loadings and errors variance; Model iv: Constrained factor loadings, error variance and covariances, and factor variances and covariances; Primary sch. = Primary school students: Secondary sch. = Secondary school students

Predictive validity of the modified Q-SPACE-Chi

In terms of predictive validity, in line with our predictions, the mean score of the PAQ-C was positively correlated with that of the Q-SPACE-Chi (r = −.10, p <.05). Particularly, all factors [social environment (r = −.12, p <.01), and school policy (r = −.14, p <.01)] of the Q-SPACE-Chi were significantly associated with PA levels, except physical environment (r = −.04, p >.05). A more favorable perception of overall school environment and specifically, social environment and school policy was associated with a higher level of student PA levels; however, their associations were relatively weak (with correlation coefficients lower than 0.20).

Test–retest reliability

The 2-week test–retest reliability of the Q-SPACE-Chi was measured by calculating the ICC, which yielded an overall value of 0.75 (95% CI: 0.70 to 0.79). The ICCs for the three factors, namely physical environment, social environment, and school policy, were 0.71 (95% CI: 0.65 to 0.75), 0.71 (95% CI: 0.65 to 0.676), and 0.69 (95% CI: 0.64 to 0.74), respectively. The overall ICC and individual ICC values were approximately 0.70, indicating moderate reliability. The Cronbach’s alpha values for these factors ranged from 0.70 to 0.72. Taken altogether, our findings support the internal consistency and test–retest reliability of the Q-SPACE-Chi.

Discussion

This study adapted and redeveloped the Q-SPACE [20] in Chinese and examine its psychometric properties for primary and secondary school students in Hong Kong. This study also hypothesized that the measurement properties of Q-SPACE-Chi was invariant among primary and secondary students. The key findings of this study are as follows: (i) Through CFA and MI analyses, a three-factor model of the Q-SPACE-Chi with 18 items was established. This model exhibited satisfactory factorial validity for primary and secondary school students in Hong Kong. (ii) Predictive validity demonstrated that the overall Q-SPACE-Chi were significantly associated with PA levels. (iii) The test–rest reliability and internal consistency of the Q-SPACE-Chi were moderately good, supporting the reliability and internal consistency of the Q-SPACE-Chi. (iii) MI further confirmed its validity for primary and secondary school students.

Factorial validity

The results confirmed the factorial validity of the three-factor modified Q-SPACE-Chi, with all items demonstrating significant loadings in the physical environment, social environment, and school policy dimensions. Although the initial CFA indicated a misfit in the modification indices, after the five items of misfit were excluded, the remaining 18 items demonstrated a good fit to the collected data. The initial exclusion of items SP3, P2, SP1, SP2, and S2 as presented in Table 2, led to the final modified Q-SPACE-Chi version comprising 18 items, with a good fit. The need to eliminate some items from the initial CFA analysis is in agreement with the approach taken by Robertson-Wilson, Lévesque and Holden [20], who similarly excluded eight items with poor factor loadings, ultimately leading to the final 18-item version of the Q-SPACE. The emergence of the three-factor Q-SPACE-Chi highlights the impact of school policy on students’ PA levels in the literature [16, 17]. Furthermore, the evaluation of the psychometric properties of our current modified three-factor questionnaire, including the measurement of the invariance, is crucial for establishing significance. This evaluation of the Mis [35] (configural, metric, scalar, and residual) provides substantial evidence supporting the reliability and validity of the Q-SPACE-Chi, offering a robust tool for assessing the perceived school PA environment among primary and secondary school students in Hong Kong.

Predictive validity

The predictive relationship between the modified Q-SPACE-Chi and PA levels among primary and secondary students was confirmed through a positive correlation observed between the mean score of the Q-SPACE-Chi and the PAQ-C. Although it was expected that the three dimensions of the Q-SPACE-Chi—physical environment, social environment, and school policy—would exhibit strong correlations with students’ PA, the observed associations between social environment, school policy, and PA were relatively weak, with correlation coefficients less than 0.20. This weaker correlation might be attributed to the impact of the COVID-19 pandemic on PA opportunities and PE classes. During the late 2022 and early 2023, in Hong Kong, the government announced the whole-school resumption of face-to-face class arrangements. The implementation of school-based curricula caused school management teams to establish various restrictive school policies for PE classes and PA opportunities (e.g., no after-school PA program or no Physical Education class) to suit their unique school contexts. Thus, the relatively low associations observed between the two school environment factors and students’ PA levels in this study can be attributed to the pandemic restrictions on students’ PA, which affected the social aspects of the school environments as well as school policy. The same explanation also went to the insignificant correlation between physical environment and physical activity in students. Perhaps, with different restrictive school policies for PE classes and PA opportunities, students in some schools might not be allowed to use school physical facilities to participate in physical activity. Another possible explanation of these weak correlation might go the discrepancy between student’s perceived and actual school PA environments. Study showed that the student’s objective and perceived availability of facilities was found to be low in their agreement [36]. Perhaps, PA associates more with objective school PA environments. However, in overall, our study confirmed the validity of Q-SPACE-Chi with a positive correlation between the overall mean of the Q-SPACE-Chi and the PAQ-C.

Test–retest reliability

The test–retest reliability of the modified Q-SPACE-Chi, assessed at a 2-week interval, demonstrated moderate reliability. Additionally, the resulted Cronbach’s alpha coefficients for the factors closely aligned with the findings of another study [20]. These results collectively support the internal consistency and test–retest reliability of the Q-SPACE-Chi.

Conclusions

This study examined the factorial validity, predictive validity, and the test–retest reliability of the modified Q-SPACE-Chi. One of the limitations in this study is that the correlation between the Q-SPACE-Chi and PA was weak due to our study timing coinciding with the COVID 19 pandemic. The pandemic-induced restrictions on students’ access to the physical environment, coupled with the lack of adequate social connections and stringent school policies, may have contributed to these weaker or insignificant correlations. More research and study replication could be done to investigate the correlation between the Q-SPACE-Chi and PA at the post-pandemic period. Next, it is recommended that the future researchers may examine both subjective and objective measurement of school environments in physical, social environment and social policy as student’s subjective perception and objective measurement of environments may be different. Even students and teachers were involved throughout the questionnaire development process, providing them with a general understanding of both the school policy and the questionnaire, students may not have a comprehensive grasp of the actual school policies. Consequently, a more in-depth investigation into the discrepancy between students’ perceptions and the realities of school policies is necessary for the future. Additionally, as questionnaire is acquired in this study to measure school PA environment, its validity may be undermined by recency and social-desirable bias. In this case, students might overestimate their perceived PA barriers and underestimate their PA under pandemic situation at that time. Other than predictive validity, future researchers may examine concurrent validity of Q-SPACE-Chi with using an observational approach in order to examine its psychometric properties. Incorporating objective assessments may also contribute to a more comprehensive evaluation in subsequent studies.

In conclusion, the validity and reliability of the three factors indicates the suitability of this scale for assessing students’ self-reported experiences within school PA environments. By incorporating these insights, this study aimed to enhance the understanding of students’ perceptions of the school environments in a more comprehensive manner. The modified Q-SPACE-Chi was the first of its kind in Hong Kong and can be used for adequately measuring PA within the context of school environments as the measurement properties of Q-SPACE-Chi was invariant among primary and secondary students. Furthermore, the addition of the third factor (school policy) is an improvement from the original Q-SPACE that comprised two factors; this additional factor highlights the pivotal role of school policies when assessing PA in the context of school environments. By encompassing the three dimensions of physical environment, social environment, and school policy, the Q-SPACE-Chi provides a comprehensive understanding of the intricate interplay of factors shaping students’ PA experiences. This scale can serve as a valuable tool for future research in both Hong Kong and mainland China.

Abbreviations

AVE

Average variance extracted

CFI

Comparative fit index

CR

Composite reliability

CFA

Confirmatory factor analysis

GH

Gamma hat index

ICC

Intraclass correlation coefficient

MI

Measurement invariance

NNFI

Non-normed fit index

PA

Physical activity

PAQ-C

Physical Activity Questionnaire for Older Children-Chinese

PARA

Physical Activity Resource Assessment

PE

Physical education

Q-SPACE

Questionnaire Assessing School Physical Activity Environment

Q-SPACE-Chi

Questionnaire Assessing School Physical Activity Environment in Chinese

RMSEA

Root-mean-square error of approximation

SFLs

Standardized factor loadings

SRMR

Standardized root mean residual

Author contributions

The only author, K.M.L. contributed to research conceptualization and study design, literature review, data collection, data analysis, and writing of the manuscript.

Funding

The study is funded by the Start-up Research Grant, Education University of Hong Kong [RG 73/2019-2020R].

Data availability

The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The purpose and details of the study were explained to all participants at the beginning of data collection. After obtaining informed consent from all participants, he or she completed a questionnaire during the site visit. Permission for human research was obtained from the University Institutional Review Board (Reference No. 2019-2020-0303) prior to the commencement of the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

ORCID

0000-0003-1409-109X.