An example of a social environmental factor related to health at the interpersonal level:

Participants and Procedure

The study sample was conveniently taken from the four schools in Rawalpindi city of Pakistan. The population of the study is composed of all the students enrolled in these four schools. Participants were invited to participate on the basis of informed consent. After the research proposal was approved from institutional review board of Zhejiang University, data were collected during October–November 2020, when schools were re-opened during COVID-19. All formal permissions were taken from the local authorities in Rawalpindi city. Data collection involved the collaboration of a local researcher. Seven hundred questionnaires were distributed among students in different schools. Of these, 664 questionnaires were returned, resulting in a response rate of 94.86%. The survey was not affected by COVID-19, as the government allowed the re-opening of schools, thus ensuring the students’ attendance. The results may be different from the normal conditions. Data were checked for missing values, and 28 cases were removed. Then, the researcher screened data and conducted all analyses in IBM SPSS v.20 [IBM, Armonk, NY, USA]. Mahalanobis distance was calculated, and 8 outliers were found, and the cases were deleted. Finally, the sample consisted of 618 school adolescents [51.5% males, 48.5% females] with an age range of 11 to 14 years [SD = 0.97]. Socioeconomic status was measured in terms of social classes based on 3-point scale containing lower class, middle class, and upper middle class. A total of 48.7% of the students were from lower-middle class, 50.2% were from middle class, and 1.1% were from upper-middle class. The students from rural areas were 49.7%, while those from urban areas were 50.3%.

3.1. Demographic Factors

Demographic characteristics of the students included demographics [5 items, i.e., age, gender, grade, weight, height], GC [1 item responded as selecting from two areas: 1 = rural and 2 = urban], and socioeconomic status [1 item, responded as selecting from social classes: 1 = lower-middle class, 2 = middle class, and 3 = upper-middle class]. BMI was measured through self-reported weight and height by means of the formula: BMI = weight [kg]/height2 [m] [48].

3.2. Intrapersonal Factors-Related Measurements

Self-efficacy. The exercise self-efficacy scale by [49] was adopted in this study. The scale is part of the self-efficacy scale for school adolescents, containing 9 items. The scale was used in its original English version, as the official and institutional language in Pakistan is English.

Internal and external motivation. Internal and external motivations for PA were measured with scales developed in a previous study by [50] based on SDT [51,52,53]. The internal motivation scale consists of three items. An example item is, “I enjoy it”. The external motivation scale also contains three items. The example item is, “My parents, other family members, or friends tell me to do it”. All items are measured with options 1 = not at all true to 5 = very true.

Attitude. Attitude toward exercise was measured using 5-point bipolar adjective scales developed by [54]. We assessed both instrumental [useful–useless, harmful–beneficial, wise–foolish, bad–good] and affective [enjoyable–unenjoyable, boring–interesting, pleasant–unpleasant, stressful–relaxing] components of attitude using adjectives that are commonly employed in the exercise domain [e.g., [55]. All the items are rated on 5-point bipolar adjective scale. The statement that preceded the adjectives was, “For me to participate in regular physical exercise is...”

3.3. Interpersonal Factors-Related Measurements

To measure social support for PA, the scale used was an adapted scale from a student survey that Amherst Health and Activity Study [56] used and validated in the previous studies with some modifications [57,58]. Initially, the scale consisted of two components: family and friends only. Researchers devised the item for teachers’ support by using the words from friends’ and family’s support [57]. It was rated on a five-point scale from 1 = never to 5 = every day. For this purpose, students were asked to answer questions related to how often they get social support from friends, family, and teachers. The item for friends/ peer support is “During a typical week at school, how often do your friends do PA or play sports with you?”. The item for teachers’ support is “During a typical week at school, how often does your teacher encourage you to do PA during recess or lunch breaks?”. The item for parents’ support is “During a week, how often does your family encourage you to do PA in your free time?”

3.4. Organizational Factors-Related Measurements

Organizational factors: Organizational factors were examined in this study to see the cognition of PA environment, PA policy, and PA culture in the school. The school PA environment contains equipment, facility, and safety component. The school environmental characteristics were assessed subjectively, using 10 items from the Questionnaire Assessing School PA Environment [Q-SPACE], which Robertson–Wilson and his colleagues [59] validated in previous studies in European and Asian cultures [60,61,62]. The questionnaire is composed of three factors that are rated on 5-point Likert scale from 1 = strongly disagree to 5 = strongly agree.

Equipment [3 items], examining the accessibility or usability of physical equipment [e.g., there is enough equipment for PA at school];

Facility [4 items], measuring the accessibility or usability of PA facilities [e.g., the school ground is wide enough for PA];

Safety [3 items], investigating perceived safety of PA equipment and facilities [e.g., it is safe to engage in PA on the grounds and in the gym at school].

The items for measuring school PA policy and PA culture refer to a scale from a Chinese study by [41]. The items were translated into English, which is the official and institutional language in Pakistan. Translation and back-translation method were used [63]. Then, two native speakers read the instrument for confirmation of accurate translation. The instrument measures the cognition of school policy and culture for PA. The example item is, “The school organizes extra-curricular exercises during breaks”. A 5-point scale is used, ranging from 1 [strongly disagree] to 5 [strongly agree]. The instrument was revalidated in the Pakistani setting, generating two factors: PA policy [7 items] and PA culture [2 items].

3.5. PA Measurements

Global School-Based Student Health Survey. There are many methods for PA measurement. The scale survey is the most economical among many measurement methods and is suitable for the measurement of large samples. Therefore, in this study, global school-based student health survey [GSHS] was used, which is specifically designed for measuring adolescents’ PA levels [64]. This questionnaire is already validated in Pakistan [64] and other Asian countries like China [65]. It includes four questions inquiring PA levels of the adolescents. The first question is “During the past 7 days, on how many days were you physically active for a total of at least 60 min per day?”. The responses ranged from “0 days” to “7 days.” Adolescents who were doing at least 1 h of PA per day were considered to be physically active [65,66]. The second question was “During the last 7 days, on how many days did you walk or ride a bicycle to or from school?”. Adolescents who walked or rode a bicycle to or from school for at least 3 days per week were considered to be active [65]. Third question was “During this school year, on how many days did you go to a physical education class each week?”. The responses ranged from “0 days” to “5 or more days.” Adolescents attended physical education class at least 5 days a week were classified as physically active [65]. The last question was about sedentary behavior: “How much time do you spend during a typical or usual day sitting and watching television, playing computer games, talking with friends, or doing other seated activities, such as surfing the Internet?”. The responses were “less than 1 h per day”, “1 to 2 h per day”, “3 to 4 h per day”, “5 to 6 h per day”, “7 to 8 h per day”, or “more than 8 h per day.” Adolescents who spent 3 or more hours sitting per day were considered inactive 65. In this study, the responses on only the first questions were analyzed to measure physical activity levels of adolescents as it was done in the previous study [65,66].

3.6. Analysis Techniques and Validation Process

IBM SPSS v.20 was used for the analyses. First, data were examined for missing values and outliers and screened accordingly. Then normal distribution was checked. Data were normally distributed, as the skewness and kurtosis were according to the acceptable standard +2 [67]. Common method bias was also tested by using common latent factor method showing no common method bias in data. Then, exploratory factor analysis was performed to ensure the underlying factor structure of all variables or resources and to check validity of instruments. Since the study was based on already developed and validated questionnaires with a determined number of factors, the number of factors was fixed to certain number as identified by α extraction. The criteria for retaining the factors were based on the eigenvalue being ≥1.0 and a value of ≥0.50 on factor loading. The 9 factors were generated under three categories. The first category included three factors: self-efficacy [9 items, α = 0.89], motivation [6 items, α = 0.85], and attitude [8 items, α = 0.87], representing individual level elements. The next factor represented interpersonal elements [3 items, α = 0.81], containing family support [1 item], friends/peer support [1 item], and teachers’ support [1 item]. However, the three items were taken as separate social support indicators. The last set of 5 factors denoted organizational elements containing school PA environment, such as equipment [3 items, α = 0.74], facility [4 items, α = 0.87], and safety [3 items, α = 0.78] and PA policy [7 items, α = 0.84] and culture [2 items, α = 0.78]. Under these criteria, one item was removed for culture factor. Composite reliability [CR] and average variance extracted [AVE] were calculated to check convergent validity of the instruments. All questionnaires have convergent validity possessing CR > 0.70 [CREquipment = 0.75 to CRSE = 0.90] and AVE above 0.50 [AVEattitude = 0.50 to AVEculture = 0.64]. Discriminant validity was also assessed in terms of AVE > maximum shared variance [MSV] [68]. AVE for all factors was found greater than MSV, hence ascertaining discriminant validity. This, collectively, indicates that the scale has an adequate validity in terms of convergent and discriminant validity. The results are presented in Table 1.

Correlation coefficients were computed to verify the possible relationships between PA and set of variables. At the end, stepwise forward regression in SPSS ver. 20 was applied to see the hypothesized effect of individual, interpersonal, and organizational factors on PA after controlling for demographic factors and to see the stronger predictors of PA. GC and SES were entered into the control variables. Gender, age, and BMI were excluded, as these were not found significantly related with PA.

4.1. Demographics

Demographic characteristics of the sample is given in Table 2.

4.2. Preliminary Analyses

In preliminary analyses, first, mean values, and standard deviations [SD] for each variable were computed. Second, an independent sample t-test was applied to see the difference of gender and GC for adolescents PA. The mean PA of urban adolescents [3.24] was a little higher than that of rural adolescents [3.10]. It is inferred that there was no significant difference between male and female adolescents in doing PA [t = −1.53, p > 0.05]. Moreover, a significant difference was found for PA of adolescents from rural and urban areas [t = −2.49, p < 0.05]. Third, one way ANOVA was applied to see the difference on PA of the adolescents from different SES. We have found students with different socioeconomic statuses were significantly different in their PA [F = 12.64; p < 0.001]. Hence, both SES and GC were controlled for the major regression analyses.

Then, correlation among the variables was calculated to see the relationship of individual, interpersonal, and organizational factors with PA. Table 3 shows the relationship between PA, age, BMI, self-efficacy [SE], motivation [MOTT], attitude [ATT], family, friends’, teachers’ support, equipment, facility, safety, policy, and culture and also means and SD values for all variables. The association between PA and individual correlates depicts that age and BMI were not related to PA of adolescents. Conversely, the correlations between PA, self-efficacy, motivation, and attitude were statistically significant [rSE = 0.26, rMOT = 0.24, rATT = 0.38]. Further, support for PA by family, friends, and teachers was significantly related to PA of adolescents [rfamily = 0.35, rfriends = 0.10, rteachers = 0.10]. However, the coefficient for family support is quite higher than those of friends and teachers. Finally, the association among PA and organizational variables was computed, highlighting that PA is significantly related to all organizational variables, such as equipment, facility, safety, policy, and culture [requipment = 0.12, rfacility = 0.17, rsafety = 0.12, rpolicy = 0.10, rculture = 0.17]. The results may be affected by the stream of COVID-19.

4.3. Regression Analysis

A stepwise forward regression analysis was computed to identify the main predictors of PA in a possible prediction model. In the present study, ATT, family support, self-efficacy, SES, MOT, facility, and GC factors were introduced as possible predictor variables for adolescents’ PA. R2 and adjusted R2 were calculated. In accordance with Ellis, as reported in a previous study [69], the adjusted R2 was reported because it represents the effect-size index that resulted after the statistical correction considering the number of participants and the variables in the prediction model.

Results of the final prediction model from the regression analysis are reported in Table 4. R2 was significant [adjR2 = 0.29; F = 36.16, p < 0.001] and explained approximately 29% of the variability observed in PA. The final model included seven predictors of PA, which are attitude, family support, self-efficacy, SES, motivation, facility, and GC. The first variable that appeared as the strongest predictor in the model was attitude towards PA [β = 0.29, p < 0.001]. Family support for PA was emerged as a second variable in the regression model [β = 0.20, p < 0.001]. The third predictor was self-efficacy for doing PA [β = 0.18, p < 0.001]. Then, SES was found as the fourth possible variable that predicted PA [β = 0.18, p < 0.001], indicating that adolescents with higher SES do more PA and vice versa. Next, motivation also played a significant role in predicting PA [β = 0.10, p < 0.05], showing that greater motivation would lead to higher PA. Additionally, facility for PA predicted PA of adolescents [β = 0.09, p