The Shirom-Melamed Vigor Measure for students and how vigor relates to academic performance, mental health and satisfaction with life: Factorial analysis and construct validity in Spanish undergraduate university students

Students suffer from a decrease in physical activity during their education period. This lower level of activity could affect, through various paths, their academic performance, mental health and satisfaction with life. In this study, we assume that vigor, a positive affect variable, would act as a mediating variable in the above relationship, and thus, we propose an instrument for evaluating vigor in academic contexts. To attain this aim, the factorial validity and reliability of the proposed vigor scale for students is tested (Study 1), along with its construct validity (Study 2).


Abstract Background
Students suffer from a decrease in physical activity during their education period. This lower level of activity could affect, through various paths, their academic performance, mental health and satisfaction with life. In this study, we assume that vigor, a positive affect variable, would act as a mediating variable in the above relationship, and thus, we propose an instrument for evaluating vigor in academic contexts.
To attain this aim, the factorial validity and reliability of the proposed vigor scale for students is tested (Study 1), along with its construct validity (Study 2).

Methods
In Study 1, 707 undergraduates (59.7% women) responded to the vigor scale adapted for students to test factorial validation (through con rmatory factor analysis) and obtain reliability indicators. In Study 2, 309 undergraduates (55.3% women) completed a questionnaire measuring physical activity, mental health, satisfaction with life, vigor and academic performance to test a structural model of the relationships between the variables to obtain construct validity.

Conclusions
Proposing an instrument such as the Shirom-Melamed Vigor Measure for students allows the opening of a research venue that is focused on the study of positive affects in academic contexts, as well as the testing of the physical activity pathways of action in obtaining positive results.

Background
In addition to the bene cial effects on physical health [1][2][3], performing physical activity (PA) in adulthood is related to a reduced risk of mental ill-health [4] and has been shown to be effective in reducing symptoms associated with depression [5,6]. Going beyond a clinical approach and considering mental health (MH) as a positive state of well-being related to an optimal psychological experience and functioning [7], there are several studies that nd a positive relationship between PA and satisfaction with life (SWL) [8][9][10], that is, the extent to which people value their present situation as an ideal situation [11].
Cognitive function is also positively affected by performing PA [12,13], and positive relationships have been found between PA and academic performance (AP) in academic contexts, such as university [14,15].
Despite the relationships found between PA and different outcomes in MH, SWL and AP, the underlying mechanisms of this association are not fully understood [16][17][18]. In this vein, conceptual models such as that of Lubans et al. [19] are of great interest in identifying the neurobiological, psychosocial, and behavioral mechanisms that mediate the effects of PA on outcomes such as cognitive function, wellbeing and ill-being. There is a rooted tradition in the study of biological and/or neuronal factors that intervene in the positive effects derived from performing PA [20] compared to the inclusion of emotional or affective factors, even though affect has shown its usefulness in the explanation of results in health, educational and work contexts, among others [21][22][23]. In the workplace, the positive and energetic affect known as vigor [24,25], which is de ned as "a positive affective response to one's ongoing interactions with signi cant elements in one's job and work environment that comprises the interconnected feelings of physical strength, emotional energy, and cognitive liveliness" (p. 143) [24], has turned out to be a mediating variable of interest in the relationship that a group of personal and work resources [26] have with work satisfaction [27][28][29], MH [30] and performance at work [28,31,32], among others. Academic work and the context in which it is performed show similarities with work activity and the framework in which it occurs, that is, the organization [33]. In fact, constructs whose origins are situated in an employment context (burnout, engagement, organizational justice, etc.) have been successfully used in academic contexts [34,35]. Considering vigor at work [24] in academic contexts could help to understand the psychosocial-type mechanisms by which PA exerts its effect on cognitive function, health and wellbeing.
Currently, there is no measure in academic contexts that allows the evaluation of vigor from the perspective of Shirom [24,25]. The Shirom-Melamed Vigor Measure (SMVM) [24] assesses vigor at work, but it is designed for workers. The SMVM has recently been adapted to different countries [29,36] but its psychometric properties have not been analyzed in academic contexts.
In sum, the aim of this paper is twofold: 1) to determine the factorial structure and reliability of the SMVM in students, namely, the SMVM for students (SMVM-S) (Study 1); and 2) to analyze the construct validity of the SMVM-S by testing a structural model of the relations among PA, AP and MH by including vigor as a mediator (Study 2).
Study 1: Validation of the factorial structure of SMVM-S This study was performed to analyze the factor structure, validity and internal consistency of the SMVM in the university context.

Page 4/19
Methods Design, participants and procedure A cross-sectional survey was performed and completed by undergraduates from several degree programs at different Spanish universities. Participation was voluntary, and informed consent was obtained. The participants were recruited by fourteen students in their last year of psychology study from the O cial Masters Program of Positive Psychology at the University of Jaén (Spain), who were instructed about the procedure and the distribution of the questionnaires. Seven hundred and seven participants (out of 709 initial; 2 were eliminated due to incomplete data) completed the SMVM-S questionnaire and the other scales during the second semester of the academic year. The participants' mean age was 21.1 years (SD=2.9; range 17 to 48 years) and 59.7% of them were female.

Demographics
The students reported their sex, gender, grade they were enrolled in, and their university of origin.

Vigor
Stemming from the Shirom-Melamed Vigor Measure (SMVM) [24] and its Spanish adaptation [36], items were modi ed by substituting the content referring to employees and/or clients with the term "class mates". The participants had to indicate how often in the past 30 days they had felt a number of different feelings while studying. The scale consisted of 12 items that comprised three dimensions: physical strength (5 items; i.e., "I feel full of energy"), cognitive liveliness (3 items; i.e., "I feel I can think rapidly") and emotional energy (4 items; i.e., "I feel able to show warmth to others"). The response format ranged from 1 (almost never) to 7 (almost always). The scale yielded an adequate reliability, with ρ values of 0.87, 0.73 and 0.83 [36].

Data analysis
Con rmatory factor analysis (CFA) determined the factorial structure and the goodness of t of the different models of measurement. The violation of the assumption of multivariate normality of the data (Mardia's coe cient=36.79) suggested the use of a robust method of maximum likelihood (ML Robust) with EQS 6.4 software. As indicators of the goodness of t of the measurement models to the data, the corrected χ 2 by Satorra-Bentler (S-Bχ 2 ), NNFI, CFI, RSMR, and RMSEA, as well as their 90% con dence intervals (90% CI) were used. An acceptable model t was de ned as NNFI and CFI values close to 0.95 or greater, a SRMR value close to 0.08 or below, and a RMSEA value less than 0.08 [37,38]. The reliability of the dimensions of the instrument were examined by ρ [39], with 0.70 as an acceptable cut-off value [40].
The composite reliability (CR) and average extracted variance (AVE) were analyzed as indicators of internal consistency and convergent validity of the variables included, respectively. Cutoff values considered acceptable were CR values over 0.70 and AVE values over 0.50 [41]. Moreover, 0.6% of the values for all the variables were missing. Missing data were imputed using the expectation maximization algorithm in IBM SPSS for Windows v22.0 after checking that the missing values were completely random using Little's MCAR test [42] (χ 2 =22.67; df=43; p=0.995). Table 1 shows the resulting adjustment values of the CFA for different measurement models. The measurement models proposed in the adaptation to Spanish of the SMVM [36] were tested, along with other models that have been con rmed in workers. Speci cally, the other models included a model consisting of a single-factor structure (M1), a hierarchical model with three rst-order factors representing Three hundred and nine participants completed this study (out of the initial 312; 3 were eliminated due to incomplete questionnaires). The participants' mean age was 22.1 years (SD=2.4; range 18 to 33 years), and 55.3% of them were female.

Demographics
The same data as those used in Study 1 were included.

Vigor
The same as that described in Study 1.

Physical activity
The International Physical Activity Questionnaire-Short Form (IPAQ-SF) [43] was administered. This instrument allows the calculation of the volume of total PA (walking, moderate and vigorous) over the last seven days by evaluating or assigning to each activity some energy requirements, which are de ned as METs (multiples of the rate of metabolic expenditure), and it allows us to obtain a result in METsminutes, that is, multiplying the result of the METs of an activity by the number of minutes during which it has been performed. A total MET-minutes per week number was obtained for each participant, where the highest values indicate a higher volume of total PA per week and therefore a higher energy consumption.
The psychometric properties of the instrument have been evaluated in different countries and have obtained optimal values of reliability and validity in the adult population [43,44]. In Spanish undergraduates, the IPAQ-SF has shown to be a valid instrument when compared to objective measurements of PA obtained using accelerometers [45].

Mental health
The General Health Questionnaire (GHQ) [46] assesses normal healthy functioning to measure minor psychiatric morbidity and the appearance of new, distressing symptoms [47]. This study includes the 12 items version (GHQ-12) [48] adapted to Spanish [49] . This scale includes items such as "Could you concentrate as well as you used to?" that are responded to in a 4-point Likert format (ranging from 0=much more than usual to 3=not at all). Despite some controversy regarding the factorial structure of the instrument, studies with the Spanish population con rm a structure of three interrelated factors [50]: dysphoria, social dysfunction and loss of con dence. Other studies also con rm these factors [51,52]. A higher score indicates lower levels of health. In general, the Cronbach's alpha for Spanish samples is 0.76 [49].

Satisfaction with life
The Spanish version [53] of the Satisfaction with Life Scale (SWLS) [54] was included in the questionnaire. This instrument has an unidimensional structure and assesses the cognitive component of subjective well-being [55]. On a 5-point Likert scale (ranging from 1=not at all to 5=absolutely), the participants responded to 5 items (i.e., "In most ways, my life is how I want it to be"). In the Spanish general population, the Cronbach's alpha value is 0.88 [53].

Academic performance
To evaluate AP, the students' grades were considered. The overall grade was determined as the weighted average of the nal scores obtained in all subjects, with a range from 0 to 10, where 10 is the highest grade. Freshman students indicated their global score in access to university exams, which was recorded on a 14-to 10-point scale. In our sample, the range was from 5 to 9.21 points, with a mean grade of 6.98 (SD=0.93).

Data analysis
All analyses were performed with EQS 6.4 software for Windows. To avoid model estimation problems due to the disparate metrics of the variables, the items were rescaled by means of the percentage of proportion of maximum scoring (POMS) [56] using the observed maximum score in each variable. Due to the deviation of the data regarding the multivariate normality (Mardia's coe cient=10.29), the option ML Robust was used. The t of the measurement model to the data was rst veri ed by CFA. The model was shaped by ve latent variables: vigor with three indicators (physical strength, cognitive liveliness and emotional energy), total PA with an indicator (MET-minutes per week), SWL with ve indicators (each item of the scale), MH with three indicators (dysphoria, social dysfunction and loss of con dence) and AP with one indicator (average grade taken from the student record). For latent variables with a single indicator, the factor loadings were set with a value equal to one, and the variance of the error was estimated to x it a priori [57]. Finally, a covariation was allowed between the error terms of two indicators of MH (dysphoria and loss of con dence) that would be theoretically justi ed [58]. The indicators of the goodness of t of the models, as well as the criteria for deciding on their t, were the same as those used in Study 1. Regarding the analysis of the structural model, following the theoretical mediational model of vigor at work [25] and theoretical proposals as Lubans' et al. [19], the t of a group of nonnested structural models based on the CAIC index was compared. Lower values of the index corresponded to a better t of the model, as long as the rest of the indices and their values met the decision criteria used in Study 1. Moreover, less than 3% of the values for all the variables on each scale were missing. Following the same procedure as that used for the missing data from Study 1, the missing values were imputed using the expectation maximization algorithm, since the missing values were completely random, and by Regarding the structural models, the M1 model or the saturated model was compared with the M2 (no direct effect), M3 (no effect from the predictor variable to the mediator) and M4 (no effect from the mediator to the outcome) models [59]. Table 3 shows the adjustment indices for each of the structural models, allowing the CAIC index, since they are nonnested models, and shows that M2 is the best model (CAIC=-308.643), that is, a model of total mediation in which the levels of PA exert their in uence on academic results and levels of SWL and MH through the levels of vigor of the students. The t indices of the M2 model show adequate values (S-Bχ 2 =108.82; df=62; p<0.001; CFI=0.92; NNFI=0.90; SRMR=0.05; RMSEA=0.05 [0.03-0.06]). Figure 1 graphically displays the mediation model with the standardized regression coe cients and the percentage of variance explained in each endogenous variable. Total weekly PA levels were positively associated with student vigor levels, and vigor was positively related to SWL levels and AP and negatively related to the levels of MH (note that a higher score in MH implies a more deteriorated state of MH). The model explained 4% of the variance in the levels of vigor, 54% of the levels of SWL, 71% of the levels of MH and 6% of the AP. The indirect effects that total PA exerted through vigor on levels of SWL, levels of MH and AP were all signi cant (β=-0.14; β=-0.16; and β=0.04, respectively), thereby con rming the important mediating role played by the students' level of vigor in the relationship between the levels of total PA and other important results in academic contexts [60].

Discussion
Vigor at work has emerged as a positive affect variable, characterized by showing physical strength, cognitive liveliness and emotional energy at work [24], and helps to explain positive consequences at work [25]. In this research, we performed two studies to analyze the psychometric properties of the SMVM-S, an instrument to assess vigor levels in the academic context. Study 1 analyzes the t to the data of different measurement models of the SMVM-S, as well as the reliability of the instrument. This study con rms that the model with the best t is that consisting of three interrelated factors, following the same structure as that used in previous studies carried out on workers [27,29,36], with signi cant correlations among the SMVM-S dimensions and adequate indices of internal consistency. Study 2 con rms the mediational model proposed by Shirom [25] in an academic context and contributes to the identi cation of affective factors that explain the effects of PA on different results [19]. The level of total PA per week would imply higher levels of vigor in students, while vigor would positively in uence academic results and SWL levels and negatively in uence the deterioration of MH. With few exceptions [61], the PA-vigor positive relationship (mainly the physical aspect of vigor) has been veri ed in different studies [62][63][64][65][66]. Vigor in university students derived from performing PA is characterized by a combination of moderate amounts of arousal and pleasure [67,68], affective experiences that have also been associated with physical exercise [69]. In the same way, it has been possible to verify that the levels of vigor in students have an effect on their average grades-as an indicator of performance-on their MH -with a lower incidence of psychological problems-and on their SWL. These results contribute to understanding psychosocial mechanisms, speci cally the experimentation of positive affect, by which PA affects the well-being and cognitive performance of university students, thus responding to the need to verify the mediational role of constructs such as the vigor that is proposed in different heuristic and conceptual models [19,25].
Providing to health and education professionals an instrument such as the SMVM-S would help to better understand the role that PA has in reducing risks and improving the well-being of university students, who show a high rate of psychological disorders [70], and a decrease in AP compared with other educational stages [71] and whose age includes the adult population group with lower levels of SWL (18-25 years) [72]. In fact, if we consider that a high percentage of university students (40-70%) can be considered as physically inactive [73][74][75][76], then the potential for PA to generate positive effects through the levels of vigor experienced is greater [77].
The results obtained herein also help to clarify the role of an affective type of variable such as vigor in one stage, namely, the university, which entails important changes in different facets [65,66] for young people.

Limitations
The current study considered total PA, without differentiating between light, moderate or vigorous PA or the different effects that these activity levels could have on the results. To make this decision, we used as a base the potential bene ts of PA, even when the minimum levels recommended by international organizations were not reached [1][2][3], and the positive effect of PA on AP, regardless of the type of PA [78]. However, we are aware that in future studies, the mediating role of vigor should be veri ed by taking into account the different levels of PA and other variables such as environmental aspects [79].
Using self-report measures for the assessment of PA could be considered another limitation. Although this instrument concurs with other objective forms of evaluation, such as accelerometry [45], it is recommended that in future studies, the relationships of the model with other types of measurement be veri ed.
Finally, all the data were collected in between exam periods, whereas di culties in managing exercise are greater during examination periods [80]. Thus, future studies should include different temporal moments to determine whether exam periods or other stressful situations in uence the relations shown in this study.

Conclusions
The analysis of the vigor approach [25] in the university context allows us to test one of the mechanisms proposed at the psychosocial level by Lubans et al. [19], whereby PA affects the AP, MH and SWL of students through the effect of PA on vigor. This study offers health and education professionals an instrument (i.e., the SMVM-S) to make progress in improving the MH, SWL and AP of university students through the promotion of PA. Note. All factor loads were signi cant.  Figure 1 Estimated standardised regression and squared multiple regression (R2) coe cients for Model 2 (no direct effect). Note. All estimated standardised regression and squared multiple regression (R2) coe cients were signi cant.