A study was conducted with an instrumental-psychometric design aimed at validating a psychological assessment instrument (Montero and León, 2002).

The population was comprised of in-person university students from three universities in Ecuador: Technical University of the North, Private Technical University of Loja, and Salesian Polytechnic University. The sample consisted of 4,524 students, selected by non-probabilistic convenience sampling; the mean age of the participants was 22 (SD = 3.15), 55% were women, 94% were single, 90% were full-time students, and in terms of ethnicity, 90% were mestizos, 5% were indigenous, 2% were Afro-Ecuadorians, and 1.2% were white.

Sociodemographic data. This section included information regarding basic sociodemographic data, such as age, sex, marital status and geographic region.

Satisfaction with Life Scale (SWLS). This instrument was designed by Diener et al. (1985) and adapted to Spanish by Atienza et al. (2000). It measures or assesses global cognitive judgments about satisfaction with life. It consists of 5 items that include a 7-point Likert-type response scale, ranging from 1 = “completely disagree” to 7 = “completely agree.” The total score ranges from 5 to 35, with higher values indicating greater satisfaction with life. The score is categorized according to the following reference points: 5–9 “extremely dissatisfied”; 10–14 “dissatisfied”; 15–19 “slightly dissatisfied”; 20 “neutral”; 21–25 “slightly satisfied”; 26–30 “satisfied”; 31–35 “extremely satisfied” (Diener et al., 1985). The study of the analysis of the psychometric properties of the scale adapted to Spanish shows that the instrument has good internal consistency (α = 0.84), a single-factor structure that explains 66% of the total variance of the construct and satisfactory convergent and divergent validity with other psychological variables such as academic satisfaction, feelings of happiness and feelings of loneliness (Atienza et al., 2000).

Patient Health Questionnaire (PHQ-9). The instrument was designed by Kroenke et al. (2001) to assess depressive symptoms according to DSM-IV criteria during the previous 2 weeks. The scale has 9 items that include a 4-point Likert-type scale for response: 0 “not at all”; 1 “several days”; 2 “more than half the days”; 3 “almost every day.” The total score can vary from 0 to 27 points that can be included in 5 categories of severity of the depressive disorder: 0–4 “none”; 5–9 “mild”; 10–14 “moderate”; 15–19 “moderately severe” and 20–27 “severe.” Regarding its psychometric properties, according to a study carried out on Ecuadorian university students, the scale presents adequate internal consistency (ω = 0.90), a hierarchical structure that includes a general factor and three latent factors: somatic, cognitive/affective and concentration/motor; the scale also possesses satisfactory convergent and divergent validity with several health indicators (López-Guerra et al., 2022). Medium to strong negative correlations were expected between PHQ-9 and SWLS scores.

Perceived Stress Scale (PSS-10). The original scale was developed by Cohen et al. (1983). The 10-item version, adapted to Spanish by Remor (2006), assesses the degree to which people perceive a lack of control in their daily lives over the previous month. The 10 items on the scale provide five response options: 0 “never,” 1 “almost never,” 2 “occasionally,” 3 “often” and 4 “very often.” To tally the result, the scores for the following items are inverted: 4, 5, 7 and 8, and finally the scores for the 10 items are added together. Higher scores indicate higher levels of perceived stress and vice versa for lower scores. Regarding the psychometric properties of the scale, in a study where it was linguistically and culturally adapted to Ecuador, good internal consistency was found (α = 0.85 ω = 0.87), a bifactorial structure that explains 56.99% of the total variance, as well as satisfactory convergent validity with multiple health indicators (Ruisoto et al., 2020).

This study was conducted within the framework of a larger project to predict drug use in university students from three institutions in Ecuador. The study was approved by the Human Research Ethics Committee (Comité de Ética de Investigación en Seres Humanos, - CEISH March 6, 2019) of the Private Technical University of Loja - UTPL, Ecuador (UTPL-DIS-2019-0088-O) and was conducted according to the principles expressed in the Declaration of Helsinki (Association, W M, 2013). Digital informed consent was obtained from all participants, who received feedback on the results of their assessment.

For the development of the study, students from three universities in Ecuador were initially invited by email to participate in the research. For 10 weeks, a process of awareness and communication was carried out aimed at students using internal institutional media, social media of the universities and campaigns on the university campuses of the three institutions in order to publicize the research and the collection of data from students who wished to participate simultaneously proceeded. The application of the instruments was carried out online, average response rate across universities was 47.80%, ranging from 39.10 to 56.10%.

Statistical analyses were performed using IBM Statistical Package for the Social Sciences (SPSS) software, version 24 (IBM Inc., Chicago, IL, United States) and the JASP program, version 0.18.3.0.

First, the factor structure was analyzed by performing an exploratory factor analysis (EFA) and a confirmatory factor analysis (CFA). Following Harrington (2009) recommendation that indicates that, to obtain the factor structure, both the EFA and the CFA should be performed in different samples, the total sample (N = 4,524) was divided into two independent and homogeneous random samples, subsamples nA = 2,264 and nB = 2,260. The chi-square statistical test did not reveal significant differences in both subsamples, so the random selection helped to maintain the same proportion of sociodemographic characteristics in each of them.

The first subsample (nA) was used to perform an EFA in order to determine the adequacy of the factor loading in each item of the SWLS and the factor structure of the scale was analyzed. For this, the feasibility of carrying out an EFA was previously evaluated using the Kaiser-Meyer-Olkin (KMO) factor adequacy criterion and Bartlett’s sphericity test, where values ≥ 0.80 for the former (Kaiser, 1970) and significance levels p < 0.05 for the latter (López-Aguado and Gutiérrez-Provecho, 2019) indicate the interrelation of the data. For the EFA, the principal axis factorization method was used in combination with oblimin rotation, retaining factor loadings greater than 0.40 in the rotated matrix (Lloret-Segura et al., 2014).

The second subsample (nB) was used to perform a CFA. Considering that SWLS is in Likert format and is an ordinal measure, a diagonal weighted least squares (DWLS) estimation method using polychoric correlations was used to perform the CFA. The DWLS method is recommended for large samples (N > 200; Freiberg Hoffmann et al., 2013). The indices selected to assess the goodness of fit of the studied models were the chi-square (χ²) ratio by degrees of freedom (df), Bentler comparative fit index (CFI), Tucker-Lewis index (TLI), standardized root mean square residual (SRMR), and the root mean square error of approximation (RMSEA). To examine the adequacy of the model, the following parameters were considered: X²/df ≤ 3 adequate, ≥ 2 optimal (Byrne, 2016); CFI and TLI ≥ 0.90 adequate, ≥ 0.95 optimal; RMSEA and SRMR ≤0.08 adequate, ≤ 0.05 optimal (Hu and Bentler, 1999).

Second, the average variance extracted (AVE) was calculated in order to assess whether a set of indicators really measures a given construct and are not measuring another different concept. The acceptance criterion is that the average variance extracted (AVE) of a construct must be greater than 0.5, meaning that the construct shares more than half of its variance with its indicators, with the rest of the variance being due to measurement error (Fornell and Larcker, 1981).

Third, factorial invariance was assessed for the total, male, and female samples in the second subsample (nB), taking into account the following models: configural invariance (M1), indicating an unrestricted factor structure (baseline); metric invariance (M2), where equivalence restrictions between factor loadings are established; scalar invariance (M3), that is, loading and intercept equivalence restrictions; and strict invariance (M4), considering the equivalence restrictions of factor loadings, intercepts, and residuals. Measurement invariance and its levels were evaluated according to the recommendations of Cheung and Rensvold (2002): ΔCFI ≤0.01 and ΔRMSEA ≤0.015.

Fourth, the reliability of the scale was analyzed through internal consistency analysis, using the McDonalds omega coefficient (ω), considering values ≥ 0.70 as satisfactory (Nunnally, 1978). The use of this coefficient was considered since it can be used in ordinal, unidimensional scales, it does not depend on the number of items and, when working with the factor loadings, it makes the calculations more stable. In addition, it is an adequate measure of reliability if the tau equivalence principle is not met, which is often violated in practice (Ventura-León and Caycho-Rodríguez, 2017; Vizioli, 2021).

Fifth, divergent validity was analyzed based on the Pearson correlation (r) between SWLS scores and scores on the depression (PHQ-4) and perceived stress (PSS-10) scales, expecting negative, moderate, and strong correlations with a significance level of p < 0.05. Cohen’s (1988) recommendations were used to establish the magnitude of the relationship between variables, thus r = 0.10 represents a weak or small association, r = 0.30 is considered a moderate relationship, and r ≥ 0.50 represents a strong or large correlation.

Finally, a descriptive analysis (mean [M] and standard deviation [SD]) of the students’ responses to the SWLS was performed. In addition, a Student’s t-test was used to assess whether there were significant differences between the means of women and men.

An analysis of the data suggests that the SWLS items showed a distribution within the limits of normality. According to Arenas et al. (2023), items are normally distributed variables when their asymmetry is less than 2 and kurtosis less than 7, and the analysis carried out in the present study showed maximum values of 0.705 for asymmetry and −1.18 for kurtosis (see Table 1).

The main tests of sampling adequacy were satisfactory (KMO measure = 0.863; Bartlett’s sphericity test: X² [10] = 6745.794; p < 0.001), so it was considered pertinent to carry out the respective EFA.

The EFA of the 5 items of the Satisfaction with Life scale using the principal axis factoring method with oblimin rotation yielded a single-factor solution that accounts for 63.2% of the total variance of the test. All loadings were values greater than 0.60 ranging between 0.645 and 0.923 (see Table 1).

To determine the factor structure of the scale, the goodness-of-fit indices of two different models were compared: (1) One factor, based on the results obtained in the EFA and the consistent results in the CFA of several previous studies including the original author of the scale and his collaborators (Diener et al., 1985), and (2) two factors, satisfaction with life with present achievements (items 1, 2 and 3) and past achievements (items 4 and 5) reported by Bagherzadeh et al. (2018), Pavot and Diener (1993) and Sachs (2003).

Based on the fit indices, both models could represent the data observed for the sample of university students. Although the two-factor or two-dimensional model (present achievements and past achievements) fits better to the empirical evidence compared to model 1 (unifactorial; see Table 2), the correlation between these two dimensions is very high r = 0.92, which makes this differentiation questionable. In addition, in the first dimension (present achievements), it loaded three items and in the other (past achievements) only two items (see Figure 1B). It is evident that a dimension with only two items should be avoided, if possible, when defining a construct (Sachs, 2003). Therefore, the one-factor model 1, which presented an optimal fit (X²/df = 3.79; CFI = 0.998; TLI = 0.996; SRMR = 0.023; RMSEA = 0.030) and is more parsimonious and consistent with the theoretical framework associated with the development of the scale as a unidimensional measure of general satisfaction with life, is proposed as the final solution for the data of the Ecuadorian students (See Figure 1A).

Regarding convergent internal validity, the result obtained was satisfactory (AVE = 0.553), which suggests that the construct explains a significant proportion of the variance observed in the responses to the items, evidencing an adequate explanatory capacity.

The single-factor model was selected for the following SWLS factorial invariance analysis for the total sample and by sex. The results are shown in Table 3, where the configural invariance (M1) can be observed showing good fit indicators (CFI = 0.999 and RMSEA = 0.016). The metric invariance (M2) resulted in good fit indices (CFI = 1.000; RMSEA = 0.009), similar to the M1 values because they presented minimal differences (ΔCFI = 0.001 and ΔRMSEA = −0.0007). The factor loadings did not vary between both sexes, so the covariances can be compared. The scalar invariance (M3) showed indices equal to the previous model (CFI = 0.999; RMSEA = 0.014) with minimal differences (ΔCFI = −0.001 and ΔRMSEA = 0.005). Invariance between thresholds is accepted. The strict invariance (M4) reflected a good fit (CFI = 0.998; RMSEA = 0.018) with minimal differences (ΔCFI = −0.001 and ΔRMSEA = 0.004), so the invariance of residuals is verified. The combined results indicate factorial invariance of SWLS in both sexes.

The internal consistency for the SWLS total score was satisfactory (ω = 0.84), which guarantees the reliability of the scale. Since the model was not bifactorial, there was no need to calculate the hierarchical omega coefficient. Likewise, there was no correlation between the errors, so the omega coefficient did not have to be corrected either (Vizioli, 2021).

Regarding divergent validity, the correlation between the scores of the Satisfaction with Life Scale (SWLS) and the scores of the Patient Health Questionnaire (PHQ-9) that measures depression and the Perceived Stress Scale (PSS-10) was analyzed with the complete sample (N = 4,524).

The results in Table 4 indicate that satisfaction with life exhibits a strong negative correlation with depressive symptoms and perceived stress, supporting adequate divergent validity.

Taking the scale as unifactorial, a summed score of all the items was calculated. The normality calculated through the asymmetry and kurtosis coefficients showed compliance with this assumption (−1.1; Zerpa et al., 2023). Regarding the analysis of this score, the university students evaluated presented moderately high satisfaction scores (M = 24.29; As = −0.484), with moderate dispersion around the mean and a platykurtic distribution (SD = 6.33; K = −0.164).

With homogeneity of variance between groups, evaluated by Levene’s test (F = 3.08; df = 1, 4,522; p = 0.08), it was found that there were no significant differences between men (M = 24.38; SD = 6.44) and women (M = 24.22; SD = 6.23) according to Student’s t-test [t(4,522) = 0.842, p = 0.400].

Is important to point out some limitations of the study. Firstly, the data must be interpreted with caution since the sample was non-probabilistic and only comprised university students, which makes it difficult to generalize the findings to the rest of the youth population. Furthermore, factorial invariance due to sociocultural factors, which are important in Ecuador, was not considered. Therefore, it is suggested that studies be carried out to further explore the psychometric properties of the scale in other Ecuadorian samples from different socioeconomic strata, ethnicities, and age groups. Secondly, the design of this study is cross-sectional, so the measurement of test–retest reliability could be assessed in future research through a longitudinal design. Thirdly, only discriminant validity was evaluated, and it is important to assess other types of validity (convergent, criterion, predictive, for example). Finally, the use of tools such as the SWLS may be subject to certain inaccuracies in data collection due to recall bias and social desirability. For future research, the use of scales intended to measure and control potential biasing sources that may influence the results is suggested (Rosenmanm et al., 2011).

Despite the limitations mentioned, this study lays the foundation for future research on satisfaction with life using SWLS in the Ecuadorian population. Satisfaction with life in university students is closely related to important aspects of the developmental stage of these young people, such as achievement of goals and definition of objectives, as well as cognitive, emotional, and motivational aspects that influence subjective well-being, such as self-acceptance, autonomy, self-determination, and an optimistic view of life (Chan Chi, 2021; García-Alandete et al., 2018). Likewise, it affects the ability to cope with adversity, responsibility, and academic performance, which are elements directly linked to the university experience, so having an instrument to measure life satisfaction is very important.

The findings confirm that the SWLS presents adequate indicators of reliability and validity in young university students in Ecuador, which positions it as a useful and appropriate tool for the evaluation of satisfaction with life in this context.

Furthermore, the results obtained support the use of the SWLS as a brief, easy-to-administer, and freely available instrument for measuring satisfaction with life in university students. Its implementation in research and practice within the university setting can significantly contribute to the assessment of this construct, allowing the design of interventions aimed at improving and promoting satisfaction with life and, ultimately, the well-being of students in the academic context. This in turn would allow the development of strategies that strengthen their psychological and academic adjustment.