Introduction
The scientific study of subjective well-being has raised great interest in the last 15 years partly due to its association with physical and psychological health. In this regard, there are studies that show a positive association between subjective well-being with longevity and various health outcomes (Diener, Pressman, Hunter, & Delgadillo-Chase, 2017). In this topic, a recent meta-analysis conducted on 29 studies found a medium-sized positive association between health status and subjective-wellbeing (Ngamaba, Panagioti, & Armitage, 2017). Similarly, other studies have reviewed the literature demonstrating that subjective well-being is associated with greater health and prolonged life (Hernández, et al., 2018). In detail, Hernández et al., (2018) summarize evidence in the scientific literature linking psychological well-being (including positive affect, optimism, life meaning and purpose, and life satisfaction) and physical health in the form of mortality, chronic disease incidence and progression. Furthermore, evidence in the literature show an association between subjective well-being and self-reports of mental health (e.g. Dobewall, Tark, & Aavik, 2018).
Subjective well-being reflects an overall evaluation of the quality of a person’s life from her or his own perspective (Diener, Lucas, & Oishi, 2018). It included both a cognitive component, such as life satisfaction, and an affective component regarding the emotional responses to ongoing life events (Diener, Oishi, & Tay, 2018). In that distinction, the cognitive component refers to the cognitive evaluation, positive or negative, that individuals make about their own life, globally or by specific domains and is usually referred to as satisfaction with life. On the other hand, the affective component includes the presence of positive affect and the relative absence of unpleasant emotional states, both momentary and long-term (Diener, et al., 2017), which is usually linked to subjective happiness. Both components have been studied in different stages of life, with a particular interest in groups that may be particularly vulnerable to developmental and environmental demands.
Two of these groups of special interest in the research of subjective well-being are adolescents and young adults, partly because of the stressful life events that are characteristic of these life stages, but also due to the presence of risk behaviours that may become potential hurdles to achieve higher levels of subjective well-being. Inspired by these and other challenges that are inherent to adolescence, Huebner (1994) developed the Multidimensional Students’ Life Satisfaction Scale (MSLSS) to provide a profile of the young population in five dimensions: family, school, friends, self and neighbourhood. Nowadays, the scale is used widely in research of populations of adolescents and university students and has been adapted to the Chilean context in the later population (Schnettler, et al., 2017).
One of the most frequently mentioned limitations when studying subjective well-being is the influence of social desirability. Milhabet, Le Barbenchon, Cambon, and Molina (2015) define social desirability as the types of affects that individuals elicit from others, or the manners in which individuals satisfy other people’s principal motivations. Under this definition, social desirability conveys the “likeableness” that individuals attribute to others in their relationships. Social desirability would pose a limitation to the measurement of subjective well-being inasmuch as participants may exaggerate or fake their responses in an attempt to impress the researchers (Fastame, Hitchcott, & Penna, 2017). In this regard, studies have found a weak but significant association between social desirability and satisfaction with life a study with university students (Miller, Zivnuska, and Kacmar, 2019), and young adults (Caputo, 2017).
Considering the influence that social desirability may have on reports of subjective well-being in adolescents and young adults, the present study aimed to determine the association between satisfaction with life and subjective happiness while controlling for social desirability and to test if general life satisfaction can be predicted by satisfaction with different life domains while controlling for social desirability.
Method
The study followed a quantitative design, non-experimental, ex post facto, and cross-sectional study (Hernández, Fernández, & Baptista 2010). Participants were asked via email to answer an online questionnaire. The online questionnaire included informed consent form mentioning general aims of the study, as well as its ethical guidelines including their right to refrain from participating and stop their participation at any point, and the anonymity and confidentiality in the treatment of their answers.
Participants
Sample is comprised of 279 university students from various Chilean universities, specifically: 58 students from the Universidad de Tarapacá (20.8 %), 39 students from the Universidad de Chile (14.0 %), 59 students from the Universidad de Talca (21.1 %), 103 students from the Universidad de La Frontera (36.9 %), and 20 students from the Universidad de Magallanes (7.2 %). Participants’ age ranged between 18 and 39 years old (M = 22; SD = 2.8). Regarding gender distribution, there were 103 male participants (36.9 %) and 176 female participants (63.1 %).
The study used a non-probabilistic sampling method by convenience, recruiting students from different university programs that were accessible to the research team via academic collaborators. Regarding distribution by program of studies: 92 participants were enrolled in Commercial Engineering (33.0 %), 86 in Psychology (30.8 %), 61 in Agronomy (21.9 %), 16 in Public and Audit Accountant (5.7 %), 15 in Biotechnology (5.4 %), and 9 in Veterinary Medicine (3.2 %). With respect to distribution of participants by the socioeconomic level: 37 students came from a high socioeconomic level (13.3 %), 81 from a high to medium level (29.0 %), 85 from a medium level (30.5 %), 57 from a medium to low level (20.4 %), and 19 from a low socioeconomic level (6.8 %).
Instrument
Means, standard deviations and reliability of the measures can be found in Table 1. The questionnaire included the following measures.
The Socioeconomic Level was measured as proposed in Adimark (2000), with the Socioeconomic Level Scale (ESOMAR). This measure combines the highest educational level achieved by the person with the highest income in the family group (e.g. high school diploma, bachelor’s degree) with the type of job this person does (e.g. occasional zero-hour contract jobs, mid-level administrative job). Participants are then classified into one of five possible socioeconomic levels: high, high to medium, medium, medium to low, and low socioeconomic level.
Satisfaction with Life Scale, or SWLS (Diener, Emmons, Larsen, & Griffin, 1985), which contains five items that evaluate global cognitive judgements that individuals make about their own life, using a Likert format. Schnettler et al. (2017) reported adequate reliability indexes for Chilean university population (. = .88). Reliability index for the present study was of .88.
The Subjective Happiness Scale, developed by Lyubomirsky and Lepper (1999), consists of a general measure of subjective happiness comprised by four items in Likert format. Two of the items ask the participants to characterize themselves, while the other two present brief descriptions and ask participants to what extent these descriptions apply to them. Schnettler et al. (2015a) reported adequate reliability indexes for Chilean university students (. = .88). Reliability index for the present study was of .80.
The Multidimensional Student’s Life Satisfaction Scale, or MSLSS, developed by Huebner (1994), and revised by Huebner, Laughlin, Ash and Gilman (1998), aims to provide a multidimensional profile of students’ life satisfaction in five different domains: Family, Studies, Friends, Neighbourhood, and Self. The instruments consider 40 items, some of them reverse-coded, on Likert format. González (2014) reported adequate reliability indexes for the overall scale (α = .90), and by domains: Family (α = .89), Friends (α = .86), Neighbourhood (α = .84), Educational Organization (α = .79), and Self (α = .86). Reliability indices for the present study ranged from .85 to .90.
Social desirability was measured with a brief version of the Marlowe-Crowne Social Desirability Scale. The brief version was developed by Saiz, Alvarado, de la Barra, Gempp, and Pezo (1993), and comprises ten items from the original scale in which participants evaluate if each of the statements reflects or not their own behaviour on a true or false (dichotomous) scale, with an acceptable reliability index of .70. High scores in the scale reflect high levels of social desirability and this brief version showed an acceptable reliability index of .60 (Mladinic, Saiz, Díaz, Ortega, & Oyarce, 1998). Reliability index in the present study was of .67, obtained through the ordinal alpha coefficient proposed by Zumbo, Gadermann, and Zeisser (2007) as a better estimator of reliability in binary data. The calculation of this coefficient was done following the guidelines and method proposed by Domínguez-Lara (2012, 2018).
Procedure
Data was collected through an online questionnaire on the QuestionPro service. Participants were invited via email to take part in the study by an academic of their respective university (a collaborator of the research team). Participants who agreed to take part in the study then received an email from the research team with a copy of the informed consent form and the link to answer the online questionnaire. The informed consent form explained in general terms the aims of the study and its ethical guidelines. Emphasis was made on the voluntary nature of the study, confidentiality of the data, anonymity of the participants, the expected absence of physical and psychological harm, and their right to stop taking part in the study at any point. Participants could opt to receive an incentive of $5,000 Chilean pesos ($8 U.S. Dollars, approximately) and they could opt to participate in a prize draw of two gift cards per university for the value of $100,000 Chilean pesos each ($160 U.S. Dollars approximately). Participants completed the online questionnaire in their own free time without supervision or knowledge of the academic who initially invited them. Data was collected over a three months’ period.
Analysis plan
The study included reliability analysis for the measures used followed by Runs Tests on the main variables of interest, partial correlations (p < .05) and multiple hierarchical regressions (p < .05). The multiple hierarchical regressions considered satisfaction with life as predicted variable, social desirability was entered in the first step of the analysis and the five domains of the MSLSS were entered as predictor variables in the second step. All analysis were done using the IBM SPSS Statistics v. 20 software.
Results
Prior conducting the main analyses of the study a Runs Test for detecting non-randomness was performed on the median of the main variables of interest, in the order data was collected. Results of these analyses show non-significant results for subjective happiness (Z = -.42, p = .68), social desirability (Z = -.15, p = .88), and for each of the dimensions of the MSLSS: satisfaction with family (Z = -1.59, p = .11), university (Z = -.17, p = .87), friends (Z = -.18, p = .86), neighbourhood (Z = -.87, p = .38), and with self (Z = -.60, p = .55). Conversely, results for satisfaction with life scale show a significant result (Z = -2.98, p ≤ .01), suggesting that participants’ responses for this variable are not randomly distributed above and below its median based on lower-than-expected number of runs obtained in this variable (R = 115).
The first specific aim was addressed through partial correlation analyses between satisfaction with life and subjective happiness while controlling by social desirability. Results of the zero-order correlations showed positive weak associations of social desirability with satisfaction with life (r(277) = .18, p < .001), and with subjective happiness (r(277) = .25, p < .001), and a strong positive association between satisfaction with life and subjective happiness (r(277) = .64, p < .001). In a similar manner, results of the partial correlation showed that the association between satisfaction with life and subjective happiness remains similar when controlling for social desirability (r(276) = .63; p < .001). Results of the zero-order correlations and partial correlation can be found in Table 2.
To address the second specific aim of the study, a multiple hierarchical regression was carried out with overall satisfaction with life as the predicted variable. In the first step of the analysis, social desirability was entered as a co-variable and in the second step of the analysis; the dimensions of the MSLSS were entered as the main predictors (i.e. satisfaction with family, university, friends, neighbourhood, and self). Results showed that the models on both steps were significant; Model 1 [F(1, 277) = 9.181, p < .01, R2 = .03, R2 Adjusted = .03], and Model 2 [F(6, 272) = 26.633, p < .001, R2 = .37, R2 Adjusted= .36]. However, when looking at significant F changes, Model 2 resulted significant [F(5, 272) = 29.181, p < .001] which indicates that variables entered in the second step produce a significant increment in explanatory power. In Model 2, the variables that significantly predicted satisfaction with life were satisfaction with family [β = .22, t(272) = 3.93, p < .001], university, [β = .13, t(272) = 2.32, p < .05] neighbourhood [β = .22, t(272) = 4.04, p < .001], self [β = .25, t(272) = 4.18, p < .001]. Results of the multiple regression analysis can be found in Table 3.
Based on the results of the multiple regression analysis it can be concluded that when considered by itself, social desirability successfully predicts satisfaction with life, but when is considered along with satisfaction in different life domains, social desirability loses its predictive power while satisfaction with family, university, neighbourhood, and self, emerge as predictors of general life satisfaction.
Discussion
Considering results of the study, it is concluded that satisfaction with life and subjective happiness are associated while controlling for social desirability, thus supporting the first hypothesis of the study. More precisely, results of the zero-order correlations suggest weak positive associations between social desirability with satisfaction with life and with subjective happiness, but results of the partial correlation analysis show a significant strong positive association between satisfaction with life and subjective happiness after controlling for social desirability.
Regarding results of the zero-order correlations showing an association between social desirability and subjective well-being, these results suggests that participants’ tendency to present themselves in a more positive way to earn others’ approval may have led them to report high indices of subjective well-being, thus undermining honesty and/or accuracy of their responses. These results go in line with hat has been proposed in the literature (Caputo, 2017). Conversely, results of the partial correlation analysis show that the association between satisfaction with life and subjective happiness is significant even when controlling for social desirability, suggesting a low shared variance between social desirability with satisfaction with life and subjective well-being. In conclusion, despite social desirability being associated with subjective well-being measures, low shared variances with them translate into participants tending to present themselves in an authentic way, with no need to show a more positive image of them than it really is when assessing their subjective well-being (Moral de la Rubia, García, & Antona, 2012).
Regarding results of the second aim of the study, a multiple regression analysis showed that social desirability predicts overall life satisfaction when considered on its own, but when including the satisfaction with different dimensions of life as predictors, social desirability loses its predictive power. These findings point out at the satisfaction with family, university, neighbourhood, and self, as predictors of overall satisfaction with life. This coincides with Diener’s (1994) approach that domains closer to the personal life of individuals are the ones that influence their subjective well-being the most.
A possible explanation for social desirability losing its predictive power over life satisfaction comes from the notion that social desirability may be minimized when data is collected through an online questionnaire, without the physical presence of an interviewer, which may result in the participants feeling less compelled to present themselves in a more positive way. This explanation is supported by Caputo (2017) who suggests that the measurement of well-being through online surveys may be affected by self-deception, i.e. the unintentional propensity to portray oneself in a favourable light, rather than by intentional falsification (which is mostly assesses by traditional social desirability scales, such as the Marlowe-Crowne Scale). Furthermore, these results are also supported by the literature showing that online questionnaires prompt participants to have higher levels of revealed information (e.g. Weisband & Kiesler, 1996), a better disposition to answer questions regarding sensitive information (e.g. Tourangeau & Smith, 1996), and lower levels of social anxiety and social desirability (e.g. Frick, Bachtiger, & Reips, 2001).
Another possible explanation for the findings regarding social desirability not affecting self-report measures of subjective well-being, is that having self-evaluations of subjective well-being that are unaffected by social desirability may be a characteristic behaviour of university students. This explanation comes from research done with university students where results report null associations between social desirability with variables that are –arguably- expected to be associated with it such as alcohol consumption (Kypri et al., 2016) academic dishonesty (e.g. Ferrari, 2005), self-complexity (e.g. Luo, Watkins & Lam, 2009) and athletic identity (e.g. Nasco & Webb, 2006). Considering this, it may be that university students’ assessments of their subjective well-being is not affected by social desirability when considered along with other variables.
Regarding limitations of the study, the relatively low reliability score obtained in the social desirability scale emerges as the main limitation. Considering the consensus establishing 0.7 as the cut-off point for reliable scales (Kline, 1999), a revision of the scale used to measure social desirability is suggested by assessing its reliability and validity, and comparing it with similar measures such as the Lie Scale of the Eysenck Personality Questionnaire or the social desirability scale SDS-17. Testing this scale will allow to conclude whether the measurement used is the optimal for these types of studies.
Another limitation of this study comes from the use of self-report measures as a source of information, which may lead to biased answers. In this matter, it is possible that information reported by participants were subject to selective memory bias, attribution bias, and/or exaggeration. It is recommended that future studies consider external sources of information (e.g. peer evaluation) or alternative methodologies (e.g. diary or longitudinal studies), to contrast the participants’ responses to subjective measures of well-being.
A third limitation of this study is the sampling method used, through which only a few university programs were selected, thus limiting the generalization of the results to other university student populations. Despite attempts to include students from a wider range of programs, the final sample resulted relatively homogenous, which may translate into experiences that are not representative from the general population of university students. In this regard, variables such as stress levels, uncertainty, and/or exigency may differ between programs, which in turn may affect the levels of satisfaction with the university domain enough to make the overall satisfaction with life fluctuate. Considering this limitation, it is suggested that future studies include a wider spectrum of university programs to extrapolate the present results and conclusions to university students of diverse areas of knowledge.
Regarding suggestions for future studies, it is suggested the inclusion of other variables that may be relevant to the prediction of subjective well-being, such as age and socioeconomic level (Kulaksızoğlu & Topuz, 2014), family support (Schnettler et al., 2015b), and autonomy support from family, friends and romantic partner (Ratelle, Simard & Guay, 2013). Another suggestion for future studies regards the comparison of the effects of social desirability on subjective well-being in different forms of administration, i.e. face to face and online surveys. This would help shed light on the possible differential effects of social desirability, and the possible benefits of one way of administration over the other in the amelioration of social desirability effects. Even if the prediction of subjective well-being was not the main aim of the present study, the inclusion of such variables may aid in the generalisability of the results and conclusions to real-life scenarios.
Conclusions
The present study explored the association between social desirability and subjective well-being. Overall, results indicate weak and positive associations between desirability is associated with the affective and cognitive dimensions of subjective well-being, subjective happiness and satisfaction with life respectively, but partial correlation analyses indicate that social desirability does not affect the association between the two. Similarly, social desirability predicts overall life satisfaction when considered by itself, but social desirability loses its predictive power when the satisfaction with life’s domains are included in the prediction. With these results in consideration it is concluded that despite social desirability being associated with subjective well-being in university students, when other variables are taken into consideration, social desirability loses its predictive power over subjective well-being. The notion that students’ evaluations of their own well-being are not influenced by social desirability, may be helpful in the measurement of subjective well-being in future research.