Validation of the Social Connectedness Scale in Argentinean adult sample

The feeling of social connectedness is an extremely important factor for the psychosocial well-being of individuals. Experiencing closeness, belonging, and mutual acknowledgement with the people who are part of the social circle, contributes to the establishment and maintenance of positive and stable relationships over time, as well as numerous benefits regarding the quality of life and mental health of individuals. In order to contribute to the field of study of interpersonal relationships, this study aims to adapt and validate the Social Connectedness Scale for Argentinean adults. The sample consisted of 399 individuals between the ages of 18 and 79 who completed the scale which was previously translated and submitted to expert judges. The results confirmed the one-dimensional structure of the scale, invariant by sex, with good internal consistency (ω = .92). Furthermore, evidence of convergent and discriminant validity was provided through correlations with measures of extroversion ( r p =.37) and social anxiety ( r p = -.61). It is concluded that the measures provided by the Argentinean version of the social connectedness scale are both valid and reliable to be used in research, as well as in clinical settings in Argentinean population.


INTRODUCTION
The feeling of belonging and social acknowledgement is an experience that gives all human beings satisfaction and purpose to their lives (Stavrova & Luhmann, 2016). This is explained by the fact that social connectedness is a fundamental human need, and it is becoming increasingly clear that this feeling bestows benefits for physical and mental health (Hutcherson et al., 2008).
Social connectedness is defined as the subjective psychological bond that people feel in relation to other individuals and social groups (Haslam et al., 2015). Van Bel et al (2009) consider it as an experience of belonging and familiarity based on personal satisfaction in terms of quantitative and qualitative evaluations of ties and relationships. Lastly, Lee and Robbins (1995) define it as an attribute of the self that reflects cognitions of a lasting interpersonal relationship with the social world as a whole. In other words, it is a cognitive structure that represents patterns of attachment of individuals (Lee & Robbins, 1998). This is based on the accumulation of individual experiences of proximal and distal relationships that provide both a sense of personal identity, as well as a sense of place in society. Nevertheless, despite the fact that the feeling of social connectedness is a stable attribute, it is not presented as rigid and static. Instead, it is nourished by the interactions and experiences that the individual considers to be the most significant and with the greatest impact throughout their lifetime (Williams & Galliher, 2006). Therefore, it is not possible to ascertain that social connectedness is a trait, or a manner of bonding, because the subjective perception of belonging (Diendorfer et al., 2021), although relatively stable, could also be dynamic. This assumption was initially demonstrated by Mitic et al. (2021), who in their proposed model demonstrated that there are multiple variables that set up social connection, in addition to the fact that cognitions and sense of self are based on many internalized factors that are supported and constantly rectified by the significant social environment. In addition, there are several studies that report how social connectedness varies throughout the different life stages, increasing towards old age (Ang, 2016;Luong et al., 2011;Moran et al., in press).
Most people who feel connected to others or who have a sense of belonging to a social group will easily identify with other people, perceive others as friendly and approachable, and participate in social groups and activities (Lee et al., 2008;Satici et al., 2016). These individuals can reevaluate relationships, make friends, and seek new social ties. On the other hand, people with low social connectedness tend to feel interpersonal distance from others and from the world in general. They may also feel misunderstood and have difficulty relating to other people (Lee & Robbins, 1995). When this state is persistent and generalized, it is psychologically disturbing and potentially debilitating, with consequences such as the inability to maintain relationships and avoidance of social activities for fear of rejection (Eraslan-Capan, 2016;Fatima et al., 2017;Tomova et al., 2021).
In their review study, Hare-Duke et al. (2019) demonstrated an increase in research on social connectedness given its relevance to psychological well-being and mental health today. Furthermore, almost two decades ago, Lee and Robbins (2000) remarked on the worrying concerns that manifested themselves in psychotherapeutic consultations related to the lack of belonging, the lack of group participation, the lack of relationships, and even the lack of connectedness with society. These factors were leading to an increase in social isolation and mistrust towards those who were outside the already established social networks.
Moreover, studies have shown the relationship between social connectedness and several disorder and pathologies (Rossi et al., 2012), including depression (Nguyen et al., 2019) and anxiety (Grover et al., 2018;Kavanagh el al., 2017). It is also considered a variable associated with physical illness (Holt-Lunstad et al., 2010). Additionally, Faro et al. (2019) found that social connectedness has an inverse and significant effect on the suffering of internalized mental problems, which in turn mediates the effect of parenting practices on these disorders. On the other hand, Hashim et al. (2019) found that social connectedness is an important predictor of negative affect linked to crime, and Savci and Aysan (2019) demonstrated the central role of social connectedness in addictions to multimedia devices. Accordingly, Duru and Poyrazli (2011) showed that high levels of social connectedness are associated with greater social adjustment and fewer difficulties in adjusting to the environment, even when presented with significant cultural differences, which has also been supported by other studies (Yeh & Inose, 2003). Lastly, recent research has also shown the effects of social connectedness to general life satisfaction (Taylor et al., 2020;Zhang et al., 2020), revealing the scope that this construct has in the various domains of individual development.
It was in this framework that the authors identified the need to have instruments intended to evaluate social connectedness and to be able to continue advancing in its study. According to the review by Hare-Duke et al. (2019), there are 21 scales that measure social connectedness from different perspectives, but few of these have studies on their psychometric properties. The authors found that most of the instruments were evaluated for internal consistency, leaving aside other important properties, such as content validity.
The Social Connectedness Scale (SCS) is one of the most widely used instruments, created in the United States by Lee and Robbins (1995). This scale measures how individuals cognitively construe interpersonal closeness with others in their social world (e.g., friends, peers, society). In its initial version, this was an 8-item scale with a one-dimensional factorial structure. However, despite its high internal consistency and its construct validity, the scale had certain psychometric limitations, such as, all of its items being written as negative affirmations (e.g., "I feel disconnected from the world around me"), which may have caused a response bias evidenced in the asymmetric distribution of the responses. As a result, Lee at al. (2001) wrote new items, in positive sense, which were incorporated into the initial version. Additionally, they modified some of the existing items to more accurately reflect the slight deficiencies in the need for belonging. This revised version consists of 20 items which respond to a unifactorial structure with adequate internal consistency (α = .92). In addition, several studies have supported the external validity of the scale, relating it to other personality variables such as extraversion, life satisfaction and stress (Lee et al., 2008), as well as anxiety and self-esteem (Lee & Robbins, 1998).
Although the SCS has been used in multiple studies from various cultures and countries, only two psychometric adaptation and validation studies have been reported. The original version was adapted for the Turkish population by Duru (2007), and the revised version was adapted for Italians by Capanna et al. (2013). In both validation studies, evidence of a unidimensional structure was provided through an exploratory factor analysis, and external validity was demonstrated through the relationships with other variables. The internal consistency and stability of the scale was also substantiated.
Considering the psychometric properties of the Social Connectedness Scale, and the importance of continuing to expand our understanding of this variable, the purpose of this study is to adapt and validate this instrument for the use of research among the Argentinean adult population, providing evidence of content, internal structure, external validity, and internal consistency. Having this tool will facilitate future research on interpersonal behavior in our country, a field of study that is simultaneously growing worldwide.

Sociodemographic data questionnaire
An ad hoc questionnaire was created to collect sociodemographic information from the participants; such as, gender, age, level of education, profession or employment, and marital status.
The Social Connectedness Scale measures the psychological sense of social belonging. Its revised version is made up of 20 items (10 positive and 10 negative) that are answered using a 6-point Likert scale (1 = Strongly disagree to 6 = Strongly agree) and high social connectedness is defined by mean scores of items equal to or greater than 3.5. Studies with American samples reveal a one-dimensional structure with excellent reliability index (α = .94).
For the scale adaptation, the translation was carried out using the reverse translation method, in which three bilingual experts translated the items from English to Spanish, and later, two other experts translated from Spanish to English. The author of the test and his team (Lee, personal communications) reviewed the final versions in English and Spanish, and the pertinent modifications were made. These modifications were made preserving the meaning of each statement while respecting the idiomatic and cultural particularities. (Goldberg, 1999) This subscale has 24 items written in sentences that state typical behaviors that allow the Extraversion factor to be measured according to the Big Five Factors Theory (McCrae & Costa, 1992). It uses a Likert-type scale with five response options ranging from Strongly Disagree with this description of myself, to Strongly agree with this description of myself, asking the participants to detail the precision with which they feel each statement describes their personality. The version adapted and validated for the Argentinean population will be used. This version presents a satisfactory internal structure analysis and a reliability index of α = .84 (Ponce, 2016).

Social Phobia and Anxiety Inventory -Brief form, SPAI -B (Garcia-Lopez et al., 2008)
This questionnaire evaluates symptoms of phobia and social avoidance. It is made up of 16 items and a 5-point rating scale. The participants will have to answer to the frequency with which the situation described in the statement occurs (1 = Never, 5 = Always). An exploratory and confirmatory factor analysis demonstrated a unifactorial structure with excellent internal consistency in Argentinean samples (α = .85) (Moran et al., 2019).

Procedures
The data was collected online using google forms platform. The invitation to participate was distributed randomly throughout adult social network accounts within the Argentinean territory. Participants were informed about the purpose of the study prior to participation, and entry to the questionnaire was not enabled until they expressed their agreement with the informed consent, wherein the goal, the voluntary nature of their participation, and the confidential nature of the study was described. The ethical issues considered were approved by Universidad Siglo 21 Research Project Evaluation Committee.
Concerning data analysis, first, expert judges evaluated the cultural, semantic, syntax and content of the items. For this purpose, an item evaluation protocol was distributed to three experts in the evaluation of constructs related to interpersonal relationships, and to experts in the construction and adaptation of psychometric tests. Aiken's V coefficient was used to determine the agreement between judges regarding the quality of the items. Additionally, to calculate the confidence intervals of each coefficient, the score method was used since it does not depend on the normal distribution of the variable, and it is highly accurate (Soto & Segovia, 2009).
Second, a confirmatory factor analysis was carried out using Mplus software and the diagonally weighted least squares (DWLS) estimation method, which is suitable for ordinal type variables. To evaluate the fit of the one factor model, the Comparative Fit Index (CFI), the Tucker-Lewis Index (TLI), the Root Mean Square Error of Approximation (RMSEA) and the Weighted Residual Mean Square (WRMR) was used. Values between .90 and .95, or higher for CFI and TLI, are considered acceptable to excellent fit. Values between .05 and .08 are expected for RMSEA and values lower than 1 for the WRMR (Yu & Muthen, 2002). To assess whether the factorial structure differs according to sex, an invariance analysis was performed for the configural model, the metric model, and the scalar model.
Third, to analyze the internal consistency of the scale, the composite reliability index (ω) was calculated. When structural equation modeling is performed, many authors recommend using this index, instead of Cronbach's alpha, because it is based on the item weights rather than covariances, thus allowing better estimations of latent variables reliability (Dunn et al., 2014;Padilla & Divers, 2016;Peterson & Kim, 2013).
Finally, the relationship between the test and other variables was analyzed using the Pearson correlation coefficient, applying Cohen's interpretation criteria (Cohen,1988) that states that a small effect size is considered when the correlation is below .10, average around .30, and large when greater than .50.

Content validity
First, the agreement between judges was calculated in relation to the quality of the translated items, of which 17 obtained satisfactory results, presenting V coefficients higher than .80 with confidence intervals (CI) between 95% and 99% (Table 1). Moreover, two items presented V coefficients of .75. These items were retained since they were considered relevant as indicators of the construct. Lastly, one item (item 20) presented a value of .42. This item was revised following the experts' recommendations, who subsequently re-evaluated it, resulting in V = 1 (CI 99% [.65-1.00]).

Descriptive analysis
Regarding the data, there were five univariate atypical cases (Z scores > ±3.29) and 18 multivariate atypical cases (Mahalanobis, p < .001), which were all retained since the elimination of these could mean a limitation in the generalization of the results (Hair et al., 2006). A descriptive data analysis was performed and descriptive statistics of the items were calculated. Additionally, the frequency of the responses was analyzed, finding that all response options were selected for all items. Regarding the distribution, as seen in Table 2, the skewness and kurtosis values were adequate and excellent according to the George and Mallery (2001) criteria, which conform to a normal type distribution.
The total scores presented a normal distribution given their skewness (-0.8) and kurtosis (0.18) values. The total mean score was 88.07 (SD = 17.69) and the mean item score was 4.40 (SD = 0.88). Regarding the social connectedness scores of the sample, it was observed that 85% of the participants presented high levels of social connectedness. Furthermore, differences in social connectedness according to sex were evaluated, finding that women had a higher mean than men, but these differences were not significant (t = .86; p = .39).
The invariance analysis of the configural, metric and scalar model by sex obtained satisfactory fit indices (Table 3), but the X2 differences between the base model (configural) and the metric model were significant. However, the chi-square test is sensitive to sample size and many researchers suggest avoiding the evaluation of measurement invariance based only on this criterion. Instead, they propose alternative fit indices depending on the invariance evaluated (Hong et al., 2003;Putnick & Bornstein, 2016).
Thus, based on Chen's (2007) criteria, it is indicated that in the metric model there were no significant alterations in the fit since there was no decrease greater than 0.10 in the CFI or an increase greater than 0.015 in the RMSEA. Lastly, the differences with the scalar model were not significant, thus it is determined that the scale is invariant for both sexes.

External validity
The relationship between the scale and other variables was evaluated. In the discriminant validity study, SCS scores were significantly and inversely correlated (r = -.61; p ≤ .001) with social anxiety scores. In the convergent validity study, the correlation with the extraversion scores was significant, moderate, and positive (r = .37; p ≤ .001).

DISCUSSION
The main purpose of this study was to adapt the Social Connectedness Scale to Argentinean adult samples and to evaluate the psychometric properties of this version. According to the standards for psychological tests proposed by American Educational Research Association [AERA] et al. (2014), the translation of a test from one language to another does not guarantee that the psychometric properties of the original version, nor some characteristics of the items, will be maintained in the new version, for which it is recommended that evidence of validity and reliability of the translated versions be collected and reported (International Test Commission, 2017).
Based on this, the scale was translated adapting the expressions to the Argentinean context whilst preserving the meaning of each item. The translated version was subjected to the evaluation of experts who indicated that the expressions and situations presented were culturally appropriate and clear at a syntactic and semantic level. This process not only provides content validity to the scale, but also allows to inquire about the compatibility of the experiences and meanings contained in each item of both versions, properties which were scarcely evaluated in the previous studies of different scales measuring social connectedness (Hare-Duke et al., 2019).
In regards to the psychometric properties of the translated scale, the data supports a one-dimensional structure coinciding with the structure presented in the original version proposed by Lee et al. (2001). The data also demonstrates the scale is invariant according to sex, both in the factorial structure, the regression weights of the items, as well as the intercept values. These results show that it is possible to determine that the SCS provides a measure of social connectedness as a global construct. However, other scales define it as a concept made up of different dimensions, for example, Relationship salience, Shared understandings, Knowing each other's experiences, Feelings of closeness, and Dissatisfaction with contact quality (Van Bel et al., 2009). In addition, there are other social connectedness scales that present factorial solutions where each dimension represents the social group the item is about (Carroll et al., 2017;Karcher & Sass, 2010). In this sense, future research and theoretical developments could further expand on the nature of the construct.
Regarding the reliability of the SCS-R, the internal consistency of the Argentinean version was satisfactory, obtaining a similar result to that obtained in American samples (Lee et al., 2001), and slightly better than in the Italian (Capanna et al. al., 2013) and Turkish samples (Duru, 2007). These values are also consistent with those found in other studies in which the scale has been used (Grieve et al., 2013;Satici et al., 2016;Sinclair & Grieve, 2017;Yeh & Inose, 2003). Even though the interpretation is similar, in this study we calculated the composite reliability, while other psychometrics studies calculated Cronbach's alpha. According to Peterson and Kim (2013), composite reliability should be calculated when structural equation modeling is used. In this approach, construct weights are allowed to vary offering better estimates of true reliability, in contrast to Cronbach's alpha which assumes equal score variance (Bacon et al., 1995).
Moreover, convergent-discriminant validity analyses were performed. A moderate and positive correlation was found between social connectedness scores and extraversion scores. In a study on subjective well-being, Lee et al. (2008) had already shown psychometrically the relationship between social connectedness and extraversion, and empirically reaffirmed that both constructs appear to be similar, but are theoretically different, since social connectedness does not include the motivational component of the behavior. The nature of the relationship between social connectedness and extraversion lies in the fact that the latter it is a personality trait, constituting a factor that predisposes an individual to certain behaviors, choices and interests. From an instrumental approach, extroverts seek to a greater extent to establish relationships and spend time with others, which broadens the possibility of exchanges and the creation of social ties that reward and contribute to the sense of social connectedness (McCrae & Costa, 1991). However, this does not indicate that they present greater relationship satisfaction when compared to introverts (Lucas & Baird, 2004). In fact, it is about the perception of belonging and availability of others (Swickert et al., 2002).
On the other hand, negative and strong correlations were found between the SCS-R, phobia and social anxiety scores, accounting for the inverse relationship between both variables, and consequently, providing evidence of discriminant validity. Some of the main consequences ensuing from social anxiety are observed in the interactions with others, negatively affecting the establishment and maintenance of positive relationships, the integration and participation in social groups, as well as acceptance by others (Rubin et al., 2009), which impacts the development of the sense of belonging, acknowledgement and satisfaction with the quality of relationships that the person establishes. Along these lines, there are numerous studies that support the negative relationship between social connectedness and social anxiety (Fatima et al., 2017;Kavanagh et al., 2017;Savci & Aysan, 2019).
In addition to the studies mentioned, further evaluations were carried out to observe whether there were differences according to sex. As a result, it was found that there are no significant differences between men and women, concurring with results reported in previous psychometric studies (Capanna et al., 2013;Lee et al., 2001), as well as in other empirical studies (Lee et al., 2002;Lee & Robbins, 2000). These findings allow us to infer that both men and women value social connectedness equally, although it does not imply that they value it in the same way (Lee & Robbins, 2000). In fact, Baumeister and Sommer (1997) stated that women develop social connectedness through intimacy and physical closeness, while men develop it by social comparison to others.
Regarding the levels of social connectedness in the sample, it was found that the majority (85%) tend to feel socially connected. This percentage, as well as the average scores, were found to be similar to the values reported by Lee et al. (2001) in American samples and by Capanna et al. (2013) in Italian samples. Nevertheless, to extend these conclusions to the Argentinean population, it is necessary to expand the sample size and to use probabilistic sampling methods in order to achieve greater representativeness (Otzen & Manterola, 2017). Moreover, the cut-off points to establish the levels of social connectedness were determined by the authors based on the range of scores that the test yields. In this case, the higher the score, the higher the level of social connectedness. Seeing as this is a construct that cannot be interpreted based on a single criterion, it would be appropriate to establish cut-off points that allow an interpretation based on norms (AERA et al., 2014).
Another limitation of this study refers to the data collection carried out online, which excludes individuals who do not participate in social media (such as Facebook or Instagram) or who do not have internet access. Therefore, future studies on the psychometric properties of the scale should include these individuals, as well as carry out evaluations that can examine the differences between online and face-to-face, pencil and paper administration. Lastly, it is essential to carry out further psychometric studies on the scores provided by the scale, for example, stability analysis, analysis between clinical and non-clinical groups, post intervention analysis, and creation of norms for the establishment of cut-off points for interpretation.
Nevertheless, with the results obtained in this study, it is now possible to use the Social Connectedness Scale to carry out further research among Argentinean adults, making progress in the field of study of interpersonal behavior and understanding of this variable in Latin America.