General self-esteem is one of the most widely studied constructs in social sciences (Bachman, O’Malley, Freedman-Doan, Trzesniewski, & Donnellan, Reference Bachman, O’Malley, Freedman-Doan, Trzesniewski and Donnellan2011), and its measurement one of the most amply discussed topics in Psychology (Rosenberg, Schooler, Schoenbach, & Rosenberg, Reference Rosenberg, Schooler, Schoenbach and Rosenberg1995). Since its inception with the tenets of William James in 1890, self-esteem has been approached from different perspectives. In 1965, Rosenberg considered a component of self-concept and defined it as a single set of thoughts and feelings about the value or importance of each individual; in other words, it is the overall positive or negative attitude people have of themselves. Rosenberg described a person with good self-esteem as someone who defines him/herself as valuable, and appreciates his/her merits, but also recognizes his/her faults. More recent studies from different theoretical perspectives complement and/or postulate the definition of general self-esteem. For instance, González-Pienda, Núñez, González-Pumariega, and García (Reference González-Pienda, Núñez, González-Pumariega and García1997) report that general self-esteem is the way in which individuals appreciate themselves due to the social feedback they receive in their various roles. Luciano, Gómez, and Valdivia (Reference Luciano, Gómez and Valdivia2002) also define self-esteem as verbal feedback (or assessment) which the person relates to his/her actions, and which is enhanced with the locus of control and the types of functions they perform.
Self-esteem is relevant in that it is a representative indicator of health and well-being, as well as an explanatory variable of human behavior (Rosenberg, Reference Rosenberg1965). For example, it has been found that low self-esteem – especially in teens – is a great predictor of deterioration in mental health (Marshall, Parker, Ciarrochi, & Heaven, Reference Marshall, Parker, Ciarrochi and Heaven2014) and is related to eating disorders (Brechan & Kvalem, Reference Brechan and Kvalem2015), suicide attempts (Wichstrom, Reference Wichstrom2000), difficulty developing positive support networks (Marshall et al., Reference Marshall, Parker, Ciarrochi, Sahdra, Jackson and Heaven2015), among others. In turn, it has been observed that high self-esteem is related, for example, to quality of life (Muñoz & Alonso, Reference Muñoz and Alonso2013), increased academic performance in adolescents (Fiz & Oyon, Reference Fiz and Oyón1998), adaptation to the social environment (Silbereisen & Wiesner, Reference Silbereisen and Wiesner2002), and emotional stability (Zeigler-Hill et al., Reference Zeigler-Hill, Holden, Enjaian, Southard, Besser, Li and Zhang2015). However, it is important to note that high or low self-esteem is not necessarily related to positive or negative effects on health, respectively (Cheng, Govorun, & Chartrand, Reference Cheng, Govorun and Chartrand2012).
Given the importance of self-esteem, a number of instruments have been proposed in order to measure it. The Rosenberg Self-Esteem Scale (RSES; Rosenberg, Reference Rosenberg1965) is the most widely validated scale (Gray-Little, Williams, & Hancock, Reference Gray-Little, Williams and Hancock1997) and the most widely used scale worldwide to evaluate self-esteem (Robins, Hendin, & Trzesniewski, Reference Robins, Hendin and Trzesniewski2001). The original RSES is a unidimensional scale that reports a global index of self-esteem. This scale has a Likert response scale, and its response options range from “strongly disagree” to “completely agree”. A high score on the RSES is interpreted as high self-esteem. The high level of acceptance of the RSES lies on the fact that it not only shows high levels of validity and reliability in different population groups (Huang & Dong, Reference Huang and Dong2012), the language used in the scale is simple, requires only basic reading skills, can be administered within minutes and the items are closely related to the construct (Gray-Little et al., Reference Gray-Little, Williams and Hancock1997). In the empirical literature exist a disagreement about the underlying factor structure of the Rosenberg Self-Esteem Scale. Inconsistency main problems are focused on whether or not the scale assesses global self-esteem as one factor (as Rosenberg proposed) composed of 5 items written in negative and 5 items in positive, or as two distinct constructs representing the positive and the negative aspects of the self-esteem (Hyland, Boduszek, Dhingra, Shevlin, & Egan, Reference Hyland, Boduszek, Dhingra, Shevlin and Egan2014; McKay, Boduszek, & Harvey, Reference McKay, Boduszek and Harvey2014).
To date, the RSES has been translated and adapted in several languages including: Canadian French, Persian, Chinese, Italian, Estonian, German and Portuguese. Additionally, although the RSES has been validated in Spanish for Spain, Chile and Argentina, and although it is a scale which is commonly used in Colombia (see: Harper et al., Reference Harper, Coleman, Perrin, Olivera, Perdomo, Arango and Arango-Lasprilla2014; Milanés & Gómez-Bustamante, Reference Milanés and Gómez-Bustamante2011; Perez-Olmos, Tellez, Velez, & Ibáñez-Pinilla, Reference Pérez-Olmos, Téllez Cruz, Vélez Traslaviña and Ibáñez-Pinilla2012; inter alia), there is currently no validated adaptation in Colombia. Besides, cultural comparisons of self-esteem have been widely used (Li, Delvecchio, Di Riso, Salcuni, & Mazzeschi, Reference Li, Delvecchio, Di Riso, Salcuni and Mazzeschi2015). In fact, implicit self-esteem does not allow for cultural comparison, although explicit self-esteem – addressed here – does (Falk, Heine, Takemura, Zhang, & Hsu, Reference Falk, Heine, Takemura, Zhang and Hsu2015). Thus, the use of unbiased evaluations between cultures is recommended for this purpose (Li et al., Reference Li, Delvecchio, Di Riso, Salcuni and Mazzeschi2015).
Consequently, the aim of this instrumental research, was to validate and adapt the Rosenberg Self Esteem Scale (1965) in Colombians, and also to show their factorial equivalence with the Spanish version.
Method
Participants
The sample of this study is divided into two groups:
The first group includes the experts; four psychologists with at least a master degree were selected – this is the number of grad-level experts recommended by Lynn (Reference Lynn1986), and they performed the cultural adaptation from the Spanish of Spain to Spanish of Colombia. All of these psychologists are Colombian nationals and residents, and they reported having lived at least two years in Spain. In addition, another four Colombian experts in psychology and/or psychometrics partook in the evaluation of the qualitative properties of the items (see procedure).
The second group was a Spanish and Colombian community sample. This group included an initial sample which consisted of a total of 1,797 participants. These participants were selected by convenience, trying to compensate for sex and age by selection. Of these – and in pursuance of inclusion criteria – a final sample of 1,139 participants that met the inclusion criteria were selected, including 633 Colombians and 506 Spaniards. Inclusion criteria in the study were being of age, accepting to partake in the study and having a Colombian or Spanish nationality, as appropriate.
Statistically significant differences were found between Spain and Colombia in terms of age, years of education, marital status and religion, as expected in community samples. No differences were observed in sex, sexual orientation and relationship. The age range for Colombia ranged between 18 and 72 years (Mean (M) = 32.21; Standard Deviation (SD) = 10.68) and in the case of Spain the range was 18 to 70 years (M = 33.93, SD = 12.71), t(1137) = 2.49, p < .01, d = .14. Following are the values observed with regard to years of study: t(1130) = 4.22, p < .01, d = 0.24; in Colombia (M = 16.72; SD = 2.81) and in Spain (M = 15.86; SD = 3.98). Differences in marital status were χ2(3, 1125) = 20.63; p < .01, η 2 = .016 (there were more Colombians in marriage by habit and repute – an uncommon choice in Spain). Following are the values for religion: χ2(5, 1139) = 186.71, p < .01, η 2 = .16 (Colombians attend a greater number of times to religious events). However, no significant differences were observed in terms of sex by country: χ2(1, 1124) = .37, p = .57; sexual orientation χ2(8, 1124) = 10.23; p = .24 or marital status χ2(8, 1124) = 10.23; p = .24.
Instruments
Sociodemographic information
Different sociodemographic characteristics of participants were evaluated using a self-administered semi-structured interview. Age, sex, educational level, sexual orientation (using the scale Kinsley), religion, and marital status were evaluated.
The rosenberg self-esteem scale (RSES; Rosenberg, Reference Rosenberg1965)
The starting point was the version validated by Martín-Albo, Núñez, Navarro, and Grijalvo (Reference Martín-Albo, Núñez, Navarro and Grijalvo2007) for Spain. The scale consists of 10 items assessing general self-esteem. Items 1, 3, 4, 7 and 10 are positive and items 2, 5, 6, 8, and 9 are negative. The questionnaire has a Likert response scale wherein items are answered on a four-point scale (1 = Strongly disagree, 2 = Disagree, 3 = Agree, 4 = Strongly agree). The total score therefore ranges from 10 to 40 points; higher scores indicate good self-esteem.
Sexuality scale (SS; Snell & Papini, Reference Snell and Papini1989)
The Sexual Self-esteem dimension of the short 15-item version (Wiedeman & Allgeier, Reference Wiedeman and Allgeier1993) validated for Colombia (Soler et al., Reference Soler, Gómez-Lugo, Espada, Morales, Sierra, Marchal-Bertrand and Vallejo-Medina2016) was used for this study. Sexual Self-esteem is the tendency to positively self-assess the ability to relate sexually with a partner. The Sexual Self-Esteem subscale were answered on a 5-option Likert scale ranging from: Disagree to Agree. Adequate Cronbach alphas (.82 in Colombia and .87 in Spain) have been observed in this study. A sample item is: “I think of myself as a very good sexual partner.”
Procedure
Cultural adaptation of the RSES was conducted following the guidelines for the adaptation of scales of the same language and cultures by Vallejo-Medina et al. (Reference Vallejo-Medina, Gómez-Lugo, Marchal-Bertrand, Saavedra-Roa, Morales and Soler2016), following the guidelines of Elosua, Mujika, Almeida, and Hermosilla (Reference Elosua, Mujika, Almeida and Hermosilla2014), Muñiz, Elosua, and Hambleton (Reference Muñiz, Elosua and Hambleton2013), as well as AERA, APA, and NCME, (2014). Adaptation was performed by four Colombian grad-level psychologists who reported having lived in Spain for at least two years.
The properties of the items in the adapted version of the scale were evaluated by four experts in psychometrics and/or psychology. The properties evaluated were: Representativeness: contribution of the item to the construct (global or general self-esteem). Comprehension: this property determines whether the item in its adapted version is properly understood. Interpretation: Level of ambiguity of the item. Clarity: how direct and concise the item is. Each expert was provided with the operational definition of the construct to assess –general self-esteem – and the battery of items created, and they were asked to evaluate the properties of each item, by means of a Likert scale, ranging from 1 (nothing … and “the feature”) to 4 (very … and “the feature”). Experts also contributed by providing an alternative wording where deemed necessary.
The battery of tests was administered simultaneously in Colombia and Spain between October the 23rd, 2014 and February the 24th, 2015. Sampling was opportunistic. The questionnaires were entered in Typeform© and could be filled out in PCs or laptops, tablets and smartphones. Distribution was carried out through personal contacts, Facebook© and Twitter©. For the convenience sampling, an online survey was used due to the similarity of answers observed between paper and online samples (de Bernardo & Curtis, Reference de Bernardo and Curtis2013).
Data analysis
A table of specifications of the items and the ICAiken program were used for qualitative analysis, thus obtaining the confidence interval for Aiken’s V. Criteria considered were those by Merino and Livia (Reference Merino and Livia2009), taking a score below .50 at the lower limit (CI = 95%) as the cutoff point for Aiken’s V and a criterion of inadequacy of the item. However, all comments of the experts were taken into account.
The SPSS 20.0 was used to analyze the psychometric properties of the items. The EQS 6.1 was used to calculate Factorial Invariance, which was evaluated progressively under a Mean and Covariances Structures (MACS) procedure (Byrne et al., Reference Byrne, Oakland, Leong, van de Vijver, Hambleton, Cheung and Bartram2009). To begin, the multivariate distribution of data was calculated using Mardia’s test, wherein values above 5 are indicative of non-normality. The test indicated results of 36.5 in Colombia and 47.4 in Spain; therefore the Maximum Likelihood Robust method (ML, Robust) was used as the estimation method as the parameter of normal distribution was not met. In terms of progressive factorial invariance, the following steps were evaluated: configural invariance: no restrictions in the model; metric – or weak – invariance: the factorial weights are restricted, thus evaluating the equivalence of the weight of each item regarding the factor; strong invariance: intercepts are restricted; and strict invariance: the variances of errors are restricted. The indices taken into account in order to evaluate the fit of the models were as follows: the Root Mean Square Error Approximation (RMSEA), the confidence interval – at 90% – and the Comparative Fit Index (CFI). Values below .08 for the RMSEA and greater than .95 for the CFI were considered indicative of good fit. A decrease no greater than .01 with regard to the least restrictive model was taken as evidence of invariance (Cheung & Rensvold, Reference Cheung and Rensvold2002). The Akaike Information Criterion (AIC) was also taken into account in the model selection, which indicates absence of FI should the increase with regard to the least restrictive model be considerable.
Results
Item analysis
Table 1 presents the qualitative evaluation performed by four experts in psychometrics and/or psychology on the 10 items of the Colombian version. The lower limit of 95% in Aiken’s V is always greater than .50. Therefore, it can be observed that there is adequate wording of the items of the Colombian Spanish version of the RSES.
Table 1. Evaluation of Characteristics of the Items of the Rosenberg Self Esteem Scale by Experts
Note: R = Representativeness; C = Comprehension; I = Interpretation; CL = Clarity; Exp = Expert; M = Mean; LL = Lower Limit; UL = Upper Limit.
Analysis of the items’ psychometric properties and reliability
Table 2 shows that the indicators analyzed are generally adequate. Both versions are reliable, and have with very similar values in both countries. Corrected item-total correlations (r it c) are always greater than .30, except in item 8 of the Colombian version. No significant increase in Cronbach’s alpha is observed if an item is removed, and the mean values of the items are slightly above the theoretical average of response (2.5). Furthermore, SDs near and/or slightly above 1 indicate adequate response variability.
Table 2. Psychometric Properties of Items in the Versions from Spain and Colombia
Note: M = mean; SD = Standard Deviation; r it c = item-total corrected correlation; α-i = Cronbach alpha if item is deleted; α: = Cronbach alpha.
Factorial invariance
Progressive factorial invariance was performed in order to evaluate the scale’s construct validity and test factorial equivalence between versions of Spain and Colombia in four different models (M1, M2, M3 and M4) (Table 4). Figure 1 and Table 3 shows the standardized results of the configural model, associated with factorial weights (λ), errors of the items and item variance of all the models. Analysis began by assessing the model’s most basic level: the configural level – without restrictions.
Figure 1. Path diagram of the standardized configural invariance models of the Rosenberg Self-Esteem Scale in Colombia and Spain. SE = General Self-Esteem; SE-P = Positive Self-Esteem; SE-N = Negative Self-Esteem; R 2 = Variance of the item which is explained by the factor. The intercepts have been removed for the sake of visual clarity (V999).
Table 3. Standarized factor loadings (λ), errors, and % of explained variance for the Spain and Colombia configural model
Note: SE = Self-Esteem; SE-P = Positive Self-Esteem; SE-N = Negative Self-Esteem; λ = Standarized wheight item-factor; R 2 = explained variance.
* Second order λ all above .85.
At this level M2 and M4 were not given any further consideration due to the poor adjust observed in this basic level. Model 3 has shown good fit index, but invariance could not be defended in the second level: weak invariance. Thus, M1 has in this sample both the best adjustment and the strongest invariance, and has the lowest starting AIC value. As it can be observed in Table 4, this model reaches good fit indexes for all levels. But item 8 was not constricted in the weak invariance as recommended by the modification indexes. Once this modification was done, we observed a strong invariance level. No strict invariance was observed for this model.
Table 4. Goodness-of-Fit Indexes for the Different Invariance Models and Model Selection
Note: S-B χ2 = Santorra-Bentler Scaled Chi-Square; df = degree of freedom; CAIC: Consistent Akaike Information Criterion; AIC: Akaike Information Criterion; CFI = Comparative Fit Index; ΔCFI = Increment of the Comparative Fit Index; RMSEA = Root Mean Square Error of Approximation; LL = Lower Limit; UL = Upper Limit; (-item 8) = ítem 8 was not constricted. M1, M2, M3 and M4 are represented in Figure 1. and Table 3.
Concurrent validity and percentile ranking scores
An analysis of concurrent validity was performed – using Pearson correlations – between the RSES and the sexual self-esteem dimension of the Sexuality Scale. Positive, moderate and significant correlations were observed with Sexual Self-Esteem p < .01; r(620) = .31; for Colombia and p < .01; r(494) = .41 for Spain. Finally, the percentile ranking scores of the adaptation of the RSES for the Colombian population were obtained, as differentiated by sex and age group; 18–30, 31–44 and over 45 years (Table 5).
Table 5. Validated RSSE Scales for the Colombian and Spanish Population, by Sex and Age Group
Note: M = Media; SD = Standard Deviation *Caution when interpreting for man +45 in Colombia and women in the same age for Spain because non-continuous and non-normality problems issues we could not reach a good variance for scores for these groups.
Discussion
This study sought to adapt and validate the RSES in a sample of Colombian individuals, and perform its factorial equivalence with a sample from Spain. The qualitative evaluation found that all items fit properly in terms of representativeness, comprehension, interpretation and clarity. Appropriate reliability indicators were obtained in the versions both of Spain and Colombia. The mean of scores of the questionnaire’s items is very similar to the theoretical mean, and its standard deviation is close to one. Different unidimensional and bidemensional models reach adequate fit indexes, but the only model that achieved a strict partial invariance was the unidimensional with all errors of the positive and negative items covariate. This indicates that this model is equivalent in terms of factorial coefficients and the values of the intercepts. The psychometric properties of the items are adequate; only item 8 (“I wish I could have more respect for myself”) has a corrected item-total correlation below .30. Furthermore, it was observed in the analysis of concurrent validity that there are positive, significant and moderate correlations between the RSES and the sexual self-esteem dimension of the Sexuality Scale.
Homologous samples were obtained in this study in terms of sex, sexual orientation and relationship between the two countries. Also, as expected due to cultural differences between Colombia and Spain (van de Vijver & Leung, Reference van de Vijver and Leung2000) significant differences between countries were observed in marital status and religion. However, the samples from Spain and Colombia could not be balanced in age and years of study. Nevertheless, the effect size of the differences observed for these two categories was low, therefore it is not expected to have an important impact in the results.
The experts who adapted the scale advised a direct wording of one of the inverse items. Thus, item 5 of the validated Spanish version (“I feel I do not have much to be proud of”), was worded positively (“I feel I have much to be proud of”). Furthermore, it was observed that this adaptation has no consequences in qualitative evaluation or psychometric features.
Qualitative analysis of the scale included a process to designate the table of specifications of the items. This analysis greatly allowed to obtain adequate content validity, as it enabled us to obtain items which were more related to the construct. This procedure is not usually utilized, although it is of utmost importance in processes of validation and/or adaptation of scales, as pointed out by Carretero-Dios and Pérez (Reference Carretero-Dios and Pérez2005). The results of the qualitative analysis showed that the items were adequately worded, since the limit below 95% of Aiken’s V was always greater than .50.
The reliability indices found (.83 for Colombia and .86 for Spain) are similar to those described by Rosenberg (Reference Rosenberg1965), and subsequently by other authors (Gray-Little et al., Reference Gray-Little, Williams and Hancock1997; Martin-Albo et al., Reference Martín-Albo, Núñez, Navarro and Grijalvo2007; Prezza, Trombaccia, & Armento, Reference Prezza, Trombaccia and Armento1997; inter alia). These reliability indices are suitable for this study, since – according to Nunnally and Bernstein (Reference Nunnally and Bernstein1995) –minimum recommended reliability by means of internal consistency reliability is .80 when the aim of the research is to find a diagnosis or classification. The mean value of scores in the questionnaire’s items is very similar to the theoretical mean, which is expected for a communitarian population, and its standard deviations near one.
Regarding the items’ psychometric properties, it was observed that all except item 8 (“I wish I could have more respect for myself”), had good psychometric properties. Although this item has corrected item-total correlation below .30, and other problems related with the invariance we consider it is not enough to dismiss this item. However, it would be important in forthcoming research to determine whether this item is actually representative of the general self-esteem construct, or it is rather an item which is representative of a specific construct of the Self, for instance “Self-respect” (Rosenberg, Reference Rosenberg1979; Wang, Siegal, Falck, & Carlson, Reference Wang, Siegal, Falck and Carlson2001), because this construct has been described as a parallel – though related – dimension to General Self-Esteem. This would explain the differences in participants’ responses, since – according to various authors (Markus & Wurf, Reference Markus and Wurf1987) – individuals tend to evaluate differently various aspects of the Self. In fact, we are not the first authors to find problems with item 8 (Leung & Wong, Reference Leung and Wong2008). And if our goal is to compare scores between Spain and Colombia this item should be excluded due to invariance issues.
Progressive evaluation of factorial invariance is relevant, as this is the only type of analysis that allows to lay the psychometric foundations for justifications of comparisons between groups (Elosua, Reference Elosua2005). As for the factorial invariance analysis conducted in the study, a strong level of partial invariance was found. This implies the possibility to compare general self-esteem scores between Spain and Colombia, with minimum bias in the measurement (Dimitrov, Reference Dimitrov2010), but as said before item 8 should be excluded for the comparison. Model 1, where all negatives and positives items were covariate within them, has the best invariance here tested. The fact that all errors where covariate implies that all of them had some similar way to be answered. Thus, these covariances suggest that all positive items have a common component (probably its positive redaction) as negative ones does. And at least in our sample this model has showed the best fit, invariance indicators and the lower AIC value in the configural level (Table 4). This data will help to include more information to the debate of the dimensionality of the RSES. As for factorial weights (λ), it was observed that all items show values which are clearly higher than .30, except for item 8 (.32), which is consistent with the results observed in this and other studies (Rosenberg, Reference Rosenberg1979; Wang et al., Reference Wang, Siegal, Falck and Carlson2001) and continues to be indicated as a problematic item.
Model 1, as the others, has been also tested in other studies. For instance, studies which have systematically reviewed the factorial structure of the Rosenberg scale, as is the case of Marsh (Reference Marsh1996) or Tomas and Oliver (Reference Tomas and Oliver1999) show how the RSES has been evaluated by means of different ways. Unidimensional structures have been proposed wherein covariances of the errors are performed in both positive and negative items of the scale – whether altogether or independently – or which have been evaluated through two-factor models, whether Positive Self-Esteem and Negative Self-Esteem or General Evaluation and Transient Evaluation, amongst other models.
Regarding the analysis of concurrent validity, it was found that there are significant correlations between the RSES and the dimension of sexual self-esteem of the Sexuality Scale. These correlations are positive and moderate. This is related to the statement above, in that general self-esteem is not equivalent or interchangeable with the specific self-esteem (Rosenberg et al., Reference Rosenberg, Schooler, Schoenbach and Rosenberg1995), since individuals usually relate to one another and evaluate themselves differently in all aspects of their lives. In addition, the findings are consistent with the postulates of Zeanah and Schwarz (Reference Zeanah and Schwarz1996), who report that people “have a sense of the sexual self which is distinct from, but contributes to, the global sense of self” (p. 2). In addition, Swenson, Houck, Barker, Zeanah, and Brown, (Reference Swenson, Houck, Barker, Zeanah and Brown2012) state that Sexual Self-Esteem and General Self-Esteem are two constructs that should be differentiated, among other reasons because the clinical implications of having a low general self-esteem, are not similar to the implications of having a low Sexual Self-Esteem. This explains how some people may have low general self-esteem and high sexual self-esteem or vice versa.
Lastly, presenting percentile ranking scores differentiated by country, sex, and age ranges can be useful in evaluating general self-esteem in these populations. We ask some caution when interpreting for man +45 in Colombia and women in the same age for Spain because some non-continuous and non-normality limitations we could not reach a good variance for scores for these groups.
In conclusion, we present a reliable scale which was rigorously adapted, in pursuance of the guidelines for adapting tests to other cultures described above, which has a stable performance in two countries with different cultures (Colombia and Spain). Moreover, this is a reliable questionnaire with a correspondence between the theoretical assumption and the dimensionality exposed, as well as with internal consistency indicators and adequate concurrent validity. This, it can be concluded that there is sufficient evidence supporting the construct validity of the scale. In addition, clear indications are presented in that the adapted scale can be used in Colombian population, both men and women of age.
However, this study is not without limitations. A non representative convenience sampling does not allow for generality of results for all Spanish and Colombian populations. Furthermore, using an internet-based sampling method does not allow for equal access to all socioeconomic strata of the population due to the reality of internet access for people. Item 8 is being problematic and future research should determine any changes on it. Therefore, it would be interesting to extend the evaluation sample in the future to different clinical realities or more complex collectives. This would allow to continue this work with the evaluation of the implications of general self-esteem in the health of the Colombian population – as has been done so far – but with a valid and reliable scale.
This work has been possible, thanks to ‘‘Fundación Universitaria Konrad Lorenz’’ funding associated to the research project (number: 55270151) granted to last author.
Appendix
Rosenberg Self-Esteem Scale (RSES; Rosenberg, Reference Rosenberg1965) – Version Validated in Colombia.
1 = Totalmente en desacuerdo; 2 = De acuerdo; 3 = En desacuerdo; 4 = Totalmente en desacuerdo
(1 = Strongly Disagree; 2 = Agree; 3 = Disagree; 4 = Strongly Disagree)
