Psychometric Characteristics of a Serbian Translation of the Unconditional Self-Acceptance Questionnaire and the Development of a Short Form

The Unconditional Self-Acceptance Questionnaire is extensively used in the assessment of one’s ability to evaluate their skills, actions, thoughts, and feelings, without a generalized evaluation of one’s worth or value. However, it has often been criticized for being contaminated with self-esteem items and, hence, theoretically incongruent. We evaluated the psychometric properties of the instrument and provided the first validation of the short Serbian adaptation of the USAQ. In Study 1 ( n = 288), we examined the latent structure of the original USAQ translated to the Serbian language and found a three-factor solution, consisting of Conditional Self-Acceptance, Unconditional Self-Acceptance, and Attitudes About Human Worth. In Study 2 ( n = 354), we sought to validate the shorter version of the scale. The two-factor solution, comprising Unconditional and Conditional Self-Acceptance, explained 51 % of the variance. Re-testing the two-factor structure on a different sample, the CFA revealed satisfactory fit indices. Apart from good internal consistency, the shortened scale showed comparable associations with appropriate convergent and discriminant constructs (e.g., explicit self-esteem, positive and negative state/trait affectivity, and the states of anxiety and depression), supporting the construct validity of the Serbian version. Further implications for research and therapeutic practice are discussed.


3
As a multidimensional concept in human psychological functioning, Self is strongly associated with mental health (McCrae & Costa, 1996). One of the critical components of Self is the evaluative component, that is, thinking about ourselves as positive or negative individuals.
Psychological science and psychotherapeutic practice recognize several constructs that include some form of self-evaluation (David et al., 2013). Among them, self-esteem is the most investigated and it is most commonly mentioned in the literature. It is defined as a positive or negative global assessment of oneself (Sedikides & Gregg, 2003). Self-esteem correlates with numerous mental health indicators such as psychological well-being (Mann et al., 2004;Zimmerman, 2000) and lower emotional distress (Sedikides et al., 2004). That is why self-esteem is so popular in educational and preventive programs for children and adults in various settings (Baumeister et al., 2003). Reviewing the literature that analyzed the impact of self-esteem on different aspects of psychological and social functioning, no evidence has been found to support the popular belief that high self-esteem is always useful (see Baumeister et al., 2003). On the contrary, Rational Emotive Behavior Therapy (REBT) recognizes self-esteem as detrimental to mental health because it is derived from an irrational global self-assessment . Global self-evaluation is considered irrational because of the lack of an objective criterion for determining the value of human beings (Chamberlain & Haaga, 2001b). As a healthier alternative, REBT offers the concept of unconditional self-acceptance (hereinafter USA), defined as full and unconditional acceptance of our beings, regardless of our competency or the respect of other people (Ellis, 1977). Unconditional self-acceptance is based on the assessment of one's characteristics and actions, but not on the global assessment of one's value because there are no objective criteria for that (Turner, 2016). 4 Cognition is the crucial determinant of human feelings according to Albert Ellis (1980), who argued that dysfunctional thinking is essential for emotional suffering. According to Ellis, cognitions have a critical role in the development and maintenance of psychological problems that arise when people hold irrational beliefs towards self, others, and life in general. In relation to other cognitive therapies, a unique characteristic of REBT is the emphasis on the evaluative component of cognitions (Dryden, 2002), that is, how a person evaluates events and situations in which he/she experiences a particular emotion. Evaluations can be both rational and irrational. In the psychotherapeutic process, we look for irrational evaluations that affect dysfunctional emotional states. Ellis (1994) suggests that emotionally healthy people accept reality even when it is unpleasant for them. REBT helps clients adopt several kinds of unconditional acceptance: self-acceptance, acceptance of others, and acceptance of life in general. The present study primarily deals with the construct of unconditional self-acceptance (USA). According to Ellis, one of the essential beliefs that form the basis of USA is that regardless of good and bad personal qualities, an individual is not worth any more or less than any other human being. When a personal preference or desire is not achieved, the rational alternative is not to criticize oneself as a person, but to criticize specific behaviors. The goal is to accept oneself as a fallible human being.
Adopting the philosophy of USA encourages people to focus on what they need to do to correct their behavior. In REBT theory, the philosophy of unconditional self-acceptance represents an alternative to the philosophy of self-downing based on a negative global evaluation of oneself due to negative acts and experiences (Van der Does, 2002). This is another unique aspect of REBT compared to other cognitive-behavioral and humanistic therapies. The essential cognitive characteristic of this attitude is the philosophy of not judging oneself, while keeping the evaluation of particular thoughts, feelings, and traits, whether positive or negative, relative to the experience (Ellis, 2001). This means that people who adopt this attitude make positive and 5 negative assessments of their skills, actions, thoughts, and feelings in a particular situation, but at the same time refuse to evaluate themselves globally, based on these individual aspects.
While USA often represents the therapeutic goal of REBT treatment, it is rarely and insufficiently empirically verified (Davies, 2008). Although it has been present in REBT theory and practice for several decades, USA research has been conducted more frequently over the past ten years and studies have mostly been correlative. In the majority of published studies, the Unconditional Self-Acceptance Questionnaire (USAQ;  was used for measuring the USA construct. These studies have confirmed the REBT theory that people with a higher level of USA are less depressed and anxious and have higher scores on the subjective well-being and life satisfaction scales. However, what is not in line with the REBT theory is a moderate positive correlation between USA and global self-esteem (average r = .50, p < .001) measured by Rosenberg′s self-esteem scale (RSE). Observed separately, USA significantly correlates with state negative affect (r = -.37, p < .01), while RSE significantly correlates with both positive (r = .45, p < .01) and negative affect (r = -.61, p < .01) (Popov & Sekulić-Bartoš, 2016). In some studies, correlations between USA and mental health variables were reduced or lost when self-esteem was statistically controlled Popov & Popov, 2011). Several experimental studies have also supported the protective role of USA in adverse emotional reactions in situations of real stressors . Still, the problem with the correlation with self-esteem remains. In another study, when tested in the same experimentally induced stressful situation, self-esteem and USA established similar (though not identical) relations with mental health variables (Popov et al., 2015). One of the reasons for such similar tendencies of constructs that are theoretically different may be the method of measurement. The USAQ, which measured unconditional self-acceptance 6 (USA) in all these studies, has often been criticized for being contaminated with self-esteem items and, hence, inconsistent with Rational-Emotive and Cognitive-Behavior theory (David et al., 2013). Another possible reason may be theoretical. Unconditional self-acceptance is considered to be a formative and continuous phenomenon with an almost unattainable positive extreme (Ellis & Dryden, 2007). Therefore, it seems reasonable to assume that people who tend to accept themselves in moments of self-evaluation are more prone to evaluating themselves positively (Ellis & Dryden, 2007).
The USAQ  is the only and thus the most widely used instrument for measuring unconditional self-acceptance from the REBT perspective. It has been used several times in studies with Serbian participants (e.g., Popov & Sekulić-Bartoš, 2016;Stanković & Vukosavljević-Gvozden, 2011). Given that the USAQ instrument has not yet been validated in any cultural or linguistic area other than the one in which it was initially constructed (although it is used very often), this study aimed to remedy this deficiency. This research aimed to examine the psychometric characteristics of the USAQ ) on a Serbian sample, as well as to propose a shorter version of the questionnaire. The study was approved by the Ethical Board of the Faculty of Sport and Tourism at the Educons University and carried out following the accepted guidelines and regulations. 7 Study 1

Aim of the study
The aim of Study 1 was the examination of the latent structure of the Serbian translation of the original USAQ , which had not been conducted before.

Sample and Procedure
The study involved 288 older adolescents (M = 17.64, SD = .84, 62% female), who were secondary school students. All respondents voluntarily participated in the research. In each of the schools, the management and parent councils signed the informed consent. All participants completed the questionnaires within their school hours, under the same instruction manual, and with the examiner's assistance when clarification was necessary.

Instruments
In this part of the study, the Unconditional Self-Acceptance Questionnaire (USAQ;  was translated into the Serbian language, using the backtranslation procedure. The instrument consists of 20 items with a 7-point Likert scale (item example: "When I am criticized or when I fail at something; I feel worse about myself as a person."). In the original study conducted by the authors of the scale, Cronbach's alpha coefficient was .72. In a previous study on a Serbian sample, it was .75 (e.g., Stanković & Vukosavljević-Gvozden, 2011). In this study, the α coefficient was .70. 8 For the examination of the external validity of the Serbian version of the USAQ, the correlation with the scores on the following instruments was used: For measuring explicit self-esteem, we administered the Serbian translation of the Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965) (item example: "I feel that I am a person of worth, at least on an equal plane with others"). The internal consistency of the Serbian translation was satisfactory both in previous research (e.g., Jovanović, 2010, α = .82), and in our sample, it was α = .79.
For assessing the states of positive (PA, e.g., enthusiastic) and negative affect (NA, e.g.,

upset), we used the Serbian Inventory of Affect based on the Positive and Negative Affect
Schedule-X (SIAB-PANAS; Novović & Mihić, 2008). In previous studies, the internal consistency was α = .90 for PA and α = .87 for NA (Popov, 2019). In this study, Cronbach's alpha coefficient for both subscales was .83.

Data Analysis
The data were analyzed in the IBM SPSS statistical package version 21, using the following procedures: an exploratory factor analysis (EFA) with principal axis factoring and a parallel analysis and a t-test for independent samples.

The Latent Structure of the USAQ on a Sample of Serbian Adolescents
To examine the latent structure of the USAQ, an EFA was used. The Kaiser-Meyer-Olkin test indicated middling sampling adequacy (KMO = .75). Items with high cross-loadings and loadings lower than .40 were initially discarded, leaving 15 items in the final analysis. After principal axis factoring, the optimal number of factors was specified by the parallel analysis with the 95th percentile criterion (O'Connor, 2000), which suggested a three-factor solution (38.93 % of the variance explained).
The extracted factors were rotated via the Promax rotation. The pattern matrix is presented in Table 1. The first factor was saturated with items directed towards conditional selfworth, which means that people value themselves depending on whether they have reached imaginary standards and expectations. These people are preoccupied with achievement and how other people see and evaluate them. To maintain the optimal thinking about themselves, they must continuously achieve successes and be accepted by the people they consider influential. In other words, this is an ego-involved form of self-regulation. In the non-REBT literature, it is called contingent self-esteem (Crocker & Wolfe, 2001). This factor was named Conditional Self-Acceptance. The content of the second factor described self-acceptance and self-worth feelings, even when personal standards are not successfully met. The factor was interpreted as Unconditional Self-Acceptance. The third factor was saturated with items that described the general philosophy about personal value and it was named Attitudes About Human Worth. The factors had low intercorrelations (Table 2).

Gender Differences in Individual Dimensions of the USAQ
To examine gender differences in the extracted USAQ factors, we performed a t-test for independent samples. Gender differences were only identified on the CSA dimension, t(286) = -.2.49, p < .01, d = 0.30. According to this result, a higher degree of conditional self-acceptance was present in the female sample.

Aim of the study
The aim of Study 2 was the validation of the shortened version of the USAQ scale in the Serbian language. The shorter version of the scale was created by first removing the items related to general self-evaluation attitudes, that is, the third factor from Study 1. We also made adjustments to the Serbian translation of the instrument and adapted it for use in the Serbian speaking area to more precisely formulate (un)conditional self-acceptance. For example, we reformulated "Making a big mistake may be disappointing, but it doesn't change how I feel about myself overall" to "Making a big mistake may be disappointing, but it doesn't make me feel less worthy as a person". Additionally, we examined the relations of the shortened version of the questionnaire with self-esteem and the measures of positive and negative trait/state affectivity, anxiety, and depression. We expected higher scores on the USAQ-short to be positively associated with higher scores on self-esteem as well as higher scores on positive trait/state affect. 11 We also expected that higher scores on the USAQ-short would be negatively associated with negative trait/state affect as well with the states of anxiety and depression.

Sample and Procedure
In this study, the sample included 354 students from the University of Novi Sad, Serbia (75% female), aged between 18 and 39 (M = 19.64, SD = 2.51). The sample in this study was independent of the Study 1 sample. The students participated voluntarily and the motivation for participation was additional course credit, in agreement with the authorities at the University.
Informed consent was obtained from all participants included in the study. The sample was randomly divided into two data subsets in order to cross-validate the factor structure of the USAQ-short.

Instruments
The short version of the Unconditional Self-Acceptance Questionnaire (USAQ-short) was

Data Analysis
An exploratory factor analysis (EFA) with principal axis factoring and a parallel analysis and a t-test for independent samples were performed in the IBM SPSS statistical package version 21. A confirmatory factor analysis (CFA) was conducted in R, using the lavaan package.

Results
The The Promax rotation was used for the extracted factors. The pattern matrix is presented in Table 3. The first factor was saturated with items that refer to conditional acceptance of oneself, such as "In some situations, I think I'm a totally worthless person". This factor could be called Conditional Self-Acceptance (α = .79). The second factor was the opposite of the first and it was saturated with items related to Unconditional Self-Acceptance, such as "I feel worthwhile even if I am not successful in meeting certain goals that are important to me" (α = .71). The extracted factors 14 significantly correlated (r = -.42, p < .001), with a higher score indicating a higher degree of unconditional self-acceptance.

Confirmatory Factor Analysis of the USAQ-short Version
We further re-examined the factor structure of the USAQ-short version by conducting a confirmatory factor analysis on a different sample (N = 149, 21% male, age M = 19.12. SD = .52). To examine the fit of the 1-and 2-factor solution for the reduced set of items, we conducted a CFA on polychoric correlations, with the diagonally weighted least squares (DWLS) estimator.
To determine the models' goodness of fit, the following criteria were used: the Comparative Fit Index (CFI) > .95, the Root Mean Square Error Approximation (RMSEA) < .06, and the Tucker-Lewis Index (TLI) > .90 (Brown, 2006). Two models were analyzedunidimensional and twodimensional models, with each item being placed on one of the two latent factors corresponding to the previously obtained EFA solution.
The unidimensional model gave unsatisfactory fit indicators, as shown in Table 4.
Secondly, the tested model yielded very good fit indices, suggesting that the proposed two-factor version of the USAQ is a good representation of the data (Table 4, Figure 1).

Gender Differences in Individual Dimensions of the USAQ-Short
After performing a t-test for independent samples, we found no gender differences in the extracted factors of the USAQ-short scale and the overall scale score.

Convergent Validity of the Serbian USAQ-Short Scale
To test the overall convergent validity of the USAQ-short scale, we correlated the obtained factors with indicators of mental health (states of anxiety and depression as well as positive and negative affect) and affectivity as a personality trait. In regard to divergent validity, there was no theoretically adequate instrument to compare the USAQ-short. The only comparable instrument was a measure of global self-esteem, for which we expected to be in a negative relationship with unconditional self-acceptance. Therefore, we can only conditionally say that this was an examination of divergent validity. Correlations with these measures are shown in Table 5.
[insert Table 5 here] In Table 5, we can see that the factor we named Conditional Self-Acceptance (CSA) was negatively related to positive affectivity (r = -.167, p < .001) as a personality trait and positively related to negative affectivity (r = .355, p < .001). The same was true for positive affect (r = -.140, p < .001) and negative affect (r = .230, p < .001), as a condition. Also, CSA was positively related to the state of anxiety (r = .255, p < .001) as well as depression (r = .292, p < .001).

Discussion
As an extensively used measure for assessing one of the key constructs in REBTunconditional self-acceptance (USA), the USAQ has provided substantial empirical support for the relationship between USA and mental health. However, due to the lack of studies on the aspects of validity and reliability, it has often been criticized for being contaminated with selfesteem items. Thus, the construct of USA has been marked as inconsistent with REBT theory.
USA should represent an alternative to global self-esteem (David et al., 2013), but the question is whether this has been confirmed in practice. We sought to address these issues regarding the evaluation of psychometric characteristics in general and to provide the first validation of the short Serbian adaptation of the USAQ scale. This research consisted of two studies conducted on two independent samples (N = 288 for Study 1 and N = 354 for Study 2).
In Study 1, we examined the latent structure of the USAQ  on a sample of Serbian adolescents and obtained a three-factor solution, with factors named Conditional self-acceptance, Unconditional self-acceptance, and Attitudes about human worth.
The content of the first two factors is self-explanatory, while the third factor with the lowest alpha contained diverse views about self and more general attitudes about human value. This 17 result is significant because previous studies mostly used the total score, although one earlier research (see Davies, 2006) also showed that the USAQ does not have a one-dimensional structure. Furthermore, the results showed that the factors were marginally correlated. To improve the scale's homogeneity and reduce the instrument to a single object of measurement (self-acceptance), we first eliminated Attitudes about human worth. The reason was that general views on human worth go beyond the notion of self-acceptance and do not necessarily define a personal relation to oneself. In therapeutic practice, we often find that clients have "double standards" regarding what is acceptable for other people and what is acceptable for them (Froggatt, 2005). For example, people can value themselves globally, although they do not do so when it comes to other people. They can also be aware of the irrationality of the global evaluation of people and life, but still prone to reacting conversely in a stressful situation, when it comes to the evaluation of themselves. Other reasons for eliminating the Attitudes about human worth factor were that its alpha coefficient was below the limit of acceptability (α = .66) and it showed high correlations with external variables (except for positive affectivity). We believe that research instruments should be more aligned with therapeutic practice, as we often use the same instruments for treatment evaluation. When it comes to the construct of unconditional selfacceptance in therapy, we are more focused on the relationship that the client holds towards himself and only indirectly on general attitudes about human value.
After adjusting the translation and wording of several items in Study 2, we examined the latent structure of the USAQ-short and extracted the optimal two-factor solution: Unconditional and Conditional Self-Acceptance, explaining 51 % of the variance. The next step was re-testing the model on a different data subset via a confirmatory factor analysis and obtaining satisfactory fit indices for the two-factor model, but not for the unidimensional solution. The two-factor 18 structure of the shortened USAQ scale, where one factor is related to Conditional Self-Acceptance and the other to Unconditional Self-Acceptance, reflects the possible nature of this construct, which can be compared to the relation of rational and irrational beliefs (Sava et al., 2011), as well as positive and negative affect (Merz et al., 2013). Following the same procedure, we should calculate the scores on the subscales separately. This result is significant for therapeutic practice, where it is essential to pay special attention to the construction of unconditional self-acceptance after the disputation of conditional self-acceptance. The relationship between self-acceptance and self-esteem is particularly important for this paper. While the association between CSA and RSE was negative and quite high (r = -.63), USA was still in a positive relation with RSE, similar to earlier studies (r = .440) (Popov, 2019). In addition to the measurement methodology discussed in previous research, these results may reflect the relationship between USA and RSE constructs in reality. At first glance, self-esteem does not appear to have been shown to be detrimental to mental health, nor separable from the USA, which is in line with previous research (e.g. Popov et al., 2020;Sava et al. 2011). In the context of REBT, if USA is seen as a continuous phenomenon, then people who tend to accept themselves in moments of susceptibility to self-worth should more often evaluate themselves in a positive rather than a negative manner . This trend of selfevaluation should be even more prominent in encounters with success than with failure. What distinguishes them, according to , is that people with high unconditional self-acceptance are not inclined to self-harm in a situation of failure. Nevertheless, our results in earlier studies (Popov et al., 2015;2016) suggest that these constructs behave quite similarly (but not the same) even in an experimental failure situation. This study also found CSA to be related to low SE, while USA was related to high SE. Given that all mentioned studies used the same instruments for measuring USA and SE (e.g. Chamberlain & Haaga, 2001b;Popov, 2019), it is possible that the used method does not represent the right way to determine whether USA is more constructive than SE in the context of mental health. People have a natural tendency towards evaluation, but also self-evaluation. The only question is how a person arrives at these assessmentswhether it is through global and conditional evaluation or through an affinity to self-acceptance (Popov et al., 2015). Rosenberg's questionnaire cannot reveal the actual nature of self-evaluation. If we want to dispute global self-evaluation, it would be more appropriate to use instruments where this process is clearly operationalized, especially the aspects of self-esteem 20 that REBT challenges (conditional evaluation and liability). In the research practice of selfesteem and non-REBT literature, several instruments that register fluctuations in self-esteem have been recognized, along with so-called contingencies or specific standards on which the individual bases their self-worth (Crocker et al., 2006;Kernis & Goldman, 2003;Kernis, 2005;Kernis et al., 2008). Finally, some authors believe that REBT should not abandon the concept of self-esteem in the context of contributing to a client's mental health. According to them, the constructs of USA and self-esteem can be understood and used as complementary, similarly as irrational and rational beliefs that are in an orthogonal relationship -an increase in rationality does not necessarily mean a decrease in irrationality (Sava et al., 2011).

Strengths, Limitations and Future Directions
What can be considered the main strength of this study is that it represents the first validation of the USAQ instrument on a foreign and culturally different population (from the original in which it was constructed), as well as the first attempt to overcome the shortcomings of the instrument identified in previous research. The main limitation of the study is related to sample selection (adolescents, mostly female and non-clinical populations). Given that this is a pioneering endeavor, this limitation may be overcome in future studies, where the new USAQshort instrument would be tested on a clinical population, different cultures, as well as in the therapeutic process itself. Given that the USAQ-short examined here has satisfying psychometric characteristics, the authors believe that it has potential for further development and contribution to research and therapeutic practice.   .70 Even when I don't get approvement from other people, I feel I am worthwhile as a person. .66 Making a big mistake may be disappointing, but it doesn't make me feel less worthy as a person. .54 I avoid comparing myself to others to decide if I am a worthwhile person. .52 Note. Component names: U-Unconditional self-acceptance, C-Conditional self-acceptance Note. χ² = chi-square; CFI = Robust comparative fit index (> .95 indicates good fit); TLI = Robust Tucker-Lewis index (> .95 indicates good fit); RMSEA = the root of the average square error approximation (< .05 points to good fit, .08 to acceptable); SRMR = Standardized Root Mean Square Residual.