As Kosovo is a newly formed country, data and research related to all aspects of its society are somewhat limited. The transition as a post-war country (Malcolm, 1998) reflected upheaval in several Kosovan social development issues — especially related to traditional family structures. These were undermined by numerous factors including rural to urban migration; the promotion of individualism through urban living; lack of localized institutions; lack of education, (mental) health, and law; and high unemployment levels. This context has posed a particular risk of increasing criminal behaviors among unemployed youths (Heiniger & Renaud, 2008).

The Juvenile Justice System in Kosovo, which is relatively new, was established with the assistance of the European Union and the United Nations in 2007. Prior to this year, Kosovan justice did not prescribe any specific guidelines for the treatment of children, adolescents, and young adults; issues related to all these social groups were processed through the adult system (Hamilton & Anderson, 2008). Currently, the Assembly of Kosovo enforces the system of law through the Juvenile Justice Code, which places young adults under the age of 21 (depending on the seriousness of the crime) under the responsibility of the juvenile system (Juvenile Justice Code, 2018). In the 10 years since its establishment, Kosovo has managed and maintained a standard of care that has been acknowledged as being one of the best practices by both the EU and the United Nations (Team, 2018).

Within this context, this study aims to provide a valid means for assessing at-risk youth, helping those already placed within the system, and helping those undergoing rehabilitation in Kosovo. Specifically, through the analysis of the psychometric properties of the Basic Empathy Scale (BES) in relation to young adults in Kosovo, the BES can provide local institutions with a valid tool for empathy evaluation, and, at an international level, extend the current literature through the inclusion of Kosovo.

Empathy

A broad body of research has acknowledged empathy as a key component for molding well-adjusted and productive members in society (Bezerra et al., 2021; de Medeiros et al., 2021; Hudson et al., 2019; Llorca-Mestre et al., 2017; Samper et al., 2021). Conversely, a deficiency in empathy can result in selfishness, anti-social and aggressive behavior, and an increased likelihood of committing offences among young offenders (Bacchini et al., 2018; Bezerra et al., 2021; Coetzee, 2019; Heynen et al., 2017, 2018; Llorca-Mestre et al., 2017; Narvey et al., 2020; Van Zonneveld et al., 2017).

As such, empathy development can play an import role in the identification of at-risk youth regarding criminal offences and assist in the rehabilitation of young offenders (de Medeiros et al., 2021; Heynen et al., 2016; Jolliffe & Farrington, 2004, 2006; Samper et al., 2021) Therefore, it is imperative to have a reliable tool for empathy measurement.

Cohen and Strayer defined empathy as “the understanding and sharing in another’s emotional state or context” (p. 523, 1996); they rationalized that this definition could allow for the application of two aspects of empathy: effective and cognitive. Effective empathy involves the mirroring of an emotional response that one has witnessed in another, whereas cognitive empathy involves the ability to comprehend an emotional response but not feel it.

Empathy evolves as individuals develop; therefore, it has been used for accurately predicting the future behaviors of individuals from a young age (Fricket al., 2014; Llorca-Mestre et al., 2017; Muñoz & Frick, 2012; van Zonneveld et al., 2017). Moreover, youth aged between 15 and 24 years (Fricket al., 2014; Llorca-Mestre et al., 2017; Muñoz & Frick, 2012; van Zonneveld et al., 2017) who have high empathy levels have better relationships with others (e.g., peers, authority figures, and parents) (Heynen et al., 2016; Van Lissa et al., 2015), unlike those with low empathy, such youth build long-lasting relationships. Youth with high empathy are less likely to be impulsive or engage in risk taking behaviors (Allemand et al., 2015; Rodriguezet al., 2019). They are also often engaged in a greater number of prosocial behaviors (e.g., volunteering, helping, and sharing) (de Medeiros et al., 2021; Eisenberg & Miller, 1987; Hoffman, 2008; Hudson et al., 2019) and show greater social competencies in adulthood (Allemand et al., 2015).

Deficits in empathy, particularly during youth, are reportedly a greater factor in offending among youth compared to adults, and young adults with reduced empathy tend to commit crimes more frequently (Llorca-Mestre et al., 2017; Narvey et al., 2020; Samper et al., 2021; Wang et al., 2017). Further, those with low empathy tend to reject authority figures, and this places them at risk of deviant peer relationships (Hudson et al., 2019).

In their longitudinal study, Allemand et al. (2015) found that both positive and negative empathy-related changes among youth affected their adult coping skills 25 years later. They also found that youth who reported a decrease in empathy were more likely to have fewer friends and suffer from loneliness and negative emotionality.

In conclusion, empathy plays a crucial role in the prosocial development of young people and can play a vital role in treatment and prevention of criminal behavior in vulnerable youth (Frick & Kemp, 2021; Heynen et al., 2018; Rodriguez et al., 2019; Samper et al., 2021; van Zonneveld et al., 2017). As such, a valid and reliable measure is required for assessing empathy in youth, and this measure should be able to predict behavior risk assessments and assistance in rehabilitation.

Rationale for the Development of the Basic Empathy Scale

In the pursuit of a valid and reliable self-report questionnaire for the measurement of empathy, Jolliffe and Farrington (2004, 2006) developed the BES. It was in response to the weaknesses that they had observed and reported in the previously used measures. Specifically, these previous measures had been suitable for measuring sympathy—an issue derived from the interchangeability between the earlier literatures on sympathy and cognitive empathy, respectively. It was also found that these tools often failed to measure both cognitive and affective empathy (Jolliffe & Farrington, 2004, 2006)—two aspects that have been generally considered as being essential for defining empathy (Cohen & Strayer, 1996; D’Ambrosio et al., 2009; Heynen et al., 2016; Jolliffe & Farrington, 2004, 2006; You, Lee, & Lee, 2018).

The identification of the two aspects of empathy is crucial, as multiple studies have shown that cognitive empathy is present in individuals and children who are prone to callous and unemotional traits—that is, traits that are common in individuals that are potentially susceptible to perpetrating violent antisocial behaviors, which can often continue into adulthood (Frick & Kemp, 2021; Frick et al., 2014; Giroux & Guay, 2021; Heynen et al., 2016; Waller et al., 2020; Zych et al., 2019). This group has been shown to be deficient in affective empathy, that is, the ability to directly share the exact emotions of others. This deficiency means that while these individuals can comprehend that their actions may cause serious distress (which can be perceived through cognitive empathy) to others, they remain emotionally indifferent to the plight of others.

The psychometric properties of the BES have been found to be adequate—to varying degrees—in different cultural contexts including Europe, America, and Asia. BES scores were positively linked with prosocial behaviors and negatively linked with certain aspects of psychopathy; consequently, it cannot be assumed that they are supported within a Kosovo context (Geng et al., 2012; Heynen et al., 2016; McLaren et al., 2019; Pechorro et al., 2015; Sanchez-Perez et al., 2014; You et al., 2018).

Methods

Participants and Design

This cross-sectional study was conducted with 476 students (38.4% male) aged 18 to 25 years (M = 20.25; SD = 1.61) from the University of Prishtina “Hasan Prishtina”. The sampled students were from the following academic fields: economics (19%), law (25%), engineering (18%), psychology (18%), and sociology (18%), representing all 3 years of bachelor studies: 1st (29%), 2nd (33%), and 3rd (31%). In the case of Law studies, the program was four years, and therefore, the 4th year of study was included (7% of total participants). In terms of location, 59% of the students hailed from urban areas (Table 1).

Table 1 Baseline characteristics of study participants (N = 476)
Full size table

Data Collection

Data collection took place in 2020 at the University of Prishtina “Hasan Prishtina”. Being a state-run University, it offers tuition-free education along with student housing that places students from all regions of Kosovo in the capital of Prishtina. The university permits access to a range of young adults. It also provides 3 years of student housing that is characterized by mixed gender and socioeconomically diversity; furthermore, it had the necessary mechanisms for conducting ethical oversight. The student populations of the chosen departments were identified as representing a broad range of student types.

The survey, once approved by the ethics committee of Faculty of Philosophy/University of Prishtina “Hasan Prishtina”, was presented to the department heads for an overview and then to teaching professors. Upon provision of consent, the students received an outline of the study in person, and surveys were distributed to the participants.

All participants voluntarily agreed to participate, signed an informed consent form, and were told that their answers would be treated confidentially and anonymously, with responses being accessible only to the researchers.

Measures

Basic Empathy Scale

The BES (Jolliffe & Farrington, 2006) is a 20-item self-report measure that assesses two components of empathy: cognitive empathy (9 items) and affective empathy (11 items); response categories are rated on a five-point Likert-type scale (ranging from 1 = I strongly disagree to 5 = I strongly agree) created in English. Upon receiving permission from the author of the BES, the English version of the BES was translated into Albanian language through multistage back-translation, a requirement for conveying the same meaning as the original text. The initial forward translation was conducted by two independent experts and was followed by reconciliation of the translation by a third independent expert, with the result being back-translated to English by a fourth independent expert who did not have access to the original text. During this comparison, no meaningful changes were found. Table 2 presents the descriptive statistics of all the items. After reverse scoring several items of the cognitive (3 items) and affective (5 items) empathy scales, higher scores were considered to be indicative of higher levels of cognitive or affective empathy. The original scale has been shown to be a valid and reliable measure for assessing cognitive and affective empathy with good internal consistency (cognitive: α = 0.79; affective: α = 0.85) (Jolliffe & Farrington, 2006).

Table 2 Item descriptives of the Basic Empathy Scale (N = 476)-Albanian language
Full size table

Inventory of Callous Unemotional Traits

The ICU scale (Kimonis et al., 2008) is a 24-item self-report measure designed to evaluate callous and unemotional traits in youth; response categories are rated on a four-point Likert-type scale ranging from 0 (not at all true) to 3 (definitely true). The ICU has parent, teacher, and self-report versions. The present study used the self-report version, which has been shown to be a reliable and valid measure for examining CU-traits in adolescent offenders (Kimonis et al., 2008). The scale is divided into three subscales: Callousness (e.g., “the feelings of others are unimportant to me”; α = 0.70), Unemotional (e.g., “I hide my feelings from others”; α = 0.64), and Uncaring (e.g., “I try not to hurt others’ feelings”; α = 0.73) (Kimonis et al., 2008). The three subscales together form a higher-order callous-unemotional dimension (α = 0.77). The present study showed good reliability for the overall ICU-factor (α = 0.75) as well as for the subscales of callousness (α = 0.72) and uncaring (α = 0.68). For the unemotional subscale, the reliability was unsatisfactory (α = 0.56).

Demographic Survey

A demographic survey was constructed to collect data on the participants’ sociodemographics. This survey included the participant variables of age, gender, field of study, year of study, and residence location (rural vs. urban).

Statistical Analysis

To examine the replicability of the two-factor structure of the original BES version, a confirmatory factor analysis (CFA) was conducted using the Lavaan package software (Version 0.6–10) in the R environment (Rosseel, 2012). Furthermore, an alternative factor model was examined by adding a method factor containing negatively worded items. In the Korean version of the BES, You et al. (2018) found that a factor model with a method factor indicated a superior fit to the original two-factor model. To account for non-normally distributed ordinal variables, the mean and variance-adjusted weighted least squares (WLSMV) estimation procedure were used (Li, 2016).

Currently, there is no consensus on power analyses in structural equation modeling and CFA (Kline, 2016). As a rule of thumb, a ratio of cases to variables between 10:1 and 20:1 is recommended. Given the relatively large sample size (N = 476), CFA was considered appropriate. Additionally, investigating the psychometric properties of the BES through CFA allowed to compare results with findings of previous BES validation studies. Assumptions of normality of the data were inspected by means of computing skewness and kurtosis values. All values were within the acceptable range of skewness between − 3 and + 3 and kurtosis within − 10 and + 10 when utilizing SEM (Brown, 2015). Additionally, SEM is recognized as a robust analytical method regarding assumptions of normality.

The goodness-of-fit of the model was evaluated by calculating several indices: the comparative fit index (CFI), Tucker-Lewis index (TLI), goodness-of-fit index (GFI), adjusted goodness-of-fit index (AGFI), and root-mean-square error of approximation (RMSEA). For the adequate model fit, cut-off values of CFI ≥ 0.90, TLI ≥ 0.90, GFI ≥ 0.90, and RMSEA < 0.08 are necessary, whereas CFI ≥ 0.95, TLI ≥ 0.95, GFI ≥ 0.95, AGFI ≥ 0.90, and RMSEA < 0.05 are indicative of good model fit (Hu & Bentler, 1999; Kline, 2016).

Cronbach’s alpha was calculated to examine the internal consistency reliability of the BES scales. Furthermore, the concurrent validity of the BES scales was examined by calculating Pearson’s r correlations between the BES scales and the ICU scales. Concurrent validity is demonstrated if the cognitive empathy and affective empathy scales are significantly and inversely correlated with the Callousness, Unemotional, and Uncaring scales of the ICU. Calculations of Cronbach’s alpha and correlational analyses were conducted using SPSS version 24.

Ethics Statement

All aspects of the current research met local ethical guidelines, including adherence to the legal requirements of Kosovo. The study was approved by the Ethics Committee of Faculty of Philosophy/University of Prishtina “Hasan Prishtina”.

Results

A confirmatory factor analysis (CFA) of the 20 items was conducted in R with Lavaan. The two-factor model that best fitted the data contained 11 items for affective empathy and 9 items for cognitive empathy; furthermore, it shared the same factor structure with the original model. The 20-item model showed a poor fit to the data: χ2(169) = 712.13, p < 0.001; CFI = 0.811; TLI = 0.787; AGFI = 0.929; RMSEA = 0.094 (90% CI = 0.087, 0.102). The standardized factor loadings of the items ranged from 0.39 to 0.63 for positively worded items and between − 0.32 to − 0.69 for negatively worded items. Notably, one item, “I tend to feel scared when I am with friends who are afraid,” did not load significantly on the factor of affective empathy.

Next, the model that had an additional method factor containing negatively worded items was examined. The results indicated an adequate fit to the data: χ2(159) = 428.22, p < 0.001; CFI = 0.906; TLI = 0.888; GFI = 0.978; AGFI = 0.963; RMSEA = 0.068 (90% CI = 0.061, 0.076). The standardized factor loadings of the items ranged from 0.39 to 0.66 for positively worded items and between − -0.32 and − 0.69 for negatively worded items. The item “I tend to feel scared when I am with friends who are afraid” demonstrated a very low, but significant, standardized loading (0.11). Removing this item from the model only slightly improved the model fit. Therefore, a model containing all 20 items and an additional method factor was accepted as the final model. A post hoc power analysis was conducted using the semPower package (Moshagen & Erdfelder, 2016) in R. The results suggested that a sample size of N = 467 with a power larger than > 99.99% (1—beta = 1—6.463622e − 09) to reject a wrong model (with df = 159) with an amount of misspecification corresponding to RMSEA = 0.050 on alpha = 0.05.

Internal consistency reliabilities were acceptable for cognitive (α = 0.68) and affective (α = 0.73) empathy. The scales correlated positively and significantly with each other (r = 0.45, p < 0.01).

In order to examine measurement invariance, the factor structure of the BES was comparable between Males and females. The model fit for the male sample (n = 183) was poor: χ2(159) = 301.05, p < 0.001; CFI = 0.787; TLI = 0.746; GFI = 0.944; AGFI = 0.905; RMSEA = 0.082 (90% CI = 0.067, 0.096); power = 0.962. However, the model fit for the female sample (n = 293) was adequate: χ2(159) = 348.89, p < 0.001; CFI = 0.897; TLI = 0.877; GFI = 0.972; AGFI = 0.953; RMSEA = 0.073 (90% CI = 0.063, 0.083); power = 0.9996). Further examination of the factor loadings of the CFA models for males and females indicated that for the female subsample, item 15 (“I tend to feel scared when I am with friends who are afraid”) of the Affective empathy factor was non-significant (standardized loading = 0.01). For the male subsample, item 1 (“My friends’ emotions don’t affect me much”) and item 7 (“I don’t become sad when I see other people crying”) did not load significantly on the Affective empathy factor. Therefore, a measurement invariance based on gender could not be further explored.

Concurrent Validity

Next, the concurrent validity of the BES was examined. The cognitive and affective empathy scales correlated negatively and significantly with ICU scales. Table 3 presents all the relevant means, standard deviations, and correlations.

Table 3 Descriptive statistics and correlations between study variables
Full size table

Discussion

This current research was primarily conducted to test the psychometric properties of the Albanian version of the BES on a sample of youth from Kosovo. A reliability analysis and confirmatory factor analysis were completed; these resulted in a reliable and valid 20-item bidimensional Albanian version of the BES, which shares the factor structure of the original model (Jolliffe & Farrington, 2006). This result differs from the work of other researcher, specifically those working with delinquency, who have reported lower itemed bidimensional versions of the BES (Heynen et al., 2016; McLaren et al., 2019; Pechorro et al., 2017; Van Langen et al., 2015).

Additionally, using the Inventory of Callous Unemotional Traits scales, concurrent validity was confirmed, with both cognitive and affective empathy scales correlating negatively and significantly with ICU (Frick, 2004). The use of ICU for establishing concurrent validity for the BES is well established, with similar results having been replicated in previous research, with subjects reporting a decrease in ICU traits when there is an increase in either aspect of empathy (i.e., cognitive, and affective) (Frick & Kemp, 2021; Heynen et al., 2016; Pechorro et al., 2017).

However, upon the examination measurement invariance, differences were found in model fit for the separate male and female subsamples. Regarding the female sample, item 15 (“I tend to feel scared when I am with friends who are afraid”) of the Affective empathy factor was non-significant. For the male sample, two negatively phrased items, namely item 1 (“My friends’ emotions don’t affect me much”) and item 7 (“I don’t become sad when I see other people crying”) did not load significantly on the Affective empathy factor, a result also reported by other researchers (Heynen et al., 2016; Pechorro et al., 2015; Salas-Wright et al., 2012; Van Langen et al., 2015).

Heynen et al. (2016) suggest that this may be the result of misunderstanding due to low education levels; however, this study’s sample was composed of university students, challenging that assumption. Alternatively, what possibly links these samples is localized interpretations of masculinity. Research has shown both a link between a masculine ideology of patriarchal societies and prisons population to destructive masculine traits such as emotional suppression, expectations of dominance/aggression, and risk-taking behaviors, traits that also place downward pressure upon empathy (American Psychological Association, 2018; Gabbiadini et al., 2016; Ingram et al., 2019; Kupers, 2005; Pemberton, 2013; Ricciardelli et al., 2015).

Limitations

As this study was conducted with a sample of students from the University of Prishtina “Hasan Prishtina” which was characterized by gender and economic diversity, all the participants had a similar education level; a factor that should be considered for generalization purposes. While the final version of the Albanian BES contained 20 items, matching the original model, one item from the CFA models for males and females from the Affective empathy factor item 15, namely, “I tend to feel scared when I am with friends who are afraid,” demonstrated a very low, but significant, standardized loading (0.11). The removal of this item only slightly improved the model fit, so it was retained. In addition, owing to poor model fit in the factor structure between the male and female subsample, a measurement invariance based on gender could not be further explored.

To conclude, this paper adds to the literature on Kosovo empathy evaluation capabilities by supporting the construct validity and reliability of the Albanian version of the BES, a result in line with previous research. That said, limitations in sample diversity and poor factor structure between the male and female subsamples suggest that this should be intended for research purposes only. Therefore, prior to the potential adoption of the Albanian version of the BES as (1) a diagnostic measure in the identification of at-risk youth and (2) a successful measure in juvenile delinquency programs that aim to reduce ICU traits, additional research is needed using larger and more diverse samples, including educational, criminality, and clinical samples.