Introduction
Empathy is defined by Farrow (2007) as the ability or process to identify and understand other person’s situation, feelings, and motives. This process implies a shared interpersonal experience to perceive and understand the emotional state and feeling with another person, and the ability to infer the origin of their emotional state. These affective and cognitive responses are essential for relationships and social behavior (Moore, Dev, Jeste, Dziobek, & Eyler, 2015; Lamm, Batson, & Decety, 2007).
Several information processes are relevant to empathy: perception and attending socially relevant stimuli, self-awareness, memory, emotion processing, executive control, perspective taking, selection of appropriate behavior, inhibition of proponent self-focus, reasoning, and Theory of Mind (ToM) (Farrow, 2007; Lamm et al., 2007). Therefore, three major clusters that include all the components of empathy have been described: a) cognitive capacity, b) affective response to understand emotional states, and c) the origin of feelings (self or others). These clusters are relevant in terms of theoretical conceptualization and for the objective assessment of the construct (Perry & Shamay-Tsoory, 2013; Kim & Lee, 2010).
Another important dimension of the concept of empathy is related to the difference between cognitive empathy and affective empathy. The first refers to the ability to perceive, interpret, infer, and explain the emotional state of another person, while the second process, affective empathy, involves inferring with another person’s mental state, which overlaps with ToM. Both systems can function with a certain independence from one another and are necessary to display and adaptive social behavior (Moore et al., 2015; Zaki & Ochsner, 2013; Pijnenborg, Spikman, Jeronimus, & Aleman, 2013; Marsh, 2018).
The assessment of empathic abilities has a critical importance for mental health. The currently available scales, such as Hogan´s scale, Mehrabian & Epstein questionnaire, and Interpersonal Reactivity index or IRI (Davis, 1983; Leiberg & Anders, 2006; Escrivá, Navarro, & García, 2004) are instruments that allow for an approach to the objective evaluation of empathy, but have several limitations to be used in clinical settings (psychotic patients, autism, personality disorders), Baron-Cohen et al. (2003; 2004) created the Empathy Quotient (EQ), a scale explicitly designed to have a clinical application and to be sensitive to the lack of empathy as a feature of psychopathology. The instrument evaluates three constructs of empathy and several studies around worldwide (Wakabayashi et al., 2006; Kim & Lee, 2010; Preti et al., 2011; Berthoz, Wessa, Kedia, Wicker, & Grèzes, 2008; Jankowiak-Siuda et al., 2017; Zhang et al., 2018) have tested its psychometric properties with adequate results. Until now, a validated self-report measure of empathy in Mexico is not available. The aim of the present study was to prove the psychometric properties of the EQ in a Mexican sample.
Method
Participants
Recruitment procedure included a non-probabilistic sampling approach: subjects from the general population of Mexico City who were available and willing to participate in the study were enlisted. After discarding any Axis I psychiatric disorder through the Mini-International Neuropsychiatric Interview MINI (Sheehan et al., 1998), a total of 200 adults were included. The sample included 124 (62%) women and 76 men (38%) with a mean age of 30.4 years (SD = 10.5, range 18-64).
Measurement and procedure
The MINI, a brief structured interview, adapted for Central and South America (Sheehan et al., 1998) was used to exclude subjects with any Axis I psychiatric disorders according to the Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM–IV–TR: American Psychiatric Association, 2000).
The Empathy Quotient (EQ) is a 60-item self-report questionnaire aimed at measuring empathy (Lawrence, Shaw, Baker, Baron-Cohen, & David, 2004; Baron-Cohen, 2012). Twenty items are used as filler, so the original scale comprises 40 items for analysis. Each item is scored initially in a 4 - point agreement scale ranging from “strongly agree” to “strongly disagree”. According to the sense of the item and the subject response, each item is then scored according to the magnitude of the empathic response, with 0 for a “non-empathic” response, and 1 for “slightly empathic” and 2 for a “strong empathic response”. The original EQ has been reduced and validity and reliability proved by Lawrence et al. (2004). This version comprises 28 items arranged in three different empathy dimensions: cognitive empathy (11 items), emotional reactivity (11 items), and social skill (6 items); for the present study, we used the version developed by Lawrence et al. (2004).
Recruitment for the study began after the translation procedure was completed in accordance with the American Research Teams (Pan & de la Puente, 2005). The EQ was first translated from English to Spanish by two independent translators and if any discrepancy arose, it was reviewed by an independent mental health professional and consensus was reached for an adequate language adaptation.
Statistical analyses
Analyses were performed with the 22.0 version of the SPSS statistical software. Skewness and kurtosis were used to test the normal distribution of the EQ 28-item scale proposed by Lawrence (2004). The procedures (Nunnally, 1994; Reyes-Lagunes & García y Barragán, 2008) for testing the psychometric properties of the EQ-28 were done as follows: First, an item by item frequency analysis was performed to determine if all answer options were attractive and was followed by a discriminative analysis by means of Student t-tests (including extreme quartiles) where items with a non-significant value (p > .05) were eliminated.
Then, an Exploratory Factor Analysis (EFA) with equamax rotation was performed. The equamax rotation was selected in view of the fact that subjects may report several EQ features concurrently. Items whose communality was higher than .40 were retained and allocated to factors, while those with a communality of less than .40 were excluded. After EFA, the internal consistency of the EQ dimensions and the total score were estimated with the Cronbach’s alpha coefficient.
A correspondence factorial analysis was done to test the relation between response options defined as: a) 0 = Non-empathic response, b) 1 = Slightly empathic and c) 2 = Strong empathic response, and the factors obtained from the EFA. Finally, we tested the congruence coefficient factor of our data with the factor loadings obtained by Lawrence et al. (2004), using the formula proposed by Wrigley and Neuhaus (1955) with > .60 coefficients considered as congruent.
Ethical considerations
The study protocol and procedures were approved by the Institutional Review Board of the INPRFM. All participants received a complete explanation of the nature and procedures of the study and those who voluntarily accepted to participate provided a written informed consent.
Results
The 28 items of the EQ scale showed acceptable values of skewness and kurtosis, where none of the values was excessively out of range (skewness ranged from -.56 to 1.41 and kurtosis ranged from -1.52 to .85) except for item 1: “I can easily tell if someone else wants to enter a conversation” with a skewness value of -1.83 (SD = .17), and kurtosis of 2.51 (SD = .34). No items were eliminated according to the criteria used in the discriminative analysis of items (average t = 7.20, p > .05).
Construct validity and internal consistency of the EQ
Using the 28 original items of the EQ, the results of the equamax rotation of the items of the EQ accounted for 35.3% with a Kaiser-Meyer-Olkin sampling adequacy of .79 and a significant Bartlett test (p < .001). Of the original 28 items used for the analysis, 22 were included in the factor model. Twenty items displayed adequate communalities in the original designated domains of the EQ (10 items from the Cognitive empathy dimension, seven from the Emotional reactivity dimension, and three from the Social skills dimension and are displayed in Table 1. Two items were related to a different domain. From the Emotional reactivity dimension, item 21, “It is hard for me to see why some things upset people so much,” displayed communality with the Social skills dimension, while item 12 “Friendships and relationships are just too difficult, so I tend not to bother with them” from the Social skills dimension, displayed communality with the Emotional reactivity dimension.
The Cronbach’s alpha coefficient for the combined 22 items was .80. As can be seen in Table 1, high reliability was also obtained for the Cognitive empathy dimension. Nevertheless, Emotional reactivity and Social skills dimensions exhibited lower internal consistency values.
Correspondence factorial analysis and congruence factor analysis
The relations between the three dimensions of the EQ obtained in the exploratory factor analysis and the scored empathic response are shown in Figure 1. Non-empathic response had a higher contribution to the Cognitive empathy factor, while the strong empathic response was more related to the Emotional reactivity factor. The social skills factor was theoretically more related to the factor of Cognitive empathy, but responses were more related to an empathic response as observed by its proximity to the Emotional reactivity factor.
Adequate congruence coefficients were obtained between the 28-item version factors of the EQ and the factors obtained in the present study. The highest coefficient was observed in the Cognitive empathy factor (.97, p < .001) followed by the Emotional reactivity factor (.91, p < .001) and finally, the Social skills factor (.89, p = .01). Cognitive empathy and Emotional reactivity factors are clearly differentiated and related to different empathic responses. The factor of Social skills is theoretically more related to the Cognitive empathy but the responses to this factor are more related to an empathic response as observed by its proximity to the Emotional reactivity factor.
Discussion and conclusion
The aim of the present study was to prove the psychometric properties of the 28-item version of the EQ in a Mexican sample. Cultural adaptation of instruments is indispensable for the development of appropriate measurements according to the specific population of interest. Therefore, it is essential to have validated instruments for the Mexican population to improve research in social cognition in psychiatry and to perform further cross-cultural studies.
An adequate internal consistency (Cronbach’s alpha = .80) of the EQ was obtained in the present study. In similar studies, adequate internal consistency has been achieved with a reduced number of items compared to the original scale. In the Japanese version of the EQ, Wakabayashi et al. (2006) used a 22-item version and obtained a .90 Cronbach’s alpha. Similar results can be found in the work of Muncer and Ling (2006) and Kim and Lee (2010) who reported the internal consistency of .84 and .78 Cronbach´s alpha in a 28-item version for the EQ scale.
Our study supports the existence of three main dimensions of empathy assessed with the EQ as those obtained by Lawrence et al. (2004), although our analysis includes only 22 of the 28 original proposed items. Cognitive empathy exhibits almost the same structure with 10 of the original 11 items proposed. Item 1 “I would be very upset if I couldn’t listen to music every day,” which was originally in this dimension was not included, possibly as it refers to a specific stimulus –music– that may play a role different to empathy in the present Mexican sample. This item, for example, loads in the Emotional reactivity factor in the Chinese validation of the EQ (Zhang et al., 2018), in the Social skills factor (with a low factor loading of .31) in the study of Lawrence et al. (2004), and is not included in the study performed by Muncer and Ling (2006). However, the remaining 10 items corresponding to the original items of this dimension explaining 20.0% of variance and with the highest reliability.
The Social skills factor, which explains 8.8% of the variance, had important differences when compared to the 28-item version of the EQ. Our factor analysis just includes three of the six items (item 4, item 8 and item 14). Item 35 “I don’t tend to find social situations confusing” and item 57 “I don’t consciously work out the rules of social situations” displayed loadings lower than the threshold of .40 and were not included in the model. Item 57 was also not included in the Chinese validation of the EQ by Zhang et al. (2018) and neither in the 28-item version of the EQ (Lawrence, 2004); this item displayed low factor loadings (.39). Nevertheless, this item, as well as item 35 were adequately located in the Social skills factor in other studies. Additionally, item 21 “It is hard for me to see why some things upset people so much,” originally displayed in the Emotional reactivity factor, loaded in the Social skills factor in a Mexican population. The Cronbach’s alpha for the Social skills factor was .60 which is lower but may be considered still acceptable given that this factor only has four items.
Similarly according to what was observed in the Social skills factor, two items of the Emotional reactivity factor, item 22 “I find it easy to put myself in somebody else’s shoes” and item 29 “I can’t always see why someone should have felt offended by a remark,” displayed loadings lower than .40 a result also observed in the study of Lawrence et al. (2004), where these items showed loadings of .38 and .33, respectively; thus, the fact that they have not been included in any dimension of the factorial analysis is to be expected. Also, item 12, “Friendships and relationships are just too difficult, so I tend not to bother with them,” loads on this factor and not in the Social skills factor as originally stated. We assume that words such as friendship and relationships may evoke an emotional meaning more than a social meaning for individuals in our sample and may be the reason why this item load in this dimension. Although the Cronbach’s alpha for this dimension is equal to the one obtained in the Social skills dimension (.60); this value should be considered as inappropriate for internal consistency as this dimension comprise eight items.
Much of what is observed in the factor analysis can be explained with the results of the correspondence analysis. As can be seen, the Cognitive empathy factor is clearly separated from the other two factors and is related to a non-empathic response. However, even the Social skills and the Emotional reactivity factors are theoretically separated from each other, and it seems that individuals respond to these factors in a more empathic and similar way, leading to what can be considered as an affective response. Therefore, we hypothesize that, for Mexican population, a mixed meaning of the items of these two dimensions may cause these inconsistencies due to cultural setting and that they should be studied further in other samples and with a confirmatory factor analysis.
Despite the above, the present version of the EQ, comprising 22 items, exhibits high congruence coefficients (> .85) with the 28-item version of the scale which means that both instruments evaluate the three dimensions of empathy in a similar way. However, the low internal consistency reported in the Social skills and Emotional reactivity dimensions emphasize the need for additional studies of these dimensions and how items are constructed to assess these important components related to empathy.
Despite the limitations found with these dimensions, we conclude that this first approach to validate the EQ show its adequacy for its use in Mexican population as the 22-item version of the EQ showed a global high reliability, consistency, and congruence with the proposed factor structure. Another limitation regarding our sample is the probable bias that can be observed in terms of sex, age and, education, where more than half of the recruited individuals were women, no information regarding educational status was obtained, and the wide age range of recruited participants. Therefore, the use of the EQ should be taken with caution considering these limitations.