Abstract
In view of the shortcomings of previous research on empathy (e.g., no dimensional differentiation, predominantly variable-centered approaches, and a lack of sample diversity), we adopted a person-centered approach to identify distinct profiles of empathy among Chinese preschool teachers, and we examined how these profiles relate to certain outcomes. We identified three profiles—moderate empathy, high cognitive empathy–moderate affective empathy, and high empathy—that varied in the level and shape of the profile indicators. Subsequent analyses showed that participants with higher levels of overall empathy scored higher in sympathy, while those with higher levels of affective empathy and lower levels of cognitive empathy scored higher in emotional exhaustion. Finally, we discussed the theoretical and practical implications of the findings and the limitations of the study.
Introduction
Empathy, which is generally defined as understanding and sharing in another’s emotional state or context (Cohen & Strayer, 1996), is a highly valued virtue in society (Winning & Boag, 2015), and it has a great influence on individual psychology and behavior. For instance, a number of studies have shown that empathy is an important motivator of prosocial behavior (Lockwood, Searacardoso, & Viding, 2014; Morelli, Rameson, & Lieberman, 2014; Sahdra, Ciarrochi, Parker, Marshall, & Heaven, 2015; Telle & Pfister, 2016). In contrast, empathic impairment has been related to problematic behaviors and characteristics such as the dark triad personality traits (Łowicki & Zajenkowski, 2017; Pajevic, Vukosavljevic-Gvozden, Stevanovic, & Neumann, 2018) and offensive behavior (Jolliffe & Farrington, 2007; Mariano, Pino, Peretti, Valenti, & Mazza, 2016). Empathy training is a core element of therapy programs for many mental health and behavioral problems (such as irascibility and domestic violence; Vachon & Lynam, 2015).
Limitations of Current Research on Empathy
As outlined above, empathy is a topic in the field of psychology that is worth investigating. Despite the many advances in research on the structure, antecedents, and outcomes of empathy, studies on empathy have had several critical limitations.
First, previous studies have tended to address empathy as a whole (i.e., they have used the total scores or the mean scores of empathy measures) and often ignored the unique role of each empathic dimension. According to many cognitive neuroscience and psychometrics studies, empathy consists of two interrelated but relatively independent components: affective empathy and cognitive empathy (Rueda, Fernández-Berrocal, & Baron-Cohen, 2015; Vossen, Piotrowski, & Valkenburg, 2015; Walter, 2012). The former refers to the ability to experience others’ emotions, whereas the latter pertains to the comprehension of others’ emotional states (Walter, 2012; Wang, Wen, Fu, & Zheng, 2017). Cognitive empathy and affective empathy often relate to their criterion variables (i.e., antecedents or outcomes) in different size or direction; however, many studies have not distinguished between these two components. For example, narcissism was negatively correlated with affective empathy but was positively correlated with cognitive empathy (Pajevic et al., 2018; Wai & Tiliopoulos, 2012). In another, physical aggression was negatively correlated with affective empathy but was uncorrelated with cognitive empathy (Vossen et al., 2015). However, many researchers have used the total empathy score rather than scores on specific dimensions of empathy when analyzing the relationship between empathy and narcissism or physical aggression (e.g., Wang, Chen, Xiao, Ma, & Zhang, 2012; Zhou, Zhou, & Zhang, 2010). Thus, the unique contributions and functions of each empathic component were concealed.
Second, although some studies distinguish between affective empathy and cognitive empathy, they all adopt a variable-centered approach, focusing individuals’ whole level of these two empathic components, and then discussing the relationship between certain empathic component and its antecedents or outcomes. A variable-centered approach assumes that all individuals belong to either a single population or multiple subpopulations with known subpopulation memberships (e.g., gender); thus, they cannot address the concern whether there are several subpopulations representing different classes of empathy.
Finally, previous studies on empathy have included limited samples. Most research has been conducted with student groups, while other studies have been conducted with helping professionals, such as medical staff and psychological consultants. The lack of sample diversity has limited the external validity of studies on empathy. In addition to the typical helping professions, there are many other vocations, such as preschool teaching, that require intensive emotional involvement. The preschool period is critical for individual development, and it has different characteristics than other developmental periods have (Li & Zhou, 2010). During this period, children’s emotions become increasingly differentiated, unstable, and easily influenced by external situations (Liu, Liu, Chen, & Huang, 2017). These characteristics call for greater empathic abilities among preschool teachers than among other teachers. Conducting research with preschool teachers is conducive to the expansion of the scope of empathic research to include more diverse samples. In addition, it can enable us to understand the emotional characteristics of preschool teachers and provide guidance for their mental health.
Current Study
The purpose of this study is to use a person-centered approach (i.e., latent profile analysis [LPA]; Lanza, Flaherty, & Collins, 2003) to identify different empathy profiles of preschool teachers. We then link these profiles to outcomes to verify their validity.
Hypotheses
Latent profiles of preschool teachers’ empathy
To the best of our knowledge, previous research has not discussed latent profiles of empathy. As a result, the current study lacks empirical references to inform predictions of latent profile membership. Thus, we adopt an exploratory approach. Based on the relative independence of the two empathic components (Decety, 2010; Nummenmaa, Hirvonen, Parkkola, & Hietanen, 2008; Shamay-Tsoory, Aharon-Peretz, & Perry, 2009; Vossen et al., 2015), we expect to find at least two of the following profiles among Chinese preschool teachers: (a) relatively high in both empathic components, (b) relatively high in cognitive empathy and low in affective empathy, (c) relatively high in affective empathy and low in cognitive empathy, and (d) relatively low in both empathic components.
Outcomes
To verify the validity of the LPA results, we examine the mean differences of three outcome variables across the various profiles. The three outcome variables are sympathy, prosocial behavior, and emotional exhaustion. Sympathy, an important psychological construct intimately tied to prosocial and moral behavior (Eisenberg et al., 1989), refers to a feeling of concern or sorrow for another person’s distress (Clark, 2010; Edwards et al., 2015) and is often confused with affective empathy (Vossen et al., 2015). However, as many studies have noted, sympathy is an outcome of empathy (Edwards et al., 2015; Geng, Xia, & Qin, 2012; Reniers, Corcoran, Drake, Shryane, & Völlm, 2011), and it positively correlates with both cognitive empathy and affective empathy (Vossen et al., 2015; Wang, Wen et al., 2017). Sympathy and affective empathy are similar in that both are emotional responses to other people, while sympathy and cognitive empathy are similar in that both include cognitive components (Wang, Wang et al., 2017). Thus, despite a lack of consensus regarding what kind of empathy is more relevant in relation to sympathy, we could reasonably hypothesize that the profile with a relatively high overall level of empathy would have a relatively high level of sympathy, while the profile with a relatively low overall level of empathy would have a relatively low level of sympathy.
Prosocial behavior refers to behaviors that are helpful, charitable, giving, self-sacrificing, and often unrewarded (Rosenhan, 1978). Past research has shown that individuals with high cognitive and affective empathy exhibit more prosocial behavior than individuals with low cognitive and affective empathy (Ding & Lu, 2016; Findlay, Girardi, & Coplan, 2006). Therefore, we hypothesized that the profiles with higher overall levels of empathy would have higher levels of prosocial behavior.
Emotional exhaustion, which is a core component of job burnout and can be affected by empathy (J. Y. Lee et al., 2016; Tei et al., 2014; Wróbel, 2013), refers to feelings of being overextended and depleted of one’s affective and physical resources (Maslach, Schaufeli, & Leiter, 2001). Although few researchers have discussed theories regarding the effects of each empathic component on emotional exhaustion, affective empathy is a type of emotional resource consumption (Tan, Zou, He, & Huang, 2011), and cognitive empathy, which lacks emotional involvement, should not be correlated with emotional exhaustion. Thus, we hypothesized that profiles with stronger affective empathy but not cognitive empathy would correspond to stronger emotional exhaustion.
Method
Participants and Procedure
The participants were 300 preschool teachers recruited in mainland China. They were between 18 and 54 years old (M = 33.65; SD = 9.09); 292 were female (97.3%), two were male (0.7%), and six did not provide information on gender (2.0%); 129 were in charge of senior class (43.0%), 83 were in charge of middle class (27.7%), 73 were in charge of junior class (24.3%), and 15 did not provide information regarding classes taught (5.0%); 122 were lead teachers (40.7%), 97 were assistant teachers (32.3%), 72 were childcare teachers (24.0%), and nine did not provide position information (3.0%); 198 were married (66.0%), 86 were single (28.7%), three reported a different marital status (1.0%), and 13 did not provide marital status information (4.3%). The participants’ years of teaching experience ranged from 1 to 33 (M = 11.88; SD = 9.28). All participants gave informed consent before filling out the questionnaire.
Measures
Measure of Empathy and Sympathy (MES)
The MES is a 12-item scale consisting of three dimensions with four items each (Vossen et al., 2015). The three dimensions are as follows: cognitive empathy (e.g., “I can tell when a friend is angry even if he or she tries to hide it”), affective empathy (e.g., “When a friend is angry, I feel angry too”), and sympathy (e.g., “I feel sorry for someone who is treated unfairly”). The first two dimensions are used to assess empathy, and the third is used to assess sympathy. Respondents express their agreement with each item on a 5-point Likert-type scale ranging from 1 (never) to 5 (always). The measure has demonstrated good reliability and validity for Dutch adolescents (Vossen et al., 2015), Turkish middle school students (Zengin, Caka, & Çınar, 2018), and Chinese college students and preschool teachers (Wang, Wang, et al., 2017; Wang, Wen, et al., 2017). In the present study, α was .69 for cognitive empathy, .71 for affective empathy, and .79 for sympathy.
Prosocial Tendencies Measure (PTM)
The Chinese version of the PTM (Kou, Hong, Tan, & Li, 2007) consists of 26 items (e.g., “I can help others best when people are watching me”). It is the most popular measure of prosocial behavior among Chinese individuals, and it has demonstrated good reliability and validity (Kou et al., 2007). The participants rated the items on a 5-point Likert-type scale ranging from 1 (does not describe me at all) to 5 (describes me greatly). In the current study, α was .88 for the PTM. Among the participants, 125 did not fill out this measure.
Maslach Burnout Inventory–Educators Survey (MBI-ES)
We employed the eight-item emotional exhaustion dimension (e.g., “I feel used up at the end of the day”) of the Chinese version of the MBI-ES (Maslach et al., 2001; Ye et al., 2017). Responses are given on a 5-point Likert-type scale ranging from 1 (never) to 5 (always). A higher mean score indicates more intense emotional exhaustion. In the current study, α was .80 for the emotional exhaustion dimension.
Analytical Approach
Our analyses consisted of two steps. First, we employed LPA to determine the number of profiles that best fit the data. Then, to verify the validity of the extracted latent profiles, a one-way analysis of covariance was conducted to test the differences in the outcome means across the various profiles. The LPA was performed with Mplus 7.4 (Muthén & Muthén, 2015), and the other analyses were performed with SPSS 23 (SPSS Inc., Chicago, IL, USA).
Results
Preliminary Analyses
Table 1 shows the means, standard deviations, and correlations of the variables. Cognitive empathy was positively related to all other variables except emotional exhaustion. Affective empathy was positively related to all outcomes.
Means, Standard Deviations, and Correlations.
Note. N = 300 except for the statistics involving prosocial behavior (n = 175).
p < .05. **p < .01. ***p < .001.
LPAs
In the LPA, we set up five competition models. Models 1 to 5 contained 1 to 5 latent profiles, respectively. Akaike’s information criterion (AIC; Akaike, 1987), the Bayesian information criterion (BIC; Schwartz, 1978), the sample-size–adjusted Bayesian information criterion (SSABIC; Tofighi & Enders, 2007), entropy (Muthén & Muthén, 2015), the Lo–Mendell–Rubin (LMR) likelihood ratio test (Lo, Mendell, & Rubin, 2001), and the bootstrapped likelihood ratio test (BLRT; Nylund, Asparouhov, & Muthen, 2007) were used to evaluate and select the models. Smaller AIC, BIC, and SSABIC indicate a better model fit. The value of entropy represents the correct rate of individual classification, and the larger the value, the better the corresponding model. The k class model is superior to the corresponding k – 1 class model when the p value of LMR and BLRT is less than .05 (Morin, Morizot, Boudrias, & Madore, 2011).
Table 2 provides the fit statistics for the LPA. From Models 1 to 5, AIC and SSABIC gradually decreased; BIC dropped the lowest in Model 2; entropy reached the highest in Model 5; the p values of the LMR for k versus k – 1 classes were only significant in Models 2 and 3; and the p values of the BLRT for k versus k – 1 classes were significant for Models 2 to 5. Taken together, Models 2, 3, and 5 were better fits than Models 1 and 4. Finally, we chose the solution with three profiles (Model 3) as the optimal model on the following grounds: (a) Model 3 was superior to Model 2 according to most fit indices (AIC, SSABIC, LMR, and BLRT); (b) Although BIC showed that Model 2 was better, some studies have found that SSABIC is a superior index to BIC (Finch & Bronk, 2011; Nylund et al., 2007); and (c) For Model 5, class sizes of three-fifths profiles were too small (n < 30) to allow generalization of the findings to the broader population (Finch & Bolin, 2017).
Fit Indices for the Latent Profile Analysis of the MES.
Note. N = 300. Boldface font indicates the selected model. MES = Measure of Empathy and Sympathy; LL = log likelihood; AIC = Akaike’s information criterion; BIC = Bayesian information criterion; SSABIC = sample-size–adjusted Bayesian information criterion; p (LMR) = p value of the Lo–Mendell–Rubin likelihood ratio test; p (BLRT) = p value of the bootstrapped likelihood ratio test.
Additional support for a three-profile solution is shown in Table 3. Values along the diagonal reflect the average probability that participants were correctly categorized in a given latent profile, whereas the off-diagonal values reflect the average probability that participants were miscategorized. For instance, a participant whose most likely latent profile membership was in Profile 1 had an 85.8% chance of being correctly categorized but only a 4.4% chance of being incorrectly categorized in Profile 3. Thus, participants had a high likelihood of being categorized in the correct latent profile and a small likelihood of being incorrectly categorized.
Average Latent Class Probabilities for Most Likely Latent Class Membership (Row) by Latent Profile (Column).
Note. N = 300. Values in bold along the diagonal reflect the average probability that participants were correctly categorized in the given latent profile.
Figure 1 plots the estimated mean of each MES item for each of the three latent profiles. Together, these item means provide a sense of what characterize the three empathy profiles among Chinese preschool teachers. The first profile (25.4%) was characterized by moderate levels of both empathic dimensions. We thus named this profile the moderate empathy group (ME). Profile 2 (33.8%) was labeled the high cognitive empathy–moderate affective empathy group (HCEMAE) because individuals with this profile had high levels of cognitive empathy and moderate levels of affective empathy. Finally, individuals with Profile 3 (40.8%) were labeled the high empathy group (HE) because they had high levels of both empathic dimensions. A subsequent one-way analysis of variance showed that, across the three profiles, the mean differences in overall levels of empathy (i.e., the mean scores of both empathic dimensions of the MES) or in the levels of both empathic dimensions were statistically significant, as shown in Table 4.
Three-profile solution of the MES.
Difference in Means of Empathy, Sympathy, Prosocial Behavior, and Emotional Exhaustion Across Latent Profiles.
Note. N = 300 except for the statistics involving prosocial behavior (n = 175). ME = moderate empathy group; HCEMAE = high cognitive empathy–moderate affective empathy group; HE = high empathy group.
p < .05. **p < .01. ***p < .001.
Relationships Between the Empathy Profiles and the Outcomes
A one-way analysis of covariance, in which we controlled for all the demographic variables (nominal variables were encoded as dummy variables), namely, age, years of teaching experience, gender, class, position, and marital status, was adopted to perform a comparison among the profiles on each of the outcome variables. The results are presented in Table 4.
The mean levels of sympathy varied across the three profiles (F = 20.77, p < .001, η2 = .15). Specifically, preschool teachers in the HE profile (M = 4.54, SE = .05) had higher mean sympathy than those in either the ME profile (M = 3.92, SE = .08; p < .001) or the HCEMAE profile (M = 4.27, SE = .06; p < .001); and teachers in the HCEMAE profile had higher mean sympathy than those in the ME profile (p < .01). Unexpectedly, the mean difference in prosocial behavior across the three profiles was not significant (F = 1.83, p > .05, η2 = .02).
Finally, the mean difference in emotional exhaustion across the three profiles was significant (F = 5.56; p < .01, η2 = .04). Specifically, the preschool teachers in both the ME profile (M = 2.54, SE = .09; p < .05) and the HE profile (M = 2.55, SE = .06; p < .01) had higher mean emotional exhaustion than those in the HCEMAE profile. No significant mean difference was found between the ME profile and the HE profile.
Discussion
Using a person-centered approach, we explored empathic profiles and examined the relationship between the extracted profiles and outcome variables. Our results allowed us to identify three latent profiles: the ME, HCEMAE, and HE latent profiles. We also found that teachers with higher levels of overall empathy scored higher in sympathy than did teachers with lower levels of overall empathy. This finding was consistent with our hypothesis and verified the validity of the LPA results.
However, no significant difference in prosocial behavior was found across the various profiles. This is not consistent with the findings of previous studies or with the current study’s correlation analysis results (without demographic variables as controls), which yielded a positive association between empathy and prosocial behavior. After comparing the correlation analysis and the analysis of covariance, we speculated that this finding of no significant difference in prosocial behaviors across the latent profiles was due to the following reasons: (a) empathy has only a low to moderate relation with prosocial behavior (see, for example, Table 1 and the meta-analysis of Eisenberg & Miller, 1987) and (b) the demographic control variables, especially those that can positively affect prosocial behavior, weaken the effect of the latent empathy profiles on prosocial behavior. 1
In addition, our hypotheses about the relation between empathy profiles and emotional exhaustion were only partially supported. We found that teachers with relatively high levels of affective empathy and relatively low levels of cognitive empathy scored higher in emotional exhaustion. Therefore, both empathic components may affect emotional exhaustion, but their effects are in opposite directions. 2 For example, the cognitive empathy levels of the HE and the HCEMAE profiles are similar, but the former has a higher level of affective empathy and higher emotional exhaustion than the latter; and while the affective empathy levels of the ME and the HCEMAE profiles are similar, the former has lower levels of cognitive empathy and higher emotional exhaustion than the latter. We explain the effect of cognitive empathy on emotional exhaustion as follows: Cognitive empathy itself does not involve emotional reactions, but individuals with high cognitive empathy can identify their and others’ negative emotions in a timely manner, and this ability will help them take the emotion regulation strategies required to avoid an excessive loss of internal resources (Hunt, Denieffe, & Gooney, 2017).
Theoretical and Practical Implications
This study offers three contributions. First, studies on empathy have mainly taken a variable-centered perspective. We extended previous studies by using a person-centered approach for the first time. Our results revealed three qualitatively different profiles and validated previous work on empathy structure, which has found that, although cognitive empathy and affective empathy are related, they are also relatively independent (Rueda et al., 2015; Walter, 2012). We found a variety of combinations of empathic components for different individuals.
Second, while many previous studies on empathy have ignored the unique role of each empathic dimension, the analyses in this study, which include correlation analysis, LPA, and analysis of covariance, distinguished between cognitive and affective empathy. We found that these two empathic components have different effects in some cases. For instance, the HE and HCEMAE profiles had significant differences in affective empathy but not cognitive empathy, and these two profiles differed in the mean score of emotional exhaustion.
Third, this study identified not only benefits of empathy but also disadvantages. This information can help preschool teachers treat and make use of empathy more effectively. This study found that preschool teachers with relatively high levels of overall empathy had higher levels of sympathy. Sympathy is regarded as the main motivator of prosocial behavior (S. A. Lee & Gibbons, 2017). This means that empathy is generally helpful for preschool teachers in their work. However, at the same time, both the correlation analysis and the analysis of covariance showed that emotional exhaustion is positively correlated with affective empathy. This suggests that too much affective empathy may bring about adverse effects on preschool teachers’ mental health. We attribute this phenomenon to the following two possible explanations: (a) Frequent experiences of affective empathy can cause excessive physiological arousal, which consumes an individual’s internal resources and thus leads to further emotional exhaustion and (b) studies have shown that individuals with high affective empathy are more likely to notice other people’s negative emotions (Chikovani, Babuadze, Iashvili, Gvalia, & Surguladze, 2015). Experiencing others’ negative emotions frequently is undoubtedly harmful to individuals’ well-being. Therefore, although empathy is generally beneficial for preschool education, it may be necessary to treat affective empathy with caution. We argue that, given the wide range of children’s emotions, preschool teachers should make better use of cognitive empathy in teaching activities, identify children’s maladaptive moods in a timely fashion, and adopt effective guidance and intervention strategies rather than always vicariously experiencing their students’ emotions.
Limitations of the Study and Future Research Directions
This study explored profiles of empathy among preschool teachers. In this professional group, there are far more women than men (in China), so it is difficult to balance the proportion of women and men in the sample. Consequently, we cannot use gender as a criterion to discuss its correlation with empathy or generalize the LPA results obtained in this study to other gender or professional groups. In the future, researchers may consider repeating this study with other groups. Moreover, we used only three covariates to verify the LPA results, and all the covariates are the outcomes of empathy, without any antecedents. This issue could be addressed by using additional covariates in future work.
Conclusion
Through LPA, we identified three empathy profiles that varied in the level and shape of the profile indicators, and we demonstrated that membership in distinct latent profiles affects critical outcomes (e.g., sympathy and emotional exhaustion). These findings highlight the diverse forms of empathy among different preschool teachers in China.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by grants from the National Natural Science Foundation of China (31771245).