TPC Journal V8, Issue 4- FULL ISSUE

The Professional Counselor | Volume 8, Issue 4 347 Vossen and colleagues suggested the AMES might be an effective measure of empathy and sympathy with alternate samples as well. Bloom and Lambie (in press) examined the factor structure and internal consistency of the AMES with a sample of emerging adult college students in the United States ( N = 1,598) and identified a 3-factor model fitted to nine items that demonstrated strong psychometric properties and accounted for over 60% of the variance explained (Hair et al., 2010). The modified 3-factor model included the same three factors as the original AMES. Therefore, we followed Bloom and Lambie’s modifications for our use of the instrument. Data Screening Before running the main analysis on the variables of interest, we assessed the data for meeting the assumptions necessary to conduct a one-way between-subjects MANOVA. First, we conducted a series of tests to evaluate the presence of patterns in missing data and determined that data were missing completely at random (MCAR) and ignorable (e.g., < 5%; Kline, 2011). Because of the robust size of these data (e.g., > 20 observations per cell) and the minimal amount of missing data, we determined listwise deletion to be best practice to conduct a MANOVA and to maintain fidelity to the data (Hair et al., 2010; Osborne, 2013). Next, we utilized histograms, Q-Q plots, and boxplots to assess for normality and identified non- normal data patterns. However, MANOVA is considered “robust” to violations of normality with a sample size of at least 20 in each cell (Tabachnick & Fidell, 2013). Thus, with our smallest cell size possessing a sample size of 115, we considered our data robust to this violation. Following this, we assumed our data violated the assumption for multivariate normality. However, Hair et al. (2010) stated “violations of this assumption have little impact with larger sample sizes” (p. 366) and cautioned that our data might have problems achieving a non-significant score for Box’s M Test. Indeed, our data violated the assumption of homogeneity of variance-covariance matrices ( p < .01). However, this was not a concern with these data because “a violation of this assumption has minimal impact if the groups are of approximately equal size (i.e., largest group size ÷ smallest group size < 1.5)” (Hair et al., 2010, p. 365). It is necessary to note that MANOVA is sensitive to outlier values. To mitigate against the negative effects of extreme scores, we removed values ( n = 3) with standardized z -scores greater than +4 or less than -4 (Hair et al., 2010). This resulted in a final sample size of 868 participants. We also utilized scatterplots to detect the patterns of non-linear relationships between the dependent variables and failed to identify evidence of non-linearity. Therefore, we proceeded with the assumption that our data shared linear relationships. We also evaluated the data for multicollinearity. Participants’ scores of Affective Empathy shared statistically significant and appropriate relationships with their scores of Cognitive Empathy ( r = .24) and Sympathy ( r = .43). Similarly, participants’ scores of Cognitive Empathy were appropriately related to their scores of Sympathy ( r = .36; p < .01). Overall, we determined these data to be appropriate to conduct a MANOVA. Table 2 presents participants’ scores by academic discipline.