The Constructs of Empathy, Mindfulness, Self-Compassion and Self-Rumination: An Exploratory Factor Analysis
Previous research has demonstrated correlation amongst empathy, mindfulness, self-compassion and self-rumination. The constructs of empathy and mindfulness, however, remain indeterminate; self-compassion is a relatively new construct; and self-rumination has only recently been shown to correlate negatively with each construct. Evaluation of construct validity is required to draw conclusions about group differences and justify particular interpretations of test scores through explanations of the behaviors the test scores summarize. The primary research objective in this study will be to examine the underlying factor structure of measures of empathy, mindfulness, self-compassion and self-rumination. The purpose of this study is to examine whether the constructs are discrete, overlapping, redundant, or indistinguishable. Three hundred forty-three participants were recruited from a large, ethnically diverse, Southwestern university (57 males and 286 females). The average age of the participants was 25 years of age (M = 24.94 years, SD = 9.193) with the youngest participant being 18 years old and the oldest participant being 62 years old. Participants’ self-identified race was as follows: 38.2% White/Caucasian, 23.9% Asian/Pacific Islander, 16.9% African-American, 14.6% Hispanic, 3.2% biracial, 2.6% Middle Eastern, and .6% American-Indian. Participants completed a demographic questionnaire and one psychological measure that was constructed using all items from five existing measures: Interpersonal Reactivity Index (IRI; Davis, 1983), the Kentucky Inventory of Mindfulness Skills (KIMS; Baer, Smith, Allen, 2004), the Rumination subscale of the Rumination-Reflection Questionnaire (RRQ; Trapnell and Campbell, 1999), the Self-Compassion Scale (SCS; Neff, 2003a), and the Marlowe-Crowne Social Desirability Scale (M-C) Short Form-C.
A series of principal-axis factor analyses with promax rotation (k=4) resulted in a comparison between 4-, 7-, and 12-factor solutions. 4-factor solution aligned with 34.1% of the overall variance. 7-factor solution aligned with 41.1% of the overall variance. 12-factor solution aligned with 48.2% of the overall variance. Overall, the 12-factor model was found to be the most interpretable, but not the most parsimonious. Across all factor solutions, the 12-factor model produced more consistency and strength across item-factor loadings, yielded the highest percentage of hyperplane counts (4-factor = 35.4%; 7-factor = 56.6%; 12-factor = 67.8%), provided good congruence with conceptually-derived organization of the measures under investigation, and accounted for the largest amount of variance. However, the 4-factor model was more parsimonious, indicated potential overlap/redundancy between constructs under investigation, and revealed that a higher-order factor may account for the several of the later factors within the 12-factor solution. These results confirm the need for future and ongoing study of these constructs and will hopefully lead to a more comprehensive understanding of empathy and mindfulness, two of the constructs that are core to the effectiveness of therapeutic work.