Examining the Interpersonal Reactivity Index (IRI) Among Early and Late Adolescents and Their Mothers

Abstract

We examined whether the Interpersonal Reactivity Index (IRI; Davis, 1980), consisting of Perspective Taking (PT), Empathic Concern (EC), Personal Distress (PD), and Fantasy (FN), is a psychometrically invariant empathy measure for early and late adolescents and their mothers. Confirmatory factor analyses demonstrated adequate properties and psychometric invariance across 2 Dutch samples (269 early adolescents, 232 late adolescents). Females scored higher than males on each subscale. Early adolescents scored lower than late adolescents on PT and FN, and higher on PD. The different groups showed similar subscale associations with psychosocial health indexes, and similar subscale contributions to a higher order empathy dimension. Most dimensions showed positive correlations between adolescents and mothers. The IRI appears adequate for examining empathy across the span of adolescence, as well as patterns between youths and mothers.

Acknowledgments

This research utilizes data from the RADAR Young and RADAR Old projects. RADAR has been financially supported by main grants from the Netherlands Organisation for Scientific Research (GB-MAGW 480–03-005), and Stichting Achmea Slachtoffer en Samenleving (SASS), and various other grants from the Netherlands Organisation for Scientific Research, the VU University of Amsterdam, and Utrecht University.

Notes

A CFI ≥ .90, RMSEA ≤ .08, and SRMR ≤ .10 indicate a model's acceptable fit to the data (Browne & Cudeck, 1993; Hu & Bentler, 1999).

We also tested four additional configurations, in which individual items were randomly assigned to parcels within their respective dimension. We increased the diversity of these alternatives by allowing no more than one parcel across all dimensions (e.g., no more than 1 parcel out of 12) to be repeated. These alternatives were tested independently for each of the three subgroups. All model fits were acceptable, with loadings > .40, CFIs ≥ .90, RMSEAs ≤ .08, and SRMRs < .07. We also examined parcels based on the item-to-construct balancing (T. D. Little et al., 2002), using the early adolescents as the reference group. Model fits were again acceptable, with loadings > .40, CFIs > .90, RMSEAs < .09, and SRMRs < .07. Further information is available on request.

The invariance tests described here were again conducted with the alternative parcel configurations referenced in footnote 2. Similar results were obtained across the different comparisons, in that no test showed a CFI ≥ –.010. Thus, these analyses further suggested that the null hypotheses of invariance should be retained.

This was also the case for a comparison of all adolescents versus all mothers, which is not reported in Table 2 because we did not combine these two groups for subsequent mean comparisons or correlational analyses.

We also examined differences in the latent means of the IRI factors, calculated within SEM. The same pattern of results was observed whether testing for differences based on observed or latent mean scores. These analyses are available on request.