Brief report: How many dimensions in the prosocial behavior scale? Psychometric investigation in French-speaking adolescents

Abstract

The purpose of the study was to evaluate the factor structure and the reliability of the Prosocial Behavior Scale (PBS). To our knowledge, no factorial validity of the multifactorial structure of PBS has been published to date. The psychometric characteristics of the PBS were examined in several samples of French adolescents (aged 11–19, N1 = 1141, N2 = 1071, and N3 = 1640) using Confirmatory factor analyses (CFA). The original four-factor structure was not confirmed due to lack of discriminant validity. CFA led us to retain a two-factor solution with a good fit and a satisfactory reliability. Furthermore, the results support the convergent validity of the PBS: helping and caring dimensions were positively correlated with empathy. In addition, partial measurement invariance across gender and grade was attested. In conclusion, the results indicate that the French version of the PBS is a useful instrument for the assessment of prosocial behaviors in adolescence.

Keywords:

• Prosocial behavior
• adolescence
• confirmatory factor analysis
• measurement invariance

Notes

¹ Owing to the WLSMV estimator used here, the change in χ² and degrees of freedom cannot be calculated in a straightforward fashion. “The difference in chi-square values for two nested models using the […] WLSMV chi-square values is not distributed as chi-square” (Muthén & Muthén, 2017). Therefore, a scaling correction (DIFFTEST function) is used, of which only the p-value should be interpreted.

² The Δχ² and ΔCFI may sometimes suggest different conclusions. Clear rules on how to proceed in such a situation are lacking (Byrne & van de Vijver, 2010). Researchers can opt to describe the conclusions of both approaches or choose one over the other, based on the admissibility of the solution and examination of the modification indices.

³ Specifically, looking at the factor loadings, item 2 “I share what I like with my friends” has the highest factor loading on Factor #2 in the first analysis and on Factor #1 in the second analysis; and item 11 “I easily lend money or other things” has the highest factor loading on Factor #1 in the first analysis and on Factor #2 in the second analysis. All other items have their strongest loading on congruent factors, so if we delete these two items, we would say that the factor structure of the scale meets the basic level of replication. The next step was to look at the squared differences in the factor loadings; these were within reasonable range (.0000–.0324), indicating that the largest difference between the standardized factor loadings is |.18|.

⁴ Open source software JASP was used to calculate omega coefficients.