Assessing Father Engagement in Home Visiting Services: Measurement Invariance between Fathers and Mothers

Background:

Home visiting programs have begun to promote active father engagement in services. However, the field lacks valid instruments to measure the important preconditions and outcomes of father engagement in home visiting services, including parents’ beliefs about father engagement in home visiting, father engagement in home visiting, and co-parenting alliance. Furthermore, although the same instruments are often used for both mothers and fathers, it is unclear whether or not measures can be considered equivalent for both groups. This study seeks (1) to validate three instruments that evaluate positive beliefs about father engagement, father engagement in home visiting, and co-parenting alliance; and (2) test measurement invariance between fathers and mothers.

Methods:

This study analyzed longitudinal intervention data collected from 304 participants (163 fathers and 141 mothers) who participated in a study of the Dads Matter-Home Visiting (HV) program designed to optimize fathers’ roles in home visiting programs. The first instrument analyzed in this study consists of 10 items (1=strongly disagree to 5=strongly agree) to assess parents’ positive beliefs about father engagement in home visiting at baseline. The second instrument was developed to assess the frequency of father engagement in home visiting services at the 4 month follow-up using 7 items (1=never to 5=always). The last instrument (Abidin & Konold, 1999) included 17 items that measured the quality of co-parenting alliance at the 12 month follow-up (1=strongly disagree to 5=strongly agree). Exploratory Factor Analysis (EFA) was first conducted to identify common factor structures and items that are equivalent to fathers and mothers. Next, Multigroup Confirmatory Factor Analysis (MCFA) was used to confirm the initial results and test measurement invariance between the two parent groups.

Results:

EFA extracted the original three independent measures, although some items were deleted if they were not loaded together as the same factor structures for both fathers and mothers. MCFA also confirmed this configural invariance model (𝝌² (232)=361.147, p < .001, CFI=.954, RMSEA=.043). Next, MCFA indicated a partial metric invariance (𝝌² (245)=388.795, p < .001, CFI=.948, RMSEA=.044), suggesting that all factor loadings of the positive beliefs about father engagement and co-parenting alliance were invariant between fathers and mothers. However, the factor loadings of father engagement in home visiting services were not equivalent between the two groups; mothers had lower factor loadings than fathers. Similarly, scalar invariance was partially supported (𝝌² (260)=411.618, p < .001, CFI=.946, RMSEA=.044), given that the intercepts of the father engagement items were different between fathers and mothers.

Conclusions:

This study suggests three valid instruments that can be used to examine some key preconditions and outcomes of father engagement in home visiting services. Furthermore, this study shows that the two instruments that measured positive beliefs about father engagement and co-parenting alliance were satisfied with all configural, metric, and scalar invariance. Accordingly, they are useful to investigate the mean differences between fathers and mothers. However, findings indicate that caution is needed when the instrument to measure father engagement in home visiting services is used with both mothers and fathers.