This study investigated whether the Filial Piety Scale applies to both high school and college students. We had to test measurement invariance, which involves a series of increasingly restrictive hypotheses. Different levels of invariance are assessed through a hierarchical procedure that compares four multi-group models in a fixed order, from the least to the most restrictive model.
The Filial Piety Scale included two filial factors: reciprocal and authoritarian. Reciprocal filial piety is characterized by gratitude and respect for parents' efforts based on affection, while authoritarian filial piety emphasizes hierarchical relationships and obligations that require children's compliance with their parents. The current study's sample consisted of 509 students, including 250 college students and 259 high school students.
An initial baseline model (Model 1) has no between-group invariance constraints on estimated parameters. Groups have the same form without restricting any non-fixed parameters across models. In Model 2, the model is the same across the groups and the factor pattern coefficients (loadings) are identical across groups because the pattern coefficients carry the information about the relationship between latent scores and observed scores. In Model 3, the factor loadings and measurement error variances were constrained to be equal in the two groups. Model 4 in this hierarchy was the most restrictive model, in which all three parameter matrices (factor loadings, intercepts, and residual variances) were simultaneously tested for equality. The results of these analyses revealed that Models 1, 2, 3, & 4 all had a good fit for RMSEA, GFI, and CFI. In the case of strict factorial invariance, the measurements across groups were unbiased and identical in terms of construct validity and reliability, as captured by the latent variables. The results show that the two factors in the Filial Piety Scale remained invariant across groups of university students and high school students.