Purposes: Use multidimensional polytomous item response modeling (MPIRM) to evaluate the psychometric properties of a television (TV) parenting practices (PP) instrument. Perform differential item functioning (DIF) analysis to test whether item parameter estimates differed across education, language, or age groups.
Methods: Secondary analyses of data from three studies that included 358 children between the ages of 3 and 12 years old in Houston, Texas. TV PP included 15 items with three subscales: social co-viewing, instructive parental mediation, and restrictive parenting. The multidimensional partial credit model was used to assess the performance. DIF was used to investigate the differences in psychometric properties across subgroups.
Results: Classical test theory analyses revealed acceptable internal consistency reliability (Cronbach's α: 0.72 to 0.83). More items displaying significant DIF were found across children's age groups than parental education or language groups. A Wright map revealed that items covered only a restricted range of the distribution, at the easier to respond end of the trait.
Conclusions: TV PP scales functioned differently on the basis of parental education, parental language, and child age, with the highest DIF among the latter. Additional research is needed to modify the scales to minimize these moderating influences. Some items may be age specific.