Addressing Internal Reliability and Predictive Validity Concerns of Survey Reports of Father Involvement
Rebekah Levine Coley, Boston College
This study assesses measurement issues in survey reports of father involvement. analyzes assessed the internal reliability and predictive validity of a multi-dimensional measure of father involvement, reported separately by fathers and mothers (N = 227). Reliable composites of father involvement were structurally similar across father versus mother reports and across resident versus nonresident and African American versus Latino fathers. Both father and mother reports, as well as a combined reporter composite, showed significant relations to children's cognitive achievement and, less consistently, to children’s behavioral functioning. Multiple group structural equation models indicated invariance in the predictive validity of father versus mother reports. Results suggest that relatively simple survey questions can be used to create reliable and valid measures of father involvement.