Comparing artificial neural net with multiple regression in a biodata criterion validation study.

摘要

This research compared Artificial Neural Nets (ANNs) to multiple regression in a Biodata criterion validation study. Using four constructs derived from 15 Biodata questions, shared variance associated with oral interview scores were measured. We proposed: Biodata preselection inventory will predict Food Server oral interview success (H1); Using sequential regression, two Education constructs will predict Food Server Oral Interview success (H2); and (step 2) two Experience constructs will account for substantial incremental variance beyond that accounted for by two education constructs (H3); ANN will account for more shared variance compared to multiple regression (H4). Findings supported H1, H2, H3, and H4. Additional analysis of ANN should be conducted to clearly validate this technique.