A COMPARISON OF THE PREDICTIVE ACCURACY OF A POOLING AND A SUBGROUPING PREDICTION STRATEGY

Area of Science:

• Psychology
• Statistics

Background:

• The subgrouping strategy involves partitioning a sample into multiple subgroups.
• Separate regression analyses are performed within each subgroup.
• Previous studies indicate some subgroups yield higher multiple correlations than the total sample regression.

Purpose of the Study:

• To investigate the efficacy of the subgrouping strategy in regression analysis.
• To determine if subgroup-specific regression results are maintained under cross-validation.

Main Methods:

• Review of existing studies employing the subgrouping strategy.
• Analysis of subgroup partitioning and individual group regression.
• Assessment of cross-validation for subgrouping results.

Main Results:

• Multiple studies have utilized subgrouping strategies.
• Identification of subgroups with superior multiple correlation compared to the total sample is possible.
• The maintenance of these enhanced results under cross-validation remains a key question.

Conclusions:

• The subgrouping strategy offers potential for improved regression analysis by identifying specific high-performing subgroups.
• Further research is needed to validate the cross-applicability of these subgroup-specific findings.