Physical Ability-Task Performance Models: Assessing the Risk of Omitted Variable Bias
Abstract
The physical capacities of job incumbents limit performance on occupational physical tasks. While muscle strength is logically an important performance-relevant physical ability, omitted variable bias may cause its importance to be overstated. This bias occurs when a causal variable in a model correlates with other causal variables that are omitted from the model. The impact of omitted variable bias on the strength-performance association was evaluated in a study of simulated job performance in men and women. The study measured 4 major abilities, Static Strength (SS), Dynamic Strength (DS), Anaerobic Power (AP), and Aerobic Capacity (AC). Performance measures were simulated lifting and carrying tasks. Analysis showed moderate to strong relationships among the ability measures. All four ability measures were significantly related to lifting and to carrying performance. However, construction of a series of alternative predictive models led to adoption of a final model, with SS and AC as the only predictors. The absence of AP and DS from the model indicates that omitted variable bias can be expected whenever these ability factors are studied in isolation from SS and AC. The practical implication is that physical training can be mistakenly focused on abilities that have no impact on job performance.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 15, 2008
- Accession Number
- ADA515128
Entities
People
- James A. Hodgdon
- Marcie B. Beckett
- Ross R. Vickers
Organizations
- Naval Health Research Center