Physical Strength and Performance of Moderate Duration Physical Tasks

Abstract

SUMMARY ProblemValid models of physical task performance could be useful for selection into Navy jobs and for computer simulations of combat performance. Simple models are desirable, but of little value if they are inaccurate. Previous studies indicated that strength predicted more than 90% of the variance in performance on a wide range of physically demanding U.S. Navy tasks. This finding implies that strength is the only ability to consider for Navy selection and modeling purposes. However, the tasks studied previously lasted at most a few minutes. Work physiology principles predict that longer lasting tasks will have a more complex causal structure in which strength and endurance both are important. ObjectivesThe primary goals of this study were to (a) demonstrate that strength is a less powerful performance predictor when tasks last longer than 1 min, and (b) evaluate the hypothesis that the initial strength-performance estimate was biased by the omission of other physical abilities from the predictive equation. ApproachStructural equation modeling was applied to data from a study of steelworkers. Task performance measures included lifting, carrying, and shoveling tasks lasting 5 min to 15 min. Physical ability measures included the static strength dimension from earlier research and a dynamic strength dimension. Structural equation models were constructed to estimate the relationship between physical abilities and performance. ResultsStatic strength strongly predicted performance (β = .86), but this association was significantly (p < .001) lower than the estimate obtained in prior studies of shorter tasks. Adding dynamic strength (i.e., sit-ups, pull-ups) improved overall criterion prediction slightly (semipartial r = .13) even though it lowered the estimated effect of static strength by 25% (β = .69). ConclusionsThe effects of physical abilities on task performance can be estimated accurately only after careful selection of tasks and ability tests. The analysis procedures must provide methods of formulating and testing specific models that combine theoretical considerations with prior empirical findings. Most prior studies of physical abilities and task performance do not meet these criteria. The resulting estimates of the impact of physical ability on performance are likely to be biased. The biases can undermine the accuracy of screening batteries and/or lead to suboptimal performance enhancement interventions. Strength and Moderate Duration Tasks

Physical Ability-Task Performance Models: Assessing the Risk of Omitted Variable Bias

“…Bias, therefore, will occur if these other abilities affect task performance and are omitted from the model. Vickers (2003a) demonstrated the potential for omitted variable bias in the estimates of strength effects in a reanalysis of Arnold et al's (1982) steelworker data. Arnold et al (1982) included variables representing two strength dimensions in their study.…”

“…On the criterion side, brief (i.e., <1 min) lifting tasks and moderate duration carrying tasks were treated as separate performance dimensions. Earlier work involved either brief tasks (Vickers, 1995(Vickers, , 1996 or moderate duration (5 min to 15 min) tasks (Vickers, 2003a), but not both. Models that considered all four ability dimensions as predictors of both performance dimensions were compared to better evaluate the complexity of the abilityperformance interface for physical tasks.…”

A Latent Trait Model of Simulated Combat Performance

“…The ME + AC model could be ruled out because the estimated effect of ME was impossibly large and because neither of the estimated effects of ability on GP was statistically significant (i.e., t < 2.00). 3 The AP + ME model was ruled out for a different reason. Only ME was a significant predictor of GP.…”

“…The resulting models based on general dimensions have adequately summarized the covariation of physical ability tests with task performance. [1][2][3][4][5] The appropriate level of analysis remains an open question despite recent findings. Those findings are limited to specific combinations of tests and tasks.…”

“…A model with fewer parameters would be more parsimonious. 3 Previous modeling efforts suggest that as few as 2 parameters can adequately characterize the ability-performance interface.…”