1Health Promotion Program, Montana State University - Northern, Havre, MT 59501
2Kinesmetrics Lab, Montana State University - Northern, Havre, MT 59501
American Journal of Applied Mathematics and Statistics.
2018,
Vol. 6 No. 6, 224-231
DOI: 10.12691/ajams-6-6-2
Copyright © 2018 Science and Education PublishingCite this paper: Peter D. Hart. A New and Simple Prediction Equation for Health-Related Fitness: Use of Honest Assessment Predictive Modeling.
American Journal of Applied Mathematics and Statistics. 2018; 6(6):224-231. doi: 10.12691/ajams-6-6-2.
Correspondence to: Peter D. Hart, Health Promotion Program, Montana State University - Northern, Havre, MT 59501. Email:
peter.hart@msun.eduAbstract
Background: The five components of health-related fitness are cardiorespiratory endurance, muscular strength, muscular endurance, body composition, and flexibility. To assess an individual on all five components can be time consuming. Thus, it would be useful to fitness specialists if a simpler and valid fitness assessment was available to measure overall health-related fitness. The purpose of this study was to employ honest assessment predictive modeling to find a parsimonious set of variables that can predict overall health-related fitness. Methods: Data used for this study came from college students who completed a fitness test battery. An overall health-related fitness score (T-score) was constructed using maximal oxygen consumption (VO2, ml/kg/min), 1RM bench press (BP, lb), maximal push-up repetition (PU, #), and percent body fat (PBF, %). The set of possible predictor variables consisted of participant age (yr), sex (male/female), body mass index (BMI, kg/m2), waist circumference (WC, cm), 1RM leg press (LP, lb), countermovement vertical jump (VJ, in), flexed arm hang (FAH, sec), physical activity rating (PAR, 0 thru 10), and sit-and-reach (SNR, cm). The honest assessment predictive modeling procedure comprised three steps: 1) development of competing models using a TRAINING dataset, 2) selecting an optimal model using a separate VALIDATION dataset, and 3) assessing fitness score construct validity using a final SCORING dataset. Results: Stepwise model selection with Schwarz Bayesian criterion (SBC) on the TRAINING data resulted in five possible models including sex, VJ, PAR, and WC. Results on the VALIDATION data indicated a three-variable model had the lowest average squared error (ASE) and consisted of sex, VJ, and PAR (F=107.8, p<.001, R2=.82, SEE=3.09). Finally, predicted values from the SCORING data showed that athletes (Mean=54.9, SD=5.1) had a significantly (p<.001) greater mean fitness score than non-athletes (Mean=39.8, SD=4.8). Conclusion: This study presents a valid equation that can simply predict overall health-related fitness in college students.
Keywords