DEVELOPMENT AND CROSS-VALIDATION OF AEROBIC

DEVELOPMENT AND CROSS-VALIDATION OFAEROBIC CAPACITY PREDICTION MODELSIN ADOLESCENT YOUTHbyRyan Donald BurnsA dissertation submitted to the faculty ofThe University of Utahin partial fulfillment of the requirements for the degree ofDoctor of PhilosophyDepartment of Exercise and Sport ScienceThe University of UtahDecember 2014

Copyright Ryan Donald Burns 2014All Rights Reserved

The University of Utah Graduate SchoolSTATEMENT OF DISSERTATION APPROVALThe dissertation ofRyan Donald Burnshas been approved by the following supervisory committee members:James C. Hannonand by, ChairTimothy A. Brusseau, MemberBarry Shultz, MemberPatricia Eisenman, MemberMatthew T. Mahar, MemberJames C. Hannonthe Department/College/School of8/26/14Date Approved8/26/14Date Approved8/26/14Date Approved8/26/14Date Approved8/26/14Date Approved, Chair/Dean ofExercise and Sport Scienceand by David B. Kieda, Dean of The Graduate School.

ABSTRACTCardiorespiratory endurance is a major component of health-related fitness testingin physical education. FITNESSGRAM recommends the 1-mile Run/Walk (1-MRW) orthe Progressive Aerobic Cardiovascular Endurance Run (PACER) to assesscardiorespiratory endurance by estimating aerobic capacity, or VO2 Peak. No research todate has cross-validated prediction models from both 1-MRW and PACER using currentFITNESSGRAM criterion-referenced (CR) standards. Additionally, new predictionmodels for 1-MRW without a body mass index (BMI) term are needed to attenuate theproblems incorporating this index into an aerobic capacity model. The purpose of thisdissertation was to cross-validate various prediction models using 1-MRW and PACERand to develop alternative 1-MRW aerobic capacity prediction models for adolescentyouth. Participants included 90 students aged 13 to 16 years. Each student completed the1-MRW and PACER, in addition to a maximal treadmill test to measure VO2 Peak.Multiple correlations among various models with measured VO2 Peak were consideredstrong (R 0.74 to 0.78). CR validity, examined using modified kappa (Κq), percentageof agreement (Pa), and phi was considered moderate among all models (Κq 0.25 to 0.49;Pa 72% to 79%; phi 0.38 to 0.65). Two new models were developed from 1-MRWtimes, one linear and one quadratic model. The linear and quadratic models displayedmultiple correlations of R 0.77 and R 0.82 with measured VO2 Peak, respectively. CRvalidity evidence was considered moderate with (Kq 0.38; Pa 73%; phi 0.57) using

the linear model and (Kq 0.34; Pa 70%; phi 0.54) using the quadratic model. Theaccuracy of these models was confirmed using k-fold cross-validation. In conclusion, theprediction models demonstrated strong linear relationships with measured VO2 Peak,acceptable prediction error, and moderate CR agreement with measured VO2 Peak usingFITNESSGRAM’s CR standards to categorize health groups. The new 1-MRW modelsdisplayed good predictive accuracy and moderate CR agreement with measured VO2 Peakwithout using a BMI predictor. Despite evidence for predictive utility of the new models,they must be externally validated to ensure they can be generalizable to largerpopulations of students.iv

TABLE OF CONTENTSABSTRACT.iiiLIST OF TABLES.viiLIST OF FIGURES .viiiACKNOWLEDGEMENTS .ixChapters1INTRODUCTION.1History of Youth Fitness Assessment. 2Health-Related Physical Fitness.3Criterion-Referenced Standards .5Development of Fitness Zone Cut-Off Scores.7Field Test Equating.12Statement of the Problem.14Research Questions.15Study ns.18Limitations.18Definition of Terms.192CROSS-VALIDATION OF AEROBIC CAPACITY PREDICTION MODELSIN ADOLESCENTS.21Introduction .21Methods.25Results .31Discussion .383DEVELOPMENT OF AN AEROBIC CAPACITY PREDICTION MODELFROM 1-MILE RUN/WALK PERFORMANCE INADOLESCENTS.54

UMMARY OF FINDINGS.79Cross-validation of Mahar PACER (2014).79Cross-validation of Cureton et al. (1995).81Development of New 1-MRW Models.82Limitations.83Future Research Directions.84Conclusions.85AppendicesA. INSTITUTIONAL REVIEW BOARD LETTER APPROVAL.87B. PARENTAL CONSENT.89C. ASSENT TO PARTICIPATE IN A RESEARCH STUDY.93REFERENCES.97vi

LIST OF TABLES2.1Descriptive data for the total sample and within sex groups.322.2Pearson correlations among cardiorespiratory endurance and body compositionparameters.362.3Cross-validation of the aerobic capacity models against measured VO2 Peak.372.4CR agreement into FITNESSGRAM’s three Healthy Fitness Zone scheme.443.1Model development employing hierarchical block-wise entry.633.2Parameter estimates for linear model . . .643.3Parameter estimates for quadratic model . . .643.4K-fold cross-validation of the 1-MRW linear model . . . .703.5K-fold cross-validation of the 1-MRW quadratic model .713.6CR agreement with measured VO2 Peak into Healthy Fitness Zones . .73

LIST OF FIGURES1.1A sample ROC curve plot depicting sensitivity on the y-axis and (1-specificity)on the x-axis.111.2Conceptual illustration of the primary field-test centered equating method forsetting cut-off scores (from Zhu, Plowman, & Park, 2010) .132.1Procedure flow chart for collection of cardiorespiratory endurance data.292.2Scatterplot and line of best fit showing the linear relationship between measuredVO2 Peak and 1-MRW times.342.3Scatterplot and line of best fit showing the linear relationship between measuredVO2 Peak and PACER laps .352.4Residual against fitted plot with trend line using the Cureton Model .392.5Residual against fitted plot with trend line using the New PACER Model.402.6Residual against fitted plot with trend line using the Mile-PEQ.412.7Residual against fitted plot with trend line using the linear PACER model .422.8Residual against fitted plot with trend line using the quadratic PACER model .432.9Distribution of Healthy Fitness Zone classification for each aerobic capacityModel. .453.1Scatterplot and best fit line showing the curvilinear relationship betweenmeasured VO2 Peak and 1-MRW times .613.2Normal probability plot of residuals for the linear model . . .653.3Normal probability plot of residuals for the quadratic model . .663.4Residual against fitted plot using the new 1-MRW linear model .673.5Residual against fitted plot using the new 1-MRW quadratic model .683.6Distribution of Healthy Fitness Zone Classification for each 1-MRW aerobiccapacity model.73

ACKNOWLEDGEMENTSI would like to thank several people who have helped me through my PhDprogram, supporting me personally and professionally throughout these 4 years. First, Iwant to thank my Mom and Dad who provided me the emotional support to go after mygoals. Without them I would not have accomplished the difficult task of completing adoctorate, and their unconditional love has meant everything to me. Second, I want tothank my son Rylan who provided me the incentive and inspiration to persevere throughthese past 4 years. He has grounded me and constantly reminds me of what is importantin life. Third, I would like to thank my committee: Drs. James Hannon, TimothyBrusseau, Patricia Eisenman, Barry Shultz, and Matthew Mahar. Each committeemember has positively and uniquely contributed to the creation of this document a

score to a reference population to interpret fitness levels, as was the case employing the old Presidential Fitness program, FITNESSGRAM classifies students into one of three Healthy Fitness Zones by relating a fitness test score to a health-criterion measure (Welk, Going, Morrow, & Meredith, 2011). A child can use the Healthy Fitness Zone