Norm-Referenced and Criterion-Referenced Reliability of the Push-Up and Modified Pull-Up

Abstract

This study determines the test-retest reliability and equivalence reliability (i.e., defined as consistency of 2 tests designed to measure the same construct) of the push-up and the modified pull-up tests from both norm-referenced and criterion-referenced frameworks. Sixty-two (30 boys and 32 girls) 5th- and 6th-grade students (mean age = 11.4 0.9 years) were administered 2 test trials for both the push-up and modified pull-up tests following a 4-week fitness unit. The norm-referenced test-retest reliability estimates, using intraclass correlations from a one-way analysis of variance (ANOVA), were consistently high for 2 trials of both the push-up and modified pull-up (R = .99) and also high for 1 trial (R = .98 for push-up, R = .97 for modified pull-up). Moderately high correlations between the push-up and modified pull-up were obtained (r ≧ .74). Criterion-referenced reliability was estimated with proportion of agreement (Pa), modified kappa (Kq), and Phi coefficient using the FITNESSGRAM(r) standards (Cooper Institute for Aerobics Research, 1999). Criterion-referenced test-retest reliability estimates were high for the push-up (Pa = .97, Kq = .94) and the modified pull-up (Pa = .95, Kq = .90). Equivalence reliability estimates were considerably lower between push-up and modified pull-up test Trial 1 (Pa = .69, Kq = .38, and Phi = .50) and test Trial 2 (Pa = .71, Kq = .42, and Phi = .52). Norm-referenced and criterion-referenced test-retest reliability estimates in this study were acceptable. However, criterion-referenced equivalence reliability findings were not acceptable. Equating of standards between fitness tests designed to measure the same component of fitness should be examined.