The purpose of this investigation was to develop an instrument that has scores that are valid and reliable for measuring students' attitudes toward fitness testing. A second purpose of the study was to determine the attitudes of secondary students toward fitness testing. A review of literature, an elicitation study, and a pilot study were conducted. The pilot study included 427 student completed instruments from three schools. Pilot study data analyses were conducted resulting in a proposed model for the final study. Participants for the final study were 1199 students from 13 schools which consisted of 524 boys and 675 girls. The data fit a four factor model for measuring secondary school students' attitudes toward fitness testing with the following four factors: cognitive, affective-enjoyment, affective-feelings, and affective-teacher. The fit statistics from the CFA indicated an overall good fit of the data to this model. GFI, AGFI, RMSEA, Bentler's CFI, and Bentler & Bonett's NFI scores were .892, .862, .080, .920, and .910, respectively. The G-C alpha reliability coefficient for the entire model was .902. The four factors and their reliability scores were: cognitive (α = .919), affective-enjoyment (α = .887), affective-feelings (α = .865), and affective-teacher (α = .801).

Secondary school students had an overall neutral to slightly positive attitude toward fitness testing (M = 3.11, s.d. = 0.71). Highest attitudes toward fitness testing were reported in the cognitive factor (M = 3.36, s.d. = .983) while the lowest attitudes occurred in the affective-enjoyment factor (M = 2.52, s.d. = 1.074). A MANOVA indicated significant differences for grade (Wilks' Lambda=.950, F(12, 3143)=2.611, p<.001) and gender (Wilks' Lambda=.902, F(4, 1188)=2.611, p<.01) with an interaction effect between grade and gender (Wilks' Lambda=.974, F(12, 3143)=2.611, p=.002). A stepwise DFA completed on the interaction means showed that the affect-feelings factor (Wilks' Lambda=.927, F(7, 1191)=18.035, p <.001) followed by the affect-enjoyment factor (Wilks' Lambda=.904, F(7, 1191)=13.345, p<.001) were the best predictors of these differences. Boys reported higher attitudes toward fitness testing (M = 3.28, s.d. = 0.74) than girls (M = 2.97, s.d. = 0.66). Boys overall attitude mean scores fell from 3.52 (s.d. = 0.70) in 9^th grade to 3.15 (s.d. = 0.72) in 12^th grade. Girls overall attitude scores dropped from 3.01 (s.d. = 0.58) in 9^th grade to 2.86 (s.d. = 0.77) in 12^th grade. A MANOVA indicated significant differences between FitnessGram and the President's Challenge fitness tests (Wilks' Lambda=.985, F(4, 1194)=4.431, p=.001). DFA indicated that the cognitive factor was the predictor of these differences (Wilks' Lambda=.989, F(1, 1197)=13.597, p=.001). Students' whose school administered the FitnessGram reported higher overall attitudes toward fitness testing (M = 3.14, s.d. 0.69) than students whose schools administered the President's Challenge fitness test (M = 3.04, s.d. = 0.74).