Validation of two test anxiety scales for physics undergraduate courses through confirmatory factor analysis and Rasch analysis

The assessment of test anxiety has received increasing attention in educational research due to the potential negative effects of anxiety on student performance. Traditionally, test anxiety scales have been developed for mathematics, but few studies have focused on physics. In this study, we validated two test anxiety scales for undergraduate physics courses: the Test Anxiety Inventory for Physics (TAIP) and the Abbreviated Test Anxiety Inventory for Physics scale (ATAIP), which were adapted from existing instruments. A convenience sample of 361 engineering students enrolled in a first-semester introductory physics course participated in the study. Confirmatory factor analysis and Rasch analysis were used to establish the construct validity of both scales. Convergent validity for the TAIP scale was established by examining its correlation with a scale adapted from the math anxiety scale. Criterion-related validity for both TAIP and ATAIP was established by analyzing the relationship between students’ Rasch scores on the two scales and their performance on two conceptual tests. Finally, measurement invariance of TAIP and ATAIP scales was established using both multigroup and differential item functioning analyses to reliably investigate gender differences in the corresponding Rasch measures. The study confirms a robust four-factor structure of the TAIP. The four subscales, Worry, Emotionality, Interference, and Lack of Confidence, demonstrate good reliability (McDonald’s $\omega =0.78,0.86,0.87,87$, respectively). Rash analysis also confirms that, for each subscale, the rating scale functioning was consistent with the item difficulty and person measures. The TAIP also demonstrates adequate convergent and criterion-related validity, as well as measurement invariance with respect to gender. The ATAIP also demonstrates good reliability (McDonald’s $\omega =0.84$), a well-functioning rating scale, and sufficient criterion-related validity. Additionally, it exhibits measurement invariance with respect to gender. Overall, the study supports that both the TAIP and ATAIP scales are reliable instruments for measuring students’ test anxiety in an undergraduate physics course. Implications for physics instruction at the university introductory level are briefly discussed.

Figure 1

CFA of the 19-item TAIP instrument.

Figure 2

PCA of residual plots for the TAIP subscales: (a) whole instrument; (b) Worry, Emotion, and Interference factors; (c) Worry and Emotion factors.

Figure 3

Probability plots for the four scales of the TAIP.

Figure 4

DIF analysis of the four subscales of the TAIP. DIF contrast ranges: $<0.43$ = negligible; (0.43; 0.64) = moderate; $>0.64$ = large.

Figure 5

Probability plot for the ATAIP.

Figure 6

DIF plot of the ATAIP items. DIF contrast ranges: $<0.43$ = negligible; (0.43; 0.64) = moderate; $>0.64$ = large.