Measurement Invariance and Latent Profile Analysis of the Test Anxiety Inventory

Abstract

<p style="text-align: justify;">Researchers have recognized the need for updates of test anxiety scales for more measurement accuracy. However, studies that investigated the measurement invariance of the Test Anxiety Inventory (TAI), and identified the latent profiles remain scare not withstanding its wide usage in Nigeria. This might have an impact on how generalizability and reliance on outcomes from such an instrument are handled. We investigated the measurement invariance of TAI and the latent profiles among Nigerian adolescents. Gender constituted our focus in the measurement invariance given its significance in test anxiety research. Adolescent students (n=539) formed the sample of our study. We employed the confirmatory factor analysis (CFA) and the latent profile analysis (LPA) to measure the invariance and identify the class profiles. We found evidence of a measurement invariance across gender in both competing models given that we obtained strict measurement invariance. We also identified a four-class profile model for both male and female students: low (17.50%; 19.00%), moderate (17.50%; 34.40%), high (36.80 %; 32.80%), and very high (12.60%; 13.80%) test anxiety profiles respectively. We concluded that while the composed factors are equivalently scaled across gender that test anxious students were not monolithic given the identified profiles.</p>