Psychological determinants of test motivation in low-stakes test situations: A longitudinal study of singletrait–multistate models in accounting

Abstract

Background: While several studies show that examinees’ test motivation biases their results in low-stakes tests, studies that investigate the predictors of motivation when taking low-stakes tests are rare. Moreover, little evidence exists on whether test motivation represents a state-like or trait-like construct. Research into these matters needs statistical models that allow distinguishing inter-individual from intra-individual variability of motivation across a range of test situations. The present study is located in a vocational school setting and aims to explain variations in vocational students’ low-stakes test motivation. We draw on Urhahne’s (Psychologische Rundschau 59:150–166, 2008. https://doi.org/10.1026/0033-3042.59.3.150) synthesis of motivation theories in academic learning contexts to substantiate potential predictors. Since we concentrate on person-dependent characteristics, relevant predictors are types of self-determined/external behavioural regulation, achievement motivation, academic self-concept, and grit. In line with Eccles et al. (In: Spence JT (eds) Achievement and achievement motives: psychological and sociological approaches. Freeman, San Francisco, pp 109–132, 1983) and Sundre (The Student Opinion Scale (SOS). A measure of examinee motivation: test manual. https://www.researchgate.net/publication/238741273_The_Student_Opinion_Scale_SOS, 2007), we further discern value attributions to and invested effort in each test as two separate dimensions of test motivation. Method: Empirical analyses utilize longitudinal questionnaire and test performance data of students (N = 852) from full time vocational schools, who participated repeatedly in a low-stakes accounting test at the end of each school year (5 years in total). Latent singletrait–multistate (STMS) models serve to disentangle trait-like and state-like components of students’ test motivation and their respective associations with trait- and state-components of the assumed predictor variables. Results: Findings from STMS models indicate that approx. 30% of variation in students’ test motivation is stable over time, whereas approx. 25% is attributable to the test situation, leading to a high portion of 45% residual variance. Bivariate STMS analyses show that students’ achievement motivation and type of behavioural regulation predict the value dimension of their test motivation. This pattern appears for the trait-like components of investigated constructs (here: time-invariant, person-specific levels of predictors and criteria) as well as for the state-like components (here: intra-individual increases or declines in predictors and criteria). Regarding the effort dimension of test motivation, similar associations among the trait-like components appear. Students’ self-reported effort is positively predicted by type of behavioural regulation and achievement motivation. Moreover, and in contrast to the value dimension, students’ perseverance plays a crucial role when predicting test-taking effort. Finally, a multivariate intercept-only growth model was estimated to analyse the relative contributions of different psychological determinants of test motivation. It shows that (a) students’ introjected regulation predicts the value dimension and (b) students’ identified behavioural regulation and perseverance (as an essential aspect of grit) predict the effort dimension of test motivation. Conclusions: Students’ motivation to master low-stakes tests represents an equally trait- and state-like construct—at least with respect to the investigated test format (paper–pencil), content (accounting) and examinee population (vocational students). Our finding that comparably stable motivational dispositions of students, namely person-specific types of behavioural regulation, achievement motivation, and grit, are highly predictive of their test motivation in multiple test situations implies that test motivation bias can be reduced substantially by controlling for general student motivation (in the relevant academic domain).