Investigating the Multiple Components of Test-Taking Motivation in a Large-Scale Assessment Context: The Importance of Expectancy for Success

• In Event: The Impact of Test-Taking Motivation and Test Consequences on the Validity of Test Score Inferences

Abstract

Test-taking motivation (TTM), as a special form of achievement motivation, is conceptualized within the expectancy-value theory (EVT). According to EVT, expectancy for success and the perceived value of the test affect test-taking behavior i.e., expended effort and actual test performance (Wigfield & Eccles, 2000). Most research in this area considers the value component (Thelk, Sundre, Horst, & Finney, 2009) but neglects expectancy for success, assuming that test-takers can’t conceive ‘success’ in tests without any feedback or consequences (Cole, Bergin, & Whittaker, 2008). However, other studies emphasize the importance of the expectancy component in low-stakes tests (Asseburg, 2011). Thus, our aim was to investigate both components of EVT in a large-scale assessment. More specifically, we examined the complex relationship between the aspects of the entire test-taking motivation construct (including expectancy, value, and effort) and performance.

We explored TTM in a large-scale testing context in Germany, where mathematical and scientific literacy were assessed in a representative sample of N = 42,298 ninth-grade students. To measure both EVT components, we used the Questionnaire on Current Motivation (Freund, Kuhn, & Holling, 2011) with four subscales: probability of success, challenge, interest, and anxiety. The expectancy component is represented by probability of success; the value component is represented by the remaining three subscales. Additionally, we measured test-taking effort (Eklöf, 2010) because it can be regarded as the outcome of value and expectancy. We collected all five TTM subscales before and after the test to compare the intended and invested test-taking motivation.

First, a measurement model of the five TTM factors was tested to verify that expectancy, value, and test-taking effort were independent components of TTM. Second, we explored the relationship of expectancy, value, test-taking effort, and performance by analyzing five latent regression models. The results of the first two models indicated that at both time points, the value subscales predicted the reported test-taking effort very well (R² ~ 60%). Adding the probability of success did not improve the prediction of effort. The comparison of the remaining models predicting test performance indicated that Model 5 should be preferred because Model 3 and 4 showed a negative effect of interest on test performance. Importantly, this effect disappeared when effort was modeled as a mediator. In the mediation model, reported effort and expectancy collected after the test explained ~ 25% of the variance in performance scores; almost twice as much as measured before the test. Here, expectancy for success was the strongest predictor of test performance.

To summarize, expectancy and value, mediated by effort, predicted test scores. Thus, to accurately model the entire TTM construct, all three aspects should be considered when evaluating the effects of TTM on test score validity. Additionally, the results showed that the predictive utility of the expectancy component increased over the time points, whereas the predictive utility of the value component remained stable. This finding suggested that measuring expectancy, value, and effort after test completion provides a better measurement of TTM than measuring intended TTM.

Authors

• Christiane Penk, Institute for Educational Quality Improvement

• Stefan Schipolowski, Humboldt University

• Claudia Poehlmann, Institute for Educational Quality Improvement