Do Students Rapidly Guess Repeatedly over Time? A Longitudinal Analysis of Student Test Disengagement, Background, and Attitudes

ABSTRACT

Considerable research has examined the use of rapid guessing measures to identify disengaged item responses. However, little is known about students who rapidly guess over the course of several tests. In this study, we use achievement test data from six administrations over three years to investigate whether rapid guessing is a stable trait-like behavior or if rapid guessing is determined mostly by situational variables. Additionally, we examine whether rapid guessing over the course of several tests is associated with certain psychological and background measures. We find that rapid guessing tends to be more state-like compared to academic achievement scores, which are fairly stable. Further, we show that repeated rapid guessing is strongly associated with students’ academic self-efficacy and self-management scores. These findings have implications for detecting rapid guessing and intervening to reduce its effect on observed achievement test scores.

Notes

¹ One should note that the TSO model is quite similar to the simpler univariate state-trait-error model outlined by Kenny and Zautra (1995). According to their model, the variance of the observable state variable is decomposed into three sources: trait variance, occasion variance, and measurement variance. One complication with using the Kenny and Zautra (1995) model in our context is that we use ordinal variables, and the variance in their model has to be set to one. Nonetheless, the similarity of the models bears mention.

² As a robustness check, we re-estimated models using only particular cohorts in elementary and middle schools. We did this check to help ensure that conflating cohort differences was not influencing the state-trait balance. Results did not change substantively (more than 1–2 percentage points in the proportion of state variance).