Abstract:

Assessment and performance are bidirectionally related. Processing large amounts of gathered educational data have potential to clearly determine and evaluate what students already know and set the boundary between that and what they need to learn. In order to offer improved Computer Based Assessment (CBA) services, we should consider the parameters that lead to improved performance, and thus, construct more accurate predictive models. Students’ beliefs of what they have learnt as well as their goal-setting are important because they reflect on students’ effort, (self-)awareness and achievement-related behaviors. Consequently, students’ perceptions of performance and goal-expectancy are two parameters that should be explored when dealing with prediction of performance. This knowledge would be valuable in providing appropriate feedback.

In this paper we present a case study of tracking:
a) students’ perceptions of performance and goal expectancy before taking a computer-based test,
b) their perception of performance after taking the test,
c) their actual performance as it is calculated by the testing environment itself and d) their time-spent behavior during test.

Our goal is to explore whether students' time-spent behavior during computer-based testing – expressing “(un-)certainty” - can reveal any differences between what they believe they know, and what they actually know. Furthermore, we investigate the correlation between students' goal-expectancy and their “(un-)certainty”. We conducted a case study with a simplified version of the LAERS assessment environment. 96 students from Secondary Education participated in the case study, from 2nd to 7th of October 2013.

We used statistical methods and Structural Equation Modeling (SEM) for the construction of a predictive model which explains the results, based on students’ time-spent on answering each question of a multiple choice quiz. Initial results indicate that:
a) students’ perceptions of performance and their actual performance significantly differ, both pre and post test,
b) students’ temporal behavior can explain satisfactorily what they actually know, and
c) goal-expectancy has an indirect effect on students’ “(un-)certainty”.

Keywords:

Prediction of performance, temporal behavior, computer-based testing, learning analytics, goal expectancy, certainty.