First-year dropout: The role of student self-evaluation

In Organisation for Economic Co-operation and Development (OECD) countries about one-third of the enrolled university students do not finish their undergraduate programme, with a negative impact on job career expectations and, in general, on the labour market.

This phenomenon, referred to as ‘dropout’ in scholarly literature, is affected by socio-economic factors such as family background, financial issues and uncertainties about the future, but also by factors strictly related to the individual such as age, gender and nationality.

In addition, recent studies suggest that the critical decision to drop out is often taken by first-years after failing one or more exams of their university study programme. Therefore, an early monitoring of the students’ performances may be crucial in preventing dropout.

In STEM undergraduate programmes, introductory courses in physics and chemistry are often foundational exams to be taken by first-year students. Performance in these exams could, therefore, represent a good litmus test of the students’ future study career.

But what factors do performance in first-year chemistry and physics exams depend on?

Maths scores in high-stake tests and perceived self-efficacy – the judgement of one’s own capability to perform a task – at the end of high school are traditionally considered strong predictors of academic performances.

However, these factors do not account for what actually happens during a first-year physics or chemistry university course. Typically, these courses feature written midterm class assignments and homework. These tasks are powerful learning opportunities for students when used as formative assessments.

But do students really activate suitable cognitive strategies based on their performance in these tasks? And, more importantly, what happens when there is a misalignment between their self-belief about their performance and the results they actually achieve?

The self-belief of being successful in a specific enacted cognitive behavioural act, such as performance on a given task, is called confidence. A relevant aspect of confidence is its accuracy, namely whether the confidence judgement significantly differs from the actual performance.

When a student overestimates her/his actual ability after performing a specific task we talk about overconfidence bias. When there is an underestimation we talk about underconfidence. An example of overconfidence bias is the well-known Dunning-Kruger effect.

Ideally, students should be ‘calibrated’, namely, the difference between their confidence rating and their performance score, which we call accuracy of self-evaluation, should be around zero. When this difference greatly differs from zero, there is evidence of an over- or underconfidence bias.

The interest in this construct is that the misalignment between confidence and actual performance is correlated to inappropriate decision-making skills, ineffective self-regulation and low achievement.

Does accuracy of self-evaluation predict likelihood of academic success?

To answer this question, we focused our attention on a sample of 206 Italian students: 81 on a biology degree course (20 boys and 61 girls) and 125 on an engineering degree course (97 boys and 28 girls). The biology students were attending a general chemistry course, while the engineering students were attending an introductory calculus-based general physics course.

In each course, as part of written midterm classwork, we submitted a multiple-choice test and, after each question, we asked the students to indicate the extent to which they were confident of their given response.

The accuracy of students’ self-evaluation was measured for each student based on the total performance score and the confidence rating. General self-efficacy in chemistry and physics, as well as the high school final examination score, chemistry and physics grades, and pass/fail of the exam, were also measured.

These are our findings:

• Overconfident students are more likely to fail the exam

We found that the accuracy of self-evaluation was a significant predictor of the likelihood of success in the chemistry and physics exams. In particular, failing students were overconfident about their performance in the midterm exam, while successful students were underconfident.

• Underconfident students activate more effective strategies

A possible interpretation of our findings is that underconfident students were more aware that they were ‘not well enough’" prepared and, therefore, they put more cognitive efforts into their studies and enacted better self-regulation strategies during the courses to cope with their difficulties.

• Male students were more overconfident than female ones

Among failing students, male students who failed the exam were more overconfident than female students. In particular, girls who failed the exam were less confident in their responses and hence more aware of their difficulties and more accurate in their self-evaluations.

• Girls are underconfident in physics, even if they perform better

In physics, girls outperformed boys in high school achievement but systematically underestimated their performance in the proposed tasks at university.

Therefore, gender differences in confidence are not likely to be due to actual preparation, but may be linked to perceived stereotypical judgments on behalf of instructors and peers, especially in courses where girls are underrepresented (as was the case in the physics assessment in our study). In contrast, when girls are not underrepresented, as was the case for the chemistry course, girls and boys are expected to be equally confident in their performances.

Improving students’ calibration

Making university instructors more aware of the metacognitive biases of their students is crucial. Only if instructors become more familiar with these aspects will it be possible to effectively implement suitable systematic pedagogical approaches to improve students’ accuracy of self-evaluation.

Among the possible approaches, the metacognitive scaffolding featured in inquiry-based teaching-learning sequences – a student-centred approach where the teacher guides the students through questions posed, methods designed and data interpreted by the students – can likely help students develop more accurate outcome expectations.

Italo Testa and Silvia Galano are based at the department of physics (‘Ettore Pancini’), University of Naples Federico II, Naples, Italy, and Oreste Tarallo is based at the department of chemical sciences, University of Naples Federico II, Naples, Italy. The complete study is available in the International Journal of Science Education: ‘The relationships between freshmen’s accuracy of self-evaluation and the likelihood of succeeding in chemistry and physics exams in two STEM undergraduate courses’.