Applying the Rasch model to analyze the effectiveness of education reform in order to decrease computer science students’ dropout

Abstract

Attrition is an important issue in higher education, especially in the field of computer science (CS). Here, we investigate to what extent an education reform affects the attrition of students by analyzing the pattern of grades of CS students’ academic achievement from 2010 to 2018 by IRT, based on Rasch-model analysis. We analyze data from 3673 undergraduate students of a large public university. In 2016 an education reform—as an intervention—was added to our BSc program: all theoretical lectures became compulsory to attend and we introduced a learning methodology course for all first-year students. According to our results, after the education reform most subjects became accomplishable, and students with lower levels of ability also tried to take exams. We succeeded in retaining 28% of our students. Analyzing students’ results could help administrators develop new programs in order to increase retention.

The impact of the first wave of COVID-19 on students’ attainment, analysed by IRT modelling method

“…According to the results, the dropout level decreased by 28%. The most important benefit of the education reform was that most subjects had become accomplishable (Takács et al, 2021 ). 1…”

“…According to the results, the dropout level decreased by 28%. The most important benefit of the education reform was that most subjects had become accomplishable (Takács et al, 2021). 1 Hypothesis 1 claims that the online transition due to COVID-19 during the second semester of the 2019 academic year did not result in a change in the requirement system of the subjects.…”

The development of molecular genetics concept test for senior high school students using Rasch analysis

“…They took this test in their schools. Table 2 provides the fit analysis, including standard errors, infit-outfit mean-square, point-biserial correlation, and reliability to determine the acceptance, validity, and reliability of items of MGCT [16].…”

“…All items have outfit and infit mean-square values in the acceptable range of 0.7 to 1.3. Infit-outfit mean-square values of more than 1.3 shows that the seen items had 30% more variety than was predicted by Rasch and if the outfit mean-square values less than 0.7 shows that the seen items had 22% less variety than was predicted by Rasch model [16]. Outfit (outlier-sensitive fit) is the criteria that more sensitive ISSN: 2252-8822 …”

“…In our study, the average item difficulty was set to zero. Placing the item positively far above zero is the most difficult, while the item difficulty negatively far below zero is the easiest [16]. Herrmann-Abell and DeBoer [17] stated that if a measure of student achievement is at the same level on the map as a measure of item difficulty, students have a 50% chance of answering the item correctly.…”