No abstract

No abstract

If teachers knew in advance whether their students are at risk of school failure, they would have the opportunity to supply these students with additional or special instruction. In Luxembourg, the likelihood of failure in school is particularly high. Taking this result into account, this paper deals with the identification of variables of primary school students that might help predict school failure in Luxembourgish secondary school. Failure was defined as (a) descending from a higher track to a lower track, (b) repeating a class, or (c) showing insufficient achievements in two main subjects. First, we chose variables from a cohort of N = 2787 students who finished primary school in 6 th grade and started secondary school in 7 th grade in school year 2008/2009 for further analyses which were shown to be effective in predicting school failure in past investigations. These variables entailed both information about students' achievements and their social background. We then examined similarities and differences in these variables between students who failed and those who succeeded. Additionally, logistic regression analyses showed that primary school achievements in mathematics and languages were the strongest predictors of failure in secondary school, followed by students' age and students' school-related behaviors. Finally, we could show that the same accuracy of prediction of school failure (mean  = .183) was obtained when a fast and frugal algorithm, containing only three predictors or less, instead of a linear regression model was used. The findings support the hypothesis that poor academic achievement is one of the strongest predictor of school failure, and that accurate predictions can be made without using complex regression models.

No abstract

No abstract