A New Paradigm In Modeling Educational Count Data: Poisson Regression
The current study focuses on how to apply Poisson Regression in modeling class participation (dependent count variable) along with three predictors (independent variables) attendance, scores in the subject and whether the students take private tuition or not. Poisson Regression model is a part of the broad generalized linear models (GLM) and is used only when the dependent variable consists of count data. The data for the study consisted of 33 students from an undergraduate business management class studying in their third semester at King Abdulaziz University. The data was continuously collected by the authors throughout the semester. Results highlighted the fact that students who took tuition participated very less in the class whereas those whose attendance and scores in English were high tend to participate more during the lectures. The results of the study will provide an impetus to administrative and pedagogical staff in addressing the issue of ensuring class participation of students from a different perspective rather than just adhering to ‘stick and carrot method’. Easy to follow approach has been adopted so that the procedure followed in the current study can be replicated in other subjects at all levels.