Overcoming Overdispersion on Direct Mathematics Learning Model Using the Quasi Poisson Regression

Abstract

The study aims to overcome the overdispersion in the final exam score of Calculus subject. The occurrence of a gap in the score is inseparable from the influence of internal factors and external factors. The internal factors are factors that originate within students themselves. In contrast, external factors are factors that exist outside of students that can affect students' success or failure, such as learning environment, the physical environment where the exam takes place, infrastructure and facility that are owned and used by students, and their condition during the exam. The connection of these factors to the gap in final exam scores can be approached with Poisson regression analysis because Poisson Regression assumes a Poisson distribution of response variables with equidispersion. It is necessary to conduct a research to find out the factors that cause the diversity of the final score in order to overcome, that is by implementing direct learning. The results obtained show that with the application of direct mathematics learning, there is no overdispersion, while the low score of the final exam results in an increase, resulting in the variance of final exam results to be smaller than the average score.

Rainfall Modeling Using the Geographically Weighted Poisson Regression Method

“…The presence of cases of overdispersion will result in the deviation value of the regression model being large. Apart from that, cases of overdispersion can also result in the standard error of the resulting regression parameter estimates tending to be lower than it should be, so if the Poisson regression model is still used in overdispersion conditions, then the estimated parameters, which should not necessarily be significant, will be considered significant [17].…”