MODELLING READING CULTURE IN NIGERIA USING FRACTIONAL BETA REGRESSION DATA FROM SOUTH WESTERN NIGERIA

Abstract

Without mincing words, it is obvious that culture of reading is fading away from our societies in Nigeria. Reading hours have reduced drastically among most students in our various institutions of higher learning. This research is therefore aimed at finding out major covariates and factors responsible for this problem, thereby constructing a model for predicting the reading hour proportion among the teaming youths of our dear country. The beta regression models is useful for cases where the response variable is continuous and restricted to the standard unit interval (0, 1). Through the diagnostic tests carried out on the models, the best fitting model is a beta regression model with log-log link function having the maximum log likelihood value and minimum values from the AIC, BIC. Also, using error metrics MAE, MSE, RMSE, and MAPE tests proved that log-log model has the best predicting power. Results revealed that females in the institutions considered read more than their male counterparts with 2.9 and 2.8 hours per day respectively. The Federal Polytechnic, Ilaro (FPI) record the highest average number of reading hours. The outcome of beta regression analysis using log-log model at 5% level show that CGPA, institutions, and zone contribute significantly to proportion of reading hours among students in the south western states of Nigeria.

Keyword: Reading, Fractional beta regression, Reading culture, Modelling, log-log model