Poisson and Negative Binomial Regression Models Application to Model the Factors of Car Ownership in Akure, South West, Nigeria

Abstract

Increase in number of cars without commensurate increase in the number of transport facilities and infrastructures has led to diverse traffic problems in many Nigerian cities like Akure. Factors which contribute to increase in the numbers of cars owned in Akure metropolis were investigated in this study. The study area was divided into three density zones namely High, Medium and Low while, data was collected using well-structured household questionnaire survey distributed amongst residents; with the survey yielding a return of 1002 questionnaire out of the 1181 distributed. Results from field findings gave the average number of cars owned per household in the study area as 0.62. Results of the Poisson Regression Model show that a change in the number of employed household members will decrease the number of cars owned in the study area by 9% while, a unit increase in the number of driver’s license holders in the household, academic qualification and average monthly income of the household will increase the number of cars owned by 60%, 26% and 30% respectively. The negative binomial model indicates that a change in the number of employed household members will decrease the number of cars owned by 10% whereas a change in the number of driver’s license holders in the household and monthly income will lead to an increase in the number of cars owned by 101% and 24% respectively. The test of model effects affirm that all the predictor variables are statistically significant indicating a good fit for the model predicted. Out of the two models, Poisson regression model is found to be a superior model due to a higher log likelihood ratio Chi Square and improved statistically significant variables. The findings in this research will assist government agencies to plan future transportation infrastructure development.

School Violence and Safety Policies and Practices in Urban and Rural Communities: Does Location Matter?

“…To further test the effect of school location on the relationship between school violent crimes and safety policies and practices, two separate models for urban and rural schools were constructed. The values of goodness of fit were 1.785 in urban schools and 1.988 in rural schools, and both p-values for the omnibus tests were less than .001, suggesting good model fit of the data (Oyedepo & Etu, 2016).…”

“…Table 3 displays the results of negative binomial regression testing the effects of school characteristics, security measures, and school programs on incidences of violent crimes at schools within the whole sample. The value of goodness of fit was 1.852, and the p-value of the omnibus test was less than .001, indicating that the model fit the data well (Oyedepo & Etu, 2016). Ten of the independent predictors were significant in the model.…”

Analysis Factors Influencing Motorcycle and Car Ownership in Medium-Sized Region in Developing Countries (A Case Study of Badung Regency, Bali-Indonesia)

“…Other aspects that influenced motorcycle ownership include the limitation of using a car, the leniency of using a motorcycle, the friction of living quality, and level of education. An investigation on factors affecting car ownership has been conducted in Akure, South West, Nigeria based on a multinomial logit model [16]. They stated that car ownership was influenced by the household income level and members, where an increase in income and family members by 50% was followed by an increase in the car ownership by 51% and 0,8%, respectively.…”

“…For example [18], have applied Poisson regression to study the relationship between the travel times of the individuals and the number of motorized trips. [16] used Poisson and negative binomial regression to model car ownership in Nigeria. They found that car ownership was influenced by income and household members.…”

Generalized Linear Model Approach for Road Accident Patterns in Kerala, India

“…A α value of 0 or very close to it condenses the Negative Binomial Model to Poisson [9,[11][12][13].…”