ISBIS4 Abstract

Contact Author's Name: Dr. Shafiqur Rahman
Title of Abstract: Population distribution and projection model for Papua New Guinea
Author(s): MS Rahman, S Nahar and M Taminza
Affiliation: University of Papua New Guinea

In order for effective planning, running and management of a country one must know the total population as well as the population distribution of that country. There is no guaranteed way of knowing the exact population for future years. Therefore some estimation is needed. Estimation of total population as well as for different sections of the total population must be known. Almost all countries of the world use some form of mathematical model for estimating future population. The population distribution of different counties in the world follow different probability distributions and PNG is no exception.
A projected model is simply an estimated mathematical formula that is used to project future population figure. From the graphical observation of census population figure of PNG first a simple linear regression model was fitted. Using the method of least squares the parameters were obtained. The significance of its parameters was tested using Student’s t-tests. Then by constructing residual plots and normal probability plots residual analysis was performed to check the validity of the assumptions of the general linear model. Residual analysis confirmed that linear regression model was not a good model. Then four models such as logarithm model, exponential model, polynomial model and logistic model were considered by using some transformations in the data set. The parameters of each model were estimated using the method of least squares and their significance were tested using Student’s t-tests. Also residual analysis was performed for each model. It was observed ! that out of the four models three models were statistically significant and satisfied all the basic assumptions of a linear model. Then the model selection criteria usually known as information criteria were applied to identify the best model. Statistical tests and model selection criteria supported logistic model as the best model for the population projection of PNG. Applying this model the estimated population of PNG for the year 2004 to 2010 were obtained. Population cannot be meaningfully compared with respect to other traits until age has been controlled for. Age often produces variations in the statistics. Mortality rates, economic activities and marital status all vary by age. Social relationships within a community are affected by relative ages. Population composition, social and economic characteristics vary with age groups. Therefore it is important to know population figure for different age groups. From the observed distribution by age group of PNG population we see that the variances are grater than the means. Out of all the known distributions we know that there are three distributions, which have their variances greater than their means. These are the negative binomial, geometric and the exponential distributions. Statistical analysis suggested that the PNG population distribution by age group follows the exponential distribution.