Projecting populations that have sparse or unreliable data, such as those of many developing countries, presents a challenge to demographers. The assumptions that they make to project data-poor populations frequently fall into the realm of "educated guesses", and the resulting projections, often regarded as forecasts, are valid only to the extent that the assumptions on which they are based reasonably represent the past or future, as the case may be. These traditional projection techniques do not incorporate a demographer’s assessment of uncertainty in the assumptions. Addressing the challenges of forecasting a data-population, I have to project the population of Bangladesh using a Bayesian approach. This approach incorporates a demographer’s uncertainty about past and future characteristics of the population in the form of elicited prior distributions. It is argued that due to the explicit incorporation of subjective elements in order to address the uncertainty issue, and due to the properties of the estimates and predictions, the Bayesian methods are a valuable tool for population forecasting.

This thesis addresses selected methodological aspects of population forecasting of Bangladesh from 2001 to 2051. I try to present a comparison between usual cohort-component method of population projection and Bayesian forecasting of population using fitted model of growth curves. The objective is to present an overview of the existing methods and to propose an alternative based on the Bayesian statistics, combining the formality of inference with the subjective expert opinion. Uncertainty and judgment in population forecasting are followed by an introduction to Bayesian statistics, and an overview of the existing forecasting methods.

This thesis analyze the population growth data obtained from censuses of Bangladesh and East Pakistan from 1901 to 2001 in Bayesian perspective. A logistic growth curve has been used to fit these data and to make the projections of the population of Bangladesh between 2001 and 2051. The analysis has been made using MCMC technique for Bayesian Inference available with the software WinBUGS. Convergence diagnostic techniques available with the WinBUGS software have been applied to ensure the convergence of the chains necessary for the implementation of MCMC.