ESTIMATION OF PARAMEETERS

The next step in model building is to estimate the parameters of the model. For this purpose, demographers have to use a proper estimation technique. For example, if a dependent variable is expressed as a linear combination of a number of independent variables, it contains certain parameters which may be estimated using the least square method or method of maximum likelihood. Thus the researcher has to use one method from a number of available methods of estimation given for a particular context. The researcher must know the merits and demerits of all the methods of estimation and choose on the basis of knowledge and logic.

VALIDATION

Validation refers to examination of whether the model predicts well. In other words, validation is an examination of how close the values of dependent variable are to the values of the variable which are estimated using the model and available data on independent variables. In regression analysis R-square is used as a measure of goodness of model and a high value of R-square suggests that the model can be used for prediction and explanation.

FORECASTING

Once the model is ready, it can be used for forecasting. It must be noted that an advanced and complex model is not a guarantee of good forecasts. Simple models do sometimes result in better prediction than complex models. The accuracy of forecasts depends on a number of factors: (a) model specification; (b) accuracy of data; (c) estimation procedure: (d) stability of the system; and the (d) length of the projection period. Long term forecasts cannot be as good as short term forecasts because the longer projection period is, the greater is the likelihood of changes in the system dynamics (i.e., relationships between variables).