Hi Govinda,

Could you please make your question clearer. What do you mean exactly by the "problem of variation of sample size"?

Govinda,

If the problem you are talking about is related to encountering non-constant variance in your model, I would recommend weighted least square approaches.

Hope this helps.

(link: https://onlinecourses.science.psu.edu/stat462/node/186).

My Dr. Govinda .. what you mean of sample size (small, moderate,Large)

you can use simulation experiment to digonastic this varation.

In regression analysis it is important for the sample sizes to be the same for the dependent and the independent variables. That is, a sample size of n is given by

(yi, x1i, x2i, ... , xmi), I = 1, 2, 3, ..., n

for the variables Y, X1, X2, ... , Xm.  Do you mean that the sample sizes of the variables could be unequal?

You might try weighting each data group by the inverse of its sample variance (1/s^2). Google: weight "inverse of the sample variance" for more information.