Thanks to Harry, Ann and John.
There are 109041 respondents (records) in the dataset.
I am interested in analyzing 1175 of these records having a
certain response (recoded as a particular variable. I believe
no variable has more than about 5 or 6 distinct responses
(most are not continuous variables, certainly not the few of
Ann, are you saying you have analysed Demographic and
Health Survey (or National Family Health Survey) data for
India? What kind of machine did you use?
I feel this should run on an ordinary laptop. I closed
Chrome and like I said, the performance improved but was still
slow (got "non response"), but got results. Now we shall try
some further calculations.