I think you can go for systematic sampling. You can choose to collect data from every kth school, beginning with a random start between first and kth school. The number k can be decided in the following manner:

Lets say you have a list of 100 schools. And either some statistical consideration, or time and cost considerations, or both, make you decide a sample size of 20 schools. Then k would be 100/20 = 5. And then, you may choose the first school randomly between school number 1 and school number 5 (lets say, you chose school number 3). Then onward, systematic sampling means you will collect data from every 5th school (i.e., 8th, 13th, 18th, and so forth).

A couple of caveats:

1. I am assuming that the serial number of schools in the list is not arranged as per any consideration (size, ranking of schools, etc.)

2. In case schools in a particular locality are all different from one another, and the localities across the geographical area (district, state, or country) are homogeneous, you may also think of cluster sampling. That would mean that you collect data from all schools in a sampled cluster (i.e., geography).

Hope it could help.

Best,

Rajiv

Tqvm Prof..

a little bit confusion.

for the above technique, can we make generalization?

That should not be a problem, knowing what I know. But it would be helpful if you take a second opinion.

Tqvm Prof..