My study will be a quantitative study and I will use an online survey. I have access to the whole population which is approximately over 200 instructors. I am wondering which sampling technique I should use.
When the population size is small as in your case ,
Case study is the technique where the universe / population is your sample.
I think Purposive sampling should be done. A simple random sampling by lottery system can be applied.
Since the universe/ Population is small it is advisable to take entire universe as sample.
Hi Mr Almansoor,
I agree with Srivastava. However, if you have access to the whole population and you dont want to involve them all in your study, then the appropriate sampling method is the "Simple Random Sample". A simple way to do that is to list all your 200 individuals in an Excel sheet and than use the function '=Randbetween' to select your sample among them.
Good Luck !
I really appreciate your responses all. This is very helpful and considerable.
Respectfully to all,
Ali
Ali -
If your population is well-known, then you may know how to efficiently stratify it. Stratified random sampling is generally more efficient than simple random sampling, meaning an estimated standard error for an overall total or mean or proportion will generally be smaller for the same sample size. (As measured using deff by Leslie Kish.) Best stratification may vary by variable (question asked), so that needs to be kept in mind. Often there are natural categories which work well.
If you have started a simple random sample, and the burden seems difficult, and especially if you have run into administrative issues and want to start again, you might use what you have now as a pilot study, and design a stratified random sample next.
If you have regressor data for the entire population, then a model-based approach could do better.
A census was suggested because it was assumed that N=200 is small enough to be able to carefully collected your data from each member of the population, with the resources you have, and that may or may not be true. Sometimes a census is not feasible with given resources, and sometimes trying to do one will likely result in more nonsampling error than the total error (from sampling error and nonsampling error) you would have from a sample. But you will have to judge that yourself, perhaps with input from some good sources.
Cheers - Jim
Given the small size of the population I would try to include them all in the study, this is known as a census.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey