Ok, I'm trying to define the sample population for my PhD but I'm a bit lost.

Assuming that I'm doing research on students behavior on Faculty of X (e.g. Economics). Would it be enough to just get the sample from Faculty X in Uni A or do I need to get other students from different Unis as well to avoid bias?  Also, would it be necessary as well to get students from other Faculties that's related to X (e.g Engineering that uses Economics or Applied Mathematics that uses Economics - to get more meaningful data

Many thanks
Dan

Sampling frame depends on the question you are trying to answer. If you want a generalised view of Economics students' behaviour then it may be inappropriate to just sample from one uni. But if you're primarily interested in one uni then one faculty in one uni is fine (unless you want a comparison/control group)

So have a clearly defined research qu first, and then your sampling frame should reflect that.

======= Date Modified 13 Jan 2012 13:59:48 =======
[deleted duplicate post]

Ah, thank you for the reply. Yes I was thinking just now if I just selected the single Uni, then it would be bias to that Uni only wouldn't it? So say for example, I would like to avoid bias, would it be suitable to have sample population from selected Unis from the selected country? I'm trying to find this method but I would need samples for it. So at the moment I'm trying to define the population to test it with but I've no proper procedure on how to define the population properly. I do know 'who' the population is, but I do not know how far I need to go to get to population; would it be just with the faculty in the uni or faculties from different unis.

But even if I did select the country, wouldn't that be bias for the selected country and selected unis within that country?

Some biases are universal but some biases depend on the question being asked.

For example, if the qu is
"Are students in Dept X happier than those in Dept Y at Uni Z?"
then only sampling students in Dept X and Dept Y at Uni Z is totally appropriate

but if the qu is
"Are students who study X happier than those who study Y?"
Then yes, you could get biases from only studying one uni, and there would be a limit to how far you could generalise your findings.

So, first explicitly define your research question, THEN work out the sampling frame.