======= Date Modified 24 Feb 2012 15:39:42 =======
Hi all,
I am currently researching immigrant entrepreneurs. Due to the absence of any reliable database to work with I choose the snowballing sampling technique.
My question how big my sample should be?
Thnx in advance

it depends on what you want to do with it. If you want to run statistics on it, certain types of statistics require different minimum sample sizes than others. Generaly the bigger the better though.

======= Date Modified 24 Feb 2012 15:37:55 =======
Thnks for a speedy response. I intend to do descriptive statistics, and since I would be comparing different immigrants groups, may be ANOVA tests would be done.

You sample size needed would depend on the size of effect you expect to see, as well as the standard deviation of whatever you are measuring in the different populations. There are loads of free online calculators for (attempting) to work out sample size...and I say attempting as often you do not know either the effect size of sd but have to estimate these based on previous literature. I found G-power one of the best and most straightforward calculators to use, though I was doing it based on a t-test rather than ANOVA, but I think it can do any test. Another option is to look at other similar studies and use their sample size as a bench mark (or ask your supervisors what they reccomend) though based in the UK they like to see a power calculation on the ethics application. (even if based on whatever rubbish numbers you want to use!) I went with how many people do I want to test...what does the power calculation say...fudge the numbers a bit till they match! :$ hmm maybe I shouldn't admit that though!

Hey! Yeah, it sounds like you need to do a power calculation- I used the online calculator for that too but they can be a bit confusing so I did check with the departmental statistician as well! I calculated sample size for 80% power and .05 significance level based (rather loosely) on expected correlations found in previous research studies. To be honest I think a lot of people fiddle the power calculations a bit so that they come up with a realistic and achieveable sample size- when I checked the literature my sup had cited in the power calculation used for my study, turned out the figures she had taken from these papers didn't quite match up to the ones she had used in the power calculation, so I had a very fun hour with the statistician trying to sort that one out! Good luck with it, KB

Thanks to all who have replied. Your experience would be very usefully. I was doubting what to do, but now I can decide with degree of faith.