I intend to research (quantitatively) "The impact of COVID-19 lockdown on the 'psychosocial landscape' of the economically middle-class population (the number of middle-class population is around 550000) in Jaffna district, Sri Lanka.
I prepared a structured questionnaire, which inquires how people have complied with the advice and guidelines of the WHO and the government authorities; how are they affected psychologically, how do they economically cope with the current crisis, etc. So far, I have got a response from randomly selected participants. Is the sample size enough to draw the result?
The sample size depends on the power of results that your study wants to demonstrate. Therefore, the results do affect the sample size you need.
You could use GPower to calculate the required sample size for the power you want to achieve or you could also look what power would result from the 40 people you already have. Furthermore you need to know what statistical model you want to use (t-test, F-test etc.). Do you have a specific hypothesis or certain groups you want to compare?
Here you can get GPower, it's open source: https://www.psychologie.hhu.de/arbeitsgruppen/allgemeine-psychologie-und-arbeitspsychologie/gpower.html
Need to focus on your objectives and study design too. The bigger sample will help you sure.
It's difficult to find normal distribution with a very small sample size that may affect your inferential analysis..
Joseph -
With such a large population, you might need to stratify. In general, stratified random sampling is helpful in quantitative studies to arrive at a lower overall standard error, but I think you would probably then want to look at each stratum which would require a larger sample.
The only way that your sample would be adequate would be if there is very little difference in responses across the population. That will vary by question, but I expect there will be large differences across the population for all questions (variables). Also, I suspect it is really not a random sample. You probably have a large number of nonresponses, so even if you had a way to randomly select your sample, I suspect that you had a much bigger sample size from which only 40 responded. If that is the case, then you are probably assuming that the average nonresponse is the same as the average response, which is likely not the case. The standard deviations could be substantially different also.
You might consider your current sample for a pilot study in preparation for a larger effort.
Best wishes - Jim