I'm doing a study to evaluate the adherence rate of clinicians to a clinical practice guideline using clinical vignettes (an online survey). The study only includes physicians who are dealing with a specific disease (a rare one). Unfortunately, there is no database, a record for the number of the registered physicians who's dealing with this case, or previous similar study.
The adherence will be presented as a percentage and a comparison between different variables (such as subspecialty and etc...) will be evaluated.
When I stared to collect data, I found that there is only few knew about this disease. In addition, most of the participants did not submit a complete response.
The survey was posted for two months and I only got 30 complete responses.
So, what is the acceptable minimum sample size for my study that I can reach and still gets a meaningful result ?
There is no "ideal number", as it depends on the size of the "Target Population" to study and the level of confidence we want to work at; yes - although only for "statistical purposes" - a "large" sample is considered when its "n" is greater than 30 subjects or elements and "small" when it is equal to or less ... this translates into the statistical tests to be used in each case and, above all, in the reliability tests of the calculated statistics; but it does not mean that it is one "n" better than the other. In short, the "adequate number" is the one that is sufficiently representative, at a pre-established level of confidence, of the "Target population" to be analyzed.
I have encountered similar challenges in the course of research, so I'll suggest the following:
1. Rather than relying solely on Data bases, why not approach an association or professional agency covering these professionals? This way, you will get their approximate or actual number and you can then cross ref this with the number of responses in your analysis/result. Besides, a recommendation letter obtained from appropriate authorities can help you gain improved access...
2. The question of 'Ethics in Research' also sets in. It is only ethical you report what you derive from your field study as it is. ....
While I will strongly recommend you try option 1 above first, you might as well instead ..
[a] Get the number of registered clinicians in your study area [rather than just generalising]
[b] Get a letter of recommendation/access to the registered clinicians from appropriate authorities in your country/county/region to improve your access and credibility. This should help improve your response rate. If not...
[c] Report what you have as it is against the details obtained from [a] above in your analysis/result.
.....
I hope you find the suggestions useful a bit. If there are further questions, let me know.
Thanks
ODJI Ebenezer
Sample size calculations depend on the design of your study and the size of the effect that you wish to detect. With any given questionnaire there will be many different study designs you can generate. You therefore need to start with your research question, not your questionnaire, and consider what combination of items from your questionnaire best reflect this question in the form of a hypothesis.
Any sample more than 30 is considered OK but more the sample better it would be to take care of sampling errors and validity.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey