Hello!
I am working on securing a grant for a study design on participants' willingness to participate in clinical trials. We were hoping to keep surveys anonymous and have them unpaired/unmatched. We would do a pre and post-survey for patients. I'm wondering--what statistical analysis would be best for analyzing those results. I've read mixed reviews on unpaired analyses.
Our potential sample size is 231.
Cody Ingle, there is a difference between anonymous and confidential participation. If you are able to maintain confidentiality of your data, then you will be able to use the preferred matched (or paired) data analysis. This matches the pre-intervention survey results to the subsequent post-intervention survey results based on a unique identifier. Obtaining IRB approval might actually aid in securing the confidentiality.
With anonymous data, you have no choice but to perform an unmatched analysis which compares the first group of survey participants against subsequent groups of survey participants. The unmatched data analysis approach is often used to compare results by cohort, as in, Cohort 1 versus Cohort 2, for example.
In the case where there is a pre and a post survey, then you can compare the first group of survey participants who participated in the pre-intervention survey against the second group of survey participants who participated in the post-intervention survey. Since there is no way to know which participants experienced growth (or loss) over the intervention timeframe, this provides a weaker form of evidence than the matched data analysis.
If the data you are gathering is from a Likert-type survey, then mean comparisons should be conducted at the scale level as opposed to the item level using a simple t-test. Report upon the difference in means, the p-value, and the effect size.
David Sul - Thank you so much! That is also what I was thinking. Yes - The PI wanted to keep surveys anonymous and confidential (unmatched), which I did inform of weaker evidence. Thank you for your help!
David Sul - Sorry for this late reply, but thank you for the clarification there as well. This makes sense for study design and the anonymous vs confidential is helpful!
- Badges
- Science topic
- Similar topics
- Computer Science
- Research Topics