Hi,
I am conducting a survey in which user's will be asked series of questions to identify effect of user's knowledge on user's attitude on digital privacy.
Section 1 is "knowledge section" in which users will be asked multiple choice questions to measure their knowledge and the data will be collected as a score.
Section 2 is "user's attitude on digital privacy" and the data will be collected from responses based on 7-point likert scale from strongly agree to strongly disagree.
Problem: If the knowledge section responses are in MCQs and Attitude section is based on likert scale then implementing a pearson correlation coefficient method is not a viable solution. How can I find the correlation between knowledge and attitude.
You can always define a distance, and correlation between discrete scales. Go back to the definition of correlation E(XY), use normalised values where possible (think of a scalar product, cos(v,w) = /(||v||.||w||, do the same E(XY)/ sqrt (E(X^2).E(Y^2)))
Hi,
Short answer: if you want to play it safe, you can use a non-parametric correlation test: https://www.statisticssolutions.com/correlation-pearson-kendall-spearman/
Depending on your data there may be ways to use pearsons (treating likert scales as interval data is a debated subject). If your sample is quite big, I would not be scared out of parametric tests (even if I would also do and report on an non-parametric to satisfy everyone and increase validity).
Another way to go, if your data allows would be to create indexes from your likert-questions. The common way to do likert-scales is to have question sets anyway, and each set could be turned into an index and then you can correlate the index with your knowledge score using parametric tests.
reading tip: Norman, G., Likert scales, levels of measurement and the “laws” of statistics. Advances in health sciences education, 2010. 15(5): p. 625-632.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude