In a co-relational study of "high school students' study habit, learning stress and academic performance" the available sample size is only 30. Pearson's linear correlation coefficient requires minimum 50 samples. We have tried Scatter Diagram to note the direction. Is there any alternative way to find its relationship?
As Prof. Gash recommends, go ahead with Pearson. But examine better the peculiaritires of the measures you are dealing with: parametric? Nonparametric? Perhaps it would be also the case of the classic nonparametric Spearmam rank correlation.
I am thankful to Hugh Gash, Gunnar Magne Økland and Maria Lucia Faria Moro for this help.
All your answers have opened up my thinking to new dimensions. I would try to solve the issue and let you know the outcome. This is a Master's thesis which i am guiding.
I would also recommend non-parametric alternatives to Pearson coefficient. First, you probably work with ordinal data. Second, Pearson coefficient measures LINEAR association, whereas Spearman just MONOTONIC. Your small sample size should not disappoint you so much, it will be reflected just by poor confidence of estimated statistics. Consider rather how much your sample is REPRESENTATIVE to your population, you generalize your results to. This is much more important in my point of view. What if you have beautiful p-values, having results not being generalized? Please take into consideration multiple hypothesis testing. It increases errors of I type. If necessary, adjust it with e.g. Benjamini and Hochberg procedure.
Good luck!
Thanks Michal S. Karbownik for your suggestions. We have stuck to Pearson. You are right about the REPRESENTATIVE samples to population which is doubtful because of Convenience Sampling technique. After a thought we have developed another tool to try triangulation. I will let you know the result. We are enjoying the fun of trial & error
Thanks, Ratna!
I like your last sentence. This is the most important: to enjoy what you do!
Yes, research is excitement of exploration to unknown and the extreme joy it brings if the findings are useful to help someone to change their situations.
I don't know whether Research Gate can be a forum for expressing concern in poetic form, as I realist that all the members are great scientists of true nature, and so wise in sharing thoughts - i feel so fortunate to be in this. Please see the attachment if you like to ...
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning