Hello everybody. I have a dataset of 100 individual specimens reactive (value: 1) or not (value: 0) to Compound A and Compound B. I am interested in knowing if there is any correlation between reactivities to both compounds or not. I did a Pearson correlation analysis and I got a positive correlation (which makes sense when I look at the data), but reading a bit more, I encountered that Pearson might not be the correct statistical analysis for this type of data. Do you maybe have any insight you could share with me regarding this?
You want to find the correlation between two binary variables. The usual statistic for this is phi. Mathematically, phi is exactly the same as Pearson's r. So you can calculate the correlation with either statistic, but you may want to report that you calculated it as phi, to avoid any uniformed criticism.
Thanks so much for your help, Dr. Mangiafico! I really appreciate it!
To add to previous answer: There are several methods,
https://www.biz.uiowa.edu/faculty/nstreet/research/tkdd14.pdf
but I would go with PHI. Its also known as Matthews correlation coefficient.
- Badges
- Science topic
- Similar topics
- Human-Computer Interaction
- Collaboration