# Measuring the presence of a correlation

#### suranjithe

##### New Member
1. Given a frequency data set (collected from 10 users by letting them watch certain videos and asking some questions about the videos) how do I determine any correlation between two variables (Correctly guessed and Correctly memorized) and calculate the significance level?

Video id | Correctly Guessed | Correctly Memorized | etc..
----------|-------------------|---------------------|------
1 | 5 | 3 |
2 | 2 | 2 |
3 | 7 | 5 |
4 | 4 | 6 |
. | . | . |
. | . | . |
. | . | . |
----------|-------------------|---------------------|------
35

2. Even though in Matlab corrcoef returns the Correlation coefficients with the P-value by calculating the t-statistics, is it correct to assume a normal distribution for such a frequency data set with a small sample size (10 users)? If not, what sort of statistics should be used?

Last edited: