Hey squidders.
How big is your data sample? If you have enough samples then expectations (i.e. means) will have an approximate normal distribution.
Helloooo
I am assessing the correlation between two variables;
For pearsons rank you need both variables to be of normal distribution and be continous data?
1 of my variables are median dates between two time points, there are 140 pieces of data and they are not at regular intervals.
The other variable is continous and normally distributed.
Can I use pearsons?
Thanks
Ali
If you have enough points in the distribution, you may be able to assume continuity in the distribution.
Given you sample size I think you might be able to use the normal correlation co-efficient.
I'm basing this on the CLT where the distribution of the mean (of some arbitrary distribution) approaches a Normal distribution regardless of the input distribution.