1. ## Correlation help.

Hello forum,

I stumbled upon this forum while looking up Google for information on my problem. I'm not sure if i'm using the right tool but let me state the problem first.
From a survey feedback, i want to see if there is any correlation between a question that results in a "Yes" or "No" type data and another question that is on a scale of 1 to 5. The hypothesis is that when ever there is a "Yes", we are likely to see a higher score on the scale.
How i prove this?

-R

2. Originally Posted by +rex
Hello forum,

I stumbled upon this forum while looking up Google for information on my problem. I'm not sure if i'm using the right tool but let me state the problem first.
From a survey feedback, i want to see if there is any correlation between a question that results in a "Yes" or "No" type data and another question that is on a scale of 1 to 5. The hypothesis is that when ever there is a "Yes", we are likely to see a higher score on the scale.
How i prove this?

-R
Take the 1-5 score and multiply it by +1 if y, and -1 if n.

Then if there is no correlation the distribution of modifed 1-5 score is uniform on {-5,-4,-3,-2,-1,0,1,2,3,4,5}. Use a chi-sq test to check this.

This test is stronger that you asked for, you can weaken it by looking at the sample mean and testing it against 0 as a null hypothesis, and >0 for the alternative hypothesis.

RonL

3. ## Thank you,

Thank you, will give it a shot and let you know.
-R