Hi, MAY90,

(it looks like we're compatriots), it doesn't make sense to talk about correlation between a qualitative, also known as **categorical** or **nominal** variable, and a quantitative (i.e., an integer or continuous variable), because correlation, intuitively, answer the following question: "if A increases, does B increase, decrease or stay the same, **on average**?" with a number which goes from -1 to 1. Now, of course it doesn't make sense to ask if a categorical variable increases or decreases: is "brown" or less than "green"? Is "drama" more or less than "comedy", "sci-fi", etc.? Also, it doesn't make sense to talk about the average of a nominal variable. Thus, you can see that there's no hope of recovering a symmetric measure of association between a nominal and a continuous variable, unlike the case of two continuous variables, where the coefficient of correlation of A and B is obviously the same as the coefficient of correlation of B and A. Because of this, from now on I'll use a notation which is not symmetric on purpose, indicating with A the nominal variable and Y the continuous variable

What makes sense, though, is to ask **if** the average of **Y**, *conditional* on the level of **A**, changes significantly, and by how much. Both questions are answered by a simple linear regression (or equivalently one-way ANOVA). To measure only the strength of correlation you could use intraclass correlation. I have to run now, but feel free to ask for more details.