I would suggest you use Goodman-Kruskal's Gamma, it tends to be more robust for ordinal data, like in your case. Get more information on this from searches, you might like it.

Hello Phi,

If it's just pairwise relationships you're interested in, then gamma or spearman rho could suffice. If, on the other hand, you were interested in what ensemble of the "emotion" variables accounts for differences on the reported intent to purchase, then ordinal regression might be suitable.

Good luck with your work.

I wonder if you really want to correlate all "emotion" items with all "purchase" items pairwise? Or do you want to create one "emotion" variable and one "purchase" variable and correlate these two new variables? Or do you want to explore the relationship between for example one (or one compound) variable for purchase and how this is related to all emotion variables, i.e. a multiple regression, where the intercorrelations between the predictor (different aspects of emotions) variables are taken into account?

Please tell us more about your model you want to explore and what your hypotheses are.

My basic hypotheses is that positive emotions (happy, interest, inspiration etc) have a positive influence on the consumer's intent to purchase. So I suppose I am looking to see if the emotions influenced intent.

I'll look into the analyses that you all suggested, thank you!

based on your hypothesis you may have multiple options:

1- convert your ordinal variables into single single nominal variable (good or bad) and do simple chi-square test

2- convert into dummy variables and input them into regression models

3- I agree with David Morse answer, though fitting the best model depends basically on your hypothesis. In addition to ordinal regression model, you may want to do general linear model with ANCOVA if you have any additional outcome variable for which you adjust.

so you have to choose the most statistically reliable method with carefully interpreting the result of your model.