1. No, you don't need to do this. If your scores are very badly skewed you might want to use a data transformation (e.g., square root) or just run nonparametric (rank-order) analyses, but it's unlikely to be necessary.

2. It makes no difference at all as far as the results of statistical analysis are concerned. Just state clearly, in your paper, which you did, and please report the basic descriptive statistics for each measure.

3. Sure, good old Pearson's r makes sense here, unless you decide to use rank-ordered correlation (Spearman's rho or Kendall's tau). If you end up with more than one predictor for a given step (e.g., both high Openness and low Conscientiousness predicting idea generation), you might want to follow up with a multiple regression analysis to see if the combination enhances predictive power.

I guess one could make a case for canonical correlation in a case like this, but as far as I can see very few people ever use that method.

Stephen Joy Appreciated for the quick feedback, and clear answers!

A follow-up question to #3 - I did indeed find that there are sometimes multiple predictors for one step (and namely high Openness & (interestingly) high Conscientiousness). Unfortunately, I am not too familiar with multiple regression, but will do some research on it. As a start, I was curious if there are any unexpected pre-requisites I should take note of (i.e. standardized data, minimum sample size, etc.)?

I have also decided to see if age / tenure would moderate the interactions between Big 5 traits and stages of innovation. I was going to use Andrew Hayes' PROCESS for the moderator analysis, do you think there are additional steps I should take to measure the moderator effect?

Thanks for your time!

George Wang

Interesting topic. Amar Bhide has conducted scholarly work on the relationship between personality and creativity. I suspect that there would be more research testing different personality traits at different stages of the "Innovation Value Chain." Out of interest, what is "the big 5 personality" test?

Kheepe Moremi The Big 5 Personality test is one of the most studied, and (somewhat) proven personality assessment tests. Feel free to have a look here: https://www.psychologytoday.com/us/basics/big-5-personality-traits

George Wang

Thank you. I made an error in the previous response. The scholar who has done most work on creativity and personality traits is Teresa Amabile. See some of the work below - Article Perspective on the Social Psychology of Creativity

Multiple regression is easier than moderator analysis, in my opinion! You do need to make decisions about the method (forward entry, backward, simultaneous - just avoid stepwise), but any stats program will run the analysis for you, even a free menu-driven one like JASP. You should have a larger sample than for zero-order correlations, naturally; it would be best if you had at least 100 participants, vs. as few as 40-60 for plain old correlations.

I do want to add (in response to Mr. Moremi) that Amabile's very important, very fine work on creativity mostly has little to do with personality traits. In fact, for some years she maintained that her method (the Consensual Assessment Technique) would not be useful in studies of individual differences. My own (less influential) research showed that it could, in fact, be useful in a personality-based study, and others have found the same. If you want research on the relationsh personality traits (such as openness) and creativity, I'd say to start with Robert McCrae's papers from the 1980s (that's when "openness" first got that name; see

McCrae, R.R. (1987). Creativity, divergent thinking, and openness to experience. Journal of Personality and Social Psychology, 52, 1258-1265), check out a meta-analysis of that literature (Feist, G.J. (1998). A meta-analysis of personality in scientific and artistic creativity. Personality and Social Psychology Review, 2, 290-309), and maybe grab a copy of the Cambridge Handbook of Personality and Creativity.

1. The calculation procedure for the correlation test is based on transforming the results into z values, free of the unit of measure, which is why they should not be standardized in advance.

2. Correlation testing involves two or more variables. When testing the correlations between several variables, the program will generate a very complex correlation matrix, difficult to track. In order to avoid that each question becomes a variable, I recommend grouping them into factors.

3. It is very important to define the number of independent variables, because the sample volume and implicitly the correlation test that can be used depend on them. For more information, please see Samuel B. Green (1991) How Many Subjects Does It Take To Do A Regression Analysis.

Also, before selecting a correlation test, it is advisable to go through the Explore procedure in SPSS.

Best,

Thank you for the additional insights everyone! Will follow-up on the additional resources :)