Be useful if you say what you mean by "significant." Perhaps just above some size correlation between each of those variables and mitogenesis response. Are you interesting in their bivariate relations, or something else?

Hello Cheyenne R. Smith In such cases, I am more interested in using regression models (their types) such as Quantile Regression. So I suggest you enter the world of regressions along with everything else you do

Cheyenne R. Smith -

If you are looking at predictors for a multiple regression, remember that each predictor's relative importance depends partly on the other predictors present. Collinearity and other considerations impact their behavior. There are methods for looking at this. Forward and backward elimination are in disrepute now because the composition of predictors depends on the order. Also, different results could be obtained, regardless of method, when different samples are used. Thus both "graphical residual analyses" and "cross-validations" are important. I suggest comparing model performances (for different samples if possible) using a graphical residual analysis, one scatterplot for each sample, with the different models together on each scatterplot. For candidate models, I suggest using your subject matter expertise as to what is needed. A model should have all it needs and no more. From your graphical analyses, you may determine usefulness of various predictors.

You might want to consider the different categories separately.

It may be helpful for you to consider the following open access version of a paper by Galit Shmueli:

Galit Shmueli:

https://www.researchgate.net/publication/48178170_To_Explain_or_to_Predict

In the appendix, there is an example I recommend that you read.

I hope that some of the above is of some use to you.

Best wishes - Jim