Chosing Predictor and Response Variable

Question

I understand that response variables are dependent on predictor variables.

For an easy case like Baking a cake, we know that Baking time, oven temperature are predictor variables while Moisture of the cake, thickness of the cake are dependent variables.

However in the following example, I find it very tricky to distinguish between Predictor and Dependant variables. I state the problematic statement below.

Ten sales staff in a large company were given a creativity test (scores range from 0 to 20, with higher scores indicating greater creativity) and were evaluated regarding sales growth performance (a score of 100 indicates an average performance, and larger scores indicate better performance). The creativity scores and sales growth performance scores are given below.

The manager wants to investigate whether creative people tend to perform better with regard to sales growth.

Could you please help me figure the dependent variable and predictor variable in the above example?

Answer 1

I'm not sure that this is properly a prediction problem (the manager might be looking for an association, and not a predictive relationship between scores, for example). The full problem and context surrounding it would help.

Regardless, the final excerpt has the information you need. The manager wants to investigate whether or not creative people (those with higher creativity scores) exhibit better sales growth (a higher sales growth score).

The information the manager starts with is the creativity score, and the manager wants to find out the association between that creativity score and sales growth. So, using your terminology, creativity score would be the predictor and sales growth score would be the dependent variable.

As a side note, I think that those terms might be contributing to your confusion. I personally prefer "regressor" to "independent variable" (especially in a model that has multiple regressors, where "independent" is kind of a misnomer), and "regressand" to "dependent variable". Regressor and regressand are nicely generic and always make the relationship between variables clear. In any case, "dependent variable" and "predictor" don't fit especially well together in my mind.