I have data that has 3 different predictors for marketing a product (radio, tv, newspaper), depending how much I invest in one of those I will earn more money or less.

So, if I want to know which of the 3 gives more money, what would be the best way to do an inference? Using 3 simple regressions or a multiple regression including the 3 predictors?

  • $\begingroup$ you should probably add self-study tag to your question. you are using a scholastic dataset, aren't you? $\endgroup$ – carlo Mar 19 '17 at 11:59

Using only a single predictor, you introduce omitted variable bias due to excluding the other two (assuming they are all useful predictors). This means that the coefficient on your predictor will not be as accurate as in the other case.

You want to use multiple regression and watch for variable significance.


Do a regression with as many predictors as you need. Look at the p-values for each coefficient. If they are <0.05, the variable is most likely having an influence. Then look at the size of the coefficient. Minding units of measurement, see which variables have large coefficients—they are the most important in influencing (sales, I assume).

(For example, statsmodels in Python gives p-values)

  • $\begingroup$ @YuvalSp, do not edit someone's answer to 'correct' it. If you think something is incorrect, leave a comment, downvote, post your own (correct) answer, etc. $\endgroup$ – gung - Reinstate Monica Mar 19 '17 at 11:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.