I want to see if there is any correlation between number of pages in a paper and the citation it receives. If I just plot number of pages vs number of citations of a paper, I might get a correlation, but that might be wrong, because there are other variables such as age of the paper, venue impact factor etc. How to get around this problem? If I do a multiple regression with some known variables and report the coefficient for my variable, is that sufficient?

  • $\begingroup$ "test" $\neq$ "report"; do you mean "test the coefficient for my variable"? Or are you more worried about measurement of the effect than testing it? $\endgroup$ – Glen_b Sep 2 '14 at 2:11
  • $\begingroup$ at this moment, i want to see if there is any significant correlation. so i guess i am more focused on the test part. $\endgroup$ – rivu Sep 2 '14 at 4:45
  • $\begingroup$ Well, multiple regression is what I'd do. If you specifically want a correlation that accounts for other variables you might calculate a partial correlation. $\endgroup$ – Glen_b Sep 2 '14 at 4:54
  • $\begingroup$ as the predictor variables do not follow normal distribution, should i use linear multiple regression or something like logistic regressiom? $\endgroup$ – rivu Sep 2 '14 at 15:34
  • $\begingroup$ Why would the predictors need to be normal for anything? $\endgroup$ – Glen_b Sep 2 '14 at 23:06

Your Answer

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

Browse other questions tagged or ask your own question.