0

I have cross-sectional data where the dependent variable shall be regressed on age. If I include age and age^2 in the regression, coefficients are highly significant. When I add age^3 as well, all coefficients (that of age, age^2, and age^3) become much smaller and insignificant. At this point, I rather expected that if age^3 does not play a role in the model, it is simple insignificant but the coefficients on age and age^2 remain significant. But this is obviously not the case here. ;) Do you have a reasoning how/why this happens? Thanks a lot! :)

user45516
  • 45
  • 3
  • 2
    Please search the site before posting a question here. This is a frequently asked question that has been addressed many times. – gung - Reinstate Monica Jun 09 '14 at 15:10
  • 1
    They're fighting with each other for pretty much the same share of the market. Plot these powers against each other for your data to see that they are highly correlated. – Nick Cox Jun 09 '14 at 15:16
  • 1
    Thanks @gung. Another good candidate for a duplicate is http://stats.stackexchange.com/q/34488. – whuber Jun 09 '14 at 15:17

0 Answers0