Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Collinearity in FE models

    Dear All;
    My research is based on the influence of cultural dimesions (6 in total) on the innovation success of countries. I have collected the innovation success rates for five years. It cover more than 50 countries.Independent variables are cultural dimesions and they are fixed. So when I run longitudal regression wth FE model, DV'a are omitted because collinearity, but no problem with RE models. The results from RE model are all in line with SPSS. I have read that we can you RE models if we have chosen our sample on a random basis.My questions are 1) Is there a " trick " in statao avoid
    collinearty (like normaliation of the values in SPPS or what can I do else ? 2) What is your opinion about using RE model in my senario

    Thanks in advance,,
    Mutlu

  • #2
    Mutlu:
    - as per FAQ, please post what you typed and what Stata gave you back. Thanks;
    - -fe- machinery wipes out time-invariant predictors. Hence, there's nothing you (or Stata) can do to avoid it;
    - -hausman- can give you some indications about preferring -fe- vs -re- specification.
    Kind regards,
    Carlo
    (Stata 19.0)

    Comment


    • #3
      Carlo, thank you so much for the prompt reply, below is my commands;

      xtset ID YRS, yearly
      panel variable: ID (strongly balanced)
      time variable: YRS, 2013 to 2017
      delta: 1 year

      .
      . xtreg GEI PD IND MAS, fe

      Stata replied as follows

      note: PD omitted because of collinearity
      note: IND omitted because of collinearity
      note: MAS omitted because of collinearity

      I can run Hausman test , becuse it repuires a fe regression and re regression results for storing.

      What prevents me using RE Regression fory model, because I ran it several times all results are in line
      with SPSS, performed on a year by year basis ?
      Thank you so much indeed




      Fixed-effects (within) regression Number of obs = 405
      Group variable: ID Number of groups = 81

      R-sq: within = 0.0000 Obs per group: min = 5
      between = . avg = 5.0
      overall = . max = 5

      F(0,324) = 0.00
      corr(u_i, Xb) = . Prob > F = .

      Comment


      • #4
        Mutlu:
        there's something peculiar with your regression specification.
        Can you please post and excerpt/example of your data?
        Kind regards,
        Carlo
        (Stata 19.0)

        Comment

        Working...
        X