Announcement

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

  • comparing two results

    Hi every one
    I got a revision regarding to my analysis result
    i showed my results by distinguishing two samples ( government sample and non-government) both results are significant.
    But a reviewer said we have to use type of analysis to compare the two results to make sure that the difference is significant.
    this is the command of first sample
    HTML Code:
    reg csr_rating Log_CEO_ExcPayGap Boardsize Profmale Pro_IndDir CEOdulity  proBoardshare sales_log CAssets FinancialLeverage firmage    TobinQA BTMA i.year i.industry if SOE==1
    and this is for second sample
    HTML Code:
    reg csr_rating Log_CEO_ExcPayGap Boardsize Profmale Pro_IndDir CEOdulity  proBoardshare sales_log CAssets FinancialLeverage firmage    TobinQA BTMA i.year i.industry if SOE==0
    therefore
    please what are the tests that can be used for comparing and which one is the more accurate?
    how can i implement it ?
    thanks in advance

  • #2
    You can Google "chow test" and "site:stata.com" to find a couple of FAQs on this, say, here and here. There's not a whole lot to say about setting up interaction terms, and so there's some overlap.

    Comment


    • #3
      @Joseph Coveney thank you so much.
      but let me show you what i have understood and applied
      i first estimated the results of govern_sample by using those commands

      HTML Code:
      reg csr_rating Log_CEO_ExcPayGap Boardsize Profmale Pro_IndDir CEOdulity  proBoardshare sales_log CAssets FinancialLeverage firmage    TobinQA BTMA i.year i.industry if SOE==1
      estimates store state1
      then i also estimated the results of secon group by those commands
      HTML Code:
      reg csr_rating Log_CEO_ExcPayGap Boardsize Profmale Pro_IndDir CEOdulity  proBoardshare sales_log CAssets FinancialLeverage firmage    TobinQA BTMA i.year i.industry if SOE==0
      estimates store state0
      and then i sued both results as one dataset by this command
      HTML Code:
      suest state1 state0
      and i run the following command
      HTML Code:
      test [ state1_mean]Log_CEO_ExcPayGap-[state0_mean]Log_CEO_ExcPayGap = 0
      which means i used the mean of g_1 multiply independent variable + the mean of g_2 multiply independent variable against zero.


      finally
      i got the following results of Chi2

      HTML Code:
                 chi2(  1) =   27.78
               Prob > chi2 =    0.0000
      here my a new question
      can i interpret this (0000) Chi2 value that the difference is significant or not?
      because my senior told me chi2 must equal 10% unless the difference is not significant.

      please explain to me whether my calculation process is right or not and the results are significant or not

      thaaaaaanx a lot

      Comment


      • #4
        @Joseph Coveney sorry for bothering you but i wanna ask about Chow test
        to my humble knowledge, chow test is only for examining whether it is better to run the model for whole sample at once or to run it based on subsamples separately.
        according to so, the reviewer gave me a comment that i have to use a test which enable me to compare between the results of two sub-samples. my question is if i use this test and report that there is significant difference between the results. am i right?
        best regards
        Alkebsee Radwan

        Comment


        • #5
          Originally posted by ALKEBSEE RADWAN View Post
          my senior told me chi2 must equal 10% unless the difference is not significant.

          please explain to me whether my calculation process is right or not and the results are significant or not
          I don't know what that means, but show what you got to your senior. If your senior is happy with it, then smiles all around.

          Comment


          • #6
            ok thanks a lot

            Comment

            Working...
            X