Don't understand what marginsplot is doing.

[Dick Campbell]

I have run a simple regression model followed by a request for margins as follows:

. regScores ib3.Time##ib2.Group
command regScores is unrecognized
r(199);

. reg Scores ib3.Time##ib2.Group

      Source |       SS           df       MS      Number of obs   =        42
-------------+----------------------------------   F(5, 36)        =      0.57
       Model |  71.5570238         5  14.3114048   Prob > F        =    0.7250
    Residual |  909.384167        36  25.2606713   R-squared       =    0.0729
-------------+----------------------------------   Adj R-squared   =   -0.0558
       Total |   980.94119        41  23.9253949   Root MSE        =     5.026

--------------------------------------------------------------------------------
        Scores |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
---------------+----------------------------------------------------------------
          Time |
            1  |     -.7875      2.513    -0.31   0.756      -5.8841      4.3091
            2  |     1.5875      2.513     0.63   0.532      -3.5091      6.6841
               |
   1.GroupCorE |      1.125    2.71435     0.41   0.681    -4.379957    6.629957
               |
Time#GroupCorE |
          1 1  |   1.770833   3.838671     0.46   0.647    -6.014351    9.556018
          2 1  |   .4458333   3.838671     0.12   0.908    -7.339351    8.231018
               |
         _cons |     56.275   1.776959    31.67   0.000     52.67116    59.87884
--------------------------------------------------------------------------------

. margins Time##GroupCorE

Predictive margins                              Number of obs     =         42
Model VCE    : OLS

Expression   : Linear prediction, predict()

--------------------------------------------------------------------------------
               |            Delta-method
               |     Margin   Std. Err.      t    P>|t|     [95% Conf. Interval]
---------------+----------------------------------------------------------------
          Time |
            1  |   56.72857   1.343255    42.23   0.000     54.00432    59.45282
            2  |   58.53571   1.343255    43.58   0.000     55.81147    61.25996
            3  |   56.75714   1.343255    42.25   0.000      54.0329    59.48139
               |
     GroupCorE |
            1  |   58.40556   1.184639    49.30   0.000       56.003    60.80812
            2  |   56.54167   1.025928    55.11   0.000     54.46099    58.62234
               |
Time#GroupCorE |
          1 1  |   58.38333   2.051856    28.45   0.000     54.22198    62.54469
          1 2  |    55.4875   1.776959    31.23   0.000     51.88366    59.09134
          2 1  |   59.43333   2.051856    28.97   0.000     55.27198    63.59469
          2 2  |    57.8625   1.776959    32.56   0.000     54.25866    61.46634
          3 1  |       57.4   2.051856    27.97   0.000     53.23864    61.56136
          3 2  |     56.275   1.776959    31.67   0.000     52.67116    59.87884
--------------------------------------------------------------------------------

. marginsplot

  Variables that uniquely identify margins: Time GroupCorE

.
In the resulting plot, I get this:

I don't understand why I have the GroupCorE=asobserved and I don't understand how to get \rid of it.

[Clyde Schechter]

Well, your variable GroupCorE takes on two values, 1, and 2. And the two curves on the graph labeled accordingly show you the expected value of Scores at each time for each of those conditions. The curve for GroupCorE = asobserved gives you the expected value at each time not conditional on GroupCorE. That is, it is giving you what what you would have gotten with -margins time- followed by -marginsplot-.

If you don't want that one included, I think you will get what you want with -margins Time#GroupCorE- (note single #, not double ##) followed by -marginsplot-.

[Dick Campbell]

Thanks Clyde, exactly what I want.. It reminds me of one of the best lines I ever read in the acknowledgements for a book: "When you don't know what you're doing it's good to have friends who do."

[Clyde Schechter]

Happy to count you as a friend!

[Walid Al-Soneidar]

If you don't want that one included, I think you will get what you want with -margins Time#GroupCorE- (note single #, not double ##) followed by -marginsplot-.

I ended up here after a Google search. So, one more thank you, Clyde!