How to add reference level(s) for regression interaction in Stata

[Andrzej Andrzej]

Hi,
I have this data (attached - my_data.csv ) and I want to add a reference levels (if possible) for interaction term to be explicitly visible
in regression summary table (something like extended coefficients table in eg. SPSS) ?
I want to use model specified like this in R:
model3 <- lm(response ~ time * treatment, data = my_data)
so response is DV and categorical predictors are time and treatment. Time has got two levels:
Pre and Post, and treatment has two levels as well: Control and Treatment.

I want to achieve something like this:



or this:


Can it be done in Stata ?

[Ken Chui]

Use option allbaselevels:

sysuse nlsw88, clear
regress wage i.race##i.married, allbaselevels

Results:

      Source |       SS           df       MS      Number of obs   =     2,246
-------------+----------------------------------   F(5, 2240)      =      7.22
       Model |   1180.1563         5   236.03126   Prob > F        =    0.0000
    Residual |  73187.8111     2,240    32.67313   R-squared       =    0.0159
-------------+----------------------------------   Adj R-squared   =    0.0137
       Total |  74367.9674     2,245  33.1260434   Root MSE        =     5.716

--------------------------------------------------------------------------------
          wage | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
---------------+----------------------------------------------------------------
          race |
        White  |          0  (base)
        Black  |  -2.194947    .415727    -5.28   0.000    -3.010198   -1.379697
        Other  |  -.4967496   2.037457    -0.24   0.807    -4.492251    3.498752
               |
       married |
       Single  |          0  (base)
      Married  |  -1.204674    .309034    -3.90   0.000    -1.810697   -.5986511
               |
  race#married |
 White#Single  |          0  (base)
White#Married  |          0  (base)
 Black#Single  |          0  (base)
Black#Married  |   1.439186   .5661142     2.54   0.011     .3290224    2.549349
 Other#Single  |          0  (base)
Other#Married  |   1.375469   2.448432     0.56   0.574    -3.425963    6.176901
               |
         _cons |   8.929288   .2590186    34.47   0.000     8.421347     9.43723
--------------------------------------------------------------------------------

[Andrzej Andrzej]

Thank you, how to set base level for a specific variable ? I have attached my_data, can you show me how to use it, step-by-step please ?

[Ken Chui]

The general practice is that we don't download data through attachment, due to safety and other reasons. (see FAQ section 12) A much more preferred way to share sample data is by posting the results of a command called dataex. More recent versions of Stata has it by default. If yours does not have it, you may be able to download it from ssc.

Whenever we cannot get the actual data, then we usually use a sample data set in Stata to show the concept, then the original poster needs to adopt the codes themselves.

To set a particular level as your base group, you'll first need to find out the coding scheme of that variable using codebook.

sysuse nlsw88, clear
codebook race married

And the schemes are:

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
race                                                                                                                                                                                        Race
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

                  Type: Numeric (byte)
                 Label: racelbl

                 Range: [1,3]                         Units: 1
         Unique values: 3                         Missing .: 0/2,246

            Tabulation: Freq.   Numeric  Label
                        1,637         1  White
                          583         2  Black
                           26         3  Other

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
married                                                                                                                                                                                  Married
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

                  Type: Numeric (byte)
                 Label: marlbl

                 Range: [0,1]                         Units: 1
         Unique values: 2                         Missing .: 0/2,246

            Tabulation: Freq.   Numeric  Label
                          804         0  Single
                        1,442         1  Married

Stata uses the lowest one by default. If we'd like to use "Black" in race and "Married" in married as base group, note down their code (2 and 1), and revise the code as:

regress wage ib2.race##ib1.married, allbaselevels

This is the effect:

      Source |       SS           df       MS      Number of obs   =     2,246
-------------+----------------------------------   F(5, 2240)      =      7.22
       Model |   1180.1563         5   236.03126   Prob > F        =    0.0000
    Residual |  73187.8111     2,240    32.67313   R-squared       =    0.0159
-------------+----------------------------------   Adj R-squared   =    0.0137
       Total |  74367.9674     2,245  33.1260434   Root MSE        =     5.716

--------------------------------------------------------------------------------
          wage | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
---------------+----------------------------------------------------------------
          race |
        White  |   .7557614   .3842609     1.97   0.049     .0022168    1.509306
        Black  |          0  (base)
        Other  |    1.63448   1.390834     1.18   0.240    -1.092978    4.361938
               |
       married |
       Single  |  -.2345117    .474324    -0.49   0.621    -1.164672     .695649
      Married  |          0  (base)
               |
  race#married |
 White#Single  |   1.439186   .5661142     2.54   0.011     .3290224    2.549349
White#Married  |          0  (base)
 Black#Single  |          0  (base)
Black#Married  |          0  (base)
 Other#Single  |   .0637171   2.474732     0.03   0.979    -4.789291    4.916725
Other#Married  |          0  (base)
               |
         _cons |   6.968853   .3453187    20.18   0.000     6.291675    7.646031
--------------------------------------------------------------------------------

[Andrzej Andrzej]

Thank you very much indeed, I am still beginner in Stata.It helps a lot.
Is it possible to export output window to ms word somehow ?

[Maarten Buis]

see help etable