Levels on interaction terms

[anniejw6]

via Murtaza

In Stata:

regress y i.sex#i.group or regress y i.sex i.group
.   margins sex#group

Adjusted predictions                            Number of obs     =      3,000

Model VCE    : OLS

 

Expression   : Linear prediction, predict()

------------------------------------------------------------------------------

             |            Delta-method

             |     Margin   Std. Err.      t    P>|t|     [95% Conf. Interval]

-------------+----------------------------------------------------------------

   sex#group |

     male#1  |    50.6107   1.367932    37.00   0.000     47.92852    53.29288

     male#2  |   61.98514   .7772248    79.75   0.000     60.46119    63.50908

     male#3  |    72.2295   .8074975    89.45   0.000     70.64619     73.8128

   female#1  |   72.23577     .63942   112.97   0.000     70.98203    73.48952

   female#2  |   78.75863   .9434406    83.48   0.000     76.90877    80.60849

   female#3  |    87.7697   2.468947    35.55   0.000     82.92869     92.6107

------------------------------------------------------------------------------

Current best workaround

with(margex,table(sex,group))

 

margex$sexgrp <-with(margex,interaction(group,sex))

margex$sexgrp <- factor(as.numeric(margex$sexgrp) - 1,

                        labels=c("fem.grp1", "fem.grp2","fem.grp3",

                                 "mal.grp1","mal.grp2","mal.grp3" ))

with(margex,table(sexgrp))

 

g <- glm(y ~ as.factor(sexgrp) , data = margex)

summary(g)

> modmarg::marg(mod = g, var_interest = "sexgrp", type = 'levels')

[[1]]

              Label   Margin Standard.Error Test.Stat       P.Value Lower CI (95%) Upper CI (95%)

1 sexgrp = fem.grp1 72.23577      0.6394200 112.97077  0.000000e+00       70.98203       73.48952

2 sexgrp = fem.grp2 78.75863      0.9434406  83.48022  0.000000e+00       76.90877       80.60849

3 sexgrp = fem.grp3 87.76970      2.4689473  35.54944 3.100836e-231       82.92869       92.61070

4 sexgrp = mal.grp1 50.61070      1.3679321  36.99796 4.247584e-247       47.92852       53.29288

5 sexgrp = mal.grp2 61.98514      0.7772248  79.75187  0.000000e+00       60.46119       63.50908

6 sexgrp = mal.grp3 72.22950      0.8074975  89.44857  0.000000e+00       70.64619       73.81280