Announcement

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

  • Interpreting AME after margins command

    Dear all,

    I am using the -margins- command to estimate the impact of several binary explanatory variables on log of contribution (in $)(all my explanatory variables are binary). I use the following command:

    Code:
    margins, dyex(*) predict(ystar(0 .) eq(#3)) atmeans
    And receive the following result:
    HTML Code:
    Conditional marginal effects                    Number of obs     =      4,974
    Model VCE    : Robust
    
    Expression   : E(lny3*0<lny3), predict(ystar(0 .) eq(#3))
    dy/ex w.r.t. : max_10 max_20 line_5 age40_65 above65 female below100000 _149999    _199999    married
    envdonor farmeryes perfield sughigh
    at           : max_10          =    .3365501 (mean)
    max_20          =    .3329312 (mean)
    line_5          =    .4989948 (mean)
    age40_65        =    .6410097 (mean)
    above65         =    .3049866 (mean)
    female          =    .2427064 (mean)
    below100000     =     .153114 (mean)
    _149999         =    .3368508 (mean)
    _199999         =    .2563625 (mean)
    married         =    .7696129 (mean)
    envdonor        =    .3187746 (mean)
    farmeryes       =    .5028146 (mean)
    perfield        =          .5 (mean)
    sughigh         =    .5006031 (mean)
    
    
    Delta-method
    dy/ex   Std. Err.      z    P>z     [95% Conf. Interval]
    
    max_10    .0631488   .0587649     1.07   0.283    -.0520282    .1783258
    max_20    .0939937   .0572245     1.64   0.100    -.0181643    .2061518
    line_5   -.0602471   .0689658    -0.87   0.382    -.1954175    .0749234
    age40_65    .7288097   .3987188     1.83   0.068    -.0526647    1.510284
    above65    .3905804   .1965155     1.99   0.047     .0054171    .7757436
    female   -.0632363   .0924284    -0.68   0.494    -.2443926      .11792
    below100000    .0536459   .0590108     0.91   0.363    -.0620131    .1693049
    _149999    -.174858   .0965162    -1.81   0.070    -.3640264    .0143103
    _199999    -.060281   .0737858    -0.82   0.414    -.2048985    .0843366
    married    .0042625   .3035701     0.01   0.989    -.5907239    .5992489
    envdonor           0  (omitted)
    farmeryes    .0747624   .0698082     1.07   0.284    -.0620592    .2115839
    perfield   -.0302773   .0667833    -0.45   0.650    -.1611702    .1006156
    sughigh    .0542216   .0663194     0.82   0.414    -.0757621    .1842053
    My question is, how do I interpret the coefficients? For example, say for the variable max_10 (I know that it's not significant), do I interpret that max_10 increases the expected contribution by 6 percent keeping all other variables constant at their means? Would this also be the average partial effect for the population rather than the average marginal effect?

    Thanks,
    Anwesha
    Last edited by Anwesha Chakrabarti; 24 Nov 2018, 14:11.
Working...
X