Logistic structural mean model

Read in binary outcome data; y outcome, x exposure, w covariate, z* instrumental variables (genotypes).

use ivbinoutdata, clear

Fit the model with a single instrumental variable.

ivlsmm y (x = z1)
Final GMM criterion Q(b) =  2.37e-30

note: model is exactly identified.

GMM estimation 

Number of parameters =   5
Number of moments    =   5
Initial weight matrix: Unadjusted                 Number of obs   =      2,500
GMM weight matrix:     Robust

------------------------------------------------------------------------------
             |               Robust
             | Coefficient  std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
       /xb_x |   1.375182   .0563156    24.42   0.000     1.264805    1.485558
      /xb_z1 |  -.1698859   .1079229    -1.57   0.115    -.3814109    .0416391
         /b0 |  -2.076602   .1038597   -19.99   0.000    -2.280163   -1.873041
     /cmxb_x |   1.196517   .1133081    10.56   0.000     .9744374    1.418597
        /ey0 |   .1448039   .0310372     4.67   0.000     .0839722    .2056356
------------------------------------------------------------------------------
Instruments for equation 1: x z1 _cons
Instruments for equation 2: z1 _cons

  Test of overidentifying restriction:

  Hansen's J chi2(0) = 5.9e-27 (p =     .)

Note: test cannot be performed because there are no
      overidentifying restrictions.

Causal odds ratio for: x

 ( 1)  [cmxb_x]_cons = 0

------------------------------------------------------------------------------
             |     exp(b)   Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   3.308574   .3748881    10.56   0.000     2.649676     4.13132
------------------------------------------------------------------------------

Fit the model with multiple instruments.

ivlsmm y (x = z1 z2 z3)
Final GMM criterion Q(b) =  .0007498

GMM estimation 

Number of parameters =   7
Number of moments    =   9
Initial weight matrix: Unadjusted                 Number of obs   =      2,500
GMM weight matrix:     Robust

------------------------------------------------------------------------------
             |               Robust
             | Coefficient  std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
       /xb_x |   1.497546   .0629276    23.80   0.000     1.374211    1.620882
      /xb_z1 |  -.3828314   .0697615    -5.49   0.000    -.5195614   -.2461015
      /xb_z2 |   -.361644   .0634738    -5.70   0.000    -.4860504   -.2372376
      /xb_z3 |  -.3727515   .0639573    -5.83   0.000    -.4981055   -.2473976
         /b0 |  -1.692298   .1204232   -14.05   0.000    -1.928323   -1.456273
     /cmxb_x |   1.113372   .0673101    16.54   0.000     .9814465    1.245297
        /ey0 |   .1778203   .0213404     8.33   0.000     .1359939    .2196468
------------------------------------------------------------------------------
Instruments for equation 1: x z1 z2 z3 _cons
Instruments for equation 2: z1 z2 z3 _cons

  Test of overidentifying restriction:

  Hansen's J chi2(2) = 1.87445 (p = 0.3917)

Causal odds ratio for: x

 ( 1)  [cmxb_x]_cons = 0

------------------------------------------------------------------------------
             |     exp(b)   Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   3.044607   .2049329    16.54   0.000     2.668313    3.473968
------------------------------------------------------------------------------

Fit the model with multiple exposures, and instruments, and adjusting for w.

ivlsmm y w (x1 x2 = z1 z2 z3)
Final GMM criterion Q(b) =  .0006646

GMM estimation 

Number of parameters =  11
Number of moments    =  12
Initial weight matrix: Unadjusted                 Number of obs   =      2,500
GMM weight matrix:     Robust

------------------------------------------------------------------------------
             |               Robust
             | Coefficient  std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
      /xb_x1 |   1.270939   .0709528    17.91   0.000     1.131875    1.410004
      /xb_x2 |   .7280662   .0577047    12.62   0.000     .6149671    .8411654
      /xb_z1 |  -.6069302   .0915561    -6.63   0.000    -.7863769   -.4274835
      /xb_z2 |  -.6409243   .0906894    -7.07   0.000    -.8186723   -.4631763
      /xb_z3 |  -.0478545   .0956419    -0.50   0.617    -.2353092    .1396002
       /xb_w |   .6417589   .0844396     7.60   0.000     .4762603    .8072576
         /b0 |  -1.973388   .1356534   -14.55   0.000    -2.239264   -1.707512
    /cmxb_x1 |   1.193709   .1028285    11.61   0.000     .9921685    1.395249
    /cmxb_x2 |   .1719145   .1120826     1.53   0.125    -.0477634    .3915923
     /cmxb_w |   .7090571     .11249     6.30   0.000     .4885808    .9295334
        /ey0 |   .1486804   .0181689     8.18   0.000     .1130701    .1842907
------------------------------------------------------------------------------
Instruments for equation 1: x1 x2 z1 z2 z3 w _cons
Instruments for equation 2: z1 z2 z3 w _cons

  Test of overidentifying restriction:

  Hansen's J chi2(1) = 1.66158 (p = 0.1974)

Causal odds ratio for: x1

 ( 1)  [cmxb_x1]_cons = 0

------------------------------------------------------------------------------
             |     exp(b)   Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   3.299294   .3392614    11.61   0.000     2.697077    4.035978
------------------------------------------------------------------------------

Causal odds ratio for: x2

 ( 1)  [cmxb_x2]_cons = 0

------------------------------------------------------------------------------
             |     exp(b)   Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   1.187576   .1331066     1.53   0.125     .9533594    1.479334
------------------------------------------------------------------------------

Causal odds ratio for: w

 ( 1)  [cmxb_w]_cons = 0

------------------------------------------------------------------------------
             |     exp(b)   Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   2.032074    .228588     6.30   0.000     1.630001    2.533327
------------------------------------------------------------------------------