Coefplot with second y-axis

[Sumedha Gupta]

Dear All,
I am running a fixed effect neg binomial regression and want to plot some of the regression coefficients with 95% CI using coefplot.

. eststo raw11: xtnbreg sumnalaxone  int3-int4 int6-int9 i.year i.treat age if year>2012, fe  exposure(exposure)   /* WORKS*/
note: 2281 groups (2281 obs) dropped because of only one obs per group
note: 2529 groups (6872 obs) dropped because of all zero outcomes

Iteration 0:   log likelihood = -394.85682  
Iteration 1:   log likelihood = -378.89183  
Iteration 2:   log likelihood = -373.71977  
Iteration 3:   log likelihood = -373.29432  
Iteration 4:   log likelihood = -373.28541  
Iteration 5:   log likelihood = -373.28387  
Iteration 6:   log likelihood = -373.28352  
Iteration 7:   log likelihood = -373.28343  
Iteration 8:   log likelihood = -373.28342  

Conditional FE negative binomial regression     Number of obs     =        685
Group variable: studypersonid                   Number of groups  =        220

                                                Obs per group:
                                                              min =          2
                                                              avg =        3.1
                                                              max =          7

                                                Wald chi2(14)     =      86.89
Log likelihood  = -373.28342                    Prob > chi2       =     0.0000

------------------------------------------------------------------------------
 sumnalaxone |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        int3 |   1.150222   1434.497     0.00   0.999    -2810.411    2812.712
        int4 |   -.650432   1.213165    -0.54   0.592    -3.028191    1.727327
        int6 |   2.006028   .5589152     3.59   0.000     .9105741    3.101481
        int7 |   2.109025   .5748433     3.67   0.000     .9823525    3.235697
        int8 |   2.130793   .5859116     3.64   0.000     .9824274    3.279159
        int9 |   2.272671   .7126006     3.19   0.001     .8759997    3.669343
             |
        year |
       2014  |   14.12338   638.2786     0.02   0.982     -1236.88    1265.126
       2015  |   14.04019   638.2786     0.02   0.982    -1236.963    1265.043
       2016  |   14.69761   638.2786     0.02   0.982    -1236.305    1265.701
       2017  |   14.98768   638.2786     0.02   0.981    -1236.015    1265.991
       2018  |   15.10221   638.2786     0.02   0.981    -1235.901    1266.105
       2019  |   14.59542   638.2787     0.02   0.982    -1236.408    1265.599
             |
     1.treat |  -.3597424   .6071373    -0.59   0.554     -1.54971    .8302249
         age |   -.041732   .0211664    -1.97   0.049    -.0832173   -.0002467
       _cons |  -22.66602   638.2789    -0.04   0.972     -1273.67    1228.338
ln(exposure) |          1  (exposure)
------------------------------------------------------------------------------

. coefplot, keep(int3 int4 int6 int7 int8 int9) mcol(cranberry) msym(o) ciopts(lcolor(gs8)) ///
>  coeflabels( int3="2013" int4="2014" int6="2016" int7="2017" int8="2018" int9="2019" , angle(45))  mlabsize(huge) ///
>  title("", color(black) size(large)) ysc(r(-4 4) noext) ///
>  yline(0, lcol(black)) xline(2.5, lwidth(2.2) lcolor(gs10))   ///
>  ytitle(Estimated coefficient, size(medium) margin(small)) vertical omitted

Since the CI on int3 is so large, we cannot see the CI of remaining coefficients. Is there a way I could plot int3 on yaxis(1) and the others on yaxis(2)? Will be grateful for any guidance to modify the coefplot code to utilize both y-axes.
Many thanks.
Sumedha.

[Ben Jann]

You could place 2013 and 2014-2019 into separate graphs and then combine them, something like this:

sysuse auto
regress price weight mpg turn trunk
coefplot ., keep(weight) vertical fxsize(25) name(a, replace)
coefplot ., keep(mpg turn trunk) vertical name(b, replace)
graph combine a b

Alternatively, some hints on how one could clip CIs are given at http://repec.sowi.unibe.ch/stata/coe...vals.html#h-7.