RE: st: RE: GLLAMM: logitic regression interaction non signficant but lincom gives signficant result

["Jenniffer Solorzano Mosquera" <Jenniffer.Solorzano.Mosquera@jibs.hj.se>]

Hi everyone,

I run -mhbounds- command to test for sensitivity my ATT estimator for the binary outcome positive2_EBITDA_mu1. According to my -psmatch2- results, my estimator indicates a positive but not significant effect for the firms treated. Taking a look to the results presented below for -mhbounds-, does it mean that the effect will never be significant no matter how much hidden bias there is? I have to look at the p-critical value with positive sign, right? If a look at the negative critical value that would mean . I am concluding based on The Stata Journal paper 'Sensitivity analysis for average treatment effects', 2007. Thanks for your insights,

 -mhbounds positive2_EBITDA_mu1, gamma(1.0 (0.05) 2.0) -

Mantel-Haenszel (1959) bounds for variable positive2_EBITDA_mu1

Gamma         Q_mh+     Q_mh-     p_mh+     p_mh-
-------------------------------------------------
    1        .544743   .544743   .292965   .292965  
 1.05        .426745   .667194   .334783   .252324  
  1.1        .312281   .782046   .377413   .217094  
 1.15        .202975   .891958   .419577   .186208  
  1.2        .098361   .997366   .460823   .159293  
 1.25       -.001963   1.09865   .500783    .13596  
  1.3       -.098352   1.19614   .539174   .115821  
 1.35       -.191118   1.29014   .575784   .098501  
  1.4       -.128199    1.3809   .551004   .083655  
 1.45       -.042557   1.46866   .516973   .070963  
  1.5        .040176   1.55362   .483976   .060137  
 1.55        .120201   1.63598   .452162   .050921  
  1.6        .197698   1.71591   .421641   .043089  
 1.65        .272832   1.79355   .392491   .036443  
  1.7         .34575   1.86905   .364765   .030808  
 1.75        .416588   1.94253    .33849   .026037  
  1.8        .485468   2.01411   .313672   .021999  
 1.85        .552504   2.08389   .290302   .018585  
  1.9        .617797   2.15198   .268354     .0157  
 1.95        .681444   2.21846   .247795   .013262  
    2        .743531   2.28341    .22858   .011203  

Gamma : odds of differential assignment due to unobserved factors
Q_mh+ : Mantel-Haenszel statistic (assumption: overestimation of treatment effect)
Q_mh- : Mantel-Haenszel statistic (assumption: underestimation of treatment effect)
p_mh+ : significance level (assumption: overestimation of treatment effect)
p_mh- : significance level (assumption: underestimation of treatment effect)
 


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/