Title
isa_search -- Wrapper Program for -isa- (Imbens' Sensitivity Analysis)
Syntax
isa_search depvar treatmentvar [indepvars] [if] [in] , tau(real) or tstat(real) [options]
options Description ------------------------------------------------------------------------- Model * tau(real) the target size of the coefficient of the treatment variable. * tstat(real) the target size of the t-value of the treatment variable. maxalpha(real) the maximum value of alpha in the search; default is maxalpha(10) minalpha(real) the minimum value of alpha in the search; default is minalpha(0) maxdelta(real) the maximum value of delta in the search; default is maxdelta(5) mindelta(real) the minimum value of delta in the search; default is mindelta(0) increment(real) the increment of the size of alpha and delta for the next iteration; default is increment(.5)
SE/Robust vce(vcetype) vcetype may be oim, robust, cluster clustvar, or opg ------------------------------------------------------------------------- * Either tau(real) or tstat(real) is required. * The value of tau(real) or tstat(real) must be positive and smaller than the original treatment effect. At least one indepvars is required. treatmentvar must be binary.
Description
isa_search is a wrapper program of -isa- Imbens' (2003) Sensitivity Analysis. This program should be used when researchers do not have any idea about how to set minalpha, maxalpha, mindelta and maxdelta in -isa-. The program automatically check where in a given region a contour line exists. Once the program ends, it will generate the diagram that consists of + (in blue), triangle (in red), and dot (in black). The contour line should exist somewhere between + (in blue) and triangle (in red) if these symbols do not represent the location of irrelevant contour, which is usually obvious from the figure. In running isa, a researcher should set the region defined by minalpha, maxalpha, mindelta and maxdelta in the following way. First, the region should be defined as small as possible as long as it produces a good contour. Second, the region should not include the areas that produce the irrelevant contours. Please read the quick guide for the further detail (Click here.)
Example
. sysuse nlsw88 . xi i.race . rename _Irace_2 black . rename _Irace_3 other . xi i.industry . rename _Iindustry_5 mnfctr . isa_search wage union age black other grade married south c_city mnfctr, tau(.314)
If the user wants to know the more detailed results, set increment(real) at the value smaller than .5.
. isa_search wage union age black other grade married south c_city mnfctr, tau(.314) incr(.25)
Saved results
isa_search saves the following variables:
isas_tau the treatment effect when alpha = isas_alpha[_n] and delta = isas_delta[_n]. isas_tstat the t-statistics of the treatment effect when alpha = isas_alpha[_n] and delta = isas_delta[_n]. isas_converged identifier of whether the estimation achieved convergence or not isas_alpha the value of alpha when quantity of interest is estimated. isas_delta the value of delta when quantity of interest is estimated. isas_plus indicator that the coordinate(isas_alpha[_n], isas_delta[_n]) is located marginally below the contour. isas_minus indicator that the coordinate(isas_alpha[_n], isas_delta[_n]) is located marginally above the contour.
Reference
Here is the link for the quick guide by the author of the program.
If you use this program, please cite:
Imbens, Guido W. 2003. "Sensitivity to Exogeneity Assumptions in Program Evaluation." The American Economic Review 93(2):126-132.
Harada, Masataka "Generalized Sensitivity Analysis." Working paper.