摘要 :
Placebo analysis:replicates the primary analysis with theoutcome replaced by a pseudo-outcome that is known not to beaffected by the treatment.Thus,the true value of the estimandfor this pseudo-outcome is zero,and the goal of thes...
展开
Placebo analysis:replicates the primary analysis with theoutcome replaced by a pseudo-outcome that is known not to beaffected by the treatment.Thus,the true value of the estimandfor this pseudo-outcome is zero,and the goal of thesupplementary analysis is to assess whether the adjustmentmethods employed in the primary analysis,when applied to thepseudo-outcome,lead to estimates that are close to zero.Theseare not standard specification tests that suggest alternativespecifications when the null hypothesis is rejected.Theimplication of rejection here is that it is possible the originalanalysis was not credible at all.
收起
摘要 :
Sample selection:sample is not representative of the populationof interest.exogenous sampling:sampling is based on conditioning variable(s is a deterministic function of x).endogenous sampling:sampling is based on response variabl...
展开
Sample selection:sample is not representative of the populationof interest.exogenous sampling:sampling is based on conditioning variable(s is a deterministic function of x).endogenous sampling:sampling is based on response variable(sis a deterministic function of y).IPW applies generally to any estimation problem that involvesminimization or maximization.IPW M-estimator can be used in linear and nonlinear models(such as probit or tobit models).Use bootstrap to make accurate inference.
收起