Publisher Summary This chapter discusses the estimation of sensitivity analysis. Sensitivity analyses can be carried out in several ways. One is to re-estimate the parameters of interest using different assumptions than those employed in the original analysis. When the additional over-identifying information required to re-estimate models is not available, it will almost always be possible to focus on a range of plausible values for a parameter assumed in the original model to have a given value. When this can be done, it is possible to re-estimate the model by varying the assumed value of this critical parameter within the range considered plausible. In analyzing a given set of panel data, the researcher may discover on the basis of cross-correlogram analysis that the causal lag appropriate for the research problem can be constrained only within a relatively wide range. If the causal model estimated embodies all the correct causal relationships but is estimated with an incorrect causal lag, parameter estimates can be biased. If the true lag lies within a range that is longer than the interval between observations but shorter than the total time interval covered in the multiwave panel, no special difficulty arises.