a two-stage regression analysis used to estimate parameters and standard errors in an unbiased way when scores for one aspect of the sample may be affected by a nonrandom selection process. An example of data that could be appropriate for this method is women’s salaries: Some aspects of the data could be modeled using standard regression methods, but the preponderance of zero salaries (reflecting a decision by some women not to work) would also need to be incorporated into the analysis. See also truncated distribution.