This article modifies an earlier study by Bond and Fleisher that employed regression analysis to predict House members' support of Bush in 1989. The sample frame used in that investigation (1959-1974) to produce estimates is found to be inappropriate, and consequently, an alternative sample is offered. Drawing support from the larger literature on Congress that documents significant institutional change since the reforms of the 1960s and 1970s, as well as a variety of empirical data, a case is made for substituting an updated sample, one that better reflects the conditions that affect presidential support in the late 1980s. The differences in model parameters and predicted scores that result from this substitution are striking and clearly challenge previous findings. More generally, these results illustrate House dynamics and suggest that an accurate assessment of presidential success hinges critically on proper specification of the congressional context.
