RETROSPECTIVE VOTING AND POLITICAL MOBILITY

Research on political mobility has ignored the impact that the past po licy decisions of officials may have on the probability of achieving a higher elected office. A large theoretical and empirical literature s uggests that voters and party members will evaluate incumbents based o n past behavior, that is, retrospectively. We test the extent nonincum bent candidates anticipate that voters in primaries and general electi ons will choose retrospectively, utilizing behavior in one elected off ice as a basis for voting for a candidate for a ''higher'' elected off ice. We explain U.S. House member decisions to run for the U.S. Senate in 1960-88 and the probability that a representative will be selected as the party's candidate. We find that those representatives whose AD A scores are closet to the estimated successful state ADA score for th eir party are most likely to seek higher offices and to be selected as their party's standard-bearers. Our evidence further suggests that a divergent platform specification fits the data better than a specifica tion in which party positions are predicted to converge to the platfor m desired by the median voter. We replicate many of the important find ings in this literature and also produce some evidence that mobility i nto the Senate is less common for those who have some power in the Hou se and is more common for those who are wealthier.