What is so special about special elections?

Some political scientists have regarded special elections as referenda on t he approval of presidents-and therefore as products of national forces-whil e explaining regularly scheduled elections as the product of not only natio nal political forces, but also constituency and candidate attributes specif ic to particular districts. In this paper we examine whether outcomes in sp ecial elections and their nearest counterpart, open-seat elections, are dri ven by similar or different forces. We used district-level data on U.S. Hou se special elections and open-seat elections from 1973 to 1997 to test a mo del that integrates constituency, candidate, and presidential approval vari ables. The results of this analysis indicate that special elections are a s ubset of open-seat elections, with both types of contests strongly impacted by candidate and constituency influences. We found no evidence of a substa ntial presidential-approval effect in special elections. The absence of suc h a relationship underscores the importance of candidates and constituent p references in structuring elections and indicates the inappropriateness of drawing national implications from special House contests.