District #.  Name of Incumbent (Party), Year Elected Prediction for 2000
  '98 Results '96 Results '94 Results '92 Results direction of partisan change District Partisanship

Categories of Prediction

"Landslide" A "landslide" victory is one by a margin of at least 20% (such as 60% to 40%) by a candidate of the party currently holding the seat
"Comfortable" A "comfortable" victory is one by at least a margin of 10% by a candidate of the party currently holding the seat
"Win" A "win" is a prediction that the candidate will win, but that the race may be decided by less than 10%
"Vulnerable" "Vulnerable" means that the candidate of the party currently holding the seat may be defeated
"Vulnerable" An italicized "vulnerable " indicates that the candidate of the party currently holding the seat is particularly vulnerable to defeat


* An asterisk indicates that if the incumbent were not running, the other major party would have a chance to win
   An indicator of a relatively significant change in district partisanship -- of at least 3%. If the arrow faces up, the direction of change is toward the party currently holding the advantage in district partisanship not the party of the incumbent). If the arrow faces down, the direction of change is against the party currently holding the advantage.
R D I "R" for Republican, "D" for Democrat, "I" is for Independent
U L C "U" indicates that the race was uncontested by the other major party; "L" indicates a "landslide" win despite a sub-60% percentage, while "C" indicates a "comfortable" win despite a sub-55% percentage
[Party Initial] [Number] The party initial ("D" or "R") and the two-digit number on the far right of a listing indicates the district partisanship -- meaning a measure of the partisan voting stance of voters in the district. "R 56" would mean that, all things being equal, a Republican candidate would likely win an open seat in that district by a percentage of 56% to 44%.
'98 '96 '94 '92 Percent of the votes the representative received in those elections or, if elected during that period, the preceding representatives of that district
[year] Listed after incumbent representatives, it indicates when they first were elected in their current district


How We Make Predictions 

The Center’s model for making predictions in congressional races has proven very successful in 1996 and 1998. Although we factor in past incumbent performance, our most important indicator is what we call “district partisanship” We have found a very reliable guide to district partisanship in U.S. House races by determining the difference between the presidential election results in the district and the presidential election results nationally. As Democrat Bill Clinton was winning 49% of the national vote in the presidential race in 1996, Republican and Democratic U.S. House candidates were evenly splitting the popular vote. The average district in the country was one in which Bill Clinton won 49% and the Democratic and Republican candidates each took half the vote. If Clinton were to receive more than 49% in a district, then, the average Democrat would win more than 49%. If Clinton received less than 49%, the average Democrat would win less than 50%.

In 1996, for example, Clinton won 44% of the vote in Washington’s congressional district 5, ahead of Bob Dole’s 43%. Thus, even though Clinton defeated Bob Dole in the district, he did less well than he did on average – 5% less, to be exact. Our district projection would be “Republican 55%” (or R 55 in our list of predictions). Indeed Nethercutt has averaged just under 55% in his three wins in the district in 1994-1998.

Note that the presidential election data is not a measure of Bill Clinton's coattails. Rather, it is simply a reliable means to take a snapshot of the partisan nature of the district. Using 43% as the Clinton benchmark in 1992, it is remarkable how consistent most districts were around the country between 1992 and 1996. Certain districts shifted, but Clinton’s share of the vote rose by about 6% in the majority of districts. Between 1992 and 1996, no district in Ohio fluctuated by more than 3% in our measure of district partisanship, for example.

We suspect our measure of district partisanship would need to be adjusted if the popular vote split between Democrats and Republicans was not so closely divided. We also suspect it would be less reliable if voters were not factoring in partisan control of the House in their vote. Some of the biggest changes in the 1990s – many southern voters shifting to Republicans and many northeast voters shifting to Democrats – clearly are a reflection of voter attitudes toward the congressional leadership of the respective parties rather than the quality of local candidates. But as it is, we think we have arrived a formula that works very well. It has been a reliable gauge in open seat races in 1996 and 1998, for example, while in 1998, Republican and Democratic incumbents both ran on average about 7% ahead of their projection – meaning that we cut the difference between the parties well, without leaning toward one or the other. (And while 7% is significant, it pales next to the average incumbent winning percentage of some 69%.)

back to Summary page 

top of page

Copyright © 2000 The Center for Voting and Democracy
6930 Carroll Ave. Suite 901    Takoma Park, MD  20912
(301) 270-4616 ____ [email protected]