A well-known political science article by Edward Tufte argued that the popularity of the president and the state of the economy can largely explain aggregate midterm election outcomes. Midterm elections are often viewed as referendums on the president; typically, the president's party loses seats. However, when a president is relatively popular, and/or the economy is in good shape, the loss tends to be less substantial. Thus, it's hard to draw a firm conclusion from political conditions heading into the 2002 midterm elections: George W. Bush remains a popular president, but the economy is not rated positively.
Bush's most recent approval rating, from an Oct. 14-17 Gallup Poll*, is 62%. This is one of the highest approval ratings posted by any president going into a midterm election since 1938. Only three presidents have had higher ratings -- Franklin Roosevelt's was 70% in 1942, Bill Clinton's was 66% in 1998, and Ronald Reagan had a 63% approval rating in 1986.
A review of the data shows that presidents with Bush's level of approval tend to lose fewer seats in the House of Representatives than less popular presidents. Roosevelt's Democratic Party lost 45 seats in 1942, which is high by current standards but much lower than the 71-seat loss for the Democrats in 1938, when Roosevelt's approval rating was 52%.
Among other presidents with approval ratings above 60%, only one suffered a midterm election-seat loss in the double digits (18 for Dwight Eisenhower and the Republicans in 1954). The last three presidents with 60% or higher approval ratings saw their parties lose no more than five seats in the House, and in 1998, the Democrats gained five seats.
The Republicans are somewhat more vulnerable on the economy. Currently, 26% of Americans rate the economy as excellent or good**, which is low compared to average economic ratings since late 1993. By comparison, 66% rated the economy as excellent or good in 1998. In 1994, a year when the incumbent Democrats lost 54 seats, 30% of Americans gave the economy an excellent or good rating. (Clinton's approval rating was 46% at the time.)
Gallup data on Americans' satisfaction with the way things are going in the United States stretches back to 1979. These ratings usually correlate highly to ratings of the economy (see the comparisons for 1994 and 1998 below). So it isn't surprising that the number of seats lost in a midterm election is related to the percentage of Americans who are satisfied with the state of the nation.
For example, in 1982 in the midst of an economic recession, only 24% of Americans said they were satisfied with the way things were going, and the Republicans lost 26 House seats. In contrast, the Republicans lost just five seats in 1986 when 58% of Americans were satisfied with the condition of the nation.
Satisfaction ratings, economic ratings and presidential approval generally move in the same direction, but this year has been an exception. Despite the low ratings of the economy, satisfaction is at 47% and presidential job approval is at 62%. This election closely resembles the 1990 election in this respect. At that time, only 32% were satisfied with the way things were going in the country, but then-President George Bush still had majority job approval. The two ratings may have counterbalanced each other and minimized the change in House seats. And while the current poor economic situation works in the Democratic Party's favor, Bush's high approval ratings are a major plus for the Republicans.
The 2002 elections present an interesting case for the referendum model of midterm elections. The fact that a popular president is presiding over a sluggish economy may negate the political advantage of either party. The elections could closely resemble the 1990 elections, when many Americans were dissatisfied with the state of the country but most still approved of then-President Bush's handling of his job. It is true that national conditions may have less impact on congressional elections now than they did 40 or 50 years ago, as a variety of factors have conspired to make many House districts safe for incumbents and/or one of the two major parties. (The average seat loss from 1938 to 1966 was 39 seats; from 1970 to 1998 it was just 19.) The current election is also similar to 1990 in that some political observers expect few competitive elections among the 435 House races.
*Results are based on telephone interviews with 1,002 national adults, aged 18 and older, conducted Oct. 14-17, 2002. For results based on the total sample of national adults, one can say with 95% confidence that the maximum margin of sampling error is ±3%.
**Results are based on telephone interviews with 1,502 national adults, aged 18 and older, conducted Oct. 3-6, 2002. For results based on the total sample of national adults, one can say with 95% confidence that the maximum margin of sampling error is ±3%.