MID-TERM ELECTION FORECAST: A DEMOCRATIC CONGRESS

            By Michael S. Lewis-Beck, University of Iowa

and

Charles Tien, Hunter College 

          Upcoming mid-term elections will change the balance of power in Washington, giving Democrats control of the House and Senate.  According to the political economy models of Lewis-Beck and Tien, the House will record a net Democratic gain of eight seats, returning Democrats to the majority there. In the Senate, the Democrats will gain three seats, thereby making the Democrat majority less narrow.

 

          The models, which are based on data gathered on elections across the post-World War II period, take into account two important substantive factors:  the popularity of the president and the increase in disposable income.  Summer estimates of both are somewhat favorable for the party in the White House.  (In June, President Bush’s popularity in the Gallup poll equaled 70 percent, which is the second highest June midterm election rating since 1950;  disposable income growth over the first two quarters of 2002 was 2.21 percent, which is the largest midterm growth in disposable income since 1986).  

 

          However, these positive numbers will be overwhelmed by the historic pattern of mid-term losses for the president’s party. The party controlling the White House has lost an average of 24 House seats and 4 Senate seats during midterm elections since 1950. The equation for the House forecast is:

 

HC = -35.31 + 0.61*P + 6.32*E - 28.86*M

          (-2.46)     (2.25)     (3.50)    (-4.20)

 

HC = seat change in House of Representatives for president's party.

P = Gallup presidential popularity rating in June of election year (70 for 2002).

E = change in real disposable income during the first six months of the election year (2.21 for 2002).

M = midterm (0 = pres. election year, 1 = midterm election year).

Figures in parentheses are t-scores.

R-square = .65

Adj. R2 = .60, SEE = 17.60, n = 27

* = significant at .05 two tails

 

          With respect to the Senate, these relatively positive numbers will help the Republicans, but not enough to overcome the losses they will sustain from having an unusually large number (20) of Republican seats being contested. 

 

The Senate forecast model is as follows:

SC = 4.20 + 0.11*P + 0.99*E - 1.96**M - 0.75*X

         (1.38)    (2.48)    (3.33)       (-1.76)      (-5.79)

 

SC = seat change in Senate for president's party.

P = Gallup presidential popularity rating in June of election year (70 for 2002).

E = change in real disposable income during the first six months of the election year (2.21 for 2002).

M = midterm (0 = presidential election year, 1 = midterm election year).

X = Seats exposed for the party of the president, or the number of seats the president's party has up for reelection (20 for 2002).

Figures in parentheses are t-scores.

R-square = .71

Adj. R-sq = .65, SEE = 2.84, n=27

* = significant at .05 two tails

** = significant at .10 two tails.