Predicting Presidential Elections and Other Things, Second
2012, Stanford University Press.
Amazon order form
The data for duplicating the results in Box 4-3 on page 74 are in the file: cong1.txt. Given the notation in this file, you need to create GI = G*I, PI = P*I, ZI = Z*I, INDEX1 = .672*GI - .654*PI + .990*ZI, VCCL250 = VCCL2 - 50, and CNST = 1. Then have the students regress VC on INDEX1, DPER, I, WAR, CNST, and VCCL250. This is for the period 1916-2008, which is 24 observations. This is the on term congressional vote equation used on the site for the 2012 forecasts.
The data for duplicating the results in Box 4-4 on page 78 are in the file: cong2.txt. Given the notation in this file, you need to create PCCI = PCC*I, ZCCI = ZCC*I, INDEX2 = - .654*PCCI + .990*ZCCI, VPL250 = VPL2 - 50, VCL250 = VCL2 - 50, and CNST = 1. Then have the students regress VCC on INDEX2, I, WAR, CNST, VCL250, and VPL250. This is for the period 1918-2010, which is 24 observations. This is the mid term congressional vote equation.
The data for duplicating the extramarital affairs results in Box 5-2 on page 94 and in Box 5-3 on page 95 are in the file: affairs.txt. There are 602 Psychology Today observations and 6,366 Redbook observations. You need to use Table 1 in the original article, A Theory of Extramarital Affairs, to go from the notation in the dataset to the notation in Chapter 5 in the book. Also, the coefficients in the two boxes are estimated using the Tobit estimator, so you can't duplicate the results by just using OLS. This is thus advanced material for most students.
The data for duplicating the wine results in Box 6-2 on page 105 are in the file: wine.txt. Have the students regress LPRICE on TEMP, RAIN1, RAIN2, AGE, and CNST using the 27 observations.
The data for duplicating the football results in Table 10-1 on page 165 and in Table 10-3 on page 170 are in the file: football.txt. For example, for the regression in Table 10-3 have the students regress SPREAD on LV, H, SAG, BIL, COL, DUN, and REC (no constant term) using the 1,582 observations.