17
Table 1. The Economy, Political Sophistication, and Presidential Choice
in Five American Elections (Logit Models)
Independent
Variables
1988 1992 1996 2000 2004
Sophistication -0.76*
-1.52**
0.38
0.49
-1.21*
(0.43) (0.43) (0.46) (0.42) (0.59)
Sociotropic (when Sophistication=0)
0.77*
1.83**
2.46**
0.94**
1.90**
(0.42) (0.42) (0.62) (0.31) (0.52)
Pocketbook (when Sophistication=0)
-0.04
-0.12
-0.83
-.01
0.56
(0.32) (0.32) (0.52) (0.34) (0.40)
Sophistication × Sociotropic
-0.49
-0.98
-0.77
-0.41
-0.50
(0.80) (0.73) (0.90) (0.65) (0.95)
Sophistication × Pocketbook
0.52
0.97
1.57*
0.25
-0.11
(0.64) (0.60) (0.80) (0.77) (0.79)
Conditional Effect:
Sociotropic (when Sophistication=1)
0.28
0.85*
1.69**
0.52
1.39**
(0.47) (0.42) (0.44) (0.45) (0.57)
Pocketbook (when Sophistication=1)
0.48
0.84*
0.73*
0.25
0.45
(0.41) (0.37) (0.38) (0.56) (0.52)
Constant and 11 Control Variables
a
[not shown]
Pseudo-R²
0.50 0.51 0.59 0.50 0.64
Log
Likelihood
-431.04 -452.91 -318.44 -337.76 -209.51
N 1246
1407
1143
973
847
* p < .05; ** p < .01 (one-tailed test)
Entries are logit regression coefficients with standard errors in parentheses below.
a
Control variables included but not shown are the same as in Gomez and Wilson (2001) (see text; full results
available on request).
Convention