Comments on Group IV, Was Gore Hurt by the Clinton Legacy?" (PowerPoint presentation)

advertisement
Was Gore hurt by the Clinton
legacy in 2000?
Group IV
Ambreen Amjad
Jessica Brodkin
Clay Martin
Kevin Nazemi
Basic Logic
• Counterfacual analysis
– Develop a baseline model of Gore’s performance
• Decide unit of analysis
– individual
– state
• Find measure of Gore’s performance
– votes
– affect
What the group did
• Unit of analysis: individual
– American national election study provides
buckets o’ data
• Model combines affect and votes
Clinton’s
Effect on
Economy
Clinton’s
Effect on
Moral
Climate
Party
Affiliation
Clinton’s
Thermometer
2000
Gore’s
Thermometer
2000
Gore’s
Votes
2000
Feelings about Clinton
Prof. of voting for Gore
Feelings about Gore
Logic of counterfacual analysis
Feelings about Gore
Bivariate relationships
Goretherm
Fitted values
Gore Thermometer
100
Vote, 0=Bush 1=Gore
Fitted values
1
0
100
Clinton Thermometer
Vote in Election
0
0
0
100
Gore Thermometer
2
3
4
Clinton’s
Effect on
Economy
Clinton’s
Effect on
Moral
Climate
Path analysis
10.6
9.4
(3.8)(.623)
1
Clinton’s
Thermometer
2000
.013
Gore’s
Thermometer
2000
Gore’s
Votes
2000
4.91
Party
Affiliation
Table 2 results
PATH
Change in
Gore’s Votes
(0 to 1 scale)
1
0.031
2
0.323
3
0.287
4
0.150
Comments
My model:
Bush’s
Thermometer
2000
Clinton’s
Effect on
Economy
Clinton’s
Effect on
Moral
Climate
Party
Affiliation
Clinton’s
Thermometer
2000
Gore’s
Thermometer
2000
Gore’s
Votes
2000
The variables
Variable name
Coding
Mean
s.d.
Clinton’s handling of the
economy
1=strongly approve
2.14
1.44
3.58
1.00
2.73
2.06
5=strongly disapprove
Clinton’s impact on moral
climate
1=much better
5=much worse
Party affiliation
0=strong Democrat
6=strong Republican
Clinton thermometer
0-100
55.48
29.69
Gore thermometer
0-100
57.57
25.65
Bush thermometer
0-100
56.14
24.86
Vote for Gore
0=Bush
0.52
1=Gore
The regressions
Clinton
therm.
Bush
therm.
Gore therm.
Gore vote
Clinton’s handling
of the economy
-7.38
(0.54)
—
—
—
Clinton’s impact on
moral climate
-6.67
(0.79)
—
—
—
Party affiliation
-4.83
(0.39)
6.69
(0.24)
-3.10
(0.26)
—
Clinton thermometer
—
—
0.49
(0.02)
—
Gore thermometer
—
—
—
0.0085
(0.0004)
Bush thermometer
—
—
—
-0.0084
(0.0043)
109.43
(2.59)
37.85
(0.83)
39.24
(1.59)
0.52
(0.04)
N
794
1,736
1,745
1,105
R2
.54
.31
.56
.59
Intercept
The path diagram
Party
Affiliation
Bush’s
Thermometer
2000
6.69
-3.10
.41
-4.83
Clinton’s
Effect on
Moral
Climate
.43
-0.0084
Clinton’s
Thermometer
2000
-6.67
.36
Gore’s
Votes
2000
0.0085
-7.38
Clinton’s
Effect on
Economy
0.49
Gore’s
Thermometer
2000
Two counterfactuals
• What if Clinton’s thermometer rating in 2000 = his
thermometer rating in 1996?
– 1996 = 59.3 / 2000 = 55.5
• Gore therm rises by .49 x (59.3-55.5) = 1.86 pts.
• Gore vote rises by 1.86 x 0.0085 = 1.58%
• Effect cut in half by including Bush therm.
• What if effect on moral climate had improved by
½ point?
– Clinton term rises by 6.67 x ½ = 3.34 pts.
– Gore therm rises by 3.34 x .49 = 1.64 pts.
– Gore vote rises by 1.64 x .0085 = 1.4%
Throw everything into one
regression
Clinton
therm.
Clinton’s handling
of the economy
-0.018
(0.010)
Clinton’s impact on
moral climate
-0.061
(0.015)
Party affiliation
-0.090
(0.008)
Clinton thermometer
0.0011
(0.007)
Gore thermometer
0.0036
(0.0008)
Bush thermometer
-0.0042
(0.0006)
Intercept
1.01
(0.10)
N
488
R2
.71
A word about linear probability
vs. logit
gorevote
Pr(gorevote)
Fitted values
gorevote
1.06954
-.201828
0
100
V000360
Comparison of functional forms
• Linear probability
– gorevote = a + b*goretherm
• Logit
– gorevote = (1+e a + b*goretherm)-1
Download