Online Appendix for “Hail to the Pork?: The Influence of
Federal Spending on Presidential Elections”
List of Tables
1
2
3
4
5
6
The electoral consequences of the change in federal grant spending per capita. These
six models replicate the results from Tables 1-4 in the manuscript using the change in
per capita grant spending in a county as the independent variable of interest, instead
of the percentage change in grant spending in the county. The results are substantively identical to those presented in the manuscript. Increased per capita grant
spending in a county boosts the incumbent party’s prospects in the next presidential
election, particularly in counties from competitive states and in counties that are
represented in Congress by members of the president’s party. Also consistent with
theory, the relationship is significantly stronger in liberal and moderate counties than
in conservative counties. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The electoral consequences of percent change in grant spending, excluding state
capital counties. Block grants which are given to the state governments and then
distributed to other parts of the state are assigned to the state capital county. To
insure that this is not skewing our results, these models replicate the results from
Tables 1-4 excluding state capital counties. All results are virtually identical to those
presented in the manuscript. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Influence of the percent change in grant spending on the incumbent administration
party’s vote share. Instead of analyzing the effect of grant spending on the change in
the incumbent administration party’s vote share, these models replicate the results
for Tables 1-4 by using the party’s actual vote share as the dependent variable and
by including it’s vote share in the preceding election as an independent variable. The
results are virtually identical to those presented in the manuscript across specifications.
The electoral consequences of percent change in grant spending, without county
fixed effects. These models employ identical specifications to those used to produce
Tables 1-4, except that they do not include county fixed effects. All of the results
are virtually identical to those presented in the manuscript. . . . . . . . . . . . . . .
The electoral consequences of percent change in grant spending, with state fixed
effects. These models employ identical specifications to those used to produce Tables
1-4, except that they employ state fixed effects instead of county fixed effects. All of
the results are virtually identical to those presented in the manuscript. . . . . . . . .
How partisan accountability mediates the influence of federal grant spending. This
model replicates Table 3, but includes data only from those counties that are located
completely within a single congressional district. The results are virtually identical
to those presented in the manuscript. . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
3
4
5
6
7
8
7
8
9
10
Base results controlling for county unemployment rate. This model replicates the
full specification in Table 1, but also includes the election year unemployment rate
in each county (data available from 1990 to 2008). The results are virtually identical
to those presented in the manuscript. . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
The electoral consequences of increases and decreases in federal grant spending.
These models replicate the results from Table 1 in the manuscript, but disaggregate
the percentage change in grants measure into increases and decreases. Voters reward
the incumbent president for increases in grant spending and punish him for decreases. 10
Individual-level analysis of the influence of district grant spending on vote choice,
excluding state fixed effects. These models replicate those presented in Table 5, but
exclude state fixed effects. All results are virtually identical to those presented in
the manuscript. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Individual-level analysis of the influence of district grant spending on vote choice,
excluding likely voter weights. These models replicate those presented in Table 5, but
they do not employ Gallup’s likely voter weights. All results are virtually identical
to those presented in the manuscript. . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2
3
−0.305∗∗∗
(0.078)
17, 799
0.500
−6.648∗∗∗
(0.126)
0.196∗∗∗
(0.032)
0.067∗∗∗
(0.012)
0.212∗∗∗
(0.019)
0.013∗∗∗
(0.003)
−0.483∗∗∗
(0.125)
−0.263∗∗∗
(0.074)
−0.320
(1.232)
0.057
(0.108)
0.341∗∗
(0.140)
0.792∗∗∗
(0.168)
All
17, 799
0.509
0.695∗∗∗
(0.116)
0.923∗∗∗
(0.118)
1.778∗∗∗
(0.136)
−7.313∗∗∗
(0.146)
0.075
(0.268)
0.058
(0.190)
0.419∗∗
(0.186)
1.292∗∗∗
(0.208)
0.186∗∗∗
(0.032)
0.065∗∗∗
(0.012)
0.220∗∗∗
(0.018)
0.012∗∗∗
(0.003)
−0.443∗∗∗
(0.124)
−0.244∗∗∗
(0.072)
−0.022
(1.219)
All
4, 216
0.542
−1.915∗∗∗
(0.337)
0.524∗∗∗
(0.112)
0.063∗
(0.032)
0.315∗∗∗
(0.047)
0.002
(0.006)
−0.690∗∗∗
(0.233)
−0.278∗∗
(0.110)
1.703
(3.754)
0.444
(0.271)
Lib
5, 808
0.613
−2.730∗∗∗
(0.271)
0.270∗∗∗
(0.066)
0.039
(0.025)
0.202∗∗∗
(0.032)
0.021∗∗∗
(0.005)
−0.692∗∗∗
(0.210)
−0.344∗∗∗
(0.108)
−1.897
(2.589)
0.495∗∗
(0.247)
Mod
7, 765
0.545
−3.516∗∗∗
(0.187)
0.099∗∗
(0.045)
0.071∗∗∗
(0.017)
0.214∗∗∗
(0.032)
0.013∗∗∗
(0.004)
−0.323∗∗∗
(0.119)
−0.370∗∗∗
(0.083)
1.699
(1.724)
0.310∗∗
(0.148)
Con
Table 1: The electoral consequences of the change in federal grant spending per capita. These six models replicate the results from
Tables 1-4 in the manuscript using the change in per capita grant spending in a county as the independent variable of interest, instead
of the percentage change in grant spending in the county. The results are substantively identical to those presented in the manuscript.
Increased per capita grant spending in a county boosts the incumbent party’s prospects in the next presidential election, particularly in
counties from competitive states and in counties that are represented in Congress by members of the president’s party. Also consistent
with theory, the relationship is significantly stronger in liberal and moderate counties than in conservative counties.
17, 799
0.500
−6.647∗∗∗
(0.122)
0.196∗∗∗
(0.032)
0.066∗∗∗
(0.012)
0.214∗∗∗
(0.019)
0.013∗∗∗
(0.003)
−0.485∗∗∗
(0.125)
−0.264∗∗∗
(0.074)
−0.291
(1.231)
0.225∗∗∗
(0.031)
−0.112
(0.075)
0.493∗∗∗
(0.112)
0.552∗∗∗
(0.111)
0.620∗∗∗
(0.112)
Observations
17, 976
17, 976
R-squared
0.479
0.481
Robust standard errors clustered on county in parentheses
*** p<0.01, ** p<0.05, * p<0.10
Constant
Three co-partisans in Congress
Two co-partisans in Congress
One co-partisan in Congress
Competitive state – within 5%
% change in county population
Iraq casualties in county 2008
Iraq casualties in county 2004
Change in pres party House vote
Campaign appearance difference
Change in per capita grants
(in 1,000s)
Change in per capita grants ×
non-competitive state
Change in per capita grants ×
competitive state
Change in per capita grants ×
0 co-partisans
Change in per capita grants ×
1 co-partisan
Change in per capita grants ×
2 co-partisans
Change in per capita grants ×
3 co-partisans
Change in per capita income
(in 1,000s)
Television ad difference
All
All
All
4
−0.112
(0.075)
−0.321∗∗∗
(0.078)
0.191∗∗∗
(0.028)
0.771∗∗∗
(0.119)
0.855∗∗∗
(0.117)
17, 665
0.501
−2.691∗∗∗
(0.143)
0.166∗∗∗
(0.029)
0.072∗∗∗
(0.012)
0.209∗∗∗
(0.019)
0.012∗∗∗
(0.003)
−0.457∗∗∗
(0.125)
−0.261∗∗∗
(0.077)
−0.137
(1.215)
0.707∗∗∗
(0.119)
All
17, 665
0.501
−2.700∗∗∗
(0.150)
0.165∗∗∗
(0.029)
0.072∗∗∗
(0.012)
0.207∗∗∗
(0.019)
0.013∗∗∗
(0.003)
−0.454∗∗∗
(0.124)
−0.258∗∗∗
(0.077)
−0.211
(1.215)
−0.000
(0.110)
0.489∗∗∗
(0.146)
1.139∗∗∗
(0.187)
All
17, 665
0.510
0.721∗∗∗
(0.121)
0.950∗∗∗
(0.122)
1.783∗∗∗
(0.139)
−3.742∗∗∗
(0.172)
0.396
(0.278)
0.312
(0.212)
0.708∗∗∗
(0.202)
1.475∗∗∗
(0.239)
0.155∗∗∗
(0.029)
0.070∗∗∗
(0.012)
0.214∗∗∗
(0.018)
0.012∗∗∗
(0.003)
−0.408∗∗∗
(0.123)
−0.241∗∗∗
(0.074)
0.102
(1.208)
All
4, 177
0.540
−2.011∗∗∗
(0.320)
0.455∗∗∗
(0.108)
0.075∗∗
(0.031)
0.309∗∗∗
(0.047)
0.002
(0.006)
−0.644∗∗∗
(0.210)
−0.258∗∗∗
(0.096)
2.999
(3.693)
0.990∗∗∗
(0.355)
Lib
5, 725
0.616
−2.694∗∗∗
(0.263)
0.254∗∗∗
(0.068)
0.043∗
(0.025)
0.203∗∗∗
(0.032)
0.022∗∗∗
(0.005)
−0.609∗∗∗
(0.191)
−0.318∗∗∗
(0.098)
−1.297
(2.576)
0.814∗∗∗
(0.267)
Mod
7, 755
0.546
−3.379∗∗∗
(0.185)
0.085∗∗
(0.036)
0.077∗∗∗
(0.017)
0.201∗∗∗
(0.032)
0.012∗∗∗
(0.004)
−0.388∗∗∗
(0.132)
−0.389∗∗∗
(0.090)
1.381
(1.700)
0.285∗
(0.155)
Con
Table 2: The electoral consequences of percent change in grant spending, excluding state capital counties. Block grants which are given
to the state governments and then distributed to other parts of the state are assigned to the state capital county. To insure that this
is not skewing our results, these models replicate the results from Tables 1-4 excluding state capital counties. All results are virtually
identical to those presented in the manuscript.
Observations
18, 170
17, 843
R-squared
0.479
0.483
Robust standard clustered on county errors in parentheses
*** p<0.01, ** p<0.05, * p<0.10
Constant
Three co-partisans in Congress
Two co-partisans in Congress
One co-partisan in Congress
Competitive state – within 5%
% change in county population
Iraq casualties in county 2008
Iraq casualties in county 2004
Change in pres party House vote
Campaign appearance difference
% change in grants ×
non-competitive state
% change in grants ×
competitive state
% change in grants ×
0 co-partisans
% change in grants ×
1 co-partisan
% change in grants ×
2 co-partisans
% change in grants ×
3 co-partisans
Change in per capita income
(in 1,000s)
Television ad difference
% change in grants
All
All
5
0.984∗∗∗
(0.003)
−2.315∗∗∗
(0.247)
0.982∗∗∗
(0.003)
−2.176∗∗∗
(0.245)
0.201∗∗∗
(0.028)
0.754∗∗∗
(0.119)
0.843∗∗∗
(0.117)
17, 959
0.909
0.980∗∗∗
(0.003)
−1.454∗∗∗
(0.292)
0.173∗∗∗
(0.029)
0.068∗∗∗
(0.012)
0.208∗∗∗
(0.019)
0.012∗∗∗
(0.003)
−0.529∗∗∗
(0.134)
−0.283∗∗∗
(0.079)
−0.311
(1.211)
0.687∗∗∗
(0.119)
All
17, 959
0.909
0.979∗∗∗
(0.003)
0.884∗∗∗
(0.189)
0.173∗∗∗
(0.028)
0.068∗∗∗
(0.012)
0.207∗∗∗
(0.019)
0.013∗∗∗
(0.003)
−0.528∗∗∗
(0.134)
−0.280∗∗∗
(0.078)
−0.393
(1.211)
−0.049
(0.108)
0.469∗∗∗
(0.146)
1.123∗∗∗
(0.187)
All
17, 959
0.911
0.890∗∗∗
(0.122)
1.242∗∗∗
(0.128)
2.267∗∗∗
(0.150)
0.960∗∗∗
(0.004)
−1.585∗∗∗
(0.297)
0.372
(0.275)
0.245
(0.213)
0.678∗∗∗
(0.201)
1.418∗∗∗
(0.237)
0.163∗∗∗
(0.029)
0.064∗∗∗
(0.012)
0.214∗∗∗
(0.018)
0.012∗∗∗
(0.003)
−0.509∗∗∗
(0.138)
−0.273∗∗∗
(0.078)
0.099
(1.202)
All
4, 290
0.882
1.060∗∗∗
(0.013)
−12.969∗∗∗
(0.834)
0.460∗∗∗
(0.105)
0.059∗∗
(0.030)
0.319∗∗∗
(0.045)
0.001
(0.006)
−0.626∗∗∗
(0.212)
−0.256∗∗∗
(0.098)
1.420
(3.712)
1.056∗∗∗
(0.348)
Lib
5, 826
0.824
0.983∗∗∗
(0.017)
−1.680
(1.057)
0.259∗∗∗
(0.068)
0.044∗
(0.025)
0.201∗∗∗
(0.032)
0.022∗∗∗
(0.005)
−0.713∗∗∗
(0.220)
−0.353∗∗∗
(0.111)
−1.919
(2.585)
0.805∗∗∗
(0.268)
Mod
7, 834
0.948
0.949∗∗∗
(0.008)
1.843∗∗∗
(0.319)
0.096∗∗∗
(0.035)
0.081∗∗∗
(0.017)
0.194∗∗∗
(0.032)
0.014∗∗∗
(0.004)
−0.440∗∗∗
(0.134)
−0.429∗∗∗
(0.092)
1.170
(1.701)
0.299∗
(0.155)
Con
Table 3: Influence of the percent change in grant spending on the incumbent administration party’s vote share. Instead of analyzing the
effect of grant spending on the change in the incumbent administration party’s vote share, these models replicate the results for Tables
1-4 by using the party’s actual vote share as the dependent variable and by including it’s vote share in the preceding election as an
independent variable. The results are virtually identical to those presented in the manuscript across specifications.
Observations
18, 464
18, 137
R-squared
0.905
0.905
Robust standard errors clustered on county in parentheses
*** p<0.01, ** p<0.05, * p<0.10
Constant
Lagged incumbent vote share
Three co-partisans in Congress
Two co-partisans in Congress
One co-partisan in Congress
Competitive state – within 5%
% change in county population
Iraq casualties in county 2008
Iraq casualties in county 2004
Change in pres party House vote
Campaign appearance difference
% change in grants ×
non-competitive state
% change in grants ×
competitive state
% change in grants ×
0 co-partisans
% change in grants ×
1 co-partisan
% change in grants ×
2 co-partisans
% change in grants ×
3 co-partisans
Change in per capita income
(in 1,000s)
Television ad difference
% change in grants
All
All
6
−0.084
(0.074)
−3.202∗∗∗
(0.092)
0.152∗∗∗
(0.022)
0.633∗∗∗
(0.098)
0.690∗∗∗
(0.097)
17, 959
0.419
−2.886∗∗∗
(0.113)
0.136∗∗∗
(0.023)
0.058∗∗∗
(0.010)
0.165∗∗∗
(0.013)
0.012∗∗∗
(0.002)
−0.157∗∗∗
(0.058)
−0.167∗∗∗
(0.045)
2.234∗∗∗
(0.523)
0.609∗∗∗
(0.099)
All
17, 959
0.420
−2.949∗∗∗
(0.116)
0.134∗∗∗
(0.023)
0.061∗∗∗
(0.011)
0.162∗∗∗
(0.014)
0.012∗∗∗
(0.002)
−0.153∗∗∗
(0.057)
−0.165∗∗∗
(0.045)
2.103∗∗∗
(0.524)
0.155∗∗
(0.073)
0.395∗∗∗
(0.117)
1.036∗∗∗
(0.169)
All
17, 959
0.427
0.516∗∗∗
(0.090)
0.877∗∗∗
(0.089)
1.347∗∗∗
(0.113)
−3.741∗∗∗
(0.136)
0.220
(0.224)
0.118
(0.180)
0.681∗∗∗
(0.168)
1.505∗∗∗
(0.204)
0.130∗∗∗
(0.023)
0.059∗∗∗
(0.011)
0.167∗∗∗
(0.013)
0.011∗∗∗
(0.002)
−0.124∗∗
(0.057)
−0.151∗∗∗
(0.043)
2.340∗∗∗
(0.525)
All
4, 290
0.406
−4.553∗∗∗
(0.187)
0.300∗∗∗
(0.076)
0.053∗∗
(0.024)
0.198∗∗∗
(0.030)
0.001
(0.005)
−0.184∗
(0.111)
−0.129∗∗
(0.058)
0.743
(1.467)
1.138∗∗∗
(0.259)
Lib
5, 826
0.453
−7.342∗∗∗
(0.180)
0.164∗∗∗
(0.051)
0.017
(0.018)
0.171∗∗∗
(0.020)
0.022∗∗∗
(0.004)
−0.142∗
(0.079)
−0.187∗∗∗
(0.044)
1.835∗
(1.070)
0.683∗∗∗
(0.203)
Mod
7, 834
0.431
1.754∗∗∗
(0.099)
0.077∗∗∗
(0.027)
0.068∗∗∗
(0.015)
0.159∗∗∗
(0.022)
0.012∗∗∗
(0.003)
−0.151∗
(0.082)
−0.321∗∗∗
(0.059)
3.173∗∗∗
(0.649)
0.260∗∗
(0.119)
Con
Table 4: The electoral consequences of percent change in grant spending, without county fixed effects. These models employ identical
specifications to those used to produce Tables 1-4, except that they do not include county fixed effects. All of the results are virtually
identical to those presented in the manuscript.
Observations
18, 464
18, 137
R-squared
0.402
0.405
Robust standard errors clustered on county in parentheses
*** p<0.01, ** p<0.05, * p<0.10
Constant
Three co-partisans in Congress
Two co-partisans in Congress
One co-partisan in Congress
Competitive state – within 5%
% change in county population
Iraq casualties in county 2008
Iraq casualties in county 2004
Change in pres party House vote
Campaign appearance difference
% change in grants ×
non-competitive state
% change in grants ×
competitive state
% change in grants ×
0 co-partisans
% change in grants ×
1 co-partisan
% change in grants ×
2 co-partisans
% change in grants ×
3 co-partisans
Change in per capita income
(in 1,000s)
Television ad difference
% change in grants
All
All
7
−3.240∗∗∗
(0.089)
−3.201∗∗∗
(0.089)
0.151∗∗∗
(0.022)
0.636∗∗∗
(0.094)
0.697∗∗∗
(0.093)
17, 959
0.458
−2.792∗∗∗
(0.115)
0.130∗∗∗
(0.022)
0.073∗∗∗
(0.011)
0.203∗∗∗
(0.017)
0.010∗∗∗
(0.002)
−0.209∗∗∗
(0.066)
−0.183∗∗∗
(0.049)
0.505
(0.521)
0.588∗∗∗
(0.094)
All
17, 959
0.458
−2.802∗∗∗
(0.121)
0.130∗∗∗
(0.022)
0.073∗∗∗
(0.011)
0.202∗∗∗
(0.017)
0.010∗∗∗
(0.002)
−0.208∗∗∗
(0.066)
−0.181∗∗∗
(0.049)
0.529
(0.520)
−0.003
(0.098)
0.385∗∗∗
(0.114)
1.004∗∗∗
(0.155)
All
17, 959
0.467
0.618∗∗∗
(0.094)
0.967∗∗∗
(0.100)
1.703∗∗∗
(0.120)
−3.793∗∗∗
(0.139)
0.241
(0.220)
0.185
(0.171)
0.584∗∗∗
(0.158)
1.404∗∗∗
(0.191)
0.122∗∗∗
(0.023)
0.072∗∗∗
(0.011)
0.210∗∗∗
(0.016)
0.009∗∗∗
(0.002)
−0.174∗∗∗
(0.066)
−0.167∗∗∗
(0.047)
0.587
(0.520)
All
4, 290
0.445
−7.520∗∗∗
(0.296)
0.335∗∗∗
(0.076)
0.072∗∗∗
(0.026)
0.272∗∗∗
(0.038)
0.005
(0.005)
−0.245∗
(0.132)
−0.150∗∗
(0.066)
0.019
(1.504)
1.003∗∗∗
(0.254)
Lib
5, 826
0.502
−2.803∗∗∗
(0.189)
0.174∗∗∗
(0.048)
0.040∗∗
(0.019)
0.180∗∗∗
(0.025)
0.019∗∗∗
(0.004)
−0.192∗∗
(0.083)
−0.192∗∗∗
(0.045)
−0.148
(1.095)
0.684∗∗∗
(0.190)
Mod
7, 834
0.473
−3.121∗∗∗
(0.147)
0.064∗∗
(0.026)
0.076∗∗∗
(0.015)
0.194∗∗∗
(0.027)
0.008∗∗∗
(0.003)
−0.200∗∗
(0.086)
−0.346∗∗∗
(0.065)
1.091
(0.695)
0.256∗∗
(0.113)
Con
Table 5: The electoral consequences of percent change in grant spending, with state fixed effects. These models employ identical
specifications to those used to produce Tables 1-4, except that they employ state fixed effects instead of county fixed effects. All of the
results are virtually identical to those presented in the manuscript.
Observations
18, 464
18, 137
R-squared
0.438
0.442
Robust standard errors clustered on county in parentheses
*** p<0.01, ** p<0.05, * p<0.10
Constant
Three co-partisans in Congress
Two co-partisans in Congress
One co-partisan in Congress
Competitive state – within 5%
% change in county population
Iraq casualties in county 2008
Iraq casualties in county 2004
Change in pres party House vote
Campaign appearance difference
% change in grants ×
non-competitive state
% change in grants ×
competitive state
% change in grants ×
0 co-partisans
% change in grants ×
1 co-partisan
% change in grants ×
2 co-partisans
% change in grants ×
3 co-partisans
Change in per capita income
(in 1,000s)
Television ad difference
% change in grants
All
All
% change in grants × 0 co-partisans
% change in grants × 1 co-partisan
% change in grants × 2 co-partisans
% change in grants × 3 co-partisans
Change in per capita income (in 1,000s)
Television ad difference
Campaign appearance difference
Change in pres party House vote
Iraq casualties in county 2004
Iraq casualties in county 2008
% change in county population
One co-partisan in Congress
Two co-partisans in Congress
Three co-partisans in Congress
Constant
0.273
(0.295)
0.267
(0.227)
0.613∗∗∗
(0.215)
1.333∗∗∗
(0.246)
0.105∗∗∗
(0.030)
0.071∗∗∗
(0.014)
0.256∗∗∗
(0.021)
0.008∗∗∗
(0.003)
−0.667∗∗∗
(0.136)
−0.601∗∗∗
(0.087)
1.806
(1.332)
0.564∗∗∗
(0.133)
0.828∗∗∗
(0.134)
1.647∗∗∗
(0.147)
−7.224∗∗∗
(0.162)
Observations
15, 714
R-squared
0.515
Robust standard errors clustered on county in parentheses
*** p<0.01, ** p<0.05, * p<0.10
Table 6: How partisan accountability mediates the influence of federal grant spending. This model
replicates Table 3, but includes data only from those counties that are located completely within a
single congressional district. The results are virtually identical to those presented in the manuscript.
8
% change in grants
Change in per capita income (in 1,000s)
Television ad difference
Campaign appearance difference
Change in pres party House vote
Iraq casualties in county 2004
Iraq casualties in county 2008
% change in county population
County unemployment rate
Constant
0.889∗∗∗
(0.142)
0.031
(0.036)
0.058∗∗∗
(0.012)
0.173∗∗∗
(0.020)
0.012∗∗∗
(0.003)
−0.497∗∗∗
(0.123)
−0.276∗∗∗
(0.077)
−6.915∗∗∗
(1.551)
−0.382∗∗∗
(0.052)
−6.425∗∗∗
(0.253)
Observations
14, 951
R-squared
0.525
Robust standard errors clustered on county in parentheses
*** p<0.01, ** p<0.05, * p<0.10
Table 7: Base results controlling for county unemployment rate. This model replicates the full
specification in Table 1, but also includes the election year unemployment rate in each county
(data available from 1990 to 2008). The results are virtually identical to those presented in the
manuscript.
9
(1)
% increase in grants
% decrease in grants
(2)
0.773∗∗∗
(0.158)
−1.119∗∗∗
(0.312)
0.710∗∗∗
(0.160)
−0.962∗∗∗
(0.315)
0.197∗∗∗
(0.028)
−6.283∗∗∗
(0.115)
−6.408∗∗∗
(0.122)
Change in per capita income (in 1,000s)
Television ad difference
Campaign appearance difference
Change in pres party House vote
Iraq casualties in county 2004
Iraq casualties in county 2008
% change in county population
Constant
Observations
R-squared
18, 464
18, 137
0.477
0.481
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.10
(3)
0.677∗∗∗
(0.161)
−0.800∗∗
(0.313)
0.169∗∗∗
(0.029)
0.070∗∗∗
(0.012)
0.210∗∗∗
(0.018)
0.012∗∗∗
(0.003)
−0.487∗∗∗
(0.125)
−0.269∗∗∗
(0.075)
−0.377
(1.211)
−2.696∗∗∗
(0.158)
17, 959
0.499
Table 8: The electoral consequences of increases and decreases in federal grant spending. These
models replicate the results from Table 1 in the manuscript, but disaggregate the percentage change
in grants measure into increases and decreases. Voters reward the incumbent president for increases
in grant spending and punish him for decreases.
10
(1)
1.868∗∗∗
(0.252)
−0.746∗∗∗
(0.244)
0.617∗∗
(0.279)
Republican
Democrat
% change in grants
% change in grants × Republican
(2)
1.914∗∗∗
(0.255)
−0.735∗∗∗
(0.243)
0.836∗∗
(0.359)
−0.595
(0.478)
% change in grants × liberalism
Liberalism
White
Married
Black
Education
Age
Male
Constant
Observations
−0.511∗∗∗
(0.077)
0.425
(0.275)
−0.013
(0.114)
−1.254∗∗
(0.631)
−0.031
(0.051)
−0.004
(0.004)
−0.213∗
(0.111)
0.788∗
(0.413)
−0.522∗∗∗
(0.072)
0.466∗
(0.258)
−0.012
(0.115)
−1.215∗∗
(0.615)
−0.028
(0.051)
−0.004
(0.004)
−0.204∗
(0.109)
0.718∗
(0.411)
2, 074
2, 074
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.10
(3)
1.876∗∗∗
(0.239)
−0.736∗∗∗
(0.236)
−1.458∗∗
(0.660)
0.701∗∗∗
(0.238)
−0.634∗∗∗
(0.074)
0.564∗∗
(0.230)
−0.019
(0.117)
−1.097∗
(0.615)
−0.033
(0.052)
−0.003
(0.004)
−0.186∗
(0.109)
0.911∗∗
(0.437)
2, 074
Table 9: Individual-level analysis of the influence of district grant spending on vote choice, excluding
state fixed effects. These models replicate those presented in Table 5, but exclude state fixed effects.
All results are virtually identical to those presented in the manuscript.
11
(1)
Republican
Democrat
% change in grants
1.719∗∗∗
(0.170)
−0.808∗∗∗
(0.173)
0.500∗∗∗
(0.193)
% change in grants × Republican
(2)
1.736∗∗∗
(0.173)
−0.809∗∗∗
(0.173)
0.582∗∗
(0.231)
−0.237
(0.328)
% change in grants × liberalism
Liberalism
White
Married
Black
Education
Age
Male
Constant
−0.507∗∗∗
(0.052)
0.583∗∗∗
(0.189)
0.139
(0.085)
−1.239∗∗∗
(0.453)
0.011
(0.040)
0.001
(0.003)
−0.102
(0.083)
−0.299
(0.708)
Observations
2, 402
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.10
−0.508∗∗∗
(0.052)
0.585∗∗∗
(0.188)
0.140∗
(0.085)
−1.220∗∗∗
(0.450)
0.011
(0.040)
0.001
(0.003)
−0.100
(0.082)
−0.316
(0.708)
2, 402
(3)
1.721∗∗∗
(0.169)
−0.803∗∗∗
(0.173)
−0.500
(0.547)
0.348∗
(0.193)
−0.552∗∗∗
(0.056)
0.610∗∗∗
(0.184)
0.138
(0.085)
−1.207∗∗∗
(0.450)
0.012
(0.041)
0.001
(0.003)
−0.096
(0.083)
−0.238
(0.718)
2, 402
Table 10: Individual-level analysis of the influence of district grant spending on vote choice, excluding likely voter weights. These models replicate those presented in Table 5, but they do not
employ Gallup’s likely voter weights. All results are virtually identical to those presented in the
manuscript.
12