Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
1
Jen Sese Political Science 343 Prof Franzese
Unpacking Voter Participation: The Effect of Issue Salience on Voter Turnout in the US Introduction
Democracies appropriate political influence equally- one person, one vote- but the
individual choice to assert participation tempers this influence. Participation is a crucial
input into a democratic system, influencing and determining policy output. Voter turnout
is a key measure of this political participation and is the core of normative theories on
democracy. Political scientists have long seen that low voter turnout has far reaching
implications. Arend Lijphart, in his presidential address to the American Political
Science Association, sees the danger of widespread abstention because it “leads to
inequality of representation and influence that are not randomly distributed but
systemically biased in favor of more privileged citizens- those with higher incomes,
greater wealth, and better education- against less advantaged citizens (1). For this
reason, the causes and effects of voter turnout are the focus of considerable interest.
This paper argues that voter participation functions in an S curve. Participation
rates among those demographics near the limits of the S curve remain relatively constant
over time. Participation among those whose participation falls in the middle of the S
curve (middle class, youth, moderate education) is the most volatile. It will utilize Hibbs'
partisan economic theory in conjunction with the median voter theorem to argue that
economic policy cleavages foster partisanship and capture the spectrum of the electorate
whose interests and political alignment are clearly defined. Conversely, economic issues
do not capture the median vote- the middle class whose economic needs are vague. . The
2
inability of economic issues to penetrate these demographics with volatile participation
indicates that these groups mobilize around other issues. I hypothesize that participation
in these volatile demographics directly correlates with issue saliency. Using voter survey
data and public opinion polls, the paper will demonstrate that these volatile demographics
mobilize around social issues with greater participation when social salience is high.
This paper will explore the effects of this dynamic interplay between voter
participation and issue saliency. The first part analyzes trends in voter participation over
time to identify the S-curve function. The second part utilizes survey data from 1992-
2004 Presidential elections to track issue salience and voter turnout.
I. Unpacking Changes in Voter Turnout A.
Voter turnout in the electoral process is a key measure of political participation.
Turnout is the percentage of eligible voters who cast a ballot in an election. In the United
States approximately 70% of the eligible population register to vote and approximately
50% vote during presidential elections (Franklin). In the US, where voting is non-
compulsory, turnout in elections is low but it still experiences considerable variance
across elections. Participation and abstention rates are not uniform across the population
of eligible voters but correlate with specific demographic characteristics- age, race,
income, education (Matsuaka and Palda 1999).
A wealth of literature postulates the determinants of voter participation. In each
country, some parts of the electorate are more likely to vote then others. In countries
with high turnout, it is difficult to find significant differences between voters and non-
voters, but in states with low turnout, the differences between those who participate and
those who abstain can be quite vast (Franklin). There are robust findings of a “class
3
bias” in the democratic process. Countless studies using survey data in a range of
countries have found a positive correlation between various measures of economic status
(education, income) and voter participation. These theorists explain the propensity to vote
mainly in terms of voter traits, such as income, occupation, education, and partisanship.
These groups have distinct interests and exert their political influence in different
manners (Franzese 72).
In developed countries, the most important factor in voter turnout is education.
The more educated a person is, the more likely he or she is to vote, even other factors,
such as income and class that are closely associated with education level, are controlled.
Income has some effect independently: wealthier people are more likely to vote,
regardless of their educational background. There is some debate over the effects of
ethnicity, race, and gender. In recent studies, politicians found that when they accounted
for education and income differences, other demographic factors had little influence.
(Sigelman, Roeder, Jewell, & Baer 1985). There is a strong correlation between different
ethnic groups and typical levels of education and income. Other demographic factors
have an important influence: young people are far less likely to vote than the elderly; and
single people are less likely to vote than those who are married. Occupation has little
effect on turnout, with the notable exception of higher voting rates among government
employees in many countries (Sigelman, Roeder, Jewell, & Baer 1985).
B. Theory and Hypotheses
4
Generally, participation directly correlates with demographic features, but an
increase in voter participation on the aggregate level does not signify a uniform increase
in individual voter participation among all demographics. Logically, when participation
increases it will increase more among those voters with a higher initial voter propensity
but this is not the case. This paper argues that voter participation is not a directly linear
function. Instead, voter participation is an S-curve. The middle of the S-curve has the
most potential for exponential growth and decay. This growth/decay slows as the
function reaches its upward/lower bounds. Based on past research, the wealthy,
educated, elderly have a high propensity to vote (Franzese, Sigelman, Roeder, Jewell, &
Baer 1985). Their participation falls closer to the upward bound, which restricts the
volatility of their turnout. Voter participation among those demographics with the lowest
turnout (poor, uneducated, unemployed, young) falls lower on the S-curve. It is more
volatile than turnout among voters with high voting propensity over time because it has
greater room for growth but is relatively stable because it falls closer to the lower bound
of the S-curve. Over time, voting habits of these demographics remain relatively constant
regardless of aggregate turnout. Those voters with mid range voting propensity
experience the most volatility. These voters in the middle of the s-curve have the most
room for growth and decay.
5
Hypothesis 1: Turnout among those voters with highest voting propensities will
remain relatively constant and stable over time.
Hypothesis 2: Turnout among those voters with low voting propensities will be
more volatile than turnout among more participatory demographics but will remain
relatively constant and stable over time.
Hypothesis 3: Turnout among those voters whose voting propensity falls towards
the middle will be the most volatile over time.
Data The United States Census Bureau collects information on reported voting and
registration by various demographic characteristics in November of every congressional
and presidential election year. All data in this section are from the Census Bureau’s
Current Population Survey (CPS).
Results
Figure 1a Figure 1b
Figure 1a and 1b illustrate aggregate voter participation in United States
Presidential elections from 1964-2004. Figure 1a shows overall participation as a
6
percentage percentage of eligible voters. Figure1b tracks changes in participation as the
percentage difference in turnout from the previous year.
Figure 2a Figure 2b
Figure 2a and 2b show the participation rates for unemployed and employed
voters. Figure 2a shows participatory trends in each presidential election year and figure
2b illustrates the change in participation from one presidential election year to the next.
The mean participation among the unemployed is 42.95% and the standard deviation is
5.24% The mean participation among the employed population is 60.144% and the
standard deviation is 3.29%. Participation among the unemployed is more volatile over
time than participation of employed citizens. Participation among employed citizens
closely aligns with aggregate participatory trends. Employment should be a strong
positive predictor of participation.
7
Figure 3a Figure 3b
Figure 3a and 3b show the participation rates for Presidential elections by levels
of educational attainment. Figure 3a shows participation rates in each presidential year
and Figure 3b shows change from one election to the next. Participation among those
with less than a 9th grade education is perpetually lower and relatively constant. Among
the entire population, the mean participation is 39.42% and the standard deviation is
8.64%. The GED/High school diploma population experienced a mean participation of
58.38% and a standard deviation of 7.76% College Education/Associates Degree
experienced a mean of 67.72% and a standard deviation of 5.62%. Those with a
Bachelors degree or more experienced an average participation of 78.17% with a
standard deviation of 4.18%. The mean participation for those with 9-12th grade
education is 43.03% and the standard deviation is 8.92%.
Pre-1984, participation for 9-12th grade education was very volatile but post-1984
participation levels out and begins to closely align with aggregate levels. From 1988
onward participatory trends among those with high school diplomas/ GEDs, college
educations, and bachelors degrees very closely aligns with overall participatory trends
until the 2004 election when turnout increases more dramatically among high
school/GED recipients and college education (no degree)/associates degree relative to
participatory increases among those with their bachelors degree (which remains relatively
constant).
8
Figure 4a Figure 4b
Figures 4a and 4b illustrate voter turnout by age. 4a shows participation across
ages over time and 4b shows the changes in participation over from one election to the
next. Volatility inversely correlates with age. Participation among ages 18-24 is the most
volatile. The mean participation is 41.7% and the standard deviation is 6.58%. Those
25-44 experiences the second most volatile turnout. The mean turnout is 57.96% and the
standard deviation is 6.42%. Voter turnout among those 45-64 is the second most stable
a participation of 69.30% and a standard deviation of 3.72%. Voter turnout among 65+ is
the most stable with mean of 66.63% and a standard deviation of 2.38%.
Figure 5a Figure 5b
Figures 5a and 5b show turnout in Presidential elections based on racial
categories. Figure 5a shows turnout rates in each election and Figure 5b shows changes
9
in turnout from one election to the next. Caucasian turnout almost directly mirrors
changes in overall participation, with a mean turnout rate of 62.08% and a standard
deviation of 4.65%. This is largely a result of a predominantly Caucasian electorate- the
US census reports that in 2005 80.2% of the population was Caucasian. Turnout among
Asian/Pacific Islanders is less volatile than overall participation, but the sample size is
smaller (Census Data was only available from 1992 onward). African American
participation is most volatile from 1964-1988, with large increases and decreases.
Participation levels from 1988-1996, with smaller peaks and valleys. The mean is
53.54% with a standard deviation of 3.17%. The mean is 27.05%% and the standard
deviation is 2.014%. Historical context is likely an important confounding variable
affecting this demographic. Early volatility is likely a direct result of systemic
disenfranchisement, which has lessened with time. Hispanic participation is more volatile
from 1972-1984 and less volatile from 1988-2004 with a mean of 30.188% and a
standard deviation of 3.34%.
Figure 6a Figure 6b
10
Figures 6a and 6b illustrate participation broken into household income ranges for 1996-
2004. The US Census reports the median household income in 2003 was $39,318. The
most volatility occurs in the mid range incomes ($15,000-$49,000). The mean
participation for $15,000-$24,999 is 46.4% and the standard deviation is 3.27%. The
$25,000-$34,999 range is slightly more volatile, with a mean of 53.76% and standard
deviation of 3.95%, making it the most volatile identity. The mean turnout for $35,000-
$49,000 is 60.4% with a standard deviation of 3.5%.
Under $10,000 and $50,000 and above experience the most stability. The mean
participation for >$10,000 is 32.93% and the standard deviation is 1.00%. The mean for
$10,000-$14,999 is 39.6% and the standard deviation is 1.95%. The mean for $50,000-
$74,999 is 67.36% and the standard deviation is 1.90%. The mean turnout for $75,000
and above is 73.03% with a standard deviation of 1.34%.
Conclusion Generally, the above figures strongly support Hypotheses 1, 2, and 3. Trends
among educational attainment and income illustrate the s-curve of participatory trends
within a demographic. The poles of educational attainment are relatively stable, with
smaller standard deviations and changes in participation. The middle sectors (9-12th
grade, High school/GED) are the most volatile with much larger standards deviations and
means of turnout change. Participatory volatility peaks among middle income voters,
while the poles of the income spectrum experience stable turnout, similarly supporting
hypotheses 1, 2 and 3. Unemployed voters are much more volatile than employed
voters, strongly supporting the first hypothesis but hypotheses 2 and 3 rely on more than
two populations.
11
Racial identities offer weak support of the hypotheses. If you examine the racial
identities as two groups, Caucasians and minorities, data supports the second hypothesis.
Minority voting patterns are consistently lower but relatively stable and vary less than
Caucasian voter turnout. Historical context is as a significant confounding variable.
Volatility by age demographic is inversely correlated to voter participation. The most
volatile demographic are those voters 18-24 and the least volatile demographic is voters
65 and above.
Potential interference comes from both sampling and response biases. The CPS
sample does not include active-military personnel or people living in institutions
(hospitals, nursing homes, dormitories, etc). The Census Bureau conducts surveys via
telephone, in person and through the mail. These methods favor people of economic
means, and sample sizes increase in relation to socioeconomic status. Surveyors directly
ask subjects for their voting habits, allowing for response biases with a skew for over-
reporting activity (Census 2004, 2000, 1996, 1992).
II. The Political Context of Elections: What Issues Garner Greater Participation? A.
One of the more robust findings on electoral behavior has been the importance of
macroeconomic conditions on voting in US presidential elections. Numerous studies
(Tufte 1980, Hibbs 1982, L.ewis-Beck and Rice 1992) show that aggregate rates of
change in unemployment, personal income and inflation have a large effect on the vote
for the incumbent. Political scientists theorize two different types of economically
motivated voting- pocketbook voting and sociotropic voting. Pocket book voting is a
12
form of retrospective economic voting. It suggests that voters will vote for an incumbent
if their personal financial situation improves/worsens under their leadership (Kinder,
Richard & Kiewiet, Roderick D., 1981). Sociotropic voters base their actions on the
perceived state of the national economy. If the voter feels that the overall economy is
doing well they will reward the incumbent with reelection and if it is floundering, they
will punish him and take away their vote (Sigelman).
These theories only capture one motivating factor of voter behavior- economic
situation. Economic voting behavior has strong empirical support, but post-war economic
growth and stability have afforded voters the opportunity to focus on other electoral
issues. Recent Gallup Poll survey data indicates that the public is increasingly less
concerned with economic factors. Gallup Polls after the 1996 presidential election
determined that education initiatives were the top priority of the American electorate.
Crime was second followed by Social Security, healthcare and Medicare (Gallup Poll). A
2000 survey, conducted before the presidential election, noted “this year, the economy is
not the urgent concern.” A Gallup Poll in October 2004 found that at least 85% indicated
that terrorism, Iraq and the economy were very important in shaping their vote choice.
Recent public opinion surveys indicate that while the economy is important, it no longer
dominates public concern.
A. Theory and Hypotheses Election-seeking policy makers drive political competition, employing tactics that
maximize their support among the electorate, thereby increasing their chances of winning
the election. There exists a wealth of empirical support illustrating economic
performance influences electoral outcomes (Hibbs 142), implying economic
13
manipulation is a decisive and effective electioneering tool. Motivated by a myopic
electorate that rewards policy makers for economic windfall, politicians strategically
pursue economic policies that benefit voters immediately preceding an election (Tufte
57). Economic policies foster partisanship and capture the spectrum of the electorate
whose interests and political alignment are clearly defined. Conversely, economic issues
do not capture the median vote- the middle class whose economic needs are vague.
Hibbs argues that the saliency of economic issues creates a demand on politicians to
manipulate macroeconomic policy to serve specific constituencies. Politicians use two
inversely linked instruments, unemployment and inflation to manipulate the
macroeconomy. To affect the economy in the short run, politicians must choose a trade-
off between unemployment and inflation (Hibbs 43). These two tools have different
distributional costs among the electorate. “Specifically, the Democratic Party represents
the bulk of lower income labor workers. The distributive costs of high unemployment
affect this part of the electorate the most. Unemployment rates for blue-collar and service
workers tend to be approximately two to three times higher than among white-collar
workers (Hibbs 52). Democrats are willing to allow relatively higher levels of inflation to
lower unemployment rates. The Republican Party serves a constituency of higher-income
citizens who are more averse to high levels of inflation. Objectively, sustainable inflation
has few costs but psychologically the concern regarding inflation is high (131).
Distributively, the top .01% of the income distribution feels the costs of inflation (117).
These opposing preferences create a partisan bias among policy makers as they
manipulate the economy to dampen the respective costs of unemployment and inflation
costs among their constituents.
14
Analogous with the central economic principle of supply and demand, the median
voter theorem is a concrete, simplistic model of majoritarian decision-making. Each voter
has a set of single peaked preferences: consisting of an ideal point that yields a maximum
utility, with linearly decreasing utility associated with subsequent proposals that further
away from his ideal point. When all actors’ preference points aggregate on a single
dimension, the median voter is the voter whose preferences are situated so that the same
numbers of other voters fall to their left and to their right. Rational choice dictates that
each voter will choose the party or candidate that is closet to his or her ideal point. The
candidate who aligns himself the closest to the median voter will always win the election
because the median voter creates a majority. “The theorem we have proved shows that
the decision adopted by the committee becomes determinate as soon as the position of on
optimum- which we can refer to conveniently enough as the median optimum is given”
(Black 29). Implicitly, as both candidates seek to capture the median voter, their
platforms will converge. The gravitational pull of the median voter often creates
moderate policies.
Policy makers use economic policy to mobilize those sectors of the electorate
whose economic interests are clearly defined. In the social dimension, there is intense
competition to converge toward the median’s ideological ideal point. Consequently, the
parties loose their distinctive partisan hold created via polar economic standpoints. Issue
oriented mobilization is less distinct and less partisan driven than economic initiatives. In
an attempt to capture the median vote, their social policies are not as polarized as their
economic initiatives tend to be. Concurrently, the economic initiative of both the
Democratic Party and the Republican Party are well defined and believed by the
15
electorate. “The U.S parties not only differ on economic issues; they differ more on
economic issues than on other matters” (Tufte 74). Gallup surveys conducted from the
mid 1940s- early 1980s indicate that voters in the upper income classes strongly align
their interests with the Republican Party and voters in the lower income class feel that the
Democratic Party best serves their interests (qtd in Hibbs 214). “The voting cleavages
diminish toward the mid-ranges of the income and occupational hierarchies, indicating
that the broad middle class is the main battle ground of electoral competition between
parties” (Hibbs 215).
What prompts people to vote in some elections and not in others? The interplay
between social and economic saliency violates the principle assumption of the median
voter theorem- one –dimensionality. Different demographics mobilize around different
issues creating a dynamic, multi level space for election.
The most volatile demographic sectors identified in Part I are the middle class,
youth, and middle range educational attainment. These three groups share one common
feature- their economic interests are not clearly defined. Economic policies capture the
interests of the electorate situated at the ends of the spectrum but do not address the
middle class, the young or those with moderate educational attainment (often directly
related to income status). Opposing economic policies do not directly affect these sectors.
Even when economic issues are salient, the dichotomous relationship and tradeoff
between inflation and unemployment is most relevant to the polar ends of
income/age/educational demographics. The weak economic preferences of a large
portion of the electorate imply that many citizens are not strictly economic voters. This
16
implies that these groups decide to participate for other reasons. These groups mobilize
around social issues, turning out to participate when social saliency is high.
Hypothesis 1: Voter turnout among higher volatility demographics will
correspond with elections where social issues are more salient than economic issues.
B. Data/Methods
Data on voter participation and issue salience are available in a reliable and
reputable time series through the American National Elections Studies (ANES). In
presidential election years, ANES asks subjects to identify “the most important issue
facing the United States.” With open-ended responses, surveyors grouped the responses
into general themes- social issues, social welfare issues, environmental issues, foreign
affaris, economic issues, etc. I use ANES data for the 1992-2004 presidential elections to
investigate the effect of issue salience and demographic factors on individual decisions to
vote. The independent variables of interest were social issue salience and economic
salience. I recoded the respondents’ answers into three categories of interest: social,
economic, social welfare. I kept social welfare separate because it combines both
economic aspects (taxation, redistribution of wealth, benefits, etc) and social aspects
(public goods, redistribution of wealth, transfers, etc). This separation allowed further
study of what issues motivated the demographic groups of interest.
Linear prediction models analyzed the effect of issue salience on participation.
The first model ran voting behavior (vote/did not vote) in respect to demographic
characteristics (age, race, income, education, employment status). Education is a
categorical variable representing the highest level of education the respondent achieved.
17
Employment and race are binary, where employed and white are coded as 1. A second
linear prediction ran voting behavior in respect to demographic identity and the “most
important issue” response. The second model analyzes the relationship between issue
saliency and voter behavior to see the effect of different issue salience on turnout among
different demographics.
To analyze the data on an individual level, I cross-tabulated different variables of
interest. Using the results from part I to identify the most volatile sectors of different
demographics, the data was condensed and organized into three groups of interest (ages
18-34, income $15,000-$49,999 and education 9th-High school Diploma) and the possible
responses for the “most important problem” questions (social, social welfare, and
economic). I counted the number of each type of response within each group of interest
and compared voter turnout.
Results
Table 1a ------------------------------------------------------------------------------ vote92 | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- race92 | .1288282 .0439558 2.93 0.003 .0425459 .2151104 income92 | .038025 .0087963 4.32 0.000 .0207584 .0552916 age92 | .0874636 .0149518 5.85 0.000 .0581142 .1168131 education92 | .0165212 .0150148 1.10 0.272 -.0129519 .0459943 employment92 | .0042576 .0059844 0.71 0.477 -.0074894 .0160045 swelfare92 | .0634751 .0622643 1.02 0.308 -.0587456 .1856957 social92 | .0080973 .0547513 0.15 0.882 -.0993759 .1155705 economic92 | .0125564 .0580111 0.22 0.829 -.1013156 .1264284 _cons | .1514309 .0959489 1.58 0.115 -.0369104 .3397722 ------------------------------------------------------------------------------
Table 1b
18
------------------------------------------------------------------------------ vote92 | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- race92 | .1445986 .0419625 3.45 0.001 .0622416 .2269557 income92 | .0421379 .0083528 5.04 0.000 .0257444 .0585313 age92 | .0911505 .0140357 6.49 0.000 .0636036 .1186974 education92 | .0122917 .0138923 0.88 0.377 -.0149739 .0395572 employment92 | .0045562 .0059409 0.77 0.443 -.0071036 .016216 _cons | .1406882 .0737689 1.91 0.057 -.0040931 .2854695 ------------------------------------------------------------------------------
Table 1b is the linear prediction model without the factor of issue salience while
table 1a includes issue saliency. Tables 1a and b show that race, income and age are all
statistically significant factors in voting behavior with race having the largest coefficient.
According to the model, issue salience is not a statistically significant factor in voter
turnout. All three categories of issue salience have small coefficients, their confidence
intervals include 0 and large p-values.
Table 2a ------------------------------------------------------------------------------ vote96 | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- race96 | -.0256724 .0361146 -0.71 0.478 -.0966452 .0453005 income96 | -.0119213 .0065793 -1.81 0.071 -.0248511 .0010085 age96 | .0158933 .0121552 1.31 0.192 -.0079943 .0397809 education96 | .012311 .0096243 1.28 0.201 -.0066029 .0312248 employment96 | -.0014199 .0044608 -0.32 0.750 -.0101864 .0073465 swelfare96 | (dropped) social96 | .0044969 .0299278 0.15 0.881 -.0543177 .0633115 economic96 | .0069543 .0356375 0.20 0.845 -.0630811 .0769896 _cons | .0990839 .0543331 1.82 0.069 -.0076923 .2058601 ------------------------------------------------------------------------------
Table 2b ------------------------------------------------------------------------------ vote96 | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- race96 | .0020786 .0257996 0.08 0.936 -.048542 .0526993 income96 | -.0052896 .0046473 -1.14 0.255 -.0144079 .0038288 age96 | .015517 .0083604 1.86 0.064 -.0008868 .0319209 education96 | .0015554 .0028191 0.55 0.581 -.0039759 .0070868 employment96 | -.0038433 .0048223 -0.80 0.426 -.013305 .0056184 _cons | .0846593 .0365067 2.32 0.021 .0130304 .1562882 ------------------------------------------------------------------------------
19
Table 2a and b indicate that in the 1996 election no demographic factors were
statistically significant determinants of voter turnout. Issue salience was not an important
factor and had no significant or measurable effect on turnout.
Table 3a ------------------------------------------------------------------------------ vote00 | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- race00 | -.5843665 .630166 -0.93 0.355 -1.828325 .6595916 income00 | -.0160549 .1738803 -0.09 0.927 -.3592976 .3271877 age00 | -.0902172 .1455082 -0.62 0.536 -.3774529 .1970185 education00 | .0419577 .1584183 0.26 0.791 -.2707627 .3546781 employment00 | .0124551 .3171995 0.04 0.969 -.6137021 .6386122 swelfare00 | -.4229635 .4257102 -0.99 0.322 -1.263322 .4173955 social00 | .1969802 .4478586 0.44 0.661 -.6871001 1.081061 economic00 | (dropped) _cons | 1.816735 .7832403 2.32 0.022 .2706061 3.362865 ------------------------------------------------------------------------------
Table 3b ------------------------------------------------------------------------------ vote00 | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- race00 | -.1348609 .2372371 -0.57 0.570 -.6012205 .3314988 income00 | -.0016667 .0886818 -0.02 0.985 -.1759969 .1726636 age00 | .0202658 .0860926 0.24 0.814 -.1489746 .1895061 education00 | .033686 .0882683 0.38 0.703 -.1398314 .2072035 employment00 | .0095379 .1893401 0.05 0.960 -.3626659 .3817418 _cons | 1.211147 .4034894 3.00 0.003 .4179697 2.004325 ------------------------------------------------------------------------------
Table 3b indicates that none of the demographic factors were significant
determinants of voter behavior. The consideration of issue salience did not change this
and none of the issues were significant in the 2000 election.
Table 4a ------------------------------------------------------------------------------ vote04 | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- race04 | .0856343 .0471034 1.82 0.070 -.0069913 .17826 income04 | .047498 .0117913 4.03 0.000 .0243111 .0706848 age04 | .0060086 .0162585 0.37 0.712 -.0259626 .0379797 education04 | .0426071 .022877 1.86 0.063 -.002379 .0875933 employment04 | -.0667208 .0467426 -1.43 0.154 -.1586368 .0251952
20
social04 | -.0063875 .0785238 -0.08 0.935 -.1607992 .1480243 economic04 | .0742026 .0889998 0.83 0.405 -.1008093 .2492145 _cons | .3592479 .1220431 2.94 0.003 .1192585 .5992372 ------------------------------------------------------------------------------
Table 4b ------------------------------------------------------------------------------ vote04 | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------- race04 | .1030511 .0427057 2.41 0.016 .0191339 .1869684 income04 | .0402342 .0104235 3.86 0.000 .0197519 .0607166 age04 | .0191761 .0146325 1.31 0.191 -.009577 .0479291 education04 | .0690702 .0199574 3.46 0.001 .0298536 .1082867 employment04 | -.0605358 .0411516 -1.47 0.142 -.1413993 .0203277 _cons | .2362483 .0867631 2.72 0.007 .0657577 .4067389 ------------------------------------------------------------------------------
Table 4b shows that race, income and education were all important factors in
2004 electoral behavior with race being the strongest. When considering issue salience
these relationships weakened but were still statistically significant. None of the issues
were statistically significant in determining turnout.
None of the models indicate that issue salience is a statistically significant factor
in electoral turnout but their possible effects could be mitigated by the aggregate analysis.
Table 5
1992 election n=985
Total Respondents Number of Respondents that Voted
Age*Social Welfare 39 24 Age*Social 74 59 Age*Economic 57 25 Income*Social Welfare
85 62
Income*Social 227 150 Income*Economic 110 76 Education*Social Welfare
84 59
Education*Social 83 61 Education*Economic 111 73 Table 6 1996 election Total Respondents Number of
21
n=1535 Respondents that Voted
Age*Social Welfare 100 9 Age*Social 171 15 Age*Income 63 7 Income*Social Welfare
131 8
Income*Social 99 10 Income*Economic 46 3 Education*Social Welfare
16 3
Education*Social 17 4 Education*Economic 8 2 Table 7 2000 election n=416
Total Respondents Number of Respondents that Voted
Age*Social Welfare 22 17 Age*Social 32 22 Age*Income 3 1 Income*Social Welfare
4 4
Income*Social 14 10 Income*Economic 8 6 Education*Social Welfare
30 23
Education*Social 21 17 Education*Economic 12 10 Table 8 2004 election n=528
Total Respondents Number of Respondents that Voted
Age*Social Welfare 3 3 Age*Social 45 37 Age*Income 11 10 Income*Social Welfare
11 7
Income*Social 106 76 Income*Economic 25 19 Education*Social Welfare
10 7
Education*Social 103 69
22
Education*Economic
20 16
Tables 5, 6, 7, and 8 all strongly indicate that social issues are an important
mobilizing factor among demographics with high voter volatility. In every presidential
election, social issues were the most salient among those 18-34 and among the middle
class. Of these demographics, social issues were cited as “the most important problem”
with greater frequency overall and garnered the most voter activity. Based on education,
two of the four elections saw social issues as the most important issue (2004 and 1996)
while in 2000 social welfare issues were the most important and in 1992 economic issues
were the most important. This counting method, while a preliminary examination, shows
a strong correlation between social issue salience and voter turnout among volatile
demographics.
It is important to note that the sample size varies drastically across the studies. In
1996, with a sample size of 1535, 1384 participants responded that they did not vote.
The nature of the cross-tabulation terms limits the number of applicable subjects. In
further studies, one should examine the effect of issue salience on demographics with
stable turnout. This will serve as a better control than the effect of issue salience on
aggregate turnout as used in this study. The “most important issue” question from the
ANES survey data captures issue saliency, but it does not ask if that is why people voted.
This data analysis shows a direct correlation between issue salience among demographics
and voter turnout, and this paper theorizes that this issue salience causes mobilization, but
more research designed to explicitly capture why people voted is needed.
23
III. Conclusion Voter turnout, the key input of a democratic system, is a product of mobilization.
It is a topic of much interest because it has great significance on our democratic system.
This paper identified turnout trends over time and sought to explain how salient issues
mobilize particular demographics. The US two-party system is economically distinct but
similarly situated in the social realm. The importance of social issues in key
demographics is very politically valuable. This paper examined voter participation from
the electorate’s perspective. In future studies, it would be of great value to examine social
issue mobilization from the policy maker’s perspective.
24
Works Cited Alesina, Alberto, John Londregan & Howard Rosenthal. “A Model of the Polical
Economy of the United States”. American Political Science Review. 87:12-33. 1993
American National Election Studies. 1992. 1996. 2000. 2004 http://www.electionstudies.org/studypages/download/datacenter_all.htm Black, Duncan. “On The Rationale of Group Decision-making.” Journal of
Political Economy. 56: 23-34. 1948 Franklin, Mark N. “Electoral Participation” Controversies in Voting Behavior 2001 Hibbs, Douglas A. Jr., The Political Economy of Industrial Democracies, Cambridge, Mass.: Harvard University Press, 1987. Kiewiet, D. Roderick. Macroeconomics & Micropolitics: The Electoral Effects of
Economic Issues. Chicago, Illinois: University of Chicago Press. 1983 Kiewiet, D Roderick & Douglas Rivers. “The Economic Basis of Reagan’s
Apperal”. The New Direction in American Politics: Chubb, John E. and Paul E. Petersons. The Brookings Institution p 69-90. 1985
Kinder, Donald R & D. Roderick Kiewiet. “Economic Discontent and Political
Behavior: The Role of Personal Grievances and Collective Economic Judgments in Congressional Voting”. American Journal of Political Science 23:495 527. 1979
Lewis-Beck, Michael S. & Tom W. Rice. Forecasting Elections. Washington,
D.C: Congressional Quarterly Press. 1992 Lijphart Arend, “Unequal Participation: Democracy’s Unresolved Dilemma,” American Political Science Review, 91(1), March 1997, pp. 1-14. Moore, David. “Republicans and Democrats Differ on
Important Election Issues”. The Gallup Organization. January 19, 2000. Newport, Frank. “Economy, Education, Health, Crime and Morality Most on
American’s Minds this Election Year”. The Gallup Organization. June 22, 2000.
Sigelman, L., Roeder, P. W., Jewell, M. E., & Baer, M. A. Voting and nonvoting: A multi-election perspective. American Journal of Political Science, 29(4), 749-765. 1985
Tufte, Edward R., Political Control of the Economy, Princeton: Princeton University
25
Press, 1978. U.S Census Bureau. Current Population Survey. 1968-2004
http://www.census.gov/population/www/socdemo/voting.html Weatherfored, Stephen M. 1978. “Economic Conditions and Electoral Outcomes: Class
Differences in the Political Response to Recession”. American Journal of Political Science 22:917-939.
The author would like to thank Leanne Powner and Robert Franzese, Jr. Without their guidance this work would not have been possible.