Determinants of Voting Behavior and the Importance of Campaign Effects

Part 1: Determinants of Voting Behavior and the Importance of Campaign Effects

What are the factors that determine how a voter going to the polls on Election Day? What are the factors that determine how a voter chooses a candidate when they mark on the ballot? What role do political campaigns themselves play during an election season? Do these political campaigns play the role in informing the voters on candidates and issues before Election Day or do they exist for something else? These are the general research questions regarding voting behavior and the importance of campaign effects in U.S. elections that scholars such as Brady, Verba, Schlozman (1995) and Fraga (2015) on voting turnout, Campbell and the Michigan scholars (1960??), Fiorina (1981) and Lau, Redlaws (2006) on vote choice, and Jacobson (2015) and Lau, Rovner (2009) on political campaigning, have studied and built upon through their own empirical work.

This paper examines into the question of what determines voter behavior (both voter turnout and vote choice) and what role do political campaigns play in American politics through the empirical findings made by scholars that attempted to answer these general questions regarding voting, campaigns, and elections. Variations among Americans on their available resources such as time, money, and civic skills that are acquired in the process of individual development, and increasing population of ethnic minorities effects voter turnout in U.S. elections while partisan identity based upon long term political socialization, the role of cognitive heuristics in candidate identification, information shortcuts, and rational choice through retrospective voting are the factors that effects vote choice. Political campaigns, on the other hand, play a role in affecting voters’ preferences and choices through raising awareness of the candidate’s name and information to voters, raising awareness on important issues, and raising voter turnout.. The use of resources such as civic skills and increasing presence of ethnic minorities on voter turnout as well as the role of cognitive heuristics and retrospective voting on vote choice help enhance American democracy. On the other hand, partisan attachment on vote choice bring mixed results on the quality of American democracy while negative ads for political campaigns is shown to degrade the quality of American democracy.

Brady, Verba, and Schlozman made their study on finding a better explanation to the variation among Americans on degree of engagement on political participation and variation on types on engagement on political participation. They acknowledge on past scholarship works that attempted to answer this question of variation among Americans on political engagement in findings such as paucity in resources to engage, absence of psychological engagement with the political process, and isolation from networks of political recruitment (Brady, Verba, and Schlozman 1995, 271). However, Brady, Verba, and Schlozman argue that these findings to explain the variation among people on political engagement are insufficient and unreliable to find a full answer to the question (Brady, Verba, and Schlozman 1995, 271-272). Instead, they propose a new methodological design by focusing on the role of resources itself in explaining the variation (Brady, Verba, and Schlozman1995, 271).

In executing their own methodological design for the study, they utilize a two-stage survey in asking Americans on their own social and civic activities. The first stage consists of asking over 15,000 Americans in a random sample via a telephone survey (Brady, Verba, and Schlozman 1995, 272). The second stage would be to ask over 2,500 respondents that are active political/social activists, African-Americans, and Latinos through in-person interviews (Brady, Verba, and Schlozman 1995, 272). The questions to be asked in the two-stage telephone and in-person survey consists of demographic questions such as educational attainment, ethnicity, income, citizenship status, questions regarding the subjects’ own daily schedules and time being spent on daily activities, as well as questions regarding acquirement of civic skills through past experiences and involvement in school, work, church, and/or in community organizations (Brady, Verba, and Schlozman 1995, 272-273, 285-287). These questions on income and educational attainment, citizenship status, acquirement of civic skills, level of political interest, and time spent on daily activities are independent variables. As for dependent variables, it is to be measured through questions being asked on specific political activities such as voter turnout, campaign work, campaign contributions, contacting public officials, protesting, and involvement with local government (Brady, Verba, and Schlozman 1995, 285-287). Their theoretical framework for the design is based upon measuring the three resources that they outline as crucial towards political participation—time, money, and acquirement of civic skills (Brady, Verba, and Schlozman 1995, 273). The resources that people acquire and utilize are not distributed equally, as Brady, Verba, and Schlozman acknowledge that differences in socioeconomic income does effect on how resources such as time, money, and civic skills are distributed and how it would lead to variations among Americans on their level of political participation (Brady, Verba, and Schlozman 1995, 274).

The findings in their study show that the resources of time, money, and civic skills do have a strong effect on political participation. In making the data, they separate the data into data on socioeconomic status and data on participation on political activities. For the effects of money, time, and civic skills on socioeconomic status, it shows a high correlation for money and civic skills, but not for time. The data on Figure 1 in their study shows that higher education translates into higher income, which would in turn be beneficial in helping to provide money on political contributions that would count as political participation (Brady, Verba, and Schlozman 1995, 275). The average time (hours) per day that is free, on the other hand, skews in favor to those with less education, as it shows that people of higher income still have to consume time due to work and other non-political activities (Brady, Verba, and Schlozman 1995, 275). Civic skills, on the other hand, is shown with higher amount of skills being acquired when a person receives higher educational attainment, as shown on Figure 2 of the study (Brady, Verba, and Schlozman 1995, 276). These findings show that income and civic skills matter most to socioeconomic status than time. On the other hand, the dependent variables in the findings that were measured show that degree of participation in political activities such as voting, campaign contribution, and volunteering for a campaign depend highly on an appropriate resource. In terms of voting, participation in voting depends more on political interests, free time, and citizenship status, participation in campaign contribution depends more on income and educational attainment, and participation in campaign volunteering depend more on both civic skills, time, and political interests (Brady, Verba, and Schlozman 1995, 283-284).

The study by Brady, Verba, and Schlozman contributes significantly to the literature on determining U.S. voter behavior in that the acquirement of civic skills, time, and money available play a strong role in determining voting behavior (especially on voter turnout) among Americans. The measuring of acquirement of civic skills and time demonstrates a significant improvement and breakthrough in the study of U.S. voting behavior that other scholars in the field have focus only on socioeconomic status such as income and educational attainment (Brady, Verba, and Schlozman 1995, 272). This study shows that these resources of time and acquirement of civic skills and its strong effects on political participation (especially on voting) can significantly alter the quality of American democracy itself as citizens with less time to participate on civic and political issues and less acquirement of civic skills that are part of human development such as leadership, writing, and communication skills can potentially degrade the quality of American democracy itself.

In another study in finding what determines voter behavior, Fraga did a study in examining co-ethnic candidacy and racial/ethnic composition within a jurisdiction as a determinant for voter turnout. The question that he attempts to find through his study is whether or not ethnoracial contexts impacts voter turnout (Fraga 2015, 5). Fraga acknowledges on past theories on race and voter turnout such as “empowerment” and “elite mobilization” theories, where the “empowerment” theory argues that individuals will likely to participate in voting when they feel their participation will have an impact on politics while “elite mobilization” argues that voter turnout depends upon the mobilization of the electorate by politicians that are seeking election (Fraga 2015, 2-3). Other theories that Fraga also mentions argue that the presence of a co-ethnic candidate can increase voter turnout for the members of a candidate’s own ethnic group as well as another theory that argues that voter turnout of an ethnic group would increase if there is a population increase of members of an ethnic group within a jurisdiction (Fraga 2015, 3). However, Fraga points out on the problems in their methodology when trying to extract the findings in supporting these theories such as overreporting and misreporting of turnout on surveys (as in the case of African-Americans and Latinos living in their own ethnic districts) and problems of selection bias (by focusing mainly on majority-minority districts and leaving out districts with non-majority minority) that leads to bias in the findings (Fraga 2015, 4-5).

Fraga proposes a new comprehensive methodology on the study of race and voter turnout that reduces bias in the findings through individual-level turnout records of three nationwide elections (general and primary of 2006, 2008, and 2010) instead of survey estimates that limits its analysis to a small number of states (Fraga 2015, 5). His independent variable is on the presence of co-ethnic congressional candidates and population composition of ethnic/racial groups in a jurisdictional congressional district, while his dependent variable is on theimpact of voter turnout by race. His unit of analysis includes not only African-American and Latino voters but also non-Latino Whites and Asian-Americans as well (Fraga 2015, 5-6). Fraga uses voter file databases from Catalist in order to examine voter turnout by race/ethnicity, data from the American Community Survey to examine voting-age population composition by race/ethnicity, and data from the Federal Elections Commission on all of the candidates running for the House of Representatives and the Senate in the 2006, 2008, and 2010 elections with coding on ethnicity being done by Fraga (Fraga 2015, 6-8).

The findings in the study show that ethnic minority citizens are more likely to turn out to vote the most when they live in a minority-majority district. This is shown on Table 3 where there are higher percentages of voter turnout for Whites, Latinos, and African-Americans living in districts where their own ethnic group is the majority over districts where their own ethnic group is a minority (Fraga 2015, 9-10). However, turnout in general in overall minority-majority districts is much lower than in non-minority majority (meaning White-majority) districts (Fraga 2015, 10). This paradoxical finding shows that districts inhabited by ethnic minorities are still less likely to turnout to vote in comparison to White-majority districts, but there is much higher voter turnout among ethnic minorities living in districts that are minority-majority than ethnic minorities living in White-majority districts. As for co-ethnic candidates, the findings show that there is much higher turnout rate of voters among White, African-American, Latinos, and Asians if there is an appearance of their own co-ethnic candidate running for election, as shown on Table 4 of the study (Fraga 2015, 10-11). However, the findings also show that for African-Americans and Latinos, even if a member of their own ethnic group is not running for election, the percentage of voter turnout in still high if there is a large percentage of their voting-age ethnic group within in a minority-majority district (Fraga 2015, 11-12). For Asians, turnout drops when there is a larger population of voting-age Asian-Americans within a minority-majority district (Fraga 2015,12). For Whites, turnout would only increase if their own co-ethnic candidate is running (Fraga 2015, 12). District racial/ethnic demographics play a much stronger role in determining voter turnout than the presence of a co-ethnic candidate. These findings support the theory made by past scholars that voter turnout among ethnic minorities would increase only if there is an increase in demographics among ethnic minorities.

The implications for Fraga’s study show that race and ethnicity itself does play a strong factor in determining voter behavior (especially for voter turnout) in U.S. elections. This contributes significantly to the literature on determinants of voting behavior as Fraga uses an observational methodology of individual-level turnout records in order to reduce bias in the findings. The findings by Fraga does show that the increasing presence of ethnic minorities in voter turnout due to increasing demographics does enhance the quality of American democracy as the increasing involvement of ethnic minorities and immigrants in the U.S. political process allows their own ethnic community interests to be represented in the halls of local, state, and federal government in pushing for policies that would prevent discrimination and allow an equal-level playing field in terms of opportunities and interests that would benefit all Americans, including ethnic minorities and immigrants.

In terms of determining vote choice, scholars such as Campbell, Miller, Converse, and Stokes (1952), Fiorina (1981), and Lau and Redlawsk have given different arguments based on different findings in their studies in determining vote choice. Campbell and the proponents of the Michigan school such as Miller, Converse, and Stokes argues that vote choice is based upon deep partisanship attachment and dispositions, while Fiorina gave his own finding that vote choice is based mainly upon a voter’s rational self-interest. Lau and Redlawsk, on the other hand, gave evidence that shows voters make their own decisions at the polls based upon cognitive heuristics and information shortcuts.

Campbell and his team of young scholars from the University of Michigan—Miller, Converse, and Stokes—did their study on finding a plausible theory to the development of vote choice. Their theoretical framework depends on improving the previous works made by Lazarsfeld and his Columbia colleagues that found “brand loyalties” in religion, social class, and social networks as a strong determinant toward vote choice (Bartels 2008, 240-241). The methodology that Campbell and his Michigan team had done utilize national surveys that were conducted in the election years of 1952, 1954, 1956, and 1958, in contrast to Lazarsfeld’s methodology on a panel survey study that focuses only on a single community in a single election year (Bartels 2008, 240, 242). Their independent variables for measurement are mainly on demographic and personal characteristics as well as attitudes of partisan and non-partisan voters, as well as non-voters, and on “various political, sociological, and psychological factors” (Bartels 2008, 242). The dependent variable is on vote choice itself.

The findings of the study made by Campbell and his team demonstrate that the landslide victory of Dwight D. Eisenhower as the new president did not affect any long-standing loyalties of American voters that vote mainly for the Democratic Party (Bartels 2008, 242). Other findings in the study show that American voters themselves are heavily attached to one of the two parties, showing persistent long-term partisan stability attachment among voters (Campbell et al. 1960, 146). On the other hand, when it comes to shifting partisan identities and electoral strength, such a situation would only occur if the party’s own identity and its attachment towards its social groups change rather than on the details of legislative or administrative policies (Campbell et al. 1960, 546). The findings also demonstrate that partisan loyalties towards one of the two major American parties are often developed in early adulthood through socialization (Bartels 2008, 244). Political information plays little to no role in contributing towards a voter’s decision choice at the polls, as voters would remain passive on receiving political information and still maintain a consistency on cognitive decision-making through continuing deep attachment to their own party (Lau and Redlawsk 2006, 9-12).

The study by Campbell and the Michigan team brought a significant contribution to the scholarship on voting behavior as the findings opened the way to the development of a new theory on vote choice that focuses on attachment to political parties called the Michigan Model. In its implications towards the state of American democracy, the findings demonstrate the importance of political parties towards American voters and how the ideological beliefs of the two major American political parties still play a strong role in the minds of American voters as well as American elected politicians themselves. The strong role of political parties in terms of voter attachment to them when choosing a decision at the polls does not bode too well for the quality of American democracy as persistent attachment towards their own particular political party may hinder efforts towards bipartisan compromise on important legislative issues as voters would end up electing politicians that stick with the party and its ideological base over the nation. As a result of partisan voters electing strong partisan politicians, it leads to political polarization and legislative gridlock—which is also not a good development towards American democracy. However, the strong role of partisan attachment in determining vote choice can also enhance the quality of American democracy as voters’ attachment towards their own particular political party enhances their own participation and involvement in the American political process, allowing voters to have more influence and say to how their party and government should run instead of interest groups and economic elites.

The study by Fiorina, on the other hand, examines to how rational choice and political information itself play a strong role in determining how a voter chooses a candidate during Election Day as Fiorina looks to how voters use retrospective voting and the idea of a “running tally” to make their voting decisions (Bartels 2008, 248). Fiorina uses Downs’s rational choice theory, Downs’s application of the rational choice theory on party ideology and the impact of the distribution of the vote, and Key’s study on how voters determine their vote based on policy and performance evaluations as the theoretical framework basis for the study on determining voter choice (Bartels 2008, 247). The voter’s goal in vote decision-making under retrospective voting in a rational choice setting is all about maintaining the voter’s own self-interest by seeking information and weighting the costs and benefits when evaluating the information (Bartels 2008, 247-248). The methodology that Fiorina uses looks into the independent variables that represent retrospective voting such as on voter evaluations on economic conditions, performance of the incumbent president in office, foreign policy, as well as on other independent variables such as issue positions, and party identification—while the dependent variable is measured through vote choice (Bartels 2008, 248). Fiorina also adds in “prospective voting” as an additional independent variable through evaluative questions regarding hypothetical scenarios on what the world would look like if an alternative party takes power to test and see if voters can make their voting choice decision based on prospective voting (Lau and Redlawsk 2006, 34).

The findings from Fiorina’s study show that there is a strong effect of retrospective voting as a determinant towards vote choice. The findings shown that voters themselves prefer using retrospective voting over prospective voting when making their vote choice as it is much easier and efficient for voters to seek information on past incumbent and economic performance in order to determine their choice at the polls than to think prospectively on what the future holds if the alternative party takes power (Lau and Redlawsk 2006, 34, 73). Fiorina’s work that supports the argument on the importance of retrospective voting on vote choice also confirms on the theoretical framework of Downs and Key on rational choice and importance on policy evaluation as determinants towards vote choice (Bartels 2008, 248).

Fiorina’s study on retrospective voting as the determiner of vote choice provides a significant contribution to the literature on voting behavior based on vote choice. The findings that demonstrate the importance of retrospective voting over prospective voting does have implications for the quality of American democracy as retrospective voting does help enhance American democracy by making elected politicians more accountable to the voters in a reward-punish system of retrospective voting when incumbent elected politicians are incompetent and not fulfilling their needs and interests for their constituent voters. American democracy itself requires elected politicians to have mechanisms of accountability to the people that elected them, which is exemplified through the use of retrospective voting among voters as a determinant in making their vote choice at the polls.

The study by Lau and Redlawsk goes beyond the past studies on vote choice based upon partisanship attachment and retrospective voting by examining the role of information shortcuts and cognitive heuristics and its impact on vote choice. Their theoretical framework focuses on comparisons with past research on vote choice by other scholars such as rational choice/retrospective voting, partisanship attachment of the Michigan Model, fast and frugal decision making (making vote decisions based upon single-issues that the voter cares about), and bounded rationality/intuitive decision making (making vote decisions based upon cognitive shortcuts) as the theoretical basis for their study (Lau and Redlawsk 2006, 45). Lau and Redlawsk argues in their discussion of the theoretical framework that the rational choice/retrospective voting model and the partisanship attachment of the Michigan Model ignores the role on the process of acquiring information, as both the rational choice model and the Michigan model of vote choice only focuses on the end result of the vote decision-making (Lau and Redlawsk 2006, 21). They propose a new methodology on the study of bounded rationality and fast and frugal decision making by focusing on developing a new methodological approach that is “process-oriented” (Lau and Redlawsk 2006, 21). In building a new “process-oriented” methodological approach, their main theoretical basis in developing the new methodology is based upon the behavioral decision theory, where decision-making is based upon seeking a decision that is “good enough” rather than “value-maximizing” as in the case of rational choice (Lau and Redlawsk 2006, 22).

In creating the new model, the independent variable is measured by looking into the demographic background characteristics of voters, political sophistication/expertise (having expertise in focusing on important information words without going into details), and campaign factors (task demands on making a decision under time pressure, decision making based upon efficiency over accuracy) (Lau and Redlawsk 2006, 22, 37-38). These variables represent as the process in acquiring information in making a vote decision as well as other factors that influences vote choice—mainly campaign factors (Lau and Redlawsk 2006, 23). The dependent variable that would be measured is on the subjective “nature” of the decision task (Lau and Redlawsk 2006, 23). Partisan dispositions are measured as a control variable (Lau and Redlawsk 2006, 22). To deploy the study, they did several experiments in which the subjects who are vote-eligible citizens at first had to complete a political attitudes survey questionnaire in measuring demographic background characteristics of voters as mentioned as part of the independent variable, participate as part of a mock campaign experiment where the subjects have to read campaign material and vote in a mock primary election (to be done in 20 minutes) (Lau and Redlawsk 2006, 290-293). After voting, the subjects were asked on their feelings and opinions of the mock candidates and their experience in voting in measuring the dependent variable (like how difficult was it to choose and if they are certain on their decision choice) (Lau and Redlawsk 2006, 293). The subjects repeat the same task for the mock general election phase (to be done in 12 minutes) and were asked the same post-voting questions on their voting experience (Lau and Redlawsk 2006,  293-294). To gather data for political sophistication, the subjects had to write down in memory on any information on the candidates that they remember from the mock campaign materials and participate in another experiment where they were given more time and complete information on the candidates in measuring whether they had made a correct vote choice (Lau and Redlawsk 2006, 294-295).

The findings of the study demonstrate that the use of information shortcuts and heuristics by voters when making their decision is only significantly helpful when the voter is in a difficult situation in making their choice (Lau and Redlawsk 2006, 252). When the data is gathered and analyzed, it is shown that the use of information shortcuts on making a correct vote on both the primary and general elections is statistically significant only for voters who have strong background knowledge on the election as well as for voters that are in a situation of making a difficult choice due to the presence of four candidates in the ballot instead of two in the primary election (Lau and Redlawsk 2006, 250). As for voters who are novices and have low information background on the election, the use of information shortcuts and heuristics are not particularly helpful in making a correct vote choice (Lau and Redlawsk 2006, 250-251).

The results in the use of information shortcuts upon making a vote choice is particularly helpful only for voters who are highly informed on politics or for those who are still undecided in making a choice due to multiple candidates running. The implications for this study is that the use of information shortcuts does enhance the quality of making a vote choice when going to the polls as well as enhancing the quality of American democracy due to high efficiency as the average American voter usually does not have free time in their schedules in paying attention to politics due to work and personal life schedules. The work by Lau and Redlawsk is more realistic in modeling their experiment based upon how an average American voter would vote in a real election, debunking the old theories of vote choice such as rational choice due to its lack of real applicability in how the average American makes their decision to vote.

Political campaigns play an important role in the quality of American democracy and many scholars have made different studies regarding political campaigns, resulting in different findings that continue to enrich the debate on the role of political campaigns in the American political system. The first studies on political campaigns begin with Lazarsfeld’s own study (1944) on voting behavior, where his findings show that presidential campaigns have little effect on how voter make their decisions at the polls due to persistent attachment towards political parties and social groups that are connected with the parties (Jacobson 33). Later studies by other scholars (Vareck 2009) show that campaigns do matter as it has the ability to accurately predict presidential election outcomes when it is framed by presidential candidates on important salient issues such as the economy, foreign policy issues, crime, honesty issues among politicians (Jacobson 2014, 34). Specific campaign activities such as national conventions are helpful influencing vote choice among voters, as in studies made by Campbell (1992), Shaw (1999), but for some activities such as presidential debates—it is not too certain with mixed results such as the studies by Shaw (1999) that shows debates as having significant effects while for Erikson and Wlezien (2012), it is not significant (Jacobson 2014, 34-35). As for other campaign activities that stimulate voter turnout such as in a study by Green (2012), the use of door-to-door canvassing as part of the campaign stimulates higher voter turnout than other aspects of the campaign in stimulating voter turnout such as direct mail, volunteer and automated phone calls (Jacobson 2014, 38).

In all, political campaigns do play a significant role in enhancing the quality of American democracy, as it helps voters get information on the candidates and how the candidates respond to salient issues important to the voters as well in stimulating voters to get out to vote. But it can also degrade the quality of American democracy in the case of negative campaigning as negative attack ads on candidates do have a significant effect not on vote choice and turnout, but on citizens’ trust on the government and the political system as a whole—as shown in a study by Lau (2007) (Jacobson 2014, 41).

Part 2: Inequality and Responsiveness in American politics

Is the American political system with its Madisonian institutions responsive to public opinion and preference? What factors that help facilitate and/or frustrate the ability to translate the preferences and opinions of the public into public policy? These are the questions regarding inequality of public opinion/preference and its impact on public policy that scholars such as Gilens and Page (2014), Erikson (2015), and Kelly and Witko (2015) have long debated in finding answers in the relationship between public opinion and public policy through their empirical findings that they found. This paper will examine into the works made by scholars in the field of public opinion and its impact on public policy in the United States and how they attempt to answer the general questions through the methodology that they use and the findings that they extract. The findings made by these scholars show paradoxical results as the American political system is responsive to both the opinions and preferences of the economic elite and business interests, but it is also responsive to the opinions and preferences of average citizens in a variety of ways such as in the state level. Inequality and gaps in income distribution and information distribution are found to frustrate responsiveness and representative outcomes in the American public policy process but it also facilitates responsiveness as the issues of economic inequality help propel elected politicians and policymakers in different levels of government (especially in the state level) to address on it.

Gilens and Page made their own research in finding which type of income group and interest group actors play a much more influential role over public policy. In this case, Gilens and Page tested out on the four different theoretical traditions of public opinion and its relationship to public policy made from past bodies of literature in answering the overarching question on who truly governs American politics. These four theories are majoritarian electoral democracy, economic-elite domination, and the two types of interest-group pluralism—which are majoritarian pluralism and biased pluralism (Gilens and Page 2014, 564). These four theories serve as the main theoretical framework for which Gilens and Page conduct their test in the study.

Majoritarian electoral democracy looks into public opinion and its relationship towards public policy as policymaking based upon the will of the average electorate (Gilens and Page 2014, 565). This is the traditional mantra that legitimizes the American democratic political experiment since the founding days of the nation (Gilens and Page 2014, 565). Past works that argue for this theory include works on rational choice theories of electoral democracy by Hotelling (1929), Black (1948) and Downs (1957), “chaos” theory on median voter prediction by social choice theorists such as Arrow (1963), the idea of median voter and its impact on public policy such as Monroe (1979) and Erikson (2002), and retrospective voting by Key (1961) (Gilens and Page 2014, 565-566).

Economic-elite domination, on the other hand, views public opinion and its relationship towards public policy as a distorted, unequal playing field on public policy that favors wealthy individuals over the average citizen (Gilens and Page 2014, 566). This theory is supported by past works from scholars such as Mills (1959), Beard (2012), Winters (2011), Domhoff (2013), Ferguson (1995), and Burch (1980), by which all agree to some extent that economic elites with wealth and property play an influential and dominant role over the halls of American government (Gilens and Page 2014, 566).

Majoritarian pluralism characterizes public opinion and its impact on public policy as a competition for public policy influence between all mass-based interest groups run by average citizens and business elites (Gilens and Page 2014, 566). This theory is well-supported through works made by scholars such as Bentley (1908), Truman (1971), and Dahl’s “polyarchy” (1956), as well as on traditional works made by founding father/politician/scholar James Madison (1787) on “factions”, in which all agree that the public policy arena is well-represented equally and in equal competition by all interest groups that have a stake in the arena (Gilens and Page 2014, 566).

Biased pluralism, on the other hand, examines the relationship between public opinion preference of interest groups and public policy through the lenses that the playing field among interest groups and its impact on public policy is tilted in favor of interest groups and associations that are business-oriented—including corporations as well (Gilens and Page 2014, 567). Scholarly works that support this theory include the popularly well-known theoretical work of Marx and Engel’s “The Communist Manifesto” on the “bourgeoisie capture of the state” (1848), as well as other scholarly research that follow the Marxist paradigm such as Miliband (1969), Ferguson (1995), and Stigler (1971), where they all agree that the policy making process and the political institutions itself are dominated by businesses, corporations, and organizations that represent the economic interests of the wealthy (Gilens and Page 2014, 567).

In their assessment through the past literature works made by scholars that studied the four theoretical fields of public opinion and its impact on public policy, Gilens and Page acknowledge past scholars’ works and its contributions, but argues that their methodologies in making the findings that brought the development on the four theoretical fields did not take into account on assessing interest-group influence over a broad set of issues that simultaneously assesses the impact of public policy from average citizens and economic elites (Gilens and Page 2014, 568). As a result, Gilens and Page made their own methodology for the study that takes into account the problems that previous literature did not address.

The methodology that Gilens and Page use in conducting their test on non-interest group theories of public opinion and public policy is based upon conducting a national survey between 1981 and 2002 on the general public, where the general public answers questions regarding policy specifics, issues of high salience where they have to respond with a pro or con answer, as well as questions regarding a proposed policy change (of which there are 1,779 proposed policy changes in the study) where the general public have to answer in a favor/oppose response (Gilens and Page 2014, 568). Members of the general public that are selected as subjects for the study also are required to give their income level in their response, as this would be used to assess responses of subjects according to different income levels (Gilens and Page 2014, 568). The income level response data and the corresponding opinion response would then be organized according percentile estimates through a quadratic logistic regression technique (Gilens and Page 2014, 568). The percentile estimates at the fiftieth income level and ninetieth income level would be to represent “median-income” and “affluent-income” in testing the two theories of majoritarian electoral democracy (which is based upon the voter of median-income) and economic-elite theory (based upon the voter of affluent-income) (Gilens and Page 2014, 568). The policy preference data made by individuals of the general public on their income and opinions regarding policy represent measures of the independent variable (Gilens and Page 2014, 568).

As for measuring the policy preferences of interest-groups, Gilens and Page utilize a modified index (a measure utilize by Baumgartner) called the “Net Interest Group Alignment” in counting the number of powerful and mass-based interest groups that favor each of the 1,779 proposed policy changes minus the number of interest groups that oppose it (Gilens and Page 2014, 569). In order to measure whether an interest group favors or opposes one of the 1,779 policy changes, Gilens and Page relied upon outside sources on information regarding interest groups to code for any responses of the proposed policy changes for the degree of favorability (Gilens and Page 2014, 569). The data extracted from the coding for responses of “somewhat” favorable/unfavorable would be given half of the magnitude of “strongly” favorable/unfavorable in order to calculate the likelihood of diminishing returns through a logarithm between the two favorable/unfavorable sides (Gilens and Page 2014, 569).

To measure the dependent variable on public policy responsiveness to the proposed policy change by the government, Gilens and Page examine on publications of the Congressional Quaterly, scholarly papers, data published by the federal government, as well as stories and information from news sources for any adoption of the proposed policy changes that were mentioned in the survey questions (Gilens and Page 2014, 569). With all the data on both the independent and dependent variables gathered, Gilens and Page then uses a single statistical model in order to estimate the independent impact of the independent variables of the policy preferences of average citizens, economic elites, and interest groups upon the dependent variable of adoption of policy change (Gilens and Page 2014, 569).

The findings of the study show that economic elites and interest groups play a substantial and significant independent impact on public policy and policy responsiveness, while average citizens play little to no independent impact on public policy and policy responsiveness (Giles and Page 2014, 572). The findings are shown on Table 3 when Gilens and Page did their initial tests of the policy preferences and alignments of average citizens, economic elites, and interest-groups (aggregating together both business-oriented and mass-based interest groups) against four multivariate models. It shows that when all of the three independent variables were tested against four multivariate models, the estimated impact for the preference of average citizens dropped towards a non-significant level after it was tested against Model 1, while there are still levels of statistical significance for its coefficients for preferences of economic elites and interest groups (Gilens and Page 2014, 572). This finding shows that the preference of average citizens does not have much of an independent impact. In terms of measuring the impact of economic elites and interest groups’ policy preferences against adoption of policy changes through the percentages data, the findings show that adoption of policy changes occur when there is a high support for policy change within the economic elites and interest groups (Gilens and Page 2014, 572). The results thus reject the theory of majoritarian electoral democracy and supports the theory of economic-elite.

In terms of measuring the impact of policy preferences within interest groups against adopted policy changes by the government, Gilens and Page made a new multivariate analysis for the net-interest-group alignment indices for both business-oriented and mass-based interest groups separately (Gilens and Page 2014, 574). The findings for this measurement show that the coefficient for business-oriented interest groups is more statistically significant than the coefficient for mass-based interest groups (Gilens and Page 2014, 575). The findings demonstrate that business-oriented interest groups have much more independent impact on public policy than mass-based interest groups. The results from the findings thus reject the theory of majoritarian pluralism and supports the theory of biased pluralism.

The findings from the Gilens and Page study demonstrates how the American political system itself is more responsive to the opinions and preferences of wealthy individuals that make up the economic elite as well as corporations and business groups over the interests of average citizens and mass-based interest groups. This shows that possession of wealth and property in the United States can actually enhances one’s own political voice on public policy and facilitate a more responsive policy from government. But for the average citizen, the unequal level playing field on the public policy responsiveness that tilts in favor of people with wealth does not do good for average citizens who want their opinions and preferences be heard and represented in the halls of American government. The study by Gilens and Page, however, lacks further research on why there is an unequal level playing field on public policy responsiveness. This question is to be answered with the next study made by Erikson.

In another study that adds to the literature on public opinion and policy responsiveness, Erikson made his own study on the impact of income inequality and information inequality upon policy responsiveness. The theoretical framework for his study depends upon the median voter theory by Downs (1958), where policy positions of two opposing election candidates that were once highly partisan fall into the moderate preference of the median voter in order to increase their own votes for the election (Erikson 2015, 13). Erikson acknowledges this theory as a way to show that elected politicians are responsive to public opinion. However, Erikson argues that the use of the median voter theory by Downs in measuring policy representation and responsiveness is not helpful as what other scholars assume, due to the reality of election outcomes that rarely converge towards strict moderate preference of the median voter as election candidates today care more about defending their own personal ideological agendas besides the other goal of winning elections (Erikson 2015, 13). Besides those reasons, Erikson also argues that the use of Downs’s theory of the median voter does not take into account the different, unequal interests of the average citizen and those of the wealthy business and professional elite, as people with different economic interests would not pivot towards the median in terms of opinion and preference on policy (Erikson 14). Erikson does not propose an outright rejection of the median voter theory paradigm in determining policy responsiveness and representation but to improve it through a new methodological study (Erikson 2015, 14).

Erikson proposes a new methodological study in finding the impact of income and information inequality on policy responsiveness. But he wants to first find if the adoption of policy changes and legislation clearly reflects public opinion. In this case, Erikson applies a modified version of Wlezien’s thermostatic model in order to clearly depict the relationship between the public’s preference for policy change (as the independent variable) and policy change by the government (as the dependent variable) (Erikson 2015, 14-15). The findings in the study show that mood becomes more liberal under Republican presidents in power and more conservative under Democratic presidents in power (Erikson 2015, 15). However, when putting the data together, Erikson also takes into account on “lag”, which is due to the checks and balances system between the legislative and executive branches of government (Erikson 2015, 15-16). The findings when taking account on the checks and balances lag show that public policy (in the increasing number of liberal laws and laws that are popular) is responsive towards public mood (Erikson 2015, 16). Erikson sees the findings coming from his own study as not accounted towards socioeconomic income, as he believes it is possible that this finding only represents the policy preferences and opinions of the wealthy elite and not the average citizen (Erikson 2015, 16). As a result, Erikson devises another study on the impact of public opinion upon public policy that takes into account differences in income among Americans and differences on information reception among Americans.

His methodology in studying income inequality and its impact on public policy looks into how Americans turn out to vote based upon their income status. Erikson wants to test the hypothesis to see if it holds true on the assumption that citizens with different income categories will likely had different levels of political influence in proportion to their voting frequency at the polls (Erikson 2015, 18). The methodology that he utilizes are mainly on the variety of survey data such as the U.S. Census’s Current Population Survey (CPS), the National Exit Poll (NEP), and the New York Times/CBS News (NYT/CBS) surveys to estimate voting participation rates based on income levels for the 2006 and 2008 elections. The findings from the survey data of CPS and NEP/NYT&CBS show that voter turnout among American is highly dependent upon income categories, as shown on Figure 4 of the study where voter turnout above 50% correlates with 75^th percentile for income in 2006 and above 75% voter turnout for above 75^th percentile for income in 2008 (Erikson 2015, 20). The findings on the strong correlation between voter turnout and income levels opens new questions on whether or not providing more information and knowledge on politics would induce greater participation among the poor, which in turn leads to greater policy representation and responsiveness.

This leads for Erikson to conduct another study within the framework of public opinion and policy responsiveness by looking into differences in access to political information and its potential correlation with income inequality. The methodology that Erikson utilizes is through using the American National Election Studies survey (ANES) from 1952 to 2004 as the data source (Erikson 2015, 20). The ANES survey asks subjects in the survey on family income as well as questions regarding political knowledge (Erikson 2015, 20-21). The findings show that around 50% and above of respondents who identify themselves as in the high-income percentile have political knowledge scores that are one standard deviation above the mean, while low-income respondents scored lower in political information (Erikson 2015, 21).

Erikson takes the study further by looking into whether information deficit among low-income Americans inhibits their response to formulate their own political opinion and preferences. In this sub-study, Erikson utilizes ANES survey data for presidential elections from 1980-2004, but puts the unit of analysis of this sub-study only on white survey respondents (Erikson 2015, 22). The findings show that for respondents that have high political knowledge, the political opinions and preferences on the issue of social welfare spending gets more coherent as respondents that identify themselves as supporters of greater social welfare spending are low-income respondents, while those who identify themselves as opponents on the issue are high-income respondents (Erikson 2015, 22). The same results are demonstrated on the comparison findings between high political knowledge and low political knowledge respondents on presidential vote choice for a Democrat, as those who are low knowledge are not coherent in their vote choice for president—regardless of their liberal/conservative beliefs on the taxing and spending issue (Erikson 2015, 23). Those with higher political knowledge are more coherent in voting for their chosen presidential candidate—with respondents that identify themselves as liberal on taxation and spending vote mainly for the Democratic candidate as president regardless of income. The findings demonstrate that one’s own position on a salient policy issue is highly dependent upon one’s own economic standing and one’s own vote choice is highly dependent on a person’s depth of political knowledge and partisan attachment (Erikson 2015, 22-23). The lack of information does greatly affect a person’s formulation of their own political opinions and preferences.

But what do these findings by Erikson relate to the general question of the impact of differences in public opinion due to income upon policy responsiveness? The findings on the impact of income differences among Americans upon their involvement on public policy does affect highly on their representation in the halls of the U.S. government as information deficit due to lack of resources to learn about politics and political information puts low-income Americans at a disadvantage in representation and responsiveness in U.S. public policy. But the other finding on the increase of liberal laws being passed in the U.S. Congress shows the paradoxical nature on the results of the study. Erikson gave his own conclusion from the study that public opinion does influence public policy in the halls of government (with a trend moving towards more liberal laws) but the impact of public opinion upon public policy is highly skewed towards Americans with more wealth (Erikson 2015, 27). This study serves as an opportunity for the U.S. to put more effort and resources in improving public education in pushing more Americans to become more informed on politics and public policy and involved in civic life in order to close the income gap as well as the information gap.

Another study on the impact of public opinion upon public policy was done by Kelly and Witko on examining the role of subnational governments such as states and how they are highly responsive to public opinion and the issue of economic inequality. Kelly and Witko acknowledges past literature on the theoretical debate on whether or not states have the ability to influence and shape the outcome of economic distribution, with different findings by scholars that show states have a limited ability to influence redistribution while other findings demonstrate that states do have a strong ability to influence post-redistribution income as well (Freund and Morris 2005; Langer 2001). Kelly and Witko also acknowledges on past literature that show that the federal government itself has made a strong influence in allowing income inequality to happen, while the state government itself has made efforts to reduce income inequality (Kelly and Witko 2012, 414-415). However, Kelly and Witko points out that the literature does not address on the “shared” ability of both the federal and state governments to shape inequality and distribution as studies by past scholars on the issue focus their unit of analysis separately on the federal government and the state governments (Kelly and Witko 2012, 415). As a result, Kelly and Witko proposes a new methodology on the study on the role of government responsiveness on income inequality that combines both the federal and state levels of governments as a single unit of analysis (Kelly and Witko 2012, 415).

Their methodology utilizes a modified version of the power resources theory (PRT) in looking to the effects of both the federal and state governments on their impacts on tackling economic inequality (Kelly and Witko 2012, 416). The power resources theory focuses on the different interest preferences between the working, middle, and wealthy elite classes, assuming that the interests of the working and middle classes are in favor of more egalitarian income redistribution (Kelly and Witko 2012, 416). But instead of focusing on class, Kelly and Witko focuses mainly on income as the unit of analysis measure for the dependent variable (Kelly and Witko 2012, 416). The dependent variable to be measured is on household pre-redistribution inequality and household post-redistribution inequality, measured by collecting household income data from the U.S. Census Bureau in the Annual Social and Economic Supplement (ASES) (Kelly and Witko 2012, 420). Pre-distribution income consists mainly of sources of income earning minus income earned from government programs while post-distribution includes earnings from government programs (Kelly and Witko 2012, 420). The independent variable to be measured is on party control of the presidency, partisan composition in the U.S. Congress, and variations of national policies being enacted (National Policy Liberalism) for the federal level, and union density (proportion of nonagricultural workforce represented by a union), dominance of left-leaning politicians in the state government (left government power), and changes in the state minimum wage for the state level (Kelly and Witko 2012, 420). All of the data that represent independent variables are measured from 1976 to 1994 and from 1995-2006 (Kelly and Witko 2012, 424). The state unemployment rate and gross state product from manufacturing are to be accounted as control variables (Kelly and Witko 2012, 420).

The findings in their study after conducting a multivariate analysis on all the data gathered demonstrate that both the federal and state governments influence both pre-redistribution and post-redistribution household inequality, with specific results show that inequality itself is lowered when there is the presence of a strong state union as well as elected left-leaning state politicians in power (Kelly and Witko 2012, 424). Other results in their findings also show that state governments that are governed by left-leaning politicians after 1994 have seen inequality being lowered while the federal government after 1994 have gradually shifted their priorities on inequality in favor of Americans that earn higher income (Kelly and Wikto 2012, 424).

The results from the Kelly and Witko study show that government can be responsive towards the opinions and preferences of the public mainly in the state level, when left-leaning politicians in the state level has power, as well as the strong presence of unions in the state-level that can influence public policy and representation. The implications from this study shows how politics can be more responsive to average citizens if it is done in the local and state level and if elected politicians in the state and local levels are willing to tackle the issue of economic inequality—a main issue that concerns average citizens in America.

In looking in all the data and findings made by various scholars on the study of public opinion and preferences and its impact on public policy, it is with conclusion that there are suggestions that can be proposed in order to improve the responsiveness of the American political system. One suggestion that should be discussed is to improve the education system in the state and local level (especially in poorer neighborhoods) in order to promote Americans to become more involved in civic life and informed in local, state, and national politics regardless of income background. Another suggestion is to elect more left-leaning politicians into the U.S. Congress and push for further discussion and debate on income distribution as well as other salient issues that are important for average citizens such as on taxation, immigration, health care that are outside of the usual discussion and debate such as foreign policy, national security, and foreign trade and commerce that are more in tune with the professional and economic elites and elite-based interest groups. This can be done by getting more mass-based interest groups that identify themselves with issues supported by average citizens to become more involved in lobbying members of the U.S. Congress (especially on Democratic congressional representatives).