Do Voters Only See What They Want to See?: The Relationship Between Political Ideology and Research Source Choice

Anthony Torretti

S. Nichols

PSCI 301

May 16th, 2012

Do Voters Only See What They Want to See?:

The Relationship Between Political Ideology and Research Source Choice

I. Introduction

In a democracy, members of the general public are tasked with making at least some decisions

that can change the fate of a nation. Ideally, a voter would do her best to stay informed on issues of

importance. This can be a difficult task in today's media environment. For television viewers, a news

broadcast that fits their ideology is just a few channels away. Internet users have an even wider array of

sources to choose their political reality from. With such diverse perspectives and biases populating the

information age, this raises a somewhat troubling question: does the average voter seek clarity on

confusing political issues, or does the voter settle for whatever position is expressed by the rhetoric he

is most comfortable with? How influential is this rhetoric in leading a biased voter into forming her

opinion? Such questions are of vital importance when trying to understand modern democracies. Critics

of mass media often lament about how corporate news media has polarized the electorate; but if we

look at this polarized news media as a commodity, shouldn't there also be equal concern expressed over

the demand for it? The purpose of this research, ultimately, is to verify the existence of consumer-side

bias in the consumption of political information, and how that bias shapes information consumption

and the opinions people form from that information.

While there have been a number of studies on the role of selective exposure in the context of

political information, there are a few factors that differentiate this piece of research from those that

precede it. For one, the studies that precede this have focused on what role selective exposure plays in

media choice rather than trying to verify its existence. (Arceneaux et. al, 2012) (Robert & Dennis,

2005) This is a somewhat problematic, as it has been shown in other media research that the influence

of selective exposure is not absolute. (Donsbach, 2009) Thus, this research and experimental model

will focus on verifying the role of selective exposure, and measuring what extent it plays, in political

media choice.

II. Theory and Hypothesis

This notion that the individual is drawn towards reinforcements of his views while avoiding

contradictions of those views is known as cognitive dissonance, and dates back to research by Festinger

in 1957. (Westerwick & Klienman, 2011) While this theory started out as a general psychological

concept, it was gradually adopted by scholars of media and communications. Today, it is more

commonly known as selective exposure theory and serves as the theoretical basis of this research.

(Donsbach, 2009)

According to the selective exposure theory, a media consumer only exposes herself to messages

that she is comfortable with. Everyone has their way of perceiving reality, and cognitive

dissonance/selective exposure theories suggest that the individual seeks to validate this reality while

avoiding anything that invalidates it. In the context of elections and media, a voter with a specific

ideology will seek media and information that validates his political philosophy, while avoiding media

information that invalidates it. Thus, if such a voter were presented with a political issue and given the

choice to research a political issue from sources of varying ideology, her primary source choice will be

one that best reinforces her ideology. Furthermore, if she has no previous concept of the issue

presented, she will ultimately abide by whatever policy position is rhetorically framed in her ideology.

Thus, the hypotheses of this research are two-fold and are as follows.

In a comparison of voters prompted to research a political topic:

- Those with a more intense political ideology will choose more intensely partisan

sources than those with more ambivalent political views

- Those with a more intense political ideology are more likely to abide by the policy

position framed within their own ideological rhetoric when faced with an unknown

issue.

III. Data and Method

The data will be collected via a controlled experiment process of my own design. The process

uses a digital software interface, issued via a laptop or tablet device, to present the participant with a

policy question, followed by a series of links to differing perspectives on the issue. After the participant

has the opportunity to read one or more of these sources, the survey asks for the participant's opinion

on the issue at hand. Once that opinion is inputted, the survey presents a series of general opinion

questions in order to pinpoint where the participant falls on the politico-ideological scale. The variables

are recorded unobtrusively by the software, and include the participant's source choices, the time spent

on each source, the participant's opinion on the presented policy issue, and the participant's overall

political ideology.

When first presented with the survey, the participant is sent through a series of introduction

screens. (Figure A-B) These introduction screens serve to introduce the political issue, and set up the

tone under which the survey is taken. The introduction screen purports that the survey is being

conducted by Camdien Policy Institute, claiming that the data it gathers is then given to politicians and

campaign managers. This bit of misinformation helps give the survey a sense of importance,

encourcing the participant to give sincere, thoughtful answers in the hope of having actual influence on

the views of politicians. The participant is also introduced to the issue of “reapportioning nontangible

local assets into a unitary city fund”. The wordy and ambiguous nature of the terminology is deliberate.

By testing participants with a fictitious and ambiguous policy matter, we can make our observations

without the influences of the participant's personal experiences and perspectives on the policy subject.

Finally, as the participant is brought through the introduction screens, she is randomly assigned to one

of three groups.

After the introduction screens, the participant will be brought to a screen with a list of sources.

What sources he sees depends on what group he was placed in. Group 1 is the control group.

Participants in the control group are given only one source link: a basic summary of facts on the policy

issue. (Figure C) Group 2 is given five links, counting the summary of facts. These other sources range

in support and opposition of the policy measure, ranging from an extreme opposition of 1 and an

extreme support of 5. (Figure D) Political rhetoric of each source also ranges on a 1-5 scale, with 1

being the most fiscally moralist, and 5 being the most fiscally individualist. In mainstream U.S.

Politics, fiscal moralism is typically associated with liberalism, while fiscal individualism is associated

with conservatism. This distinction is important, because this allows us to control on actual ideology

rather than group identity, which is often confused with ones ideology when categorizing based on

traditional party structure. Participants in Group 3 are given a list similar to Group 2, but with one

variation: the opinions expressed by the differing sides of the fiscal-ideological spectrum are reversed.

(Figure E) In Group 1, the relationship between support and fiscal ideology is negative. As support

rises on the 1-5 scale, fiscal ideology descends in the moralist direction on the 1-5 scale. In Group 2,

the relationship is positive. In short, the moralist and individualist stances on the issue are flipped

between the two groups. This helps differentiate between opinions formed by respondents who've

attempted to analyze the issue, and respondents who simply fall in line with the rhetoric that fits their

ideology. In each group, the participant's source choices and time spent on each source is tracked.

After the participant is finished with the source list, the first survey question is posed as follows

(Figure E):

Based on what you've learned about the measure, how would you rate your support for

"funding reapportionment" policies such as CB 62A?

1 – Strongly Oppose, 2- Oppose, 3-Neutral/Unsure, 4-Support, 5-Strongly Support

This answer is then recorded, once again on a 1 to 5 scale, and the participant is presented with the

following survey questions (Figure F):

- Our national and state governments are faced with numerous fiscal problems at

this point. Many political leaders advocate large cuts to the budget whereas other

political leaders want to increase revenues. On a 1-5 scale, where would you place

yourself in that debate?

- A lot of political dialogue these days has revolved around what role corporations should

play in our economic recovery. The main conflict tends to be between those who believe

corporations being prosperous will help the economy recover, and those who believe that

corporations being responsible will be of better help. On a 1- 5 scale, which argument do

you more closely identify with?

-What fiscal role do you think government should play? Do you tend to see government as

the shaper of an ideal society or the provider for only the most basic societal needs? Rate

yourself on a 1 to 5 scale.

Each of these questions have five responses rated on a 1-5 scale. Responses scored with 1 are responses

that would be affiliated with fiscal moralism; whereas 5 responses would be affiliated with fiscal

individualism. The software then takes these responses and averages them, assigning the respondent an

overall fiscal-ideological identity for use in analysis.

IV. Results

When comparing the two variables of ideology and first source choice, we use gamma since

both the independent and dependent variables are on a 1-5 scale, and are therefore ordinal.

TABLE 1: Fiscal Ideology VS First Source Choice

# of Valid Cases (out of 158) Gamma Significance

69 0.67 0

Above, we see that the gamma yields a considerably positive relationship, and with a

significance of zero, the null hypothesis can be easily rejected. In essence, those with a fiscal-

ideological category of 1 were highly likely to choose the source that had an ideological slant of 1, and

so on. But is the high gamma caused by those with intense ideologies? Or is it all the moderate

individuals choosing the more neutral options that boosted this correlation? By establishing the base-

line source choice “3” as a missing variable, we can see how the outliers affected the gama.

TABLE 2: Fiscal Ideology VS First Source Choice (Neutral Source as Missing Variable)

# of Valid Cases (out of 158) Gamma Significance

16 0.7 0

Here, we see that the elimination of the neutral source actually increases the gamma, indicating

an even stronger correlation that exists among the outliers of fiscal ideology. Overall, the data present

seems to support the first hypothesis.

One potential weakness in the data present is the number of valid cases, or the number of

individuals who chose to read any source. While some lacking in participation was expected, the extent

to which participants chose to not read any source made a good amount of the data unusable in testing

the first hypothesis. There are a few explanations as to why so many declined. For one, choosing a

source was optional, and the importance of reading any of the source links was not emphasized.

Another possible explanation is that early participants were given a snack-size bag of chips as an

incentive to participate. Participants encouraged to participate in this way are, perhaps, more likely to

give minimal input into a survey.

Table: 3 Number of Sources Read VS Chip or Non-Chip Recipient (Chip Recipient = 1, Non-

Chip=0)

# of Valid Cases (out of 158) Gamma Significance

158 -0.63 0

The above chart indicates a highly negative relationship between chip reception and number of

sources read. Meaning that those who received chips were significantly more likely to give minimal

input than those not given chips.

The second hypothesis, fortunately, did not suffer from the same pitfalls, mainly because a

participant didn't need to actually read a source in order to be subject to the effects of narrative framing.

In the table below, we see how drastically different the opinions are of the individuals in the different

control groups.

Table 4: Fiscal Idealolgy VS Support/Opposition Among Experimental Groups

GROUP GAMMA SIGNIFICANCE CASES

Control -0.26 0.18 69

Moralist

Support/Individ Oppose

-0.96 0.02 38

Individ.

Support/Moralist

Oppose

0.9 0.05 37

While there is a negative relationship even within the control group, it is considerably low in

magnitude and the null hypothesis is stronger in that group than it is in either of the other groups. The

gamma is extremely strong in the other groups, in completely opposite directions. In short, participants

with the same ideology rating, but exposed to different headlines, gave completely opposing answers

when asked about support for the policy measure, and thereby lending extreme validity to the second

hypothesis.

V. Future Considerations

While the best steps were taken to rule out experimental error, the process was not without its

missteps upon reflection. For one, the survey did not put enough emphasis on the importance of making

at least one source selection. Had more participants been encouraged to make selections, the sample

size would have been considerably larger. The lack of diversity in the sample size may have also

limited the accuracy of the results. This survey was conducted primarily on college campuses at

California State University San Marcos, and Palomar College. Some attempts were made to conduct

the survey at locations such as Wal-Mart, but very few expressed interest in participation. As a result,

the experimental sample can be said to, at best, represent college-aged individuals but not the

population at large.

In addition to the missteps above, there are also improvements that could be made in the process

to ensure more accurate results in the future. For one, the use of a 1-7 scale as opposed to a 1-5 scale

would better capture the diversity of individual political ideology, and the use of a more complex series

of questions to determine the participant's fiscal ideology would better reveal the nuance that exists

among the fiscal views of participants. Furthermore, while attempts were made to create a sense of

importance in how the participant answered, it proved to be ineffective in motivating participants to

select sources and contemplate the issue. All in all, while the data that the experiment yielded managed

to support both hypotheses, we are a long way from being able to claim anything conclusive.

VI. Conclusion

Overall, the data yielded by the experiment managed to reinforce both hypotheses. Firstly, the

relationship between strength of ideology and source choice was shown to be positive when comparing

the two variables, but didn't draw from a large enough sample to reveal anything conclusive. On the

other hand, the relationship between ideology and opinion on the surveyed policy matter was affected

drastically by what survey group each participant was placed in. The nature of the experimental model

allows us to draw a couple of inferences from this result: Firstly, that participants were more influenced

by the rhetoric or headlines that appealed to their fiscal ideology, rather than any actual judgments the

participant made on the policy matter. If the latter were true, participants would've attempted to gain as

much information possible about the policy matter, and then make a judgment based on their policy

goals as dictated by their ideology. Instead, we see participants giving wildly diverging opinions based

on little to no reading.

One major issue the emerged was the lack of source-selection on the participant's end. While

there were a number of suggested causes, one cause that was able to be reinforced by the data was the

existence of increased satisficing caused by incentivization. It is probably that giving individuals an

incentive, and withholding that incentive until the experiment's completion, encourages participants to

give minimal input and participation into the experimental model.

Ultimately, while the hypotheses were reinforced by the data, the research presented here only

represents a remedial step in delving into the relationship between political or fiscal ideology and

selective exposure in news media. As the methodology is developed and refined, it may be possible to

find a definite answer to the question in the future.

Works Cited

Arceneaux, K., Johnson, M., & Murphy, C. (2012). Polarized Political Communication, Oppositional

Media Hostility, and Selective Exposure. Journal of Politics, 74(1), 174-186.

doi:10.1017/S002238161100123X

Donsbach, W. (2009). Cognitive dissonance theory—Roller coaster career. How communication

research adapted the theory of cognitive dissonance. In T. Hartmann (Ed.), Media choice: A

theoretical and empirical overview. 128-149. London: Routledge.

Robert, L. P., & Dennis, A. R. (2005). Paradox of Richness: A Cognitive Model of Media Choice.

IEEE Transactions On Professional Communication, 48(1), 10-21.

doi:10.1109/TPC.2003.843292

Westerwick, S., & Kleinman, S. (2011, March 16). Preelection Selective Exposure: Confirmation

Bias Versus Informational Utility. Communication Research, 39, 170-194.

doi:10.1177/0093650211400597

Supplements

Figure A and B: Introduction Screens

Figure C: Control Group Source List

Figure D: Group 1 Source List (Moralist Support/Individualist Opposition) Link Order Randomized

Figure E: Group 2 Source List (Moralist Oppose/Invidivualist Support) Link Order Randomized

Figure F: Baseline Source (3) “Summary of Findings”

Figure G: Extreme Moralist (1) Sources. TOP- Group 1 Source, BOTTOM- Group 2 Source

Figure H: Moderate Moralist (2) Sources. TOP- Group 1 Source, BOTTOM, Group 2 Source

Figure I: Moderate Individualist (4) Sources. TOP- Group 1 Source, BOTTOM – Group 2 Source

Figure J: Extreme Individualist (5) Sources. TOP- Group 1 Source, BOTTOM – Group 2 Source

Figure K: Survey Question on Proposed Policy Measure “Funding Reapportionment”

Figure L: Survey “Exit Polling” Used to Assess Fiscal Ideology