Public Opinion Exam

direct democracy

citizen participation in public affairs and commitment to democratic values above the values of the group

rational

not swayed by personal or group loyalties, but actually positive for society as a whole

democratic values

majority rule, universal participation, civil equality, rights and liberties

classical theory of democracy

an indirect democracy that has the same requirements of citizens but elects people are representatives

requirements of citizens

active in public affairs, quality participation, committed to democratic values, protect against undemocratic leaders

“least liked group” measure

asks what a respondent’s least favorite group of people is and then asks which rights they would extend to this group; measure of commitment to democratic values

contemporary theory of democracy

claims that as long as political elites are committed to democracy, it does not matter whether citizens are super committed

participatory theory of democracy

says citizens need to support democracy both in theory and in action, and must do more than only vote

Dewey v Truman 1948

prime example of why many Americans lost trust in polls

Literary Digest

correctly predicted every presidential race until 1936 when they got it massively wrong, eventually leading to them shutting down

scientific surveys

representative, high response rates, objective question wording, reliable interview mode

sampling

selection of a subset of respondents from the broader population

population

broader group we are trying to test

sample size

usually 1,000-2,000

pseudopolls

polls in which individuals self-select in, leading to unrepresentative responses

quota sampling

matching the percent of a survey sample to the know percent of the same demographic in the overall population; quota subset may still not be representative

simple random sample

each person has the same chance of being selected; best way to ensure representativeness; usually not feasible

probability sampling

every individual in the population has a specific, known probability of being selected into the sample

stratified sampling

population is divided into subsets or strata and then randomly selected from within the strata

cluster sampling

multistage random sampling; multiple in-person interviews within the same geographic areas; ANES

sampling error/margin of error

± 3 percentage points; decreases as sample size increases

95% confidence interval

95% confident that the true mean falls within this range

non-probability samples

do not know probability that any individual is part of the survey sample, so you cannot compute sampling error or generalize to a larger population

weighing

used to account for bias in both probability and non-probability samples; makes the sample’s sociodemographic characteristics more accurately reflect the population’s characteristics

item non-response

can bias results if the people who answered a question differed in some way from those who chose not to answer

factors affecting election poll accuracy

timing, treatment of undecided voters, identification of likely voters, lying

undecided voters

often break strongly for the challenger

Gallup’s index of likelihood of voting

looks at voter registration, past voting, gender, age, social group, etc

Bradley Effect

people responding to a question in the way they want to be perceived rather than how they actually feel; Clinton in 2016

2016 polls

late-deciding voters, “Shy” Trump supporters, and not enough non-college-educated white people affected the outcome

survey mode

method of taking/collecting the survey

shirking

exiting a survey in the middle of it

why does mode matter?

social desirability, response rate, cheating, shirking, representativeness, cost, speed, sample size

attrition

amount of people who shirk the survey

advantages of in-person surveys

follow-up questions, higher response rates, in-depth answers, complex questions, less cheating

disadvantages of in-person surveys

social desirability bias, interviewer bias, costly and long

advantages of mail surveys

cost-effective, not geographically constrained, anonymous, less social desirability bias, no interviewer bias

disadvantages of mail surveys

low response rates, time consuming, coverage error

coverage error

error stemming from not reaching the right people due to variations in where people live and how much access they have to their mail/phone

advantages of phone surveys

cheaper than mail and in-person, efficient, more complex questions than mail surveys

disadvantages of phone surveys

low response rates, coverage bias, social desirability bias, can be costly depending on length

advantages of internet surveys

least expensive, quick, low social desirability bias, easy

disadvantages of internet surveys

cheating, not representative, coverage bias, shirking, low attention

Westwood and AI

created an AI model that could answer surveys according to a specific demographic and it was completely undetectable

Theory of the psychology of the survey response

Tourangeau’s four components to the cognitive process of responding to a survey question; comprehension, retrieval, judgment, selection

comprehension of the question

how do respondents understand and interpret a questions? will they answer even if they do not know what it means?

retrieval of information

tendency of survey respondents to satisfice and minimize the amount of cognitive energy they spend answering a question; refer to what is at the top of their mind

judgment

respondents average across the considerations that happen to be at the top of their head

selection of an answer

respondents choose the response alternative that best fits their judgment; requires translation from judgment to actual answer

response editing

a chosen response may not reflect the true judgment made

double-barreled questions

questions that ask respondents to answer two questions at once or react to two stimuli

things to avoid in writing survey questions

double-barreled questions, overly complex language, vague questions, double negatives, loaded questions

retrieval and judgment bias

top of mind considerations can be influenced by the survey itself, leading to biased responses

authoritative sources

giving people information from what they think are authoritative sources makes people more likely to agree with the source

response acquiescence

tendency of people to agree with statements

battery of questions

asks for reactions to multiple statements, worded in opposite directions, and then reverse coded together to get an actual scale of how people feel

response alternatives

different alternatives produce different patterns of opinion and affect the degree to which questions reflect attitudes vs nonattitudes

nonattitudes

random opinions made up on the fly in order to answer the question, even if it is something they have never thought about; people may not want to admit they have not thought about it

Likert Scale

scale that measures both direction and intensity of opinion, from agree strongly to disagree strongly

branching questions

first asks direction, then uses follow-up questions to measure intensity

semantic differential questions

asks people to place themselves on a 1-7 scale where each point has an explanation associated with it; good for concrete policy stuff or things with measurable differences between points

feeling thermometers

rate political figures, parties, etc., on a 1-100 scale where 50 is neutral; better for more squishy stuff like feelings toward political figures

what should people know about politics?

institutions and processes, policy issues, major political questions, prevailing socioeconomic conditions, people

importance of political knowledge

more likely to participate, more political power, elections can better convey proper information about public preferences

reasons to expect increases in knowledge

trends in education, trends in workforce, technological advancements, larger scope of federal government

reasons to expect decreases in knowledge

less civic education, less unionization, negative coverage, less in-depth news, information overload, misinformation

discursive sophistication

being able to talk about the nuances of an answer; the how and the why

factual knowledge

straight up knowing the answer to a question; studies show that women know less on a factual level, but that men are more likely to guess

differences in knowledge by demographic

white people have more traditional knowledge but black people have more knowledge on carceral violence

effects of political knowledge

some trends show higher levels of political knowledge correlate with negative feelings toward the other party and condoning acts of political violence; other studies show increased tolerance as knowledge increases

Lau and Redlawsk

showed that people usually do vote correctly, based on predictions of what their vote should be, even if they did not have all the information

opinions

manifestation of an attitude; may reflect both attitudes and nonattitudes; Erikson and Tedin refer to this as the depth of opinion holding

attitude

deeply formed idea, stemming from values or ideologies

ideological constraint

attitudes on various issues are consistent with each other

ideological conceptualization

thinking about politics in ideological terms

components of ideological sophistication

constraint and conceptualization

stability of opinions

do opinions remain the same at different points in time or under different circumstances? panel surveys can test this

Phillip Converse and response instability

believes it stems from nonattitudes and that people’s opinions simply reflect their top of head considerations rather than real attitudes

Chris Achen and response instability

believes it is caused by measurement error caused by imperfect response options and questions; fault of the survey rather than of the respondent

simple measurement error correction

correct for response instability by summing a respondent’s views on multiple issues

Zaller and response instability

believes it is caused by ambivalence so people say whatever they want to in the moment by averaging across the considerations at the top of their head; not imprecise or mixed feelings

Freeder, Lenz, and Turney What goes with what?

found that greater knowledge of what goes with what is associated with greater attitude stability in panels, when the person agrees with their party; discuss how many people simply follow their party

levels of conceptualziation

ideologue, group benefits, nature of the times, no issue content

ideologues

can talk about why they like someone or something in abstract ideological terms ~20%

group benefits

do not use ideological language but can identify policy issues that fall on one side or another and affect certain groups ~30%

nature of the times

thinking about what is going on now ~28%

no issue content

no real reason for voting either way ~22%

Phillip Converse’s Black and White Model

some people are ideologically sophisticated, but most are completely unsophisticated with random survey responses and no real attitudes

Sides and Tesler 2016

examined how a not very conservative candidate like Trump won the Republican nomination - people who identify as conservative are not actually as conservative as we might think; they are not ideologues

Stimson and Ellis

Americans are symbolically conservative but operationally liberal

issue publics

Americans tend to be issue specialists rather than generalists; have real attitudes on a couple issues rather than all issues

creating an issue public

hypothesis that the public holds latent concerns that connect to policy issues, but they do not realize it until something happens to connect them to it

Piccioto 2024, Schaffner and Soler 2024

highlights importance of weighting education in polls going forward

satisficing

aiming for a satisfactory answer rather than the best one

political trust

confidence that authorities will observe the rules of the game and serve the general interest; some distrust is good for motivating the public to stay interested

why is political trust lower?

media coverage, negative attacks on the government from people in the government themselves

redistributive policies

take money from people and give it to others

distributive policies

take money from people including the beneficiaries

benchmark surveys

collect standard information going into a campaign on electoral strength

trial heat surveys

Question or series of questions within a survey that typically asks who a person would vote for if they were voting today