91 terms

Public Opinion: Midterm

Reasons to Study Public Opinion
1. Policy (in a democracy) should rest on public opinion
2. Respect for public opinion is a safeguard against demagoguery/dictatorship
3. Public opinion provides clues about culture
4. Public opinion, at times, must be mobilized
5. Public opinion dictates the bounds of US foreign policy
Definitions of Public Opinion
1. Public opinion is an aggregation of individual opinions
2. Public opinion is a reflection of majority beliefs
3. Public opinion is found in the clash of group interests
4. Public opinion is media and elite opinion
5. Public opinion is fiction
Direction in measuring public opinion
for or against an issue/policy
how strongly do you feel about an issue
how strong are your opinions over time
information context
how much information needed on an issue to make a rational stand
content analysis
ex. analyzing mass media, as it reflects public opinion
frequency, terminology, tone
focus groups
defined topic, open ended questions, loose discussion in a non-threatening environment
advantages of a focus group
multiple actors reduces the bias of the interviewer
can reveal process of opinion formation
social nature of opinion (react to each other)
allow respondents to use own terminology
3 stages/types of focus groups
1. discovery
2. development
3. definition
open ended, lose questions
good for topics with unknown opinion
more intervening/focused
moderator asks more focused/guided questions
narrow discussion: Might focus on alternative issue frames, question wording, response categories
Goal: Match wording to subject's language / terminology
Goal: Identify potentially confusing terms
Goal: Increase validity of questions
Public Opinion Polls
Cross-Sectional Survey
survey of people at one point or time
Method of Analysis #1: Compare responses across groups
-Example: Relationship between Income and Redistribution Preferences
-Result: People with higher incomes tend to favor lower taxes, and lower redistribution
Method #2: ask same questions at difference times to see how different time periods affect opinion
-example: asking opinion on President during different economic conditions
Panel Surveys
Survey conducted among the same population at multiple points in time
Advantage of Panel Survey
allows for better estimation of changes over time (compares same people, removing wonder of whether groups are comparable or not like in cross-sectional ones)
allows for individual level analyses (ex. does change in income change opinion of president)
Disadvantages of Panel Survey
people may drop out of study over time
Exit Poles
- Survey conducted outside of polling site
- Estimates trends in electoral opinions, vote choice
- This is how TV Networks announce electoral outcomes before official vote counts
Deliberative Polls
Survey combined with active discussion on topic
Pseudo Polls
polls that are not intended to estimate opinion
Push Poll (pseudo poll)
polls used by candidates or interest groups to promote their own agenda
ex. questions that portray an opponent in a negative light
Advocacy Polls (pseudo poll)
polls intended to misrepresent public opinion
fundraising/selling under guise of research
group of people to survey is about
people that a selected from population to survey
non-probability sample (non-random sample)
in depth interviews with select individuals
not representative of population
probability sampling (random sampling)
more representative of population
random number generator
gathering survey data by phone
- Random digit dialing
- Fast, cheap, less threatening to respondents
- Still allows for personal connection between interviewer and respondent
gathering survey date in person
- Expensive, Time Consuming
- Can allow for rapport to develop between interviewer and respondent
- Increased risk of interviewer effects
gathering survey data over the internet
Growing field
How to attract random samples through the internet?
gathering survey date through the mail
- Random addresses
- Less threatening to respondents
- Less oversight / monitoring of interviewer
- Low response rate - threatens generalizability of the sample
sampling error
Error resulting when we try to generalize from sample to full population
confidence level
Typically displays range of opinion within a 95% confidence interval
true population value
actual opinion of entire population, almost impossible to get
relationship between sample size and sample error
larger sample size means smaller sample error
unit non-response
entire observation missing from sample
causes of unit non-response
caused by inability to contact respondents, responded but unable to contact them
Most likely to respond:
- Young and Old (Not Middle Aged)
- Poor people
- Low density housing areas
How to reduce unit non-response
- Advance warning of survey
- Offer incentives for participation
- Trained interviewers
Consequences of unit non-response
- High UNR reduces sample size, increases expense
- High UNR can lead to non-representative sample
- The can cause biased estimates
item non-response
Some data present, but other data is missing for a given observation... some questions might not be answered, etc.
causes of item non-response
Respondent Characteristics
Exposure to topic
Interest in survey

Questionnaire Wording Design Issues
Sensitive topics (Examples: Income, Prejudice, Crime, Social Policy)
Confusing wording
Cognitive Difficulty of Question

Interviewer Behavior / Skill
consequences of item non-response
Skewed results
Reduced sample size
how to reduce item non-response
Increase privacy of response
Embed socially acceptable response categories for sensitive items
when a respondent declares an opinion when they really dont have an opinion
problems with non-attitudes
Reduces accuracy
Distorts estimate of public opinion
how to reduce non-attitudes
provide information before hand to make subject able to make a more informed opinion
have filter questions: allows them to admit they have no opinion
benefit of "don't know options"
reduces skewed data
types of question wording
double negative
double barreled (two issues)
argumentative/leading (framed to guide towards certain opinion)
open-ended questions v. close-ended
close-ended is more efficient, easier to code
concerns of close-ended
does it allow sufficient response options
adding an item might artificially raise salience
does adding "other" option help?
no... people rarely select it
full list v. branch
full list: are you a rep, dem, ind, etc.
branch: are you a rep... stron rep? weak rep?
full label v. non-label
non-label: pick on a scale of 0 (extreme liberal) to 10 (extreme conservative)
full label: 0 (extreme liberal), 1 (very liberal), 2(strong liberal) 3.... 10 (extreme conservative)
advantages of labels
labels provide context
disadvantages of labels
difficult to analyze
advantages of not offering a middle
forces people to pick a size
disadvantages of offering a middle
more non-attitudes
hide controversial opinions
do people tend to chose agree or disagree more often?
how to fix "agree" selection bias
alternate positive and negative framing
primary and recency effects
primary, subjects tend to chose first option when taking a paper survey
recency, subjects tend to chose last option when taking a phone survey
subgroup analysis/comparison
splitting a sample into subgroups to compare
political socialization
the manner in which we learn about and develop political opinion
early childhood political views (elementary)
pride in country
benevolent view of leaders
middle school political views
growing awareness... still generally positive
adolescent political opinoin
start to gain specific political opinions... mainly influenced by parents
life cycle effects
individual attitudes are affected by age
more impressionable during younger years, more stable when older
ex. more conservative when older, etc?
generational effect
political event/context effects entire generation, generally in impressionable years
ex. WWII, great depression, etc
period effect
salient features of political period influence/effect a majority of people regardless of age
what's Zaller's jam?
formation of political opinions
political predispositions
Stable, individual-level traits that regulate the acceptance or non-acceptance of the political communications the person receives
filter between info received and accepted
Any reason that might induce an individual to decide a political issue one way or the other
ex. opinion on war affected by views on economy, security, foreign policy, violence, etc
two types of political messages
persuasive arguments
cueing messages
persuasive arguments
arguments or images that try to persuade someone to take position of point of view
once accepted, persuasive arguments become considerations
cueing messages
Consist of "contextual information" about the ideological or partisan implications of a persuasive message
Cue helps determine which political messages to accept (as new considerations) and which to reject
what are Zaller's four axioms?
reception axiom
listening too opinions, sides, etc
taking in info
resistance axiom
resisting opposing opinions
more aware, more likely to resist
accessibility axiom
ability to think of recent consideration
salience is time dependent
response axiom
individuals respond to surveys by considerations that are most salient to them
therefor responses are not always long standing or well developed
The Zaller RAS Model
Zaller on Survey Responses
"True Attitude" does not exist
survey response is rare, not how people think
having mixed, contradictory feelings,
explaining response instability
- Information flow on both sides of the issue
- Individuals accept considerations on both sides
- Most salient considerations can vary over time
- Therefore, survey response will vary over time
question order effects
again, effects the ambivalent more
question wording effects
more aware =
more extreme responses/opinions
mainstream effect information environment
All communications take same side on issue
No cueing messages to alert people to inconsistency with political predispositions
mainstream effect outcome
all messages will be received and accepted because there are no cues provided to justify rejection
ex. most educated people take same side as elites
polarizing effect information environment
Elites disagree along clear partisan / ideological lines
Even flow of communications from both sides
Both sides provide arguments and cueing information
polarizing effect outcome
aware people will receive all messages but will be more selective with accepting
unaware people will receive all messages but me inconsistent with accepting them