Lecture 17: Public Opinion

Foucault’s Governmentality

• Since the 18th century, "governmentality" has been the dominant mode of power in Europe

• Population control takes place through

  • Institutions and the state

  • Norms

  • Self-regulation

Norms

The ideas, accepted standards, and social categories that shape how we think

Self-regulation

How individuals monitor and discipline their own behavior

Power

The ability to prevail in struggles over resources, rights, or privileges

First Dimension

Formal Decision Making - Getting someone to do something that they otherwise would not do

Second Dimension

Setting rules of the game, deciding what can be considered (ex: agenda setting)

Third Dimension

Preference Shaping

Public Opinion

Aggregation of individuals’ political views

Sampling

Polling a subset of people drawn from the population of interest

Alexander Hamilton

• Distrusted the masses

• Public opinion thought to be easily corrupted

• Public was distanced from leaders through indirect elections

Thomas Jefferson (1787)

Independent farmers could make sound political decisions if they had access to education and information (universal human reason)

Alexis de Tocqueville (1848)

Public opinion in democracies could become oppressive, with people conforming to majority views rather than thinking independently

Walter Lippman (1922)

Most citizens were too busy, uninformed, and susceptible to manipulation to truly govern themselves

Straw Polls

• An unofficial ballot designed to measure public opinion

• Considered bad science/unreliable

• Current call-in polls and some internet polls are examples of straw poll methods

Straw Polls in 1800s

• 1800s public opinion determined through straw polls

• First used in 1824, showed Andrew Jackson would win 64% of vote over John Quincy Adams and Henry Clay-Jackson did win the popular vote

1936 Presidential Election

• Sent out 10 million mail-in ballots to car registrants and people with landlines, received 2.3 million responses

• Failed because the sample was not representative - screened out the poor, and poorer people voted for FDR

Selection Bias

Some people have a better/worse chance of being polled for a specific and consistent reason

Non-random sampling (Representative sample)

Used criteria to construct the sample so that it would resemble the population in certain ways (gender, race, education, etc.)

1948 Election

• Gallop’s quota sampling failed in 1948 (Truman vs Dewey)

• Quotas missed key criteria that affected voting

• Demographic data from 1940 census, didn’t reflect post WWII urbanization

• People could change their minds in the time before the election

Interviewer Bias

Chose more accessible, cooperative people who were disproportionately Republican

Random sampling

• Drawing names from a hat—each name has a 1/N probability of selection

• You can calculate how representative your sample likely is

Non-random sampling

• Surveying people at a mall—you have no idea what probability any given citizen had of being there and willing to respond

• You can't mathematically justify generalizing to the broader population.

Central Limit Theorem

When you take repeated random samples from a population, the averages of those samples will form a normal (bell curve) distribution around the true population average—even if the underlying population isn't normally distributed

Sample size

• The number of individuals, items, or data points included in a study or experiment, selected from a larger population to represent it statistically

• More cases = more representative and lower margin of error

Margin of Error

The range of uncertainty or variability around an estimate or measurement

Telephone surveys

• Random calls from phone books

• Cell phones (if used, may double count the wealthy)

• Caller ID screening (non-responses)

Internet Surveys

• Opt-in panels, respondent driven sampling (non-probability sample)

• Digital divide: rich vs. poor, young vs. old

• Statistical methods can help correct sampling bias

Social desirability bias

Not telling pollsters the truth because of fear of judgment

Survey question wording

Affects how accurate the answers are, but hard to predict

Double-barreled questions

2 questions combined into 1

Ambiguous questions

Not specific

Leading questions

Wording encourages a specific answer

Response rate

• Proportion of participants vs. number of people contacted

• High refusal rates affected by participants believing that they won’t benefit, their distrust of strangers, too many surveys or questions are too personal

Response Rates Over Time

• Response rates diminishing over time

• In 50s/60s in person surveys had 80% response, today internet polls have <10% response

Exit polls

A poll of people leaving a polling place, asking how they voted

Focus groups

A demographically diverse group of people assembled to participate in a guided discussion on a political campaign, product, etc.

Face-to-face polls

An interviewer is physically present to ask the survey questions and to assist the respondent in answering them

Most polling today occurs

Over the phone or internet