Scientific Polls

What are Scientific Polls?

• Public opinion matters because the United States is a representative democracy built on popular sovereignty and republicanism.
• To protect democracy, government policies should reflect the wishes of the people.
• Scientific polls are a way to gauge public opinion accurately, unlike internet straw polls or push polls.
• Scientific polling uses random sampling controls to ensure statistical representation of a population.
• Unscientific polling lacks systematic sampling controls.

Methodology of Surveys

The methodology of a survey includes:

• Identification and selection of sample members.
• Development and evaluation of questions.
• Determination of poll delivery mode and response collection.
• Training and supervision of interviewers.
• Data collection from sample members.
• Assessment of the data for accuracy and consistency.
• Adjustment of conclusions to account for identified errors.
• Ensuring accuracy of reporting, clearly supported by data.

Types of Polls

• Public opinion polls survey a sample group to represent the opinions of a larger population.
• Reliability and veracity depend on the sampling, questions, and methodology.

Scientific Polls

Tracking Polls

• Repeated polls at intervals to track public opinion changes on issues or candidates.
• The first poll in a series is called the benchmark poll.
• Tracking polls use data from the past week and discard older data.

Entrance Polls

• Conducted before voters cast their votes; have a low margin of error.

Exit Polls

• Conducted immediately after voters exit polling stations.
• Conducted by private companies for media to get early election outcome indications.

Unscientific Polls

Straw Polls

• Conducted unscientifically using convenience sampling, leading to selection bias.

Push Polls

• Skew public opinion about a candidate and inform campaigns about candidate strengths and weaknesses, often using hypothetical issues.

Sampling

• How a sample is selected distinguishes scientific from unscientific polls.
• The selection method determines how well the sample mirrors the voting population.
• Probability-based sampling involves random selection.
• Non-probability sampling involves selection based on researcher judgment.
• Sampling error is the expected rate of error based on sample size; larger samples have smaller errors.

Probability-Based Sampling

Random-Digit Dialing

• Samples created randomly from telephone area codes with random digits applied.
• Household sample selection is necessary if multiple registered voters reside in the household.

Registration-Based

• Samples taken from lists of registered voters.

Stratified

• The national population is divided into fourths, and certain areas are selected as representatives.

Non-Probability Sampling

Self-Selected

• Respondents select themselves, such as in dial-in or internet-based polls.

Internet Panels

• Samples are selected from respondents who signed up to be members of an internet panel.

Quota

• Researchers structure samples with specific characteristics and select respondents who satisfy them.

Questions

• Neutral framing is essential, ensuring questions are specific and without bias.
• Open-ended questions allow respondents to frame their own answers.
• Closed-ended questions ask respondents to pick from a given set of options.
• Rating-scale questions are a type of closed-ended question.

Is it Legit? (Reliability and Veracity)

• Determine reliability and veracity to assess the quality and credibility of public opinion data.

Reliability: Is the Data Biased?

• Bias affects reliability.
• Consider who conducted the poll and how it was completed.
• Some companies conduct research for specific political parties, which may introduce bias.

Selection Bias

• Occurs when there is a different probability of selection of one demographic group that the research did not anticipate.
• Example: Calling households only in the morning samples a different demographic than in the evening.

Self-Selection Bias

• Occurs when individuals select themselves into a group.
• Common in internet polls, especially on party-specific websites.

Non-Response Bias

• Occurs when individuals do not answer calls or refuse to answer the poll.
• Often a problem with mail surveys.

Coverage Bias

• Occurs when a poll's sample is not representative of a certain population.
• Example: Telephone sampling excludes those without phones, while online surveys exclude those without internet.

Response Bias

• Occurs when responses do not reflect true beliefs because of question wording.
• Vague questions or a limited number of options can affect responses.
• Loaded questions with inherent bias can affect responses (used in push polls).

Veracity: Is the Data Accurate?

• Accuracy affects veracity and credibility.
• Sampling error indicates how closely survey results imitate reality.
• Sampling error is affected by the conditions and timing of a poll.
• Biased or emotionally charged questions can compromise accuracy.
• Polls conducted closer to an election have smaller sampling errors.
• The number of respondents, how they were chosen, the amount of background information needed, and question bias/phrasing also affect sampling error.

Data Analysis

• Analyze and interpret quantitative data in tables, charts, graphs, maps, and infographics.
• Understand different ways data can be presented (pie graphs, bar graphs, data maps, charts).

Steps for Analyzing Political Data

1. Get the Big Idea: Focus on the title and ranges or subcategories of the data.
2. Read the Prompt: Understand what information you need from the data.
3. Annotate the Data: Keep in mind the prompt and document analysis.
4. Write Your Response: Answer each part of the prompt, using the identified trends.

Do Polls Matter?

• Public opinion matters to election candidates, interest groups, policymakers, political parties, and the media.
• Poll data affects policy decision-making and how citizens think, react, and interact with the government.
• How polls are conducted can affect public opinion.
• Push polls are designed to plant negative ideas about a policy or candidate.
• Exit polls can influence media to project election winners before voting closes.

Case Studies & Examples

• 2008 presidential election: Push poll targeted Jewish voters, implying Barack Obama was anti-Jewish and anti-Israel.
• 2000 presidential election: Exit polls reported by the Los Angeles Times showed how close the major candidates were overall and whom subgroups of voters preferred.
• 2016 election: Donald Trump's Electoral College win surprised many, as tracking polls showed Hillary Clinton in the lead.

Social Desirability Effect

• Respondents may not answer honestly or not respond because they're uncomfortable sharing their opinion.

Bandwagon Effect

• Polling results convince people to support a candidate already determined to be the probable winner.

• Limitations of polls: They can misrepresent the general public's views, and people's minds can change quickly, affecting accuracy.

Deliberative Polling

• Gives the public the opportunity to think carefully about issues and their relationship to public policy.
• Small, representative samples meet with experts, and discussions are shared with the public.