Representative Surveys

 

POLSCI 4SS3
Winter 2024

Announcements

  • I asked for classroom change but haven’t heard any news yet (as of January 24)

Last week

  • Overview of course topic, goals, evaluation, expectations

  • We installed R and RStudio and explored them a bit

  • Cloud option always available if all else fails

  • Brief overview of MIDA framework

  • More details in the course website

Today

  • Start the topic of public opinion

  • Representative surveys as the standard design

  • Takeaway: Complete random sampling is the gold standard but too hard to apply in realistic settings

  • Discussion: Getting surveys right

  • Lab: Sampling and descriptive inference

Review

  • What are the elements of a research design?

  • What is a model?

  • What is an inquiry?

  • What is a data strategy?

  • What is an answer strategy?

  • Why are we doing this?

Public opinion

https://www.vocabulary.com/dictionary/public%20opinion

Public opinion

  • The study of self-reported attitudes and behaviors

  • Primarily among general public

  • Goal: Mapping self-reports to actual attitudes and behaviors

Two challenges

  1. Asking the right questions

  2. Asking the right people

Asking the right questions

Schwarz (1999)

Elements to consider

  • Literal vs. pragmatic meaning
  • Open vs. closed answer format
  • Frequency scales (e.g. more that 1 hour)
  • Reference periods (last week, last year)
  • Rating scales (0-10, -5-5)
  • Demand effects
  • Priming effects

Asking the right people

What is this?

Sampling

  • : How units are selected for a study

  • Which units?

  • How will you reach them?

  • Sampling choices are consequential to how we craft answer strategies

Some key sampling decisions

  • Mode (in-person, lab, phone, mail, internet)

  • Sampling frame

  • Sample size

  • Sampling procedure

  • Oversampling

Random sampling

  • Simple: Coin flip

  • Complete: Exactly \(n\) of \(N\) sampled with same inclusion probability

  • Stratified: Sort in groups or strata, then sample

  • Cluster random: Sample whole groups of units

  • Stratified cluster: Take a guess!

  • Multi-stage: Sample clusters, then sample units

Summary

  • We conduct surveys because we want to understand public opinion

  • Challenge: Map self-reports to attitudes and behaviors

  1. Ask the right questions (meaning, format, scales, context)

  2. Ask the right people (mode, sampling)

Why so many choices?

Next Week

Sensitive Questions

Focus on: What to do when people lie in surveys?

Break time!

 

Lab