Cookies on OyList

We use essential cookies to keep OyList working, and Google uses cookies to show and measure the ads on our lists. With your permission, Google Analytics cookies help us count visits and understand which pages are useful.

How we use cookies

Share

Embed
Share this list with friends or copy the link:

Items7

1

The Literary Digest’s 1936 Mail-In Poll

by curious_compiler  • 
The Literary Digest’s 1936 Mail-In Poll
The magazine’s mail-in poll produced millions of replies and picked Alf Landon, but Franklin Roosevelt won. The sample was drawn partly from sources such as phone directories and car registrations. Survey researcher Peverill Squire found that the people who sent their ballots back also differed from those who stayed silent. Use the example to ask who was invited and who actually answered.
More about The Literary Digest’s 1936 Mail-In Poll
2

The 1948 Truman–Dewey Polls

by curious_compiler  • 
The 1948 Truman–Dewey Polls
Several major pollsters expected Thomas Dewey to beat Harry Truman. Interviewers working under quotas could look for people of a given age or sex, while still deciding whom to approach inside each group. The post-election inquiry examined that discretion and how far before Election Day the pollsters stopped interviewing. Both are good questions to ask before taking a forecast at face value.
More about The 1948 Truman–Dewey Polls
3

YouGov’s 2017 British Election Model

by curious_compiler  • 
YouGov’s 2017 British Election Model
YouGov's 2017 model projected a Parliament with no overall majority. Multilevel regression and poststratification, or MRP, models the voting choices of different kinds of people, then combines those estimates with information about who lives in each constituency. Read this case to see why a seat estimate involves more assumptions than quoting one national vote share.
More about YouGov’s 2017 British Election Model
4

Pew’s 2023 Online Panel Comparison

by curious_compiler  • 
Pew’s 2023 Online Panel Comparison
Pew gave the same questionnaire to three probability-based online panels and three opt-in samples, then compared their results with outside benchmarks. The probability panels recruited from randomly drawn addresses. The opt-in samples used volunteers and other sources. Ask how people got into the sample before drawing conclusions from the fact that the poll was online.
More about Pew’s 2023 Online Panel Comparison
5

The 2020 National Election Pool Exit Poll

by curious_compiler  • 
The 2020 National Election Pool Exit Poll
The 2020 exit poll was more than people with clipboards outside polling places on Election Day. Edison interviewed people coming out of both early-voting centers and Election Day polling places and reached mail voters by phone. If a report says how a group voted, check how the survey reached voters who never went through an Election Day polling place.
More about The 2020 National Election Pool Exit Poll
6

Education Weighting After the 2016 Election

by curious_compiler  • 
Education Weighting After the 2016 Election
An assessment of 2016 polling found that many state polls did not adjust for education. College graduates were more likely to answer surveys, so giving their responses more influence than their share of the electorate warranted could skew the estimate. Pew’s follow-up shows how researchers can use a survey of voters to help set those weights.
More about Education Weighting After the 2016 Election
7

Iraq War Poll Questions Before the 2003 Invasion

by curious_compiler  • 
Iraq War Poll Questions Before the 2003 Invasion
Pew gathered polls in which Americans answered different questions about potential U.S. military action in Iraq. Some questions raised casualties or asked what respondents thought if other countries opposed the plan. Look at the exact wording before comparing the percentages. Questions about different conditions can give readers genuinely different information.
More about Iraq War Poll Questions Before the 2003 Invasion

Comments