Can AI Stand In for Human Survey-Takers? Not Really
Synthetic surveys using digital twins do a poor job estimating the public mood and are especially error-prone for some subgroups, including Republicans and Black adults.
Numbers, Facts and Trends Shaping Your World
Read our research on:
Arnold Lau
Research Methodologist
Publications
-
-
decoded
Why and how we’re weighting surveys for past presidential vote
This piece explains why, when and how we are weighting our surveys on Americans’ past vote.
-
report
Comparing Two Types of Online Survey Samples
Pew Research Center conducted a study to compare the accuracy of six online surveys of U.S. adults – three from probability-based panels and three from opt-in sources. On average, the absolute error on opt-in samples was about twice that of probability-based panels.
-
decoded
Assessing the effects of generation using age-period-cohort analysis
In this piece, we demonstrate how to conduct age-period-cohort analysis, a statistical tool, to determine the effects of generation.
-
short reads
Does public opinion polling about issues still work?
National polls like the Center’s come within a few percentage points, on average, of benchmarks from high response rate federal surveys.
-
short reads
Most Americans believe in intelligent life beyond Earth; few see UFOs as a major national security threat
About two-thirds of Americans (65%) say their best guess is that intelligent life exists on other planets.
-
report
Measuring the Risks of Panel Conditioning in Survey Research
A new evaluation of the Center’s national American Trends Panel finds little evidence that panel estimates are affected by errors associated with panel conditioning, a phenomenon that occurs when survey participation changes respondents’ true or reported behavior over time.
-
report
Confronting 2016 and 2020 Polling Limitations
Looking at final estimates of the outcome of the 2020 U.S. presidential race, 93% of national polls overstated the Democratic candidate’s support among voters, while nearly as many (88%) did so in 2016.
-
report
What 2020’s Election Poll Errors Tell Us About the Accuracy of Issue Polling
Given the errors in 2016 and 2020 election polling, how much should we trust polls that attempt to measure opinions on issues?
-
decoded
Assessing the consistency of online opt-in polls
In this post, we examine whether online opt-in or “nonprobability” surveys are consistent in the same ways as probability-based surveys.
