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:
Methodological Research
Sign up for our Methods newsletter
The latest on survey methods, data science and more, delivered quarterly.
All
Methodological Research
Publications
-
Comparing Survey Sampling Strategies: Random-Digit Dial vs. Voter Files
A new telephone survey experiment finds that an opinion poll drawn from a commercial voter file produces results similar to those from a sample based on random-digit dialing.
-
Q&A: How and why we studied teens and cyberbullying
Roughly six-in-ten U.S. teens have been bullied or harassed online. Senior Researcher Monica Anderson discusses the methods and meaning behind the data.
-
Sharing the code we used to study the public’s interest in the Flint water crisis
In 2017, Pew Research Center published an analysis of Google search trends related to the water crisis in Flint, Michigan. The analysis…
-
Q&A: What works and what doesn’t when studying social media data
Ever since Friendster and Myspace gained popularity in the early 2000s, social scientists have been interested in studying the impact of…
-
Q&A: The challenges of creating a religious typology
This new analysis creates a typology that cuts across denominations, sorting Americans into seven groups, or “clusters,” based on their religious practices and values, their views about religion in general, and the sources of meaning and fulfillment in their lives. Rich Morin, a senior editor at the Center, explains how the study was put together, and discusses the role of cluster analysis in creating the typology.
-
Using supervised machine learning to quantify political rhetoric
Several posts on this blog have examined unsupervised methods of natural language processing. These algorithms and models can help…
-
Validating 2016 voters in Pew Research Center’s survey data
(Related post: Validating 2020 voters in Pew Research Center’s survey data)
-
Making sense of topic models
Topic models can produce clusters of words that characterize written documents. But how do we figure out what those clusters mean, exactly?
-
An intro to topic models for text analysis
Topic models can scan documents, examine words and phrases within them, and “learn” groups of words that characterize those documents.
-
For Most Trump Voters, ‘Very Warm’ Feelings for Him Endured
In the wake of Donald Trump’s 2016 election victory, an overwhelming majority of those who said they had voted for him had “warm” feelings for him.
REfine Your Selection
Date
Years
Formats
Regions & Countries
Research Teams
Bylines
Signature Reports
A Field Guide to Polling: Election 2020 Edition
While survey research in the United States is a year-round undertaking, the public’s focus on polling is never more intense than during the run-up to a presidential election.
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.
Q&A: What is the American Trends Panel, and how has it changed?
Members of the American Trends Panel can now take our surveys online or over the phone with an interviewer.
How Public Polling Has Changed in the 21st Century
A new study found that 61% of national pollsters used different methods in 2022 than in 2016. And last year, 17% of pollsters used multiple methods to sample or interview people – up from 2% in 2016.
