Democratizing access to the best data on religion since 1997
DATA ARCHIVE
DATA ARCHIVE

Archive Navigation

Summary
Codebook
Downloads
View Question Bank

Archive Categories


Sort the above Archive Category by:



Browse Alphabetically
Browse All Categories
Browse Newest Additions

File Summaries


Search Data Archive


American National Election Studies, Social Media Study, 2020

DOI

10.17605/OSF.IO/5F3EA

Citation

Iyengar, S., Brader, T., Hillygus, S., Shaw, D., & Valentino, N. (2023, September 20). American National Election Studies, Social Media Study, 2020.

Summary

The American National Election Studies (ANES) 2020 Social Media Study is a two-wave panel survey conducted on the Internet to provide data about voting and public opinion in the 2020 presidential election and to link these survey data with data downloaded from participants' Facebook accounts. The two-wave design mirrors the ANES Time Series design, with pre-election and post-election questionnaires. This release contains only survey data and 'vote validation' data; data from the linked Facebook accounts will become available separately in the future.

Though the study features pre-election and post-election surveys, this study should not be confused with the ANES 2020 Time Series Study, which also includes pre- and post-election surveys on the Internet with a higher response rate, different and longer questionnaires, and a different and larger sample than this study.

The ARDA has added five additional variables to the original data set to enhance the users' experience on our site.

Data File

Cases: 5750
Variables: 596
Weight Variable: WEIGHT_PRE, WEIGHT_POST, WEIGHT_PRE_SPSS, WEIGHT_POST_SPSS, WEIGHT_PRE_NR, WEIGHT_POST_NR

For analysis of pre-election data only, use the variable 'WEIGHT_PRE' to generalize to the population. For pre- and post-election combined or post-election data only, use 'WEIGHT_POST'.

The variables WEIGHT_PRE, and WEIGHT_POST were developed by NORC to account for selection probability, nonresponse (in categories of Facebook user status, 2 race categories, 3 age categories, 2 education categories, and 2 gender categories), and post-stratification factors (7-category age, 2-category gender, 9-category census division, 4-category education, 4-category race/ethnicity, and an 8-category interaction term of age and gender, age and race, and race and gender). Post-stratification was based on benchmark values for adult U.S. citizens from the American Community Survey. The post-election weight was trimmed by ANES to a maximum value of 5 to limit variance. The pre-election weight should be used for analysis of the pre-election data alone. The post-election weight should be used for analysis of the pre- and post-election data together or of the post-election data alone.

The pre-election weighted design effect is 1.52; the root design effect is 1.23. This means that the sample's average statistical power is equivalent to an unweighted simple random sample size of 5750/1.52 = 3,783, and the expected sampling errors are increased by a factor of 1.23 relative to what would be expected of a sample of 5750. The post-election design effect is 1.63 and the post-election root design effect is 1.28. Note that design effects vary and will be larger or smaller for some estimates.

SPSS weights. We recommend that SPSS users not using the Complex Samples procedures use the weight variables WEIGHT_PRE_SPSS, and WEIGHT_POST_SPSS which are normed to means of 0.66 and 0.61, respectively. These variables will account for the smaller effective sample size due to sampling and weighting. When not using the Complex Samples procedures it is appropriate to use these adjusted weights because SPSS uses frequency weights that do not account for weight variance when computing sampling errors.

Weights without post-stratification. The variables WEIGHT_PRE_NR and WEIGHT_POST_NR account for selection probability and non-response but are not post-stratified or scaled to a mean of 1. They are included for methodological purposes such as re-weighting the data using different approaches to post-stratification.

Data Collection

Pre-election: Aug. 20 - Sept. 17, 2020; Post-election: Nov. 1, 2020 - Jan. 1, 2021

Original Survey (Instrument)

ANES 2020 Social Media Survey

Funded By

The study was funded by a gift to Stanford University from Facebook and by grants from the National Science Foundation.

These materials are based on work supported by the National Science Foundation under grant number SES-1835022 to Stanford University and SES-1835721 to the University of Michigan. Any opinions, findings and conclusions, or recommendations expressed in these materials are those of the authors and do not necessarily reflect the views of the National Science Foundation.

Collection Procedures

The study was conducted on the Internet using an online survey panel that was selected using probability sampling methods. This sample was provided by NORC at the University of Chicago.

Data from 5,750 completed pre-election interviews are on the file. An additional 80 participants completed the questionnaire but their data failed a quality-control check performed by NORC. Data from these cases were considered invalid and they were not weighted and were excluded from the data file due to item nonresponse (they did not answer more than 50 percent of the questions they were asked), speeding (they completed the questionnaire 33 percent faster than the median time), or nondifferentiation/straightlining (they selected the same response for all grid items).

Data are provided for 5,277 post-election interviews. An additional 52 participants completed the post-election questionnaire but their data failed a quality-control check of the type described above. For questionnaire variables, these cases have the same unit non-response code as cases of post-election unit nonresponse, but they are identified by the variable w2qual.

This release of the data replaces the initial datasets released on January 28, and March 8, 2021. The March 8 release included the pre-election and post-election survey data. This release adds 'validated' voter registration and turnout data and corrects minor errors that were previously documented; it does not include data from linked Facebook accounts, which will be provided in the future using a separate process.

With this special study, quality control review and error correction have not been provided to the same standard as with a final data release of an ANES Time Series study. Future re-releases of the data may include some changes to the data, such as error corrections, to improve accuracy or usability. ANES always wants to correct errors. If you find errors (even years after this data release) or have comments or questions about the data, please write to This email address is being protected from spambots. You need JavaScript enabled to view it.

Sampling Procedures

The study's target population was U.S. citizens age 18 or older. The sample was stratified by previously reported Facebook user status, race/ethnicity, age, education, and gender. Completion targets were established, and once the target number of Facebook users had completed the questionnaire, additional self-reported Facebook users were screened out.

The weighted household recruitment rate was 21 percent, reflecting the percentage of households where an adult was enrolled in the panel and completed a recruitment questionnaire. The retention rate from the time of initial recruitment to the time the sampling was done for this survey was 80.4 percent. A sample of 31,043 individuals was initially invited to participate in this survey. Of those invited, 35.2 percent completed a screening step that ascertained their U.S. citizenship status, obtained informed consent for the study, and determined whether they were or were not a Facebook user. Of those screened, 98.6% were eligible for the study, and 53.3 percent of those eligible completed the study. The overall response rate is the product of the recruitment, retention, screening, and completion rates, and was 3.2 percent. The post-election re-interview rate was 91.8 percent.

Principal Investigators

ANES Principal Investigators: Shanto Iyengar and Ted Brader
Associate Principal Investigators: Sunshine Hillygus, Daron Shaw, and Nicholas Valentino.

Our Sponsors

Our Affiliates

US RELIGION
WORLD RELIGION
DATA ARCHIVE
RESEARCH
TEACHING
CONGREGATIONS
ABOUT
© 2024 The Association of Religion Data Archives. All rights reserved.