2025 Cross-Sectional Engagement Survey methodology
Summary
SSRS conducted the Cross-Sectional Engagement Survey for Pew Research Center using address-based sampling and a multimode protocol. The survey was fielded from July 9 to Dec. 5, 2025. Participants were first mailed an invitation to complete an online survey. A paper survey was later mailed to those who did not respond. Additionally, the mailings provided participants a toll-free number to call if they preferred to take the survey over the phone with a live interviewer. In total, 2,705 respondents completed the survey online, 2,500 respondents completed the paper survey, and 188 respondents completed the survey over the phone (total n=5,393). The survey was administered in English and Spanish. The AAPOR Response Rate 1 was 28%.
Sample definition
The sample was drawn from the U.S. Postal Service Computerized Delivery Sequence File and was provided by MSG (Marketing Systems Group). Occupied residential addresses (including “drop points”) in all U.S. states (including Alaska and Hawaii) and the District of Columbia had a nonzero chance of selection. The draw was a national, stratified random sample, with differential probabilities of selection across the mutually exclusive strata. SSRS designed the sample plan as shown in the table above.
Mailing protocol
SSRS sent initial mailings in a 9-by-12-inch window envelope via first-class mail to the 20,751 sampled addresses. These packets included $1 bills (visible from the outside of the envelope) and a letter that asked a member of the household to complete the survey. The letter provided a URL for the online survey; a toll-free call-in number; a password to enter on the online survey’s landing page, or tell the telephone interviewers if they chose to call in; and a FAQ section printed on the reverse side. If two or more adults were in the household, the letter asked the adult with the next birthday to complete the survey. Two experiments were also embedded in this mailing. The first experiment was to test the effect of QR codes on response rates. As such, 65% of the letters included a QR code and instructions on how to use it. The second experiment was to test the effect of the cash incentive on response rates. As such, 65% of the letters included two $1 bills, and the remaining 35% included one $1 bill. The two experiments were fully and randomly crossed across all sampled addresses. Nonresponding households, regardless of experimental condition, were later sent a reminder postcard and then a reminder letter via first-class mail.
After the web portion of the data collection period had ended, SSRS sent nonresponding households with a deliverable address a 9-by-12-inch Priority Mail window envelope. The Priority envelope contained a letter with a FAQ section printed on the reverse side, a visible $5 bill, a paper version of the survey and a postage-paid return envelope. The paper survey was one 11-by-17-inch page folded booklet-style. The within-household selection instructions were identical to those used in the earlier online survey request. The same households were later sent a second envelope containing another copy of the paper questionnaire by first-class mail.
The initial mailing was sent out in two separate launches: soft launch and full launch. The soft launch made up 5% of the sample and was sent out several days earlier than the full launch. The full launch consisted of the remaining sample.
Households in Hispanic strata, and additional sample records predicted to likely be Hispanic, received all materials in English and Spanish. All other households received materials in English only. Those who completed the survey were sent a $10 post-paid incentive.
Questionnaire development and testing
Pew Research Center developed the questionnaire in consultation with SSRS. The online questionnaire was tested on both desktop and mobile devices. The test data was analyzed to ensure the logic and randomizations were working as intended before the survey was launched.
Weighting
The survey was weighted to support reliable inference from the sample to the target population of U.S. adults. The weight was created using a multistep process that includes a base weight adjusting for differential probabilities of selection and a raking calibration that aligns the survey with the population benchmarks. The process starts with the base weight, which accounted for the probability of selection of the address from the U.S. Postal Service Computerized Delivery Sequence File frame, as well as the number of adults living in the household, and incorporated an adaptive mode adjustment for cases that responded in an offline mode.
Then the base weights are calibrated to population benchmarks using raking, or iterative proportional fitting. The raking dimensions and the source for the population parameter estimates are reported in the table below. All raking targets are based on the noninstitutionalized U.S. adult population (ages 18 and older). These weights are trimmed at the 1st and 99th percentiles to reduce the loss in precision stemming from variance in the weights.
Design effect and margin of error
Weighting and survey design features that depart from simple random sampling tend to result in an increase in the variance of survey estimates. This increase, known as the design effect, or “deff,” should be incorporated into the margin of error, standard errors and tests of statistical significance. The overall design effect for a survey is commonly approximated as 1 plus the squared coefficient of variation of the weights.
For this survey, the margin of error (half-width of the 95% confidence interval) incorporating the design effect for full sample estimates at 50% is plus or minus 1.9 percentage points. Estimates based on subgroups will have larger margins of error. It is important to remember that random sampling error is only one possible source of error in a survey estimate. Other sources, such as question wording and reporting inaccuracy, may contribute additional error. A summary of the weights and their associated design effect is reported in the table below.
Dispositions
The table below reports the disposition of all sampled households for the survey.
The American Trends Panel survey methodology
Overview
Data in “How engagement is linked with Americans’ views of politics and news” and “How engagement relates to knowledge about politics” comes from Wave 179 of the American Trends Panel (ATP), Pew Research Center’s nationally representative panel of randomly selected U.S. adults. The survey was conducted from Sept. 8 to Sept. 14, 2025. A total of 5,195 panelists responded out of 5,852 who were sampled, for a survey-level response rate of 89%.
The cumulative response rate accounting for nonresponse to the recruitment surveys and attrition is 3%. The break-off rate among panelists who logged on to the survey and completed at least one item is less than 1%. The margin of sampling error for the full sample of 5,195 respondents is plus or minus 1.6 percentage points.
The survey includes an oversample of non-Hispanic Asian adults in order to provide more precise estimates of the opinions and experiences of this smaller demographic subgroup. These oversampled groups are weighted back to reflect their correct proportions in the population.
SSRS conducted the survey for Pew Research Center via online (n=4,986) and live telephone (n=209) interviewing. Interviews were conducted in both English and Spanish.
To learn more about the ATP, read “About the American Trends Panel.”
Panel recruitment
Since 2018, the ATP has used address-based sampling (ABS) for recruitment. A study cover letter and a pre-incentive are mailed to a stratified, random sample of households selected from the U.S. Postal Service’s Computerized Delivery Sequence File. This Postal Service file has been estimated to cover 90% to 98% of the population.1 Within each sampled household, the adult with the next birthday is selected to participate. Other details of the ABS recruitment protocol have changed over time but are available upon request.2 Prior to 2018, the ATP was recruited using landline and cellphone random-digit-dial surveys administered in English and Spanish.
A national sample of U.S. adults has been recruited to the ATP approximately once per year since 2014. In some years, the recruitment has included additional efforts (known as an “oversample”) to improve the accuracy of data for underrepresented groups. For example, Hispanic adults, Black adults and Asian adults were oversampled in 2019, 2022 and 2023, respectively.
Sample design
The overall target population for this survey was noninstitutionalized persons ages 18 and older living in the United States. It featured a stratified random sample from the ATP in which non-Hispanic Asian adults were selected with certainty. The remaining panelists were sampled at rates designed to ensure that the share of respondents in each stratum is proportional to its share of the U.S. adult population to the greatest extent possible. Respondent weights are adjusted to account for differential probabilities of selection as described in the Weighting section below.
Questionnaire development and testing
The questionnaire was developed by Pew Research Center in consultation with SSRS. The web program used for online respondents was rigorously tested on both PC and mobile devices by the SSRS project team and Pew Research Center researchers. The SSRS project team also populated test data that was analyzed in SPSS to ensure the logic and randomizations were working as intended before launching the survey.
Incentives
All respondents were offered a post-paid incentive for their participation. Respondents could choose to receive the post-paid incentive in the form of a check or gift code to Amazon.com, Target.com or Walmart.com. Incentive amounts ranged from $5 to $15 depending on whether the respondent belongs to a part of the population that is harder or easier to reach. Differential incentive amounts were designed to increase panel survey participation among groups that traditionally have low survey response propensities.
Data collection protocol
The data collection field period for this survey was Sept. 8 to Sept. 14, 2025. Surveys were conducted via self-administered web survey or by live telephone interviewing. For panelists who take surveys online:3 Postcard notifications were mailed to a subset on Sept. 8.4 Survey invitations were sent out in two separate launches: soft launch and full launch. Sixty panelists were included in the soft launch, which began with an initial invitation sent on Sept. 8. All remaining English- and Spanish-speaking sampled online panelists were included in the full launch and were sent an invitation on Sept. 9.
Panelists participating online were sent an email invitation and up to two email reminders if they did not respond to the survey. ATP panelists who consented to SMS messages were sent an SMS invitation with a link to the survey and up to two SMS reminders.
For panelists who take surveys over the phone with a live interviewer: Prenotification postcards were mailed on Sept. 5. Soft launch took place on Sept. 8 and involved dialing until a total of seven interviews had been completed. All remaining English- and Spanish-speaking sampled phone panelists’ numbers were dialed throughout the remaining field period. Panelists who take surveys via phone can receive up to six calls from trained SSRS interviewers.
Data quality checks
To ensure high-quality data, Center researchers performed data quality checks to identify any respondents showing patterns of satisficing. This includes checking for whether respondents left questions blank at very high rates or always selected the first or last answer presented. As a result of this checking, four ATP respondents were removed from the survey dataset prior to weighting and analysis.
Weighting
The ATP data is weighted in a process that accounts for multiple stages of sampling and nonresponse that occur at different points in the panel survey process. First, each panelist begins with a base weight that reflects their probability of recruitment into the panel. These weights are then calibrated to align with the population benchmarks in the accompanying table to correct for nonresponse to recruitment surveys and panel attrition. If only a subsample of panelists was invited to participate in the wave, this weight is adjusted to account for any differential probabilities of selection.
Among the panelists who completed the survey, this weight is then calibrated again to align with the population benchmarks identified in the accompanying table and trimmed at the 1st and 99th percentiles to reduce the loss in precision stemming from variance in the weights. Sampling errors and tests of statistical significance take into account the effect of weighting.
The following table shows the unweighted sample sizes and the error attributable to sampling that would be expected at the 95% level of confidence for different groups in the survey.
Sample sizes and sampling errors for other subgroups are available upon request. In addition to sampling error, one should bear in mind that question wording and practical difficulties in conducting surveys can introduce error or bias into the findings of opinion polls.
Dispositions and response rates
Engagement group creation and analysis
Drawing on data from the 2025 Cross-Sectional Engagement Survey, we used cluster analysis to sort Americans into four engagement groups based on the ways they take part in public life. This approach identifies people who share similar engagement patterns across politics, religion, news consumption and other civic activities.
Group assignment was based on respondents’ answers to 19 questions about their participation in these areas. The survey also includes questions on attitudes, demographics and identities, but the engagement cluster analysis only used items measuring behaviors. The items selected for inclusion in the clustering were chosen based on extensive testing to find the model that fit the data best and produced groups that were substantively meaningful. The table below lists the items used to create the engagement groups. The analysis considers all 19 behaviors together and treats each item as equally important.
Questions used in the engagement cluster analysis
| Question | Response options |
|---|---|
| VOL12_CPS: In the past 12 months, did you spend any time volunteering for any organization or association? (This includes activities you may do infrequently or for children’s schools or youth organizations.) | 1 Yes 2 No |
| GROUPS_CPS: In the past 12 months, did you belong to any groups, organizations, or associations? | 1 Yes 2 No |
| LOCAL_MTG: In the past 12 months, have you attended a local government meeting such as a city council or school board meeting, either in person or online? | 1 Yes 2 No |
| NEWS_LEVEL_NAT: How closely do you follow national news? | 1 Very closely 2 Somewhat closely 3 Not very closely 4 Not at all closely |
| NEWS_LEVEL_LOC: How closely do you follow local news? | 1 Very closely 2 Somewhat closely 3 Not very closely 4 Not at all closely |
| TALKNEWS: How often do you discuss news and current events with others, whether in person, online, or over the phone? | 1 Nearly every day 2 A few times a week 3 A few times a month 4 Less often |
| NEWSONLINE: In the past 7 days, have you done any of the following activities online (on a website, social media, or a messaging app)? LIKE: Liked or saved posts about news PUB: Publicly posted, shared, or commented about news PRIV: Privately chatted about or shared news | 1 Yes 2 No |
| DON1_CPS: In the past 12 months, did you give money or possessions with a combined value of more than $25 to a political organization, party, or campaign? | 1 Yes 2 No |
| DON2_CPS: In the past 12 months, did you give money or possessions with a combined value of more than $25 to a non-political group or organization, such as a charity, school, or religious organization? | 1 Yes 2 No |
| CIVENG_POST: Here’s a list of activities some people do and others do not. Please indicate if you have done each of the following activities over the past 12 months. VOLUNT: Worked or volunteered for a political party, candidate, or campaign DISPL: Displayed a sign or bumper-sticker or wore something to show your support for a political cause, campaign, or issue SMMOD: Used your social media bio or profile picture to show support for a cause, issue, or political candidate LETTER: Contacted any elected official MARCH: Participated in any demonstrations, protests, marches, or political rallies | 1 Yes 2 No |
| ATTENDPER: Aside from weddings and funerals, how often do you attend religious services in person? | 1 More than once a week 2 Once a week 3 Once or twice a month 4 A few times a year 5 Seldom 6 Never |
| ATTENDONLINE2: In general, how often do you watch or participate in religious services online or on TV? | 1 More than once a week 2 Once a week 3 Once or twice a month 4 A few times a year 5 Seldom 6 Never |
| VOTED2024B: In the 2024 presidential election between Kamala Harris and Donald Trump, did things come up that kept you from voting, or did you happen to vote? | 1 Voted 2 Did not vote |
Source: Survey of U.S. adults conducted July 9-Dec. 5, 2025.
PEW RESEARCH CENTER
The specific statistical technique used to calculate group membership is weighted clustering around medoids (using the WeightedCluster package version 1.8-1 in R version 4.5.2). The k-medoids algorithm assigns respondents in the survey to groups characterized by differing patterns of responses. Prior Pew Research Center typologies used the same method or a closely related method, such as cluster analysis via the k-means algorithm, to identify groups.
Decisions made at different points in the analysis can have a dramatic impact on the ultimate results of any data partitioning exercise. Different solutions are possible using the same data depending on the algorithm used for partitioning, the ways in which the variables are coded, the number of clusters, and even the order in which respondents are sorted. Refer to “How we measured Americans’ engagement in public life” for more details.
Coding of input variables
Input variables used for constructing the clusters were coded using their original response formats. Binary items were treated as binary, and scaled items were kept on their original scales. We did not standardize variables, as the clustering relied on Gower’s distance, which is designed to accommodate mixed data types when computing distances between respondents.
Data cleaning
Prior to clustering, we conducted data quality checks to assess item nonresponse across input variables. We excluded 13 respondents from the cluster analysis who did not offer answers to nine or more of the 19 items used to create the engagement groups, as their limited data did not allow for reliable classification. Among the remaining respondents (n=5,380, unweighted), there was a small amount (ranging from nine to 96 cases) of item nonresponse in the measures used. We did not impute missing values for these cases, as Gower’s distance calculates distances based on the available responses for each pair of respondents. This approach allowed us to retain respondents with partial item nonresponse while minimizing the influence of missing data on the clustering results.
Selecting the number of groups
Solutions with different numbers of groups were examined. The results presented in this report were chosen for their effectiveness in producing cohesive groups that were sufficiently distinct from one another and large enough in size to be analytically practical and substantively meaningful.
Other methodological decisions
To address the potential sensitivity of cluster analysis to the order in which cases are entered, each model was run several times. The k-medoids algorithm was robust to different initial conditions and consistently returned the same results even with different random starts.
While each model differed somewhat from the others, all of them shared certain key features. The final model selected to produce the engagement groups was judged to be strong from a statistical point of view, most persuasive from a substantive point of view, and representative of the general patterns seen across the various models run.
Projecting engagement clusters to ATP Wave 179
We also looked at how engagement groups differ in attitudes (e.g., feelings about the country, trust in news organizations) and civic knowledge. Those questions were not asked in the 2025 Cross-Sectional Engagement Survey due to limited space. However, they were included in American Trends Panel (ATP) Wave 179, which also asked the same questions about engagement behaviors.
To connect the two surveys for further analysis, we assigned ATP respondents to the derived engagement groups based on how similar their behavior patterns were to those of the most typical respondents in each engagement group from the cross‑sectional survey. We coded the behavioral variables in ATP Wave 179 in the same way and again used Gower’s distance. This approach lets us analyze attitudes and knowledge using ATP data while keeping the engagement groups rooted in the cross-sectional survey. The ATP results were broadly similar to the cross-sectional survey results in both engagement patterns and demographics.
