Group size estimation and political attitudes
Summary and takeaway
Policy decisions that are focused on particular groups are related to perceptions about the sizes of those groups. Results of this Methodology matters experiment designed to impact the way that people think about group sizes suggests that the relationship between group size estimation and policy attitudes is not causal (e.g. beliefs about the size of a group do not directly change policy attitudes). A related experiment that corrects misperceptions about the sizes of different groups similarly shows that it has very little, if any, impact on policy views.
Introduction and background
A few years ago, YouGov published the results of a survey showing how Americans – sometimes dramatically – misunderstand the demographics of their country. In general, the public’s estimates of group sizes in the American population are correlated with their actual sizes (e.g. if we rank the groups from smallest to largest in the population and compare the rank order of the estimates provided in the survey, the rankings generally agree), but the absolute sizes of the different groups are, in many instances, off by orders of magnitude. For example, in that survey (conducted in 2022) Americans estimated that 21% (fully one out of every five!) American adults are transgender. The true share is much, much smaller.
Do these errors in group size perception roll up to consequences for political attitudes? Are there things that we can do to better measure these kinds of quantities in surveys?
>View experiment's disclosure statement
Data summary
To get at these research questions, we designed an experiment (see the preregistration here) that replicated part of the earlier survey with a few additional experimental conditions. In a survey of 3,906 Americans conducted between March 18 and April 2, 2025, we asked respondents to estimate the sizes of several groups in the general public (this survey asked about five groups: transgender Americans, gun owners, Americans with at least a college degree, people who have a pet and people who own cars).
Our results were similar overall to the original survey. In general, people dramatically overestimate the size of small groups and underestimate the size of larger groups. The one exception to this pattern was the estimates of the share in the general population with a college degree where the public’s perception ended up being very close to the true share.
Experimental setup
Half of the respondents were randomly assigned to see some instructions that were meant to help anchor them to give better estimates of group sizes (a lot of research has shown that people have difficulty with numeracy). Preceding each question where they were asked to estimate group sizes, these respondents saw:
“There are about 260 million adults in the United States. For reference, 3% of adults would be 7.8 million people (or approximately the number of people who live in Washington State). 20% of adults would be one out of every five or about 52 million people (approximately the number of people who live in California and Illinois combined).”
Independent of the first randomization, another half of the sample was randomly assigned to see the correct information about the sizes of the different groups in the general public after they gave their estimates.
After doing the group size estimation exercise, respondents were asked their opinions on transgender access to bathrooms that match their gender identity and restrictions on guns.
Research hypotheses
H1: Providing anchoring information will increase the accuracy of group size estimation
Our expectation was that providing people with some way to anchor their estimates of the sizes of different groups would lead them to make more accurate guesses. The original survey on which this study is based suggested that people might have particular trouble with small groups, so the additional context that people in the anchoring condition saw was meant to help them understand both how percentages translate into something perhaps more understandable (e.g. “20% of adults would be one out of every five”) as well as providing the numbers implied by different percentage estimates (e.g. 3% is equivalent to 7.8 million people).
H2: Overestimation of the size of a group in the general public leads to more restrictive policy views about that group
Our expectation is based on the idea that larger groups might be seen as more of a threat leading to support for more restrictive policies. One argument against overly restrictive policies focused on transgender individuals in particular is that ultimately they end up being targeted at a very small number of people. Below, we look at both the correlation between group size estimates and policy attitudes as well as the causal impact of group size estimates (using the anchoring treatment).
H3: Correcting individual perceptions about the sizes of different groups in the general population will impact their political views.
If policy attitudes are related in some way to policy attitudes, can we change policy attitudes by correcting misperceptions about the sizes of groups?
Results
Hypothesis 1: Providing anchoring information will increase the accuracy of group size estimation
Providing respondents with anchoring information to help them think about what percentages of the adult population mean in terms of overall population seems to have generally led to smaller estimates across the board. This had the effect of making the estimates of small-population groups (transgender and gun owners) more accurate while decreasing the accuracy of larger-population group estimates (pet owners and car owners).
In a multivariate analysis with the average absolute error across the five different group estimates as the dependent variable, the anchoring treatment significantly increased error overall. The table below shows the regression results for the overall errors as well as the errors for each group estimation. The individual group estimates include the average of the errors made in the other groups as control variables.
Hypothesis 2: Overestimation of the size of a group in the general public leads to more restrictive policy views about that group.
We can look at this hypothesis in two ways. First, the correlation between beliefs about group size and restrictive policy attitudes.
A little less than half of the sample that was exposed to neither of the treatments (e.g. they didn’t see the anchoring information that attempted to ground their estimation and they did not see the corrective information after giving their estimates) believed that transgender Americans make up 5% or more of adults in America.
Overall, those who (mistakenly) believe that transgender people make up a substantial share of the general public are somewhat less likely to hold restrictive views about public bathroom use (contrary to our expectations).
Interestingly, the effect works in opposite directions depending on respondent partisanship. There are large partisan differences overall in these views with Democrats and those who lean toward the Democratic Party being much more likely to oppose restrictive policies for transgender individuals compared to Republicans and Republican leaners.
Democrats who believe that transgender Americans are a smaller group hold more permissive attitudes compared to those Democrats who believe that transgender Americans make up a larger share of the public. Among Republicans, the opposite pattern holds – Republicans and Republican leaners who believe that trans people make up a larger share of the public are somewhat less likely to hold restrictive views compared to Republicans who are more accurate in their perceptions about the size of the trans community.
The story is somewhat different when it comes to gun policy. The sample of people who saw neither treatment was about evenly divided between those who believe gun owners are 50% or less of the population and those who say gun owners are more than 50% of the population.
Overall, people who believe that gun owners make up a relatively smaller share of the population were more supportive of restrictive gun control measures compared to those who believe gun owners make up a larger share of the total population. There is a large partisan divide in gun attitudes, but among both Democrats and Republicans, the share holding restrictive gun control attitudes decreases as beliefs about the size of the gun owning population increases.
The results presented above are correlational. These findings can only tell us that these attitudes seem to “go together,” but we can’t say with confidence that they are causally connected.
As we saw in the analysis for Hypothesis 1, providing respondents with anchoring information did significantly decrease their estimates of the sizes of both the transgender and gun owning populations. We might expect the impact on their group size estimations to have some downstream effect on their political attitudes.
Overall the treatment appears to have had not much impact. There is some evidence in a multivariate analysis that – at least among Democrats – the treatment has an indirect effect. Democrats who increase their accuracy of their estimate of the trans community by 1 percentage point become 3 percentage points more likely to support allowing transgender people to use the bathroom matching their gender identity. For attitudes about gun control, increasing the accuracy of Democrats’ views by 1 percentage point increases their likelihood of supporting stricter gun control laws by 1 percentage point. The multivariate results showed no impact of the treatment.
Hypothesis 3: Correcting individual perceptions about the sizes of different groups in the general population will impact their political views
The corrective treatment was the most direct intervention. After giving their group size estimates, respondents in this condition were shown how their estimates compare to the actual sizes of each population. Immediately after being shown this corrective information, they were asked their policy views.
The overall results show only very minor impacts of correcting people’s group size perceptions on their policy attitudes.
General discussion
The results here show that the relationship between policy attitudes and group size perception is not straightforward for most people. Getting people to think in greater detail about the ways percentages of the general population translate into numbers of people does impact their overall estimates (the intervention used in this study had the effect of shrinking estimates on average). The results of this study suggest that (at least for attitudes about transgender rights and gun control) correcting misperceptions about the sizes of different groups in the general public is not sufficient to change attitudes overall. It is possible that different interventions that more effectively help people think about large numbers or with less polarizing issues the impact of correcting misconceptions might have a greater impact on policy views.
1 The multivariate analysis here is a two-stage least squares (2SLS) regression. This analysis controlled for age, gender, race/ethnicity. The first stage of the regression included those controls plus education which estimated the impact of the treatment on the errors made in predictions. This allows us to estimate the indirect effect of changing people’s perceptions about the size of each group through the treatment.
About the author
Brad Jones, Ph.D., Senior Research Director, Scientific Research
Jones is a survey methodologist for the Scientific Research Group. He assists clients with research and questionnaire development. Jones holds a Ph.D. from the University of Wisconsin in political science. Prior to joining YouGov, he worked for Meta generating insights about users’ needs and expectations connected to transparency and control of the advertising delivery system. While at Meta, Jones worked with a wide range of product teams and researchers across Meta’s various platforms. Before working for Meta, he spent the first part of his career at Pew Research Center working with the U.S. Politics and Public Policy team. During his time at Pew, Jones gained extensive experience working through every stage of the survey research process from questionnaire development, to project management, analysis and reporting. He earned his B.A. in Political Science from Brigham Young University.
About Methodology matters
Methodology matters is a series of research experiments focused on survey experience and survey measurement. The series aims to contribute to the academic and professional understanding of the online survey experience and promote best practices among researchers. Academic researchers are invited to submit their own research design and hypotheses.