Published on April 28, 2026
The report, Perceptions and Reality: Understanding Crime Concerns in the United States, contains two types of analyses, which draw on data from different sources. The first analysis examines crime sentiment trends from Gallup, crime rates from the FBI’s Uniform Crime Reports (UCR), and victimization rates from the National Crime Victimization Survey (NCVS). The second analysis uses individual-level responses from the Gallup Poll Social Series (GPSS), which interviews American adults (aged 18+) monthly on various topics organized thematically. This methodology focuses on the individual-level analysis.
Data
Since 2001, the GPSS survey conducted each October has measured people’s perceptions of crime, personal safety, and victimization, which are central to this study’s focus. Gallup identifies respondents via a dual-frame design to include landline and cell phone numbers, using random digit dialing to reach individuals. For landlines, Gallup selects a random respondent within the household based on the resident with the next birthday. Interviews are conducted in English and Spanish, depending on respondents’ primary language. For each survey, Gallup interviews at least 1,000 respondents and applies weights to adjust for unequal selection, nonresponse, and double coverage, as well as to align the sample with the U.S. population by gender, age, race, Hispanic ethnicity, education, region, population density, and phone status.1
Since the individual-level GPSS data include the respondent’s zip code and county-level unique identification codes, the analysis incorporates additional data using these geographic variables and merges them with the GPSS dataset. Zip code-level measures were obtained from the American Community Survey.2 County crime rates were generated using the UCR Return A.3 County-level unemployment rates were obtained from the Local Area Unemployment Statistics program.4 Across the variables, any responses in the Gallup data that were entered as “Don’t know” or “Refused” were excluded from the analysis and coded as missing, unless otherwise noted.
Dependent Variables
- National crime perceptions relied on responses to the question, “Is there more crime in the U.S. than there was a year ago, or less?” For the analysis, these variables were coded 1 to indicate a more response and 0 if the response was same or less.
- Local crime perceptions used the question, “Is there more crime in your area than there was a year ago, or less?” For the analysis, these variables were coded 1 to indicate a more response and 0 if the response was same or less.
- Fear of crime was measured using the response to “Is there any area near where you live—that is, within a mile—where you would be afraid to walk alone at night?” A yes response was coded as 1, and a no response was coded as 0.
Independent Variables
Variables in Gallup Data
- Household victimization was captured using two variables, which are the culmination of responses to seven questions about crime victimization in the 12 months leading up to the interview. These seven items are: (1) “Your house or apartment was broken into,” (2) “Money or property was stolen from you or another member of your household,” (3) “A car owned by you or other household member was stolen,” (4) “A home, car, or property owned by you or another household member was vandalized,” (5) “Money or property was taken from you or another household member by force, with gun, knife, weapon or physical attack, or by threat of force,” (6) “You or another household member was mugged or physically assaulted,” and (7) “You or another household member was sexually assaulted.” Gallup then derived two variables—one reflecting the number of times of personal victimization and the other the number of times of household victimization. Responses were originally coded to reflect Not a Crime Victim (0), One Crime (1), Two Crimes (2), continuing up to seven crimes (7). These response categories and coding were retained, with no response treated as a missing value.
- Political party affiliation was measured using the question, “In politics, as of today, do you consider yourself a Republican, a Democrat, or an Independent?” Response options included Republican/Lean Republican, Independent, No lean, and Democrat/Lean Democratic, and were coded on a three-point scale from 1 (Democrat/Lean Democratic) to 3 (Republican/Lean Republican).
- Political ideology was captured through the question, “How would you describe your political views—very conservative, conservative, moderate, liberal, very liberal?” This variable was recoded such that higher values on the five-point measure indicated more conservatism.
- Presidential approval was measured by asking, “Do you approve or disapprove of the way [….] is handling his job as president?” Approve responses were coded as 1, and Disapprove responses were coded as 0.
- Congressional approval was measured through the question, “Do you approve or disapprove of the way Congress is handling its job?” Approve responses were coded as 1, and Disapprove responses were coded as 0.
- Favoring the death penalty was measured by asking, “Are you in favor of the death penalty for a person convicted of murder?” Responses were recoded to reflect Yes, in favor (1) as opposed to No, not in favor (0).
- Christmas spending was captured by asking respondents about their anticipated investment in Christmas: “Is that more, less, or about the same amount as you spent last Christmas?” Responses were coded as More (1), Less (-1), and The same (0).
- Time to find a quality job was measured by asking respondents, “Thinking about the job situation in America today, would you say that it is now a good time or a bad time to find a quality job?” Responses were coded as Good time (1), Bad time (-1), and Don’t know (0).
- Age was measured as a continuous variable by Gallup.
- Male is the respondents’ gender originally captured with three categories, Male, Female, and Nonbinary, but was recoded to indicate Male (1) relative to all others (0).
- Education was measured in one of two ways, depending on the year the survey was fielded. From 2000 to 2011 it was captured as, “What is the last grade or class that you completed in school? Do you have any post-graduate education beyond a 4-year bachelor’s degree?” From 2014 to 2024 it was captured as, “What is the highest level of school you have completed or the highest degree you have received?” As such, the education variable includes six categories of responses: HS or less, Some college, College grad only, Post-grad, Don’t know, and Refused. This variable was recoded as a four-point ordinal variable from HS or less (0) to Post-grad (3).
- Race/ethnicity of respondents was compiled by Gallup using a combination of questions about race and ethnicity, resulting in a six-category variable. This was recoded to a four-category variable reflecting Non-Hispanic White, Non-Hispanic Black, Hispanic, and Other, with Non-Hispanic White treated as the reference category in analyses.
- Income is coded by Gallup as an 11-point scale and categorized into Less than $10K, $10-20K, $20-30K, $30-40K, $40-50K, $50-75K, $75-99K, $100-149K, $150-249K, $250-499K, and $500K and over.
- Married was measured by Gallup by asking, “Which of the following best describes your marital status: currently married, living together with a partner, widowed, divorced, separated, or never married?” Married was coded as 1, and other responses were coded as 0.
- Full-time employment was captured by asking, “Which of the following best describes your employment situation—employed full-time, employed part-time, retired, a homemaker, a student, unemployed but looking for work, or unemployed and not looking for work?” Full-time employment was coded as 1 and other responses were coded as 0.
- Child under 18 was measured through the question, “Do you personally have any children under age 18?” Responses were coded as Yes (1) and No (0).
- Religious affiliation was recorded by Gallup with the question, “What is your religious preference—are you Protestant, Roman Catholic, Mormon, Jewish, Muslim, another religion or no religion?” This originally captured 22 different religious groups or affiliations, including none or other. Gallup recoded these responses to reflect eight categories—Protestant, Catholic, Mormon, Other Christian religion, Jewish, Other non-Christian religion, Other (unspecified), None/atheist/agnostic, and No response given. This variable was recoded to reflect religious affiliation with any denomination or group (1) as opposed to no religious affiliation, i.e., None/atheist/agnostic (0).
- Gun at home was captured by asking, “Do you have a gun in your home?” Responses were coded as Yes (1) and No (0).
- Region was coded by Gallup and measured through four categories: East, Midwest, South, and West, with East serving as the reference category in the analysis.
County Level Variables
- Violent/property crime rates were generated by aggregating local crime rates to the county level for agencies that reported offense data for the full 12 months of a calendar year. The denominator used to calculate the rates was the sum of the populations covered by these agencies. Coverage within a county may change as the reporting of offenses to the FBI changes. These county-level rates were then merged with the respondent’s county of residence.
- Unemployment rates were obtained from the U.S. Bureau of Labor Statistics Local Area Unemployment Statistics program and represent the number of unemployed people as a percentage of the labor force.
Zip Code Level Variables
- Population was the total population in the zip code.
- Median income was the median household income in the past 12 months, adjusted for inflation.
- Percent juvenile was the percentage of the total zip code population that is under 18.
- Percent non-White was the percentage of the total population that is non-White and non-Hispanic.
- Percent single-headed with children was the percentage of family households in which children are present, and no spouse is present.
- Percent poverty was the percentage of the population that is under the poverty level.
- Different house than a year ago was the percentage of the population that lived in a different house a year ago.
Analytic Strategy
The analyses used pooled logistic regression models with year fixed-effects (2019 was the reference year) and Gallup-provided survey weights. Because some measures were not available in all years, independent variables were introduced sequentially across six model specifications. Each set of models was estimated for all three dependent variables. Missing observations were removed using listwise deletion.
Results
The results presented in this section supplement the results in the main report. Interpretation of findings can be found there.
Table S1 shows the descriptive statistics for the sample. The range of values reflects the recoded data (as noted above).
Table S1. Weighted Descriptive Statistics
Tables S2 through S4 display the logistic regression results. In the main report, Figures S4 through S5 show the percentage change in odds ratios for the logistic regressions; these tables are the underlying untransformed results.
Table S2. Weighted Logistic Regression Results – More Crime in the U.S.
Table S3. Weighted Logistic Regression Results – More Crime in Your Area
Table S4. Weighted Logistic Regression Results – Afraid to Walk Alone at Night
Supplemental analyses revealed that across time, the effects of some relationships generally remained consistent. Figures S1 through S3 show the predicted probabilities of partisanship, ideology, and Christmas spending, by year, for each dependent variable. These models extend into 2023 because they exclude the variable time to find a quality job, which was not collected by Gallup in 2022 or 2023.
Figure S1. Predicted Probabilities for Select Variables – More Crime in the U.S., 2009-2023
Figure S2. Predicted Probabilities for Select Variables – More Crime in Your Area, 2009-2023
Figure S3. Predicted Probabilities for Select Variables – Afraid to Walk Alone at Night, 2009-2023
Endnotes
1 For additional information on the GPSS, visit Gallup’s methodology website. https://www.gallup.com/175307/gallup-poll-social-series-methodology.aspx
2 Steven Ruggles, Sarah Flood, Matthew Sobek, Daniel Backman, Grace Cooper, Julia A. Rivera Drew, Stephanie Richards, Renae Rodgers, Jonathan Schroeder, and Kari C.W. Williams. IPUMS USA: Version 16.0. 2013 American Community Survey: 5-Year Data [2009-2013, Block Groups & Larger Areas]; 2018 American Community Survey: 5-Year Data [2014-2018, Block Groups & Larger Areas]; 2023 American Community Survey: 5-Year Data [2019-2023, Block Groups & Larger Areas];. Minneapolis, MN: IPUMS, 2025. https://doi.org/10.18128/D010.V16.0
3 Kaplan, Jacob, 2024, “Summary Reporting System (SRS) – Offenses Known and Clearances by Arrest (Return A)”, https://doi.org/10.7910/DVN/OESSD1, Harvard Dataverse, V4
4 Bureau of Labor Statistics. Local area unemployment statistics. https://www.bls.gov/lau/lauov.htm
