Why Crime Statistics Can Be Misleading

The Dark Figure of Crime

Criminologists use the term "dark figure of crime" to describe the gap between crimes that actually occur and crimes that appear in official statistics. This gap is substantial.

The Bureau of Justice Statistics (BJS) conducts the National Crime Victimization Survey (NCVS), interviewing roughly 240,000 individuals annually about their crime experiences, including incidents they did not report to police. The NCVS consistently reveals that a large share of crime goes unreported:

Several factors influence whether victims report crimes: the perceived severity of the offense, whether the victim believes police can help, trust in law enforcement, fear of retaliation, and whether insurance requires a report (which explains why motor vehicle theft has the highest reporting rate).

The implication is significant: any comparison based on FBI UCR data is comparing reported crime, not actual crime. A city with higher community trust in police may have higher reported crime without necessarily having more actual crime than a city where residents are less likely to call police.

Voluntary Reporting Gaps

FBI UCR participation is voluntary. While agencies covering approximately 98% of the U.S. population participate in some form, "participation" varies in quality:

• Partial-year reporting. Some agencies report for fewer than 12 months. The FBI prorates data for agencies reporting 6-11 months but excludes those reporting fewer than 6 months entirely.
• Missing categories. An agency might report violent crime data but not property crime, or vice versa. This creates inconsistent coverage within a single jurisdiction.
• Non-participating agencies. Some jurisdictions simply do not participate in the UCR program. Their crimes are invisible in the data.
• NIBRS transition gaps. The FBI's mandate that all agencies transition to the National Incident-Based Reporting System (NIBRS) by January 2021 caused significant disruption. Agencies that had stopped using the old Summary Reporting System but were not yet certified for NIBRS simply disappeared from the data. In 2021, the FBI received data from agencies covering only about 65% of the population, down from 97% the prior year. Coverage has since improved, but the 2021 data year remains compromised.

For more on how the UCR program works, see our guide on FBI crime statistics explained.

The Small-Population Problem

Crime rates expressed per 100,000 population are essential for comparing places of different sizes, but they can be deeply misleading for small populations. This is a statistical phenomenon, not a flaw in the data itself.

Consider this example: a town of 2,500 residents has 1 murder in a year. Its murder rate is (1 / 2,500) × 100,000 = 40 per 100,000. That is roughly seven times the national average. But it reflects a single, possibly random event. The following year, with zero murders, the rate drops to zero. Neither number tells a useful story about safety.

This is why credible crime rankings apply population minimums. The FBI itself advises against ranking cities and cautions that many factors beyond what statistics capture influence crime. When using PlainCrime, keep in mind that cities with fewer than 25,000 residents are more susceptible to rate volatility. Our rankings apply population thresholds to reduce this effect.

Methodology Differences Across Jurisdictions

Even within the UCR framework, agencies have discretion in how they classify offenses. This creates inconsistencies:

• Aggravated vs simple assault. The line between aggravated assault (a Part I violent crime) and simple assault (a Part II offense) depends on the severity of injury or the use of a weapon. Different departments may classify borderline cases differently, leading to incomparable rates.
• Larceny thresholds. Some states define "grand larceny" (felony theft) at $500, others at $1,000 or more. These thresholds affect how offenses are categorized and reported.
• Rape definition transition. The FBI updated its definition of rape in 2013. Agencies adopted the new definition at different times, making rape statistics unreliable for cross-jurisdictional comparison during the transition period.
• Arson investigation. Many fires are initially classified as accidental or of unknown origin. Jurisdictions with more thorough fire investigation may identify more arsons, appearing to have higher arson rates.

FBI Data vs BJS Data

The United States has two major systems for measuring crime at the national level. Understanding their differences is important for interpreting crime statistics:

These two systems sometimes tell different stories. The NCVS may show rising victimization while UCR data shows stable or declining reported crime, or vice versa. Neither is "wrong" -- they measure different things. Together they provide a more complete picture than either alone.

Media Perception vs Statistical Reality

Public perception of crime is shaped more by media coverage than by data. This creates a persistent gap between perception and reality:

• The availability heuristic. Dramatic crimes (mass shootings, kidnappings, sensational murders) receive intensive coverage, making them feel more common than they are. Cognitive research shows people estimate the frequency of events based on how easily examples come to mind.
• Coverage asymmetry. A single murder in a community of 50,000 may receive days of local news coverage. A year-long decline in burglaries across the same community receives little or no coverage. Bad news is news; good trends are not.
• The Gallup gap. Gallup has polled Americans on crime perceptions for decades. In most years since 2001, a majority of respondents say crime is increasing nationally, even during periods when FBI data showed substantial decreases. This perception-reality gap is one of the most consistent findings in public opinion research.
• Social media amplification. Ring doorbell videos, crime alert apps, and social media sharing create an impression of rising crime by making individual incidents far more visible than they were in the pre-social-media era, even when underlying rates have not changed.

How to Use Crime Data Responsibly

Despite these limitations, crime data remains valuable when used with appropriate care. Here are principles for responsible interpretation:

1. Look at trends, not snapshots. A single year of data is a snapshot that may be influenced by anomalies, reporting changes, or random variation. Trends over 3-5 years are far more informative.
2. Compare like with like. Compare cities of similar size, similar urbanization levels, and within the same data year. Comparing a college town of 30,000 to a major metro of 2 million is not meaningful.
3. Distinguish rate from count. Always use per-capita rates (per 100,000) for comparisons, never raw numbers.
4. Check reporting completeness. If an agency reported for only 9 months, its annual totals will be lower than a full-year reporter. The FBI adjusts for this, but raw data tables may not reflect the adjustment.
5. Separate violent and property crime. Composite "crime rate" rankings that blend the two can obscure important differences. A city with high shoplifting but low violence is a very different place than one with the opposite profile. See our guide on violent vs property crime.
6. Consider context. Crime data alone does not explain crime. Poverty rates, population demographics, policing strategies, housing patterns, and dozens of other factors influence both actual crime and reporting rates.
7. Be skeptical of rankings. Any ranking that declares one city "the most dangerous in America" is making a dramatic claim from imperfect data. The FBI explicitly cautions against this practice.

What PlainCrime Does About These Limitations

We present FBI UCR data as transparently as possible. On every city and state page, we show the data source and year. We apply population thresholds in our rankings to mitigate the small-population problem. We show violent and property crime separately rather than blending them into a single misleading number. And we publish these guides to help users interpret the data with appropriate context.

No dataset perfectly captures the complex reality of crime in America. But FBI UCR data, understood within its limitations, remains the best available source for comparing crime across thousands of U.S. jurisdictions. Use it as a starting point for understanding, not as a final judgment.

Frequently Asked Questions