When Victims Don’t Call: Police-Cited Nonreporting in the NCVS,1992–2023

Background

Data and Methods

Data from the 1992-2023 NCVS (BJS, 2024) are used to create nationally representative time series of the public’s crime reporting decisions. The NCVS uses multi-stage cluster sampling to interview respondents (age 12 or over) within households. Respondents are kept in the panel for three years and answer questions from survey staff every six months. Weighted data from the NCVS allow one to infer about national-level trends in personal and property victimization and crime reporting (Baumer & Lauritsen, 2010; Lynch et al., 2025).

Over the life of the NCVS, the federal government has changed the collection method for the survey to improve data quality and manage costs (Cantor & Lynch, 2005; Langton et al., 2017; Lynch et al., 2025; Truman & Morgan, 2025). Since 1992, the NCVS has used paper-and-pencil interviewing (PAPI), computer-assisted telephone interviewing (CATI), and computer-assisted personal interviewing (CAPI). The pandemic also modified NCVS data as the government used a mix of techniques to guard against COVID-19 infection, with most interviews occurring over the telephone (Morgan & Thompson, 2021). Changes in survey mode may influence estimates in subtle ways over the life of the survey, so one should bear this effect in mind when evaluating the results.

Using these data, one can create annual-level time series to track emerging trends in crime victimization reporting during periods of negative scrutiny which allows one to conduct an interrupted time series analysis (Shadish et al., 2002). As outlined in the de-policing literature, negative scrutiny may occur during pronounced periods of police criticism through public protests (Kim, 2023; Rosenfeld & Wallman, 2019; Shjarback et al., 2017), with recent scholarship suggesting the biggest instigators of law enforcement criticism happened after the deaths of Michael Brown or George Floyd (Capellan et al., 2020; Kim, 2023, 2024; Powell, 2023; Shjarback et al., 2017). Therefore, one might expect breaks in police-cited non-reporting to occur after these events.

Readers should bear in mind that the data used in this study stem from an incident-level file. All data are derived from instances where someone was a victim of a crime and then made a reporting decision. Victim-driven responses differ from other forms of reporting (e.g., suspicious activity, shots fired) and may not fully encompass all methods of contacting police about offense concerns. The data, while imperfect, still yield valuable insights to the behavior of crime victims. Age categories (derived from Tapp & Coen, 2025), race (Bureau of Justice Statistics, 2021), and sex demographic information for all respondents in this sample are presented in Table 1.

OPEN IN VIEWER

Table 1.

NCVS Incident-Level Victim Demographics, 1992–2023

​	Count	Pct.
Age
12–17	32,117	9.99
18–24	46,015	14.32
25–34	67,320	20.95
35–49	89,457	27.84
50–64	55,611	17.3
65+	30,864	9.6
Race
Hispanic	42,043	13.15
Black, Non-Hispanic	39,713	12.42
Other, Non-Hispanic	17,221	5.38
White, Non-Hispanic	220,845	69.05
Sex
Male	151,415	47.17
Female	169,571	42.83

Measures

In line with the hypotheses, two primary measures are used for analysis. First, as a measure of baseline civilian behavior, the Report Rate was created from responses that asked whether the police were notified of a crime and represents the share of reports divided by the sum of reports and non-reports. Second, Police-Cited Non-Reporting is measured by examining non-reported incidents that indicated the police were the most important reason for not reporting a crime to law enforcement. These include respondents stating the police are ineffective, biased, or would not bother with a crime. Police-related non-reporting answers were summed together and divided by the non-reporting rate to represent the share of law enforcement-aligned non-reporting reasons.

Outside of police-related reasons, respondents can also indicate whether other factors were most important when deciding not to report the crime. Including other non-response options allows one to create a counterfactual against the police-cited reasons to detect (dis)similiarities. Separate time series were created for Minor Issues (e.g., minor/unsuccessful), Practical Considerations (e.g., too inconvenient), Private Matters (e.g., private/personal matter), Social Influences (e.g., advised not to report), Fear (e.g., afraid of reprisal), and Other reasons. All time series were weighted using the incident-level series weight to create national-level estimates (Baumer & Lauritsen, 2010; Lynch et al., 2025). The response options used for each category are summarized in Table 2, and the time series are plotted in Figure 1.

OPEN IN VIEWER

Table 2.

Survey Questions Used for Analysis

Most Important Reason for Not Reporting

Which of these would you say was the important reason why the incident was not reported to the police?

Police-Cited Reasons

Police wouldn’t think it was important enough, wouldn’t want to be bothered or get involved

Police would be inefficient, ineffective

Police would be biased, would harass/insult respondent, cause respondent trouble

Minor Issue

Minor/unsuccessful/small or no loss/recovered property

Child offender, kid stuff

Not clear a crime or harm was intended

Private Matter

Private/personal matter

Social Influences

Did not want offender in trouble with law

Advised not to report

Reported to another official

Practical Considerations

No insurance/or loss < deductible

Police knew too late

Could not recover/identify property

Could not identify offender/lack of proof

Too inconvenient/takes too much time

Fear

Afraid of reprisal

OPEN IN VIEWER

Figure 1.

Reporting rates and reasons for non-reporting in the NCVS, 1992–2023

Results

Table 3 shows the stationarity tests for the report rate and reasons time series, while Table 4 shows the linear regression. Structural breaks are detected in the police, private matter, practical considerations, and other reasons time series, and depicted with a dashed vertical line in Figure 2.

OPEN IN VIEWER

Table 3.

Stationarity Tests for Time Series

	Dickey-fuller	Phillips-Perron	Zivot-Andrews
Variable
Report rate	−2.24	−10.59	−3.18
Police-cited reasons	−0.01	−0.42	−5.15*
Minor issues	−1.96	−9.26	−2.83
Private matter&	−3.15*	−16.54*	−5.66**
Social influences	−2.24	−10.59	−3.18
Practical considerations	−4.33***	−21.05**	−5.54**
Fear	−4.33***	−22.06***	−5.57**
Other reasons	0.05	−0.15	−4.35

OPEN IN VIEWER

Figure 2.

Police-cited, private matter, practical considerations, and other reasons for non-reporting, 1992–2023

Private matters are on a downward trend, and a break is predicted in 2018 (although 8-9% people say this was the most important reason for non-reporting). Around 18% of people cite practical concerns as the most important reason for non-reporting (and has stayed around that level since the late 1990s). Contrary to expectations, changes in police-cited reasons predate major negative scrutiny events, and have doubled since the mid-2000s. People are also less likely to cite “other reasons” for non-reporting.

Prior to the mid-2000s, 10% of people stated the police were the biggest reason why they chose not to report, but now one in five (20%) say that was the most important reason for non-reporting. It is remarkable that the time series has not regressed to a lower value, and that this effect is prevalent in the policing series alone. Put another way, police-cited reasons for non-reporting across the nation have worsened and shown no signs of improvement for about 20 years, even though reporting rates for crime victimization have not substantially changed in the same time span.

The previous figures and analyses focus on police-cited reasons for non-reporting across all offenses. One might argue that the public does not report certain crime types to the police, but otherwise believes the police are competent to handle problems. For example, some literature demonstrates that one may be more likely to call the police for less serious crimes if one perceives the police well (Garofalo, 1977; Xie & Baumer, 2019). As a sensitivity check against crime type issues, the police-cited reasons for non-reporting across crime type (property, violent, and violent crimes excluding simple assaults) and plotting the series in Figure 3 (regression results and stationarity tests are available in the Supplemental Appendix). Although the reporting rate patterns across each crime type are visually similar to the aggregated ones mentioned earlier, property crimes are the only time series with statistically significant differences in police-cited non-reporting over time.OPEN IN VIEWER

Differences Across Age, Race, and Sex

The previously reported analyses focus on general crime reporting patterns without disaggregating by important demographic subgroups that may show different police-cited reasons for non-reporting. In the interest of space conservation, the bulk of the tables and figures supporting the additional analyses are presented in the Supplemental Appendix. Highlights of the analyses are presented in the rest of this section.

Age

Using age categories used by the Bureau of Justice (Tapp & Coen, 2025), a few patterns emerged from the data. Results from interrupted time series analyses show that certain age groups were more likely to cite the police as the most important reason for not reporting a victimization. Statistically significant results were found in younger respondents (12–17, 18–24, and 65+), and estimated breaks are shown in Figure 4. Panels without vertical lines are shown for comparison, but do not have statistically significant breaks in police-cited non-reporting.OPEN IN VIEWER

Figure 4.

Structural breaks in police-cited nonreporting reasons across age categories (12–17, 18–24, 65+), 1992-2023

Race

Figure 5 shows the results of statistically significant patterns in police-cited non-reporting across race. As shown in the graph, Black, Non-Hispanic and Other, Non-Hispanic people were more likely to indicate the police as the most important reason for not reporting crime relative to Hispanic and White, Non-Hispanic respondents. Similar to the main analyses, this break appears in the mid-2000s.OPEN IN VIEWER

Figure 5.

Structural breaks in police-cited nonreporting reasons across race categories (Black, Non-Hispanic and Other, Non-Hispanic), 1992–2023

Sex

Finally, Figure 6 shows police-cited non-reporting differences between male and female respondents. According to the statistical analysis, female respondents have not experienced a statistically significant change in police-cited non-reporting reasons. However, male respondents experienced a shift in this area, with a notable break shown in 2006. OPEN IN VIEWER

Figure 6.

Structural breaks in police-cited nonreporting reasons for male respondents, 1992–2023

OPEN IN VIEWER

Table 4.

Linear Regressions of Time Series Data and Structural Break Tests

	Report Rate-D	Police	Minor Issues-D	Private matter	Social Influences-D	Practical considerations	Fear	Other - D
Variable
Lagged DV		0.99 (0.05)***		0.13 (0.16)
Trend				−0.00 (0.00)*		0.00 (0.00)***
Intercept	0.00 (0.00)	0.01 (0.01)	−0.00 (0.00)	1.37 (0.52)***	−0.00 (0.00)	−1.14 (0.32)***	0.02 (0.00)***	−0.00 (0.00)
Structural Break Test
Wald statistic	4.87	69.73***	1.84	27.01***	3.62	42.26***	5.87	11.98***
Break year	N/A	2006	N/A	2018	N/A	1998	N/A	1998

Regression results are presented with coefficients and standard errors (in parentheses). A D-indicates variable was differenced during estimation. All regressions used robust standard errors. Not all regressions required lagging the dependent variable, differencing, or a time trend which is why some cells are missing information.

Discussion

Not everyone may like the police, but calling the police to report a crime represents a basic trust exercise in law enforcement’s ability to address a civilian’s concerns about criminal activity. This paper demonstrates that there are no substantive changes in crime reporting behavior - people still call the police, reporting crimes at a similar rate over time. However, survey takers are twice as likely to cite concerns about law enforcement as the primary reason for non-reporting and, notably, this has persisted for the last twenty years. This translates to about 20% of people do not report a crime to police because of law enforcement.

After disaggregating by age, offense type, race, and sex, one can find similar effects across demographic groups. Notably, male respondents, younger people, and Black, Non-Hispanic and Other, Non-Hispanic, and victims of property crime roughly doubled their likelihood of stating the police were the most important reason for not reporting their victimization. The observed effects predate the deaths of Michael Brown and George Floyd which was contrary to expectations.

As noted before, this effect is somewhat surprising as this shift in public confidence occurred before some of the most recent police misconduct stories in America (e.g., Michael Brown, George Floyd) which have been widely studied (Capellan et al., 2020; Kim, 2023, 2024; Powell, 2023; Rosenfeld & Wallman, 2019; Shjarback et al., 2017). Coupled with the initiation of the change, the effect appears permanent and persistent over time with no sign of regression to a lower value. These data, then, suggest a long-standing new norm for non-reporting reasons.

Using an interrupted time series approach allows one to build upon prior literature that focuses on trends of crime nonreporting (Xie et al., 2024). Moreover, the emphasis on national-level time series estimates that cover the last thirty years and other non-reporting reasons allow one to assess whether any observed effect is unique to policing or is part of a broader trend in reason shifting. And, as shown by the analysis of age, sex, and race/ethnicity trends, the police-cited nonreporting trend appears across various demographics. Thus, the study adds to a growing evidence base that shows substantive changes in the reasons for not calling the police (even though people call the police at a similar rate).

As with any survey, one should evaluate whether responses reflect the true views of survey takers or if some external force changes public opinion. One might assume a social desirability bias made police disillusionment a fashionable choice for stating why someone did not report a crime. However, even if many responses are driven by a social desirability bias, the persistent citation of police-related reasons for non-reporting over the last 20 years is still remarkable.

The reporting rates offer an opportunity to unpack why the trends appear this way. One possibility lies in the public’s overreliance on police to address a growing array of issues (Lum et al., 2022; Rouhani et al., 2025). Calling the police has become easier with increased access to mobile phones in the United States, which enables the public to call anytime from anywhere (Lum et al., 2020, 2023). Further, the advent of security systems placed in homes and businesses has pressured police to attend to burglar alarms, creating a new demand on police resources to respond to outstanding security alerts (Farrell et al., 2011, 2014). Public disbelief in police caring about crime, or law enforcement’s ineffectiveness to address offending, may reflect the implicit recognition of law enforcement’s resource scarcity problem and inability to address all civilian concerns.

The growing discontent with police as it relates to crime reporting presents an opportunity to examine other public responses to police behavior. For example, a large body of research emphasizes the need for the police to work with the public to solve crime problems and foster strong community relationships (Braga et al., 2015; Gill et al., 2014; Goldstein, 1979, 1987; Hinkle et al., 2020; Weisburd et al., 2010). In some ways, the onus is placed upon police to seek out connections with the public, even though the public must also choose to cooperate with law enforcement. If one assumes that more positive attitudes correspond with a greater willingness to work with police, then one might assume negative attitudes chill public cooperation with law enforcement (McLean et al., 2024; Sunshine & Tyler, 2003; Tyler, 2004). A natural extension of this study, then, is to examine how public participation with police has changed over time. Studying this effect may illustrate whether the public, which still appears to report crimes to police at a consistent level, has chosen to disengage with law enforcement in other ways.

Public participation may take myriad forms. The public can choose to attend community policing events, report crime and other disorder to law enforcement, engage in casual conversations with law enforcement, or find another form of connection with law enforcement. Public participation patterns with law enforcement are largely unknown, along with any knowledge about public participation engagement changes after negative scrutiny events. If other forms of engagement follow a similar pattern to the findings presented in this paper, one might expect other forms of public aversion toward police.

Further, as noted by a reviewer, the study uses crime reporting as a proxy for willingness to call the police for assistance with crime problems using responses from a victimization survey. Using the incident-level file constrains the population to crime victims who may be predisposed to asking for assistance. Victim reporting may differ from other calls to police related to crime (e.g., reports of suspicious behavior, shots fired), so a more holistic measure of civilian perceptions and cooperation could use a mix of public-instigated contact with police as a measure of law enforcement avoidance.

The findings show a national-level negative shift in policing sentiment affected the reasons why people do not report crime to the police. Observing this shift in negative sentiment raises the question of an opposite effect, where positive opinions of police may transform civilian attitudes and behaviors related to law enforcement. Much of the de-policing literature examines how negative scrutiny curtails police proactivity with potentially community-endangering effects (Capellan et al., 2020; Kim, 2024; Rosenfeld & Wallman, 2019). However, if negative scrutiny can predict negative behaviors, then the opposite effect may occur if law enforcement is viewed in a positive light (say, “re-policing”). Positive views of the police may signal an endorsement of law enforcement activities and imbue the police with the security to proactively police the community to reduce crime. This effect would differ from a normal regression effect where community sentiment reverts to a baseline level where police continue a “normal” amount of proactive policing. Studying these effects would allow law enforcement to better understand how public sentiment directly impacts police department performance and improvements to police-civilian interactions.

However, the national-level aggregation used in this study does not permit such an analysis and necessarily conceals agency-level variation in the hyper-decentralized American police system. Positive effects may be most prevalent if agencies adopt community policing practices or commit to procedural justice principles (Gill et al., 2014; Mazerolle et al., 2013a; Peyton et al., 2019). Traditional survey methods may unmask trends in sentiment across the United States, but incur substantive costs and take several years to collect data. Instead, one might employ sentiment analysis techniques of police social media accounts to understand community-level variation in police sentiment and how it relates to successful policing (Fallik et al., 2020; Intravia et al., 2020; Ralph, 2022). Using these methods, then, enables one to parse out sentiment changes and police-related outcomes over time.

Although this was not the focus of the study, one should note that important variations in call reporting may occur across demographics and the lifecourse. For example, it is well-known that younger people often commit more crime and are victimized most often (Hashima & Finklehor, 1999; Johnson & Lazarus, 1989; Menard, 2012). At the same time, younger people show a greater distrust of law enforcement relative to older peers (Murphy, 2015; Sindall et al., 2017), despite a larger exposure risk to law enforcement contact. This suggests that some individuals later in life choose to either trust the police or revise their perspective on law enforcement and may be subject to a “turning point.” A change in attitudes may be ascribed to one’s willingness to trust law enforcement to protect one’s spouses and children, one’s perceived change in vulnerability to criminal activity, or one’s acquisition of property and assets that need protection. In short, one might expect a developmental process of police reliance tied to the development of one’s social ties, stakes in conformity, maturation, or some mix of these factors (Laub & Sampson, 1993; Menard, 2012; Sampson & Laub, 1992, 2016).

Independent of one’s reliance on police, the public’s ability to recognize crime may change over time. A greater awareness of potential crimes, their associated impacts to the public, and law enforcement’s role in addressing these crimes. Studying changes in this behavior is important because one can assess whether education efforts (local, state, or national) broaden awareness of emerging crimes and enable crime prevention efforts (Goldstein, 1979, 1987). This may prove especially important as one grows familiar with conventional street crimes but loses sight of new offenses on the horizon (e.g., identity theft). Exposing the gaps between public knowledge and reactions to crime may spotlight new areas for law enforcement authorities to improve crime education.

The NCVS collects survey data to unmask lurking crime trends in the United States and necessarily creates American-centric estimates. However, the focus on American–related attitudes limits the ubiquity of these findings to the broader international community. As outlined earlier in the discussion, changes in attitudes toward police may coincide with security infrastructure investments and new demands on law enforcement. Yet, despite these changes, it remains unknown whether police-attributed reasons for non-reporting have grown in other countries. This is an important question since it allows one to compare national trends in reporting practices and assess the universality of police-cited reasons for non-reporting.

Police-public partnerships have changed in a way where people often say the police themselves are the most important factor for not reporting crime. Given law enforcement’s mission to protect and serve the public, this paper shows a sort of public disillusionment with police that has persisted for several years. One of the biggest challenges for police, then, is to define the future of police-civilian relationships and gain more public trust in basic police functions, despite persistent downturns in public attitudes toward law enforcement.

Supplemental Material