Reducing Crime In and Around the University of Cincinnati

Campus is not the campus

None of the recent shootings occurred on university property. All occurred within the surrounding neighborhoods included in this analysis, areas containing student housing, nightlife, commercial corridors and the routes between them. For a parent evaluating the student experience the relevant geography therefore extends past UC's legal property line, which is a normative judgment rather than an empirical one and is offered as such. The unit of analysis here is three concentric distance bands measured from two campus centers. They are an analytical construct, not UC's property line, and every zone label in this brief means a distance band and nothing more.

The cluster is the largest in seven years, and the city around it did not move

Press coverage of a cluster is not evidence of a cluster. The test has to hold everything else constant, so it uses the same twenty-day window, August 17 to September 5, in every year for which Cincinnati has published a consistent offense vocabulary. That is seven windows, 2020 through 2026, drawn from the city's two STARS files. Nothing is spliced across vocabularies and nothing is compared to the older PDI file.

Two things are counted in each window: Part 1 violent offenses within 0.70 miles of campus, and Part 1 violent offenses across the entire city. The second is the control. If Cincinnati as a whole were having a violent late summer, the ring would rise with it and there would be nothing here specific to UC.

2026 is the highest of the seven. It exceeds every prior window, sits 76% above the six-window average of 8.5, and takes the ring's share of citywide violence to 13.2% against a prior range of 5.7% to 10.1%.

The city did not move. Citywide counts across the same seven windows run 115, 99, 111, 129, 123, 106, 114. The 2026 figure is within a point of the six-year average. The local increase was not accompanied by any citywide increase, which makes a general Cincinnati wave an unlikely explanation for the local count. It does not rule out broader spatial fluctuation between police districts, which this design cannot detect.

A longer view is available in the older PDI file, which uses a different offense vocabulary and therefore cannot be joined to the table above. Within that file alone, the same window across 2011 to 2023 shows the ring's violent counts declining steadily, consistent with finding four. It cannot be used to place 2026 on that scale, and it is not used for that here.

The hour still holds. The weekend does not.

The city's response, announced September 4, 2026 and taking effect that evening, closes Calhoun and McMillan between Vine Street and Clifton Avenue from 8pm Friday until 4am Monday, with exemptions for permit-holding residents and employees of nearby businesses.³² Whatever its operational rationale, which may involve crowd size, bar closing times, traffic, staffing economics or event schedules rather than crime distribution, the design concentrates resources on the weekend and overnight. The incident record can test whether violence itself shows the same concentration. On one axis it does and on the other it does not.

Time of day has not moved at all. In the historical record, 42.1% of violence in the Core and Ring fell between 10pm and 4am. In the current record it is 42.2%. That is about as stable as a crime statistic gets, and it is a strong mandate for an overnight response.

Day of week is a weaker story than it first appears, and it is worth being precise about how weak. Historically, Friday through Sunday carried 46.5% of the violence in these zones, above the 42.9% those three days would get by chance. In the current record they carry 40.0%. That looks like a reversal, and I originally wrote it up as one. It does not survive a significance test.

A chi-square test comparing the full seven-day distributions across the two eras returns χ² = 8.47 on 6 degrees of freedom, p = 0.21, with Cramer's V of 0.067. There is no statistically detectable change in how violence distributes across the week. What can be said is narrower: the historical weekend share of 46.5% has a 95% confidence interval of 44.0% to 49.1%, which excludes chance, while the current 40.0% has an interval of 35.3% to 44.7%, which contains it. Weekends used to be measurably elevated. In the current data they are not measurably anything.

The late-night hotspot extends beyond the barricades

Rank every block face in the Core and Ring by violent incidents occurring between 10pm and 4am since June 2024. The top of that list is not Calhoun and it is not McMillan. It is Short Vine, in Corryville, which the weekend closure does not reach. One caveat belongs with the table: these are raw counts, not standardized for block length, licensed-premises count or late-night pedestrian volume, none of which this data contains. A block with more people on it should be expected to generate more incidents.

The three Short Vine blocks carry 26 late-night violent incidents between them. The three closed blocks carry 14. That is a reason to extend the overnight footprint north, not a reason to abandon the closed blocks, and finding 3b gives a positive reason to keep them. Barricading one end of a nightlife corridor while leaving the hotter end open also creates an obvious displacement risk.

The individual incidents from this year's window make the same point. Of the fifteen, one was on West McMillan, three were on Short Vine or Van Street, six were in Avondale, and the fatal shooting on September 1 was on East Shields Street in Corryville at five o'clock in the afternoon⁵, which no overnight street closure would have touched.

Shootings and robberies are not on the same blocks

Everything above counts Part 1 violent crime as one category: homicide, rape, robbery, aggravated assault and strangulation. That is the standard classification, but it merges offenses with different causes and different remedies. A robbery of a student outside a bar at midnight and an aggravated assault between people who know each other in an apartment are both Part 1 violent, and no single deployment addresses both.

It matters here because the response being examined was triggered by shootings specifically. So the right question is not whether these blocks have high violent-crime counts, which finding three establishes. It is whether they are hotspots for the kind of violence the city is currently trying to stop. Cincinnati publishes a separate victim-level shootings file, which answers it directly.⁶³

The two measures point at different blocks, and that is a correction to finding three rather than a footnote to it. On the same three-block comparison, Short Vine carries nearly twice the late-night robbery and assault of the closed blocks, 26 against 14. The closed blocks carry twice the shooting victims, 6 against 3 through September 9. No single Short Vine block appears among the top shooting blocks, which are Burnet Avenue, Maple Avenue, East University Avenue, Blair Avenue, Calhoun Street and Flora Street.

Calhoun is inside the closure, with 4 shooting victims on the 200 block alone. On the specific question of shootings, the city's chosen geography is better aimed than I credited it in finding three, and I have left that finding standing rather than quietly softening it, because it is correct about the offense category it measures and wrong to generalize from.

Then, on Friday, September 11, 2026, a man was shot in the stomach in front of 2607 Short Vine, at 12:34 in the afternoon.⁶⁴⁶⁵ He survived. The city's crime file records it as an aggravated assault on the 2600 block; it is not yet in the shootings file, which runs through September 9. With it the Short Vine count rises from 3 to 4. That is one incident, and a count this small moves by a third on a single event, which is exactly why it should not be read as a trend. It does narrow the gap, and it means Short Vine cannot be described as a corridor without gunfire.

Within 0.70 miles of UC the refreshed file records 92 shooting victims from 2023 through September 9, 2026. Eleven are inside the innermost 0.30-mile band, most at its outer edge on Calhoun Street, plus the two on East Shields Street on September 1. Around Xavier the same radius holds 39. Overnight concentration holds, with 40.2% of shootings in the UC ring between 10pm and 4am, though several of this fall's incidents, September 11 among them, happened in daylight.

That distinction runs through the recommendations. Point one, moving patrol toward Short Vine, is aimed at late-night robbery and street assault on the Short Vine corridor, and the evidence for it is strong. It is primarily a robbery and assault measure and should not be sold as a shooting-reduction plan, even though the corridor has a shooting history of its own. The measure aimed at gunfire is point eight, focused deterrence and violence interruption on the Avondale and Burnet corridor, where the shootings actually concentrate.

The decade context argues against panic

The blocks around UC are considerably safer than they were fifteen years ago. Counting the offenses that Cincinnati classified consistently across the whole period, violence in the Core and Ring fell by roughly half between 2011 and 2023.

This matters for two reasons. It means the current cluster is a departure from a long improvement rather than the continuation of a decline, which is what makes it worth an acute response. And it means the structures already in place are not failing wholesale, which argues against tearing them up.

Xavier is the control. Washington is the robustness check.

The official framing during the cluster was that this is simply what an urban campus involves. UCPD's Lieutenant Brinker told reporters that "campus is extremely safe" and that "off-campus is safe as well, but we are on an urban campus, and things do happen from time to time," adding later that the department was "always battling that perception of safety." That claim is testable, because other universities are also on urban campuses.¹

Parents comparing colleges compare city reputations, and city reputations are built from citywide numbers. Washington DC has a worse reputation than Cincinnati and a higher homicide count. It is also home to a number of well-known universities, four of which are used here. So the useful question is not which city has more crime, but how concentrated that crime is at short distances from a campus. This measures the geography, not any individual's exposure.

The same model used throughout this brief was run around four Washington campuses, using DC Metropolitan Police incident records pulled from the District's open data service.⁵⁶ Identical distance bands, identical date window of June 3, 2024 to September 4, 2026, and offense definitions matched as closely as the two vocabularies allow: homicide, sexual assault, robbery and aggravated assault in both cities. Because the campuses cover different amounts of ground, every figure is a density, incidents per square mile, computed against the true area of each band.

Xavier does the real work here. It sits 2.5 miles from UC and it holds constant almost everything an objection could otherwise be built on: same city, same police department, same prosecutor, same state law, same reporting system, same open-data file, same offense taxonomy, same years, same economy. The only thing that differs is the neighborhood. If Cincinnati-wide conditions were the principal explanation for UC's numbers, Xavier should look substantially more similar than it does.

The Washington campuses are a secondary check, and they answer a different objection: that any urban university looks like this. Three cohorts in total, chosen so that none can be dismissed as a flattering selection.

The Washington northwest cohort is Georgetown, George Washington and American. In the records, incidents inside their inner 0.70 miles fall in Wards 2 and 3, the wealthiest and lowest-violence parts of the District. On their own they would stack the deck.

Howard is the second cohort and the hard case. Its incidents fall overwhelmingly in Ward 1. It is an urban campus in a high-homicide city with genuine violence around it. Adding it roughly halves the headline ratio, from 4.9 to 2.7.

Xavier is the third, and it is the most useful of all, because it removes every remaining objection at once. It sits 2.5 miles from UC. Same city, same police department, same recording practices, same open-data file, same era. Nothing differs except the neighborhood.

The shapes are different, and the shape is the whole argument.

Around the DC campuses the line rises gently and never jumps. Density runs 50 per square mile in the campus core and climbs gradually to 81 at a mile and a half. Moving from the innermost band outward changes almost nothing: 50.1 inside 0.30 miles, 55.0 in the next band. The District's severe violence is real, but it sits east of the Anacostia, several miles and a river from any of these four schools. A student does not walk into it by leaving the library.

Around Xavier the campus core is empty and the curve comes back down. Zero violent offenses inside 0.30 miles across the whole 27-month window, against 35 incidents of any kind. It rises to a peak of 116 in the half-to-three-quarter-mile band, out in Evanston, then falls away to 51. Xavier has a rough band near it. It does not have a rough band against it.

Around UC the line steps up at the edge of the innermost band and never returns. In the quarter-mile bands charted above, concentration runs 71.5 per square mile from 0 to 0.25 miles, then jumps to 179.4 in the 0.25 to 0.50 band, and stays between 115 and 155 all the way out. On the Core and Ring definitions used elsewhere in this brief, the same step reads 83.3 to 149.6. That band is Short Vine, Calhoun, McMillan and the near residential blocks. It is the walk home.

Broken out campus by campus, the inner 0.70 miles looks like this. The spread is wide enough that any single headline number should be treated with suspicion, which is why all of them are here.

Xavier is the finding. UC runs 1.73 times Xavier's violent-crime concentration inside 0.70 miles, with a 95% confidence interval of 1.41 to 2.13. The interval clears 1.0 comfortably, so this is not a small-numbers artifact. The campus cores are further apart than that: 47 violent offenses inside UC's innermost band, zero inside Xavier's. Any explanation offered for UC's numbers has to survive the question of why it stops two and a half miles east, inside the same city and the same police department.

Howard is the robustness check, and it is the hardest case Washington offers: urban campus, high-homicide city, real violence in the surrounding blocks. UC runs 2.39 times higher, CI 1.89 to 3.02. Whatever is happening around UC is not simply what happens around an urban university.

One figure cuts the other way and is worth stating. The share of incidents that are violent is similar at the campus cores, 6.2% at UC against 8.3% across the four DC campuses. Both are theft-dominated places, which is what a campus normally looks like. UC does not have a different kind of crime problem in its core. It has the same kind, followed by a sharp increase in concentration outside the innermost analytical band.

Pick whichever comparison you find fairest. Against the Washington average, UC's ring runs 2.7 times higher. Against Howard, 2.4. Against Xavier, in its own city, 1.7. Against the northwest campuses, 5.2. Every one of them points the same direction, and the disagreement between them is only about how far.

Does any of this depend on where I drew the lines?

A fair objection to any zone analysis is that the boundaries were chosen after seeing the data. The 0.30 and 0.70 mile cuts are walking-distance judgments, but nothing about them is inevitable, and if the findings only hold at those exact radii they are an artifact. So I re-ran everything at five specifications. The result is split, and the half that fails is worth more than the half that passes.

That correction improves the finding rather than removing it, because the underlying spatial fact does not depend on any binary cut. In quarter-mile steps the concentration runs 71.5, 179.4, 114.6, 130.2, 139.8, 154.6 per square mile from campus outward. There is a quiet innermost quarter mile, a sharp peak between 0.25 and 0.50 miles, and a sustained elevated plateau beyond it. That curve is what the chart above shows, it needs no chosen threshold, and it is the honest form of the claim.

So the statement is not "cross a line and violence doubles." It is: violence concentrates in a band that begins roughly a quarter mile from campus and does not fall away again. Wherever the university's actual property line sits relative to that band is a question this analysis cannot answer without parcel data, and it should be answered before anyone acts on a specific boundary.

Does the "police were hamstrung" explanation fit Cincinnati?

It is a fair question to put to the data, and the data gives a mixed answer that does not match either side's talking points. The budget version of the claim does not hold. The staffing version does, for reasons that have little to do with city policy.

The Collaborative Agreement, which is usually the first thing cited, dates to 2002 and grew out of a federal Department of Justice process after the 2001 riots. It changed pursuit protocol, use-of-force reporting, body cameras and mental health response.¹¹¹² It is a court-supervised settlement rather than a recent partisan choice, and both judges and officers involved in its review have described it as having improved the department's ability to work in the neighborhoods it polices.

The sharpest live dispute is about accountability after arrest, not about arrest itself. The head of the Cincinnati police union has argued that juvenile crime is out of control and that juvenile court judges should be detaining violent repeat offenders.¹³ That is a criticism of county courts, not of the city budget, and it points at a real gap this brief can measure. The friction is not abstract: in March 2026 a county prosecutor resigned from the FOP and established an officer credibility list after the union criticized his office.¹⁴ Ohio voters had already amended the constitution in 2022, through Issue 1, to require courts to weigh public safety when setting bail, so the statutory tools for holding violent repeat offenders exist.²²

Robbery is the weak point, and it is the offense that matters most here. It clears at 27.8% citywide, and it is the largest single violent category in the Ring, accounting for 132 of the 360 violent incidents there. In the Ring as a whole, only about a quarter of violent incidents end in an adult arrest. A student robbed near Short Vine has roughly a one in four chance of anyone being charged. That is the number the plan below tries to move.

What the public data can and cannot be asked

Three explanations get offered whenever a city's violence rises: drugs, homelessness and immigration. All three describe real Cincinnati policies. The question here is only whether the incident record can connect any of them to what is happening around UC, and for all three the answer is that it cannot. That is worth stating precisely, because "the data does not show it" and "the data cannot see it" are very different statements, and only the second one is true here.

Where that leaves the argument: if someone wants to test the drug claim properly, the data to request is arrest and citation counts by offense and beat, which the city holds but does not publish in these files. The homelessness claim would need shelter-system and outreach records held by Strategies to End Homelessness. The immigration claim cannot be tested against city records at all, because the city does not collect the variable, so it would have to be argued from county jail or federal data instead.

The city should publish the data that would settle this

Two of these three questions are unanswerable from Cincinnati's public files, and that is a problem the city can fix rather than a fact anyone has to accept. A plan built on data nobody will release is a guess, and brushing past the question is not the same as answering it.

Arrest data is the missing piece, and it was promised in 2016. On October 26, 2016, the city's Office of Performance and Data Analytics announced an expansion of published policing data and stated: "Data related to arrests and traffic stops are being prepared for publication on the portal very soon."³⁷

Traffic stops arrived. As of September 5, 2026 the portal carries four of them, updated daily. Arrests never did. A review of the complete Cincinnati open data catalog on September 5, 2026 returns 102 datasets, roughly twenty of them police data, and not one arrest file.³⁸ That is nearly ten years between the commitment and nothing.

To be precise about what that does and does not mean: the records exist and they are obtainable. Cincinnati Police will release arrest data in response to a public records request under Ohio Revised Code 149.43. What does not exist is proactive publication. The distinction matters, because it is the difference between data a journalist, a community council or a parent can pull up alongside the incident file, and data that has to be requested one person at a time, waited for, and re-requested every time it needs updating. Every other policing dataset in the catalog crossed that line years ago. Arrests did not.

Arrest records are the only dataset that would show whether drug enforcement in these blocks has risen, fallen or stopped, and until they are published on the same footing as the rest, the drug question cannot be argued from evidence by anyone, in either direction, including the people asserting it.

What proactive publication would require is modest, and the city already publishes a harder version of it. A joinable arrest extract needs only: date and time; charge, by statute and plain description; block-level address in the masked form already used for incidents; disposition at arrest; an adult or juvenile flag; and the incident number, so arrests can be linked to the published crime file. It needs no names, no individual addresses, no dates of birth and no demographic identifiers. Traffic stops and pedestrian stops are published today at finer personal granularity than that.

The privacy explanation does not survive contact with the catalog. Cincinnati already publishes traffic stops, pedestrian stops, contact cards, use of force broken out by incident, officer and subject, officer-involved shootings, police firearm discharges and closed Citizen Complaint Authority complaints.³⁸ Every one of those is more sensitive than an arrest record and several are far more politically contested. If those can be anonymized and released, arrests can be. Whatever is holding the arrest file back, it is not a standard the city applies to the rest of its policing data.

Two published datasets would partly close the gap, and this brief did not use them. The Calls for Service file records what residents actually phone in, including disorder complaints that never become offense reports, and the drug overdose feed is a daily, geocoded proxy for where drug activity is concentrated. Neither is analyzed here, which is a limitation of this brief rather than of the data. I do not know whether anyone else has run them against these blocks. Both should be run against these zones, and points four and nine would be the natural place to publish the result.

None of this is a demand for surveillance. It is a request for the aggregate numbers a city already holds, in the same form it already publishes six other public safety datasets. Point nine exists to make that routine rather than exceptional. Until it is, the records can be obtained by anyone willing to file for them, and the specific fields to ask for are listed above.

What the record does show is a signature that none of the three explains. The fifteen incidents in this year's window are robberies and aggravated assaults, on a handful of blocks, overwhelmingly after dark, and eight of them have no charged suspect. The pattern is consistent with concentrated predatory offending, but the public incident data cannot establish that a small number of offenders is responsible, that any of them are repeat offenders, that the victims are UC students, or that the incidents are connected to one another. None of those fields exist in what the city publishes. The lever the evidence actually supports is clearance, and clearance is what points four, five and six below are about.

What families see, and what is shooting people, are two different problems

Parents dropping students off in August describe visible street disorder on Calhoun: panhandling, people who appear to be living outside. It is a complaint I have heard from other parents and it appears in local reporting, and it is a real problem. I have no data on how common it is. It is also not the problem that produced fifteen violent incidents in twenty days, and conflating the two is the fastest way to spend the entire response on the visible half.

Why panhandling is not simply banned

This is constitutional law, not a Cincinnati policy choice. Asking for money is protected speech, and after Reed v. Town of Gilbert in 2015 subjected content-based speech restrictions to strict scrutiny, anti-panhandling ordinances were struck down across the country. Ohio cities have been repealing theirs rather than lose in court. Cincinnati has already been through it: council passed a panhandling permit scheme in 2003, and in 2007 the First District Court of Appeals upheld the time-and-place limits while striking the permit requirement as a prior restraint on speech.²⁴²⁵²⁶

What remains fully enforceable is aggressive solicitation: obstructing a footway, following someone, physical contact, soliciting at a cash machine or transit stop or after dark, threatening language. An officer watching someone hold a sign has no charge available. The same officer watching someone follow a student to a car does. Some of the public frustration here appears to rest on an assumption that none of it is illegal, when the aggressive conduct always has been.

On camping, the law moved in 2024

In City of Grants Pass v. Johnson, decided June 28, 2024, the Supreme Court held 6 to 3 that enforcing public camping ordinances against homeless people is not cruel and unusual punishment, reversing Martin v. Boise.²³ The Eighth Amendment obstacle cities had been operating under is gone. Hamilton County separately holds a permanent countywide encampment ban. On camping, the binding constraint is now shelter capacity and staffing, not legal authority.

Broken windows: the two halves have opposite evidence

The theory is usually invoked as a single thing. It is not. Wilson and Kelling's 1982 argument was about physical disorder and place management; the arrest-driven program it became in 1990s New York is a different intervention, and the research has come out very differently on each.

New York's murder count really did collapse, from roughly 2,245 in 1990 to under 300, and no honest account dismisses that. But violent crime fell about 37% nationally across the same period in cities that ran nothing like broken windows, and NYPD headcount grew by roughly a third under Giuliani and Bratton. Adding officers has good evidence behind it. Misdemeanour enforcement specifically does not. Stop-and-frisk was held unconstitutional in 2013 and curtailed, and New York's crime kept falling afterwards. Kelling himself objected to what his idea had been turned into.²⁹

So the instinct behind broken windows points at the right family of interventions, and this brief already leans on it heavily. The environmental half is in appendix A, points eleven, twelve and sixteen. The place-based policing half is points one, two and four. It is only the arrest-driven version, the part with the weakest evidence, that this plan declines to recommend.

The funding-chain argument, taken seriously

The most common version of the disorder objection is a causal chain rather than an aesthetic complaint: panhandlers spend the money on drugs, the drug money reaches dealers, and the dealers are the armed people doing the shooting. Each link deserves a separate answer, because two of them are broadly right and the chain still does not close.

Link one is substantially correct and should not be soft-pedalled. A 2013 San Francisco survey found 94% of panhandlers reported spending on food but 44% also reported spending on drugs or alcohol, with 32% describing themselves as addicted to drugs and 25% to alcohol.³⁴ Toronto research put median panhandling income at about C$300 a month. Substance use in this population is high and a real share of the money does go to it. Anyone claiming otherwise is not reading the research.

Link three is also correct as a general mechanism. Illegal retail drug markets do generate gun violence, for the structural reason that disputes inside them cannot be settled in court. That is one of the better-established findings in criminology.

Link two is where it breaks, and it breaks on the absence of evidence rather than on a counter-calculation. No public dataset used here measures the number of people panhandling on these blocks, their income, what they buy, or the revenue of any nearby illicit market. I originally attempted an estimate by combining a survey from another city with an assumed local population, and removed it, because that is four layers of extrapolation dressed up as arithmetic. The proposed funding chain simply cannot be quantified from available evidence, in either direction, by me or by anyone asserting it.

The incident record can test whether the local pattern resembles several markers commonly observed in market violence. Victims of market and gang violence often decline to cooperate with police, and Cincinnati records that disposition explicitly. These markers are suggestive rather than a validated classifier, and none of them identifies illicit-market involvement directly.

The blocks around campus show less victim non-cooperation than the city as a whole, more robbery, and a lower homicide share. That pattern is more consistent with opportunistic robbery and street predation than with violence internal to an illicit market, although the incident records identify neither student status nor the relationship between offender and victim. None of the fifteen incidents in this year's window was closed as victim-refusing-to-cooperate.

Where the funding-chain argument does fit better is the Outskirts, and specifically Avondale. Higher non-cooperation, higher homicide share, and the Reading Road blocks recurring as hotspots across fifteen years are all consistent with an entrenched market. Six of this year's fifteen incidents were there. The Avondale pattern is more compatible with that hypothesis than the campus-adjacent blocks are, though the incident record cannot establish illicit-market involvement in either. If the mechanism is operating anywhere nearby, Avondale is the likelier place, roughly half a mile from where the visible disorder is. That is why point eight puts focused deterrence and violence interruption into Avondale rather than treating it as an extension of the Calhoun problem.

Nine measures to act on, and ten to evaluate

Nine recommendations follow directly enough from the analysis above to act on now. Ten more are real but weaker, either evidence-backed elsewhere without being demonstrated here, or structural enough to need a feasibility study first, and those are held in Appendix A rather than padding the list. Each names the body that would have to act, a rough cost band, and the finding it answers. Nothing here requires a new law, and several use municipal powers Cincinnati already holds and is not currently using on these blocks.

Further interventions, for evaluation rather than immediate action

These ten are held out of the main plan deliberately. Each is either an intervention with good evidence elsewhere that this brief did not demonstrate locally, or a structural proposal reaching beyond what incident data can establish. They are worth evaluating. They are not worth committing to on the strength of the analysis above, and a decision-maker handed nineteen requests at once will act on none of them.

A note on where this came from