A few weeks ago I started poking around the city’s traffic-stop dashboard. It’s the one filed under tampa.gov/police/transparency, the one the city points to when anyone asks about who gets pulled over in Tampa. Stops by race, stops by division, stops by gender. The kind of thing you’d want a city to publish.

So I did what I do and slurped up the data and threw it on some graphs. Something looked off.

From launch in 2022 through November 2025 — Hispanic drivers showed up as somewhere between 8% and 10% of all traffic stops. Then, starting in December, it doubled. Hispanic share jumped to nearly 22%. White share dropped from 54% to 42%. Black and everyone else stayed about where they’d been.

It would be easy to jump to conclusions in the current environment.

So I asked TPD to explain the numbers. Here’s what they sent back:

In October and November, an internal audit identified that officers were correctly documenting Hispanic ethnicity during traffic stops; however, a technical issue was discovered that prevented some of that data from transferring properly between two reporting systems. Once discovered, the Tampa Police Department and City of Tampa Technology and Innovation Department worked quickly to address the issue.

Currently, they have corrected the issue for warning data. Work is still underway to correct citation data and update past records. Because these fixes are being made in stages, some current numbers may look inconsistent, but they do not reflect a change in actual police activity as our overall traffic stop numbers remain consistent.

A follow-up named the two systems:

Specifically, the state ticket reporting system (TraCS) classifies Hispanic as a race, while the TPD Records Management System classifies Hispanic as an ethnicity. Because both systems independently transmitted data using different classifications, the information was captured and displayed inconsistently.

What I take that to mean: the state’s ticket system and TPD’s own records system disagreed on whether “Hispanic” goes in the race column or the ethnicity column. The dashboard kept the race column and threw the ethnicity column away. So every Hispanic driver whose stop was logged in TPD’s records — instead of through the state ticket system — got published to the public as White.

That’s how a number stays at 9% for three years.

How many stops are we talking about? The dashboard had 261,548 traffic stops in it at the time of analysis. If the corrected Hispanic share (~22%) is right and the published share (~9%) was wrong, the gap is roughly 12 points. That’s about 31,000 stops of Hispanic drivers that the public dashboard told you were stops of someone else.

For context: Hispanic residents make up roughly a quarter of Tampa.

The part that bothers me more than the bug

Bugs happen. Two systems with two different rules for race and ethnicity getting joined incorrectly is exactly the kind of thing that does happen, and TPD is fixing it. Fine.

What I keep getting stuck on is the timeline.

• July 2022: Dashboard launches. Wrong from day one.
• October–November 2025: TPD’s internal audit catches it.
• May 6, 2026: TPD tells us about it, after we asked.
• Today: The dashboard is still serving the uncorrected historical numbers. TPD says the fix is “expected within the next month.”

Five months between “we know it’s wrong” and “the public knows it’s wrong.” And the only reason the public knows now is that someone outside TPD looked at the data and noticed.

The dashboard’s own description says it exists “for the purpose of providing vehicle traffic stop transparency to the general public.” That’s the job. If the agency operating a transparency dashboard discovers their transparency dashboard is publishing wrong race data and doesn’t say anything for five months, what is the dashboard actually for?

The disclaimer doesn’t get them off the hook

TPD pointed, fairly, to the dashboard’s standing disclaimer — that the data is preliminary, subject to revision, and “should not be used as an official reporting source or compared directly against other finalized reports.”

Sure. Except the same dashboard provides date-range search and historical filtering. It invites comparison over time. It’s filed on the city’s website under “transparency.” A boilerplate disclaimer that says don’t trust this for anything is not the same thing as a notice that says we know the race data has been wrong since 2022 and we’re working on it. Those are very different conversations.

Right now, somewhere on the city’s site, a number is wrong. There’s no asterisk on the chart. There’s no banner across the top. If you went to the dashboard this morning and pulled up “stops by race” for 2024, you’d get the wrong answer, and you would have no way to know.

A note on the numbers

I want to be careful here. Race and ethnicity are categories that government agencies impose on people, and the categories themselves are crude — federal forms treat Hispanic as an ethnicity that cuts across race, Florida’s ticket system treats it as a race, TPD’s records system goes the other way. Reasonable people identify themselves in ways that don’t map cleanly to any of these systems. None of that is the dashboard’s fault.

But “the categories are messy” is not what happened here. What happened is that the system had the information — officers were documenting it correctly, per TPD — and the publication pipeline dropped it. That’s an engineering choice, made in 2022, that nobody outside TPD got told about until 2026.

The corrected numbers, when they arrive, will come from the same pipeline run by the same agency. We’re being asked to take TPD’s word that version 2 is right after version 1 was wrong for three and a half years.

If transparency is meant to build trust and demonstrate accountability and equitable policing, this is the opposite.

Methodology: Numbers in this story come from TPD’s own ArcGIS Feature Service Traffic_Stops_Public/FeatureServer, which is the data source backing the public dashboard. We pulled the full record set on April 30, 2026 — 261,548 stops between July 2022 and April 28, 2026. Scripts are open source. Email if you want them.