About
Martin Ngoh
Justice Lens is built and maintained by one person. This page says who, why, and how to reach him.
Background
I am a data scientist, and for the past five years my work has been in econometrics and causal inference: the discipline of asking, when two groups are treated differently in the data, how much of the gap can be attributed to the decision being studied and how much to everything else that differs between the groups. The methods on this site are the standard instruments of that literature applied to public records: representation indices against explicit baselines, the veil-of-darkness design, the search outcome test, a Bayesian threshold model, interrupted time series with placebo checks, and a pre-registered difference-in-differences whose pre-trend test fails and says so.
Why I built this
The Metropolitan Police Department publishes its stop, arrest, incident, and ticket records through DC Open Data. What it does not publish is a baseline. A count of stops with no denominator is an insinuation, not a finding, and claims about policing in Washington, from every direction, tend to circulate that way: a raw number, no comparison, no method, no caveat.
I wanted one place where the standard tests were run on the public record and left running. Every figure on Justice Lens is recomputed from the latest MPD release when the page loads, the method sits beside the number, exclusions are counted rather than quietly dropped, and a direction is asserted only when a statistical test supports it. The site publishes whatever the data returns. Some results point to disparity; some favor MPD. Both kinds appear at the same prominence, because the credibility of the method depends on reporting what it finds either way.
The audience I had in mind is the reader who has to defend a number to someone else: a reporter, a litigator, a researcher, a resident at a council hearing. Each analysis page is written so that its figure can be traced to a dataset, a window, a sample size, and a definition, and so that its limits are on the page rather than in a footnote.
How the site is run
Justice Lens is free, independent, and not affiliated with, endorsed by, or operated by the Metropolitan Police Department or the District of Columbia government. It is operated by Specta LLC. The statistical methods are documented on the methodology page; the engineering, including the data pipeline, is described in how it works. The source papers behind the veil-of-darkness, outcome, and threshold tests are cited on each analysis page.
Contact
Corrections, data questions, methodological objections, and requests for the underlying queries are welcome at [email protected]. If you find a number on this site that does not trace to a labeled source, that is a bug, and I would like to hear about it.