A service change lands on your board agenda. Somewhere in the packet, someone has to answer one question: does this change fall harder on minority or low-income riders than on everyone else?
That question is Title VI. And for most small agencies, the honest answer is "we think it's fine" with nothing behind it. Let me show you the actual test, because it is simpler than the compliance language makes it sound.
The one rule to remember
FTA Circular 4702.1B asks you to compare two groups for any major service change: the protected group (minority riders, or low-income riders) and the non-protected group (everyone else). If the protected group absorbs a disproportionate share of the harm, you have a potential disparate impact, and you have to justify it or change the plan.
The common yardstick is the four-fifths rule: if the protected group is affected at less than 80% of the rate of the comparison group, that is your flag.
Say that you cut a route
Say Route 12 runs through a corridor that is 55% minority, while your system as a whole is 30% minority. You propose cutting its midday frequency.
Work it as rates:
- Riders losing service who are minority: 55%.
- Riders losing service who are non-minority: 45%.
- Ratio of non-minority impact to minority impact: 45 / 55 = 0.82.
Flip it to the burden side. The minority share of the people you are hurting (55%) is well above the minority share of the system (30%). The harm is concentrated on the protected group. That is a disparate impact you now have to address: restore some service, add a mitigation, or document a substantial legitimate justification with no less-discriminatory alternative.
Change one number and the story changes. If that same corridor were 28% minority against a 30% system, the burden would track the system and the flag would not fire. The test is only as good as the demographic data you run it against, which is exactly why it belongs on your own Census numbers, not a national average.
Why this is worth doing before you commit
Most agencies run this analysis once, after the plan is basically decided, as a document the board needs. By then the four-fifths result is something you defend, not something you use.
Run it the other way around. Test the change while it is still a sketch, see the disparate-impact result the same afternoon, and adjust the alignment before it ever reaches a hearing. The math does not get easier later. The options do get fewer.
That is the whole point of putting the four-fifths test on your own GTFS and Census data: not to produce a prettier report, but to move the equity check from the end of the process to the start of it.
Run it before you commit.
