Your guide to NYC's public proceedings.
Q&A
Council Member Won and DOT discuss injury data and statistical significance
1:10:16
·
96 sec
Council Member Julie Won presses Eric Beaton for more details on the DOT's analysis of injury data and statistical significance in their daylighting study. Beaton provides additional information on their methodology and data analysis process.
- Won asks about the relationship between increased injuries and the absence of parked cars
- Beaton clarifies that intersections with daylighting not from traffic safety studies showed higher injury rates
- Won questions the statistical methods used, particularly regarding multiple comparison corrections
- Beaton explains the various filters and tests applied to ensure statistical significance across thousands of intersections
Julie Won
1:10:16
But that's why we're asking for hardened daylighting so that people can't park there.
Eric Beaton
1:10:20
Yeah, and we are very supportive of hardened daylighting.
1:10:22
We think it is helpful.
Julie Won
1:10:25
So let me just remind you the questions.
1:10:27
How can DOT claim?
1:10:28
So for DOT, you could definitively say that an increase in injuries at these locations were the result of cars not being present and being greater visibility.
Eric Beaton
1:10:38
Was said is that locations that have this daylighting that did not come from a traffic safety study, those intersections have a higher rate of injuries than intersections that do not.
Julie Won
1:10:52
For those locations, there are just two different spots, soft daylighting and 52 foot long bus stops
Gale A. Brewer
1:10:59
with
Julie Won
1:10:59
pedestrians.
1:11:01
So can you help me understand, given that analysis includes hundreds of comparisons, did you apply multiple comparison corrections control for false positives?
1:11:09
And if not, how can you claim that any of these hydrant effects were statistically significant?
Eric Beaton
1:11:14
Yeah.
1:11:14
Absolutely.
1:11:15
In in looking at thousands of locations, we we put it through a number of different filters to try to see if anything was showing a statistically significant result.
1:11:25
And, you know, we we looked we, you know, tried to if you separate it by borough, does it it work better in some boroughs or others?
1:11:31
We looked at it by, traffic signal versus stop sign.
1:11:35
And do one of those show a difference?
1:11:38
And so when we say that we didn't look at a variable, that's not to say that we didn't look at it.
1:11:43
We did a test and looked at the statistical significance and found that it wasn't there despite comparing across these many thousand intersections.