Two Sources, One Claim: Did the Oak Street Bike Lane Cut Crashes?
Priya is writing a one-paragraph report on whether her city's protected bike lane on Oak Street reduced crashes. Source A is a city transportation department count: reported crashes on Oak Street fell from 14 in 2018 to 9 in 2019, the year the lane opened; the same report notes that Oak Street traffic volume dropped 11% in 2019 because a nearby bridge closed for repairs. Source B is a neighborhood blog post headlined "Crashes Are Up," which cites 3 crashes at the Oak/5th intersection in January 2019 versus 2 in January 2018. Priya has space for one paragraph, must keep every number she uses accurate, and her teacher has banned the word "proves."
1. Which source better supports the claim that the bike lane reduced crashes, and what are the specific limits of that support?
Read the worked answer
Source A is stronger, but neither source supports a causal claim.
- Source A is a systematic, full-year count from the agency that records crashes: 14 → 9 across 12 months. That is a real decline.
- Limit 1: the same report gives an alternative explanation - traffic on Oak Street fell 11% when the bridge closed. Fewer cars can mean fewer crashes regardless of the lane.
- Limit 2: with counts this small, a drop of 5 crashes is unstable; 2–3 fewer bad-weather crashes could produce it.
Source B is weaker on evidence, not just tone. It compares a single intersection over a single month, and the entire "increase" is one crash (3 vs. 2). No total volume, no comparison street, no method. A blog post is also not a systematic count.
Defensible sentence for Priya: "Reported crashes on Oak Street fell from 14 to 9 in the year the bike lane opened, though an 11% drop in traffic that year complicates any single cause."
2. Priya's draft reads: "The bike lane caused crashes to fall by 36 percent." Check the arithmetic, then diagnose the reasoning problem and rewrite the sentence.
Read the worked answer
Arithmetic check: (14 − 9) ÷ 14 = 5 ÷ 14 = 0.357… ≈ 35.7%, which rounds to 36%. The percentage is computed correctly off the raw counts.
Reasoning problems (two, and the percent is not one of them):
- "Caused" overreaches. The design is a before/after comparison with no control street and a known confounder (the 11% traffic drop). Before/after alone cannot isolate the lane.
- Small-number fragility. The whole 36% rests on 5 crashes. One reclassified report changes the percentage by several points.
Rewrite: "Reported crashes on Oak Street fell from 14 to 9 - about 36% - in the year the protected lane opened, although an 11% decline in street traffic that same year means the drop cannot be credited to the lane alone."
3. Priya can request exactly one more dataset before she files the report. What should she ask for, and how would it change what she can say?
Read the worked answer
Ask for the same 2018–2019 crash counts on a comparable street that did not get a bike lane (ideally one that also lost traffic to the bridge closure).
Why this one: it supplies the missing comparison. If the control street's crashes also fell by roughly a third, the lane adds nothing to the explanation. If the control street held steady or rose while Oak Street fell, the lane becomes a much more plausible contributor.
Second-best alternative: Oak Street crash data expressed as a rate per vehicle-mile rather than a raw count. That directly neutralizes the 11% traffic drop instead of leaving it as an open objection.
What Priya gains either way: she can move from "crashes fell" (descriptive, safe) toward "crashes fell more than on a comparable street" (comparative, still not proof). Without one of these, her paragraph must stay descriptive.