Via Roma: Did the Bike Lane Cause the Drop?
Via Roma, a two-lane street in a mid-sized city, carried mixed car and bicycle traffic. In 2021 the city recorded 34 cyclist injuries on it. In January 2022 the city opened a protected bike lane along its full length and, in the same month, cut the speed limit from 50 to 30 km/h. In 2023 the city recorded 19 cyclist injuries on Via Roma and announced that the lane had made cycling safer. Automatic counters on the street logged 210,000 bicycle trips in 2021 and 168,000 in 2023. The transport department now wants to extend the design to Corso Verdi, a four-lane arterial carrying roughly three times Via Roma's traffic, where no parallel quiet streets exist.
1. A councillor objects: 'Fewer people cycled, so of course there were fewer injuries - that proves nothing.' Test this objection using the counter data, and state precisely what the figures do and do not establish.
Read the worked answer
Normalise by exposure.
- 2021: 34 / 210,000 × 100,000 = 16.2 injuries per 100,000 trips
- 2023: 19 / 168,000 × 100,000 = 11.3 injuries per 100,000 trips
That is a fall of about 30% in risk per trip, not just in raw counts. So the councillor's objection fails on its own terms: reduced cycling volume explains at most part of the raw drop, and none of the rate drop.
What the figures establish: risk per bicycle trip on Via Roma was lower in 2023 than in 2021.
What they do not establish: that the lane caused it. The 30 km/h limit took effect in the same month, on the same street, and would reduce injury severity and frequency on its own. A single street with two simultaneous changes cannot attribute the effect to either. The correct conclusion is narrower: 'cycling on Via Roma became safer after both changes', not 'the lane made it safer'.
2. Which one of the following, if true, most seriously weakens the city's claim that the lane caused the reduction in injuries? (A) The lane cost €2.1 million to build and has reduced average car speeds on Via Roma by 9 km/h. (B) Two of the 34 injuries recorded in 2021 occurred on a section of Via Roma that was removed from the street's official boundary in 2022. (C) The 30 km/h limit introduced alongside the lane also applied to several streets without protected lanes, where cyclist injuries fell by about a third over the same period. (D) The automatic counters do not record cyclists who ride on the pavement.
Read the worked answer
Answer: (C).
- (A) gives a mechanism by which the lane works (slower cars) - it strengthens the claim, at whatever cost.
- (B) removes 2 injuries from the 2021 baseline. Even granting it, 32 → 19 is still a large fall; the effect is marginal, and it does not supply an alternative cause.
- (C) is a control group. If streets without lanes saw a similar one-third fall under the same speed limit, then the speed limit (or a city-wide trend) is sufficient to explain Via Roma's result, and the lane adds nothing that the data can detect. This attacks the inference directly.
- (D) is a measurement objection that applies equally to both years, so it cannot explain a change between them.
The discriminating principle: a weakener must either break the link between cause and effect or supply a rival cause that is present in the same period. Only (C) does the second.
3. The department argues: 'The lane worked on Via Roma, so it will work on Corso Verdi.' Identify the assumption this extension argument depends on, and say what feature of Corso Verdi makes that assumption hardest to defend.
Read the worked answer
The assumption: that the mechanism responsible for the Via Roma improvement is (i) known and (ii) present on Corso Verdi. Since Via Roma changed two things at once, the mechanism has not been isolated - it might be the protected separation, the 30 km/h limit, or their interaction.
Why Corso Verdi strains it:
- The speed limit is a street-level measure that plausibly did much of the work on Via Roma; whether Corso Verdi gets one is unstated, and a four-lane arterial's traffic speeds are governed by its geometry, not by a lane marking.
- With three times the traffic and no parallel quiet streets, would-be cyclists there have no low-stress alternative today, so the composition of riders - and therefore baseline risk per trip - differs from Via Roma's.
- Induced demand cuts both ways: a lane that attracts many new, less experienced riders can raise the count of injuries even while lowering risk per trip, which will be read as failure.
Sound conclusion: the extension is a reasonable bet, but the evidence base supports a pilot with its own before-and-after counts on Corso Verdi, not an inference from a street that differs on the variables that matter.