Reading the E-Scooter Pilot Report Before It Reaches the Board
Riverside University's sustainability office ran a one-semester e-scooter pilot with 40 scooters and two docking hubs, one beside each of the two largest dormitories. The office logged 1,200 rides. A survey link was emailed to all 4,000 students; 380 students completed it, and 81 percent of those respondents agreed that the pilot 'reduced my need to use a car'. Ride records show that 68 percent of trips started and ended within 300 metres of the two dormitory hubs, the average trip was 0.9 km, and only 12 percent of trips occurred between 07.00 and 09.00. The office's draft report proposes expanding the scheme to the whole campus and states that the pilot 'cut car use among students'. The student council has asked for a review of that sentence before the report goes to the board, and wants each objection tied to a specific number rather than to general doubts about surveys.
1. What is the strongest conclusion the pilot data can actually support? Rewrite the disputed sentence accordingly.
Read the worked answer
The data support a claim about uptake and geography, not about car use. In one semester the scheme produced 1,200 rides, heavily concentrated near the two dormitory hubs, with 68 percent of trips beginning and ending within 300 metres of those hubs and an average length of 0.9 km. That pattern is consistent with short trips that would otherwise have been walked. A defensible rewrite is: 'The pilot was used mainly for short trips around the two dormitory hubs.' The original sentence, 'cut car use among students', needs evidence of substitution from cars, which the ride log does not contain and the self-report survey cannot establish on its own. The 12 percent share of trips in the 07.00-09.00 window also undercuts the commuting story, because car commuting would cluster in that window.
2. Identify the main threat to the validity of the 81 percent figure, and show the arithmetic that makes the threat serious.
Read the worked answer
The threat is self-selection plus a low response rate. Completion was 380 of 4,000 students, or 9.5 percent. The 81 percent who agreed equals about 308 students, which is 308/4,000 = 7.7 percent of all students. So the headline number describes a small, self-chosen minority who bothered to open an emailed link, plausibly students already favourable to the scheme. The remaining 90.5 percent of students are unobserved, and no information is given about whether they are indifferent, hostile, or simply ignore email. The figure is not wrong as a description of respondents, but it cannot be generalised to the student body. A defensible report would state the response rate beside the percentage, or use a random sample with follow-up.
3. What evidence would best test whether the pilot reduced car use, and why is it better than the survey?
Read the worked answer
The best evidence is behavioural and comparative: counts of car entries to campus parking or barrier-gate logs before and during the pilot, ideally compared with a term without the scooters or with a campus area that had no hubs. Trip diaries for a random sample, or GPS traces showing where riders came from, would show whether a scooter trip replaced a car trip or a walk. The survey is weaker because it asks students to recall and interpret their own behaviour, and it was answered by a self-selected 9.5 percent. The ride log is also insufficient: it records that a trip happened, not what it substituted for. Parking counts directly measure the outcome of interest, and a comparison period controls for weather, timetable changes, and fuel prices.