Rimu High School crossing: judging the reliability of student-collected data
Two Year 11 students investigated whether a pedestrian crossing is needed outside the main gate of Rimu High School (roll 1,450). They stood at the gate from 8:15 to 8:45 a.m. on one Tuesday and one Thursday and tallied pedestrians and vehicles along the 50 m road frontage. Tuesday was fine: 96 pedestrians crossed or walked along the verge, and 410 vehicles passed. Thursday was wet: 58 pedestrians and 380 vehicles, but the students noted that many senior students were away at a sports fixture. They also surveyed 30 students from their own form class; 27 agreed the road was 'dangerous'. Their draft report says: 'A crossing is needed because 94% of students say the road is unsafe and nearly 500 vehicles pass each morning.' The teacher has asked them to check reliability before the report goes to the board of trustees.
1. Identify two weaknesses in the data and explain how each one could change the conclusion.
Read the worked answer
Weakness 1: the sample survey is biased. The 30 students surveyed came from the students' own form class, which is not a random or representative sample of 1,450 students. A form class is likely to be one year level and may share the same route to school. The 94% figure therefore describes one class, not the school. If the survey were repeated across year levels and routes, support could easily fall to 60-70%, which weakens the claim that there is a school-wide safety problem.
Weakness 2: the traffic counts are too few and too short. Two half-hour periods on two days is a very small sample. Thursday was wet and had a sports fixture, so pedestrian numbers dropped from 96 to 58 for reasons unrelated to safety. Neither day samples lunchtime, after school, or winter. The claim 'nearly 500 vehicles pass each morning' also adds Tuesday and Thursday together (410 + 380 = 790 over two mornings, about 395 per morning), which double-counts and overstates the traffic.
Either weakness could reverse the recommendation: a lower level of concern plus lower traffic volumes might point to improved signage and a speed table rather than a full crossing.
2. Explain how the students could improve the validity and reliability of their investigation.
Read the worked answer
To improve reliability (repeatability): count on more days, including at least one full week, both fine and wet weather, and a term-time week without fixtures. Break each half hour into 5- or 10-minute intervals so the counts can be checked, and have two observers count independently to test agreement.
To improve validity (does it measure what it claims?):
- Measure the actual risk factors, not just numbers. Record vehicle speeds (a speed gun or radar sign), the number of pedestrians who cross within 10 m of a blind corner, and how long people wait to cross.
- Count at several times: before school, lunchtime, after school and on a weekend, because the safety problem may be at 3:15 p.m. rather than 8:15 a.m.
- Survey a random stratified sample across year levels and travel modes, with a clear question such as 'Have you had a near miss on this road?' rather than 'Is it dangerous?'.
- Record the location of each crossing and each near miss on a sketch map, so the data shows where the problem is.
A crossing may still be the right answer, but these changes would show whether the danger is real, where it is, and when it occurs.
3. The board asks: does the evidence support building a crossing? Write the judgement you would give, with reasons.
Read the worked answer
Judgement: not yet - the evidence points to a problem but does not yet justify a crossing.
The strongest point in the students' favour is that the road is busy: about 400 vehicles pass in half an hour at the school gate, and pedestrians are sharing a verge with them. That is a real exposure, and it is worth acting on.
The problem is that none of the evidence measures risk. Vehicle numbers alone do not show danger; 400 slow-moving cars in a 30 km/h zone behave very differently from 400 cars at 60 km/h. There is no speed data, no crash or near-miss record, and no count at the times when students actually leave school. The 94% figure comes from one form class and is presented as if it were the whole school, which weakens the report's credibility even where its instinct is right.
The 58-pedestrian Thursday count also suggests the students have not separated weather effects from real behaviour.
What I would advise: install a temporary speed display and repeat the counts for a week, including after school, then return to the board. If speeds are high or near misses cluster at one point, a crossing or a raised speed table is justified. If speeds are already low, improved signage and a footpath may be enough and cheaper.