Considerations regarding the "Shortest Warning Time" in the range of vehicles
Some people believe that because it is not possible to observe every vehicle that might ever approach in the future at a given crossing,
we can never be certain that we have enough warning time from all the vehicles -- every crossing should always be considered a Situation of Uncertainty (Bennett, 1991).
It's true that we can never be certain about anything -- we can't even be certain that the sun will rise tomorrow -- but we can be confident it will rise as expected, based on our observations.
So, how many approaching vehicles do we need to sample in order to feel confident that the range of warning times in that situation is longer than our crossing time?
At the request of Dr. Everett ("Butch") Hill shortly before he died, Dr. Mary-Maureen Snook-Hill and the O&M Department at Peabody College did a preliminary study of this issue in 1995 (Snook-Hill, M and Sauerburger, D. 1996).
We were interested in how to ensure that in our assessments we were including the vehicles with the shortest warning time.
When measuring the detection-to-arrival time of 20 vehicles in 24 situations, we found that one of the vehicles with the shortest warning time from a given direction was among
the first 12 approaching vehicles 95% of the time.
It can take anywhere from 15 minutes to 3 hours to observe and record the warning times of 12 vehicles from each direction, which would be time well spent if we could analyze crossings once and be done.
However the situation can change from day to day and even minute to minute, so a careful analysis of the crossing at one time of the day does not tell us anything about the situation at that crossing at other times.
It is not really feasible to take that long to analyze the situation each time you want to cross so, again, how many approaching vehicles do we need to sample in order to feel confident that it will be clear to cross whenever it is quiet?
Researchers often are faced with this question -- how much data must be collected before concluding that there is no difference between the control and the experimental group?
They have to balance the ideal with what they can manage with the resources and time that they have, and decide how many samples they need in order to be confident of their results.
We must do the same when analyzing crossing situations.
We must each decide whether we have observed enough that we feel confident that the range of warning times of approaching vehicles in that situtation will be longer than our crossing time -- in other words, feel confident that whenever it is quiet in that situation, it is clear to cross.
If we are unable to observe enough samples to be confident, then we're in a Situation of Uncertainty.
Snook-Hill, Mary-Maureen and Sauerburger, Dona (1996). "Teaching students to assess safety for crossing streets which have no traffic control," in Proceedings of International Mobility Conference VIII, Tambartun National Resource Centre, Melhus, Norway, pp. 535-540
Return to "How can you be confident that the full range of warning times for approaching vehicles in a given situation is longer than the crossing time?