Calculated Traffic Counts in August & September 08
After encountering the learning curve related to the counting process in the spring semester, we began maintaining the equipment every two weeks on a very regular basis. This timetable is well within the manufacturer's design specifications for battery life. With the exception of two unrelated operator errors during a single period in July, the counters had been working flawlessly.
On Friday, August 22, in preparation for the return of students, a team of three staff members experienced in using the counting system went together out to the counting equipment locations and set them up for the coming period. To our surprise, when we checked them at the end of that cycle, we found that both counters had failed to record any data for the entire period. We have no explantion for this; we are completely baffled, especially given our coordinated extra attention to setting them up correctly in the first place.
Our solution for establishing values for the missing days in this instance is very simple:
- Look at the known values on both sides of missing period
- Compare them to last year's data for the same time frame
- Add 3.5% to the daily totals for the missing days
The 3.5% figure is arbitrary, but was selected because that is the ratio by which the average daily trips for the known dates this year is lower that for the corresponding period last year--that is, we doubled the difference in order to produce an abundant margin of error.
Because school started this year a week later on the calendar than in 2007, we shifted the dates in the database so that the first day of classes would line up for an apples-to-apples comparison.
Last year's counts, represented by the orange bars, were higher than this year's counts (green bars) on all but a few days, by a margin of 3.5% overall. The burgundy bars show the days on which a calculated value of 103.5% of last year's counts was used.