There is a certain happiness sighted when your bus comes along. It is of course a small specialized form of happiness and will never be a great thing.

-Richard Brautigan, The Old Bus

Thursday, December 4, 2008

Perfect foolishness on my part

I should just take down yesterday's post, but I will leave it up as an object lesson for myself in the dangers of sloppy work. It also serves as a cautionary tale for anyone who visits here who hasn't figured out yet that not everything you read on the web is necessarily true.

For too many years I was surrounded by editors who protected me from myself. Now I'm left alone without an excuse.

I added strike-out text to yesterday's post to mark where I carelessly swapped September for October data. That was simply sloppy. But worse than that, at least to me, was that I misread the two monthly reports. Somehow I completely missed the "Average Daily Ridership" header.

Commenter Anthony said:

Just a quick note to your comment, "All of the increase in bus ridership came from weekend riders." This is not quite true. When comparing monthly ridership, you should also check the number of weekdays in the month. Your tables show that October had 23 weekdays while Sept had 21. Because ridership is much higher on weekdays than weekends, most of the increase in monthly ridership is because October had two more weekdays in it than September. ...
To which I replied and in the process showed my total foolishness by attempting to divide the average daily weekday ridership by the total number of days. Huh?

Rather than embarrass me publicly, Anthony sent me an email that explains:

I agree that taking into account the number of weekdays strengthens your argument. My point was that the increase in weekend ridership was not sufficient to result in an increase in monthly ridership, given the reduction in average weekday ridership. If there had been an equal number of weekdays in the month, monthly ridership would have declined in spite of the increase in weekend riders. So even though average weekday ridership declined by 6% (from 69,700 to 65,400), with the extra two weekdays in October, total weekday ridership increased by 3% (from 1,463,700 to 1,504,200). And that increase of 40,000 explains most of the overall monthly increase.

In your reply, it looks like you are dividing the average daily ridership by the number of days, which I don't follow.
The next time I ask for statistics from RT, I can imagine the response I'll get.

No comments: