Tuesday, 4 April 2017

Bullet Journal statistics for March 2017: When Statistics Go Bad

Statistics: You're doing it wrong. Except if you're a professional statistician, in which case any wrong-doing would be expected to at least be intentional.

I've been tallying my tasks with a simple chart like this:

This is a compressed version of the markings in the diary. For each day, I tally the number of tasks done, migrated, or cancelled, plus the ones that got postponed. I think of it as a "wind diagram". It is a rather nice representation, for it not only pretty but also self-correcting: If any of the +N columns have a number, there must be a number as least that large down and to the left. The difference is the number of tasks of that postponement that got finished that way. It also allows me to correct mismarkings, such as when I miscount the number of dashes in a postponement marker (marked with a *). It's also very simple to make, only requiring looking at one day at a time and counting marks.

The "Abs" column is just a summation, while the "Rel" column shows the difference between the postponements - the number of tasks that have gotten done after that many days of postponement.

What I did wrong in February was not subtracting the postponed done tasks from the number of immediately done tasks. Thus where I claimed to have done 71 tasks the very same day, I had only done 37, little more than one per day. How very emberassing.

With this knowledge, my March statistics look less nice but more true that those of the last post. I finished 98 tasks, but only 48 got done the same day, less than the 71 I claimed in February (and no, I did not finish 110 tasks in February). I migrated only one task, but cancelled an outrageous 15 tasks! Some of them were done by other people (delegate!), some were the victim of changes of plans, but some I just postponed until they were not relevant any more - something I used to do a lot.

In March, I also introduced a new concept: skipping days for tasks that cannot be done on a given day. For instance, I cannot make a doctor's appointment on the weekend around here. Copying over these items needlessly is annoying. So I leave such an item undone instead of marking it with the postponement arrow, but to avoid forgetting about it entirely, I add an up-arrow to the next day. This indicates that there are leftovers on earlier days to migrate later. Obviously, I don't want to do this for more than a few days or it becomes a bother to look back every day, but then anything that should go further out should just go on the month overview. I still count these as being postponed for those extra days, to not confuse my wind diagram.