What the Measurement Showed

For nine years the department had believed that the morning shift produced fewer faults than the afternoon one. It was not a guess. It came from a table that was updated every month and shown at every review, and the difference in it was large enough that nobody thought to check.

Yusuf checked it because he was new and because the table was the first thing he was given. What he found was that faults were recorded when they were noticed, and they were noticed when somebody had time to look. The afternoon shift had time.

He wrote four sentences about this and sent them to Corinne, who had built the table. She replied within an hour asking for the raw figures, and then said nothing for six days. On the seventh she sent back a second table, built her own way, which showed no difference between the shifts at all.

What happened next was the part Yusuf remembers. Corinne presented both tables at the review herself, in that order, and explained what was wrong with the first one. She had made it. She had defended it for nine years. It took her eleven minutes.

The department did not thank her for it, exactly. Two people suggested that the new figures needed further work, which was true and was also a way of saying nothing. But the table was withdrawn, and the shift pattern that had been built around it was quietly allowed to change.

Yusuf learned two things that year, and he would put the second one first. A measurement that nobody has ever questioned is not a strong measurement; it is an unexamined one. And the person who can take an argument like that apart in public is almost always the person who built it.