If Two People Score It Differently, You Have No Measurement
Last timeChoosing the Cases
Most evaluations fail here: the rule for what counts as a good answer was never written, so every score is partly a measurement of who happened to mark it.
The rule two people share
Here is a test that takes an afternoon and that very few teams have run. Take
thirty answers from your system. Give them to two colleagues separately, with
whatever instructions your evaluation currently uses, and ask each to score
them. Then compare.
The usual result is that they agree on about two thirds, which sounds
reasonable and is not. If both markers call three quarters of answers
acceptable, then two people answering at random in those proportions would
agree on 62 per cent of items without reading anything. Two thirds is barely
above guessing.
| step | marker A | marker B | agreed | what they were given | what happened |
|---|---|---|---|---|---|
| 1 | good | poor | no | score the answer for helpfulness | Marker A rewarded the thorough explanation. Marker B marked it down for not answering until the third paragraph. Both are defensible readings of the word helpful. |
| 2 | 2 of 3 | 2 of 3 | yes | three checks: answers the question, stat | Same answer, same two people. The checklist removes the question of what helpful means and replaces it with three questions about the text. |
What the test gives you is not a verdict on your colleagues. It is a
measurement of your instructions, and the instructions are the part of an
evaluation that almost never gets reviewed.
Checklists instead of adjectives
The repair is mechanical. Take the adjective and ask what observable fact about
the text would make you apply it.
Helpful becomes: does it answer the question that was asked, rather than a
nearby one. Does it state the limitation if there is one. Does it give the
reader something to do next. Three yes-or-no questions about the words on the
page, none of which requires a theory of helpfulness.
The lesson stops here
5 more paragraphs to go
You have read the opening. The rest of the argument, the problems that check whether it landed, and the lines worth keeping at the end all come with a plan.
The first lesson of every course in the library reads the whole way through, free, so you can see exactly what the rest of them are.
See the planThe contentsThis is the reading half
Starting the course gives you your own copy of it. Every idea on every page has problems standing under it, marked with a reason rather than a tick, and any sentence you do not believe can be opened and argued with. None of that can happen on a page nobody owns.
The contents