Teachers were asked by their school district to complete a survey.  The experiment was a 2×2 design (one of our faves):

The result?  A nothingburger.  No difference in survey participation between A-D, randomly assigned groups.

To quote Paul Harvey, “And now for the rest of the story.”

Researchers noticed a difference in survey participation on one of their external, control variables – teacher effectiveness.  More effective teachers were more likely to fill out the survey.

This led to a post-hoc analysis that found what we often find; test ideas work differently for different people.

Priming the Identity of the teacher by referencing “teacher” in the email survey invite did work but only for those teachers who are  good teachers.

Identities matter when they matter.  Calling me a donor or dog lover or conservationist or researcher or father only matters if,

  • I am that thing and
  • That thing matters to me

A proxy for the latter is the person is good at it.