Back in 1980 as a re-entrant in my mid 30s, I was on my Secretarial course the Towers when the whole course was shunted to the Taj Mahal (Stn Educ Centre) for an as yet undeclared purpose. Seems someone had suggested ascertaining the psychological make-up of those who had been selected previously, and making that the selection "model". The guarantee was that there would be no looking at ACRs to see how performance later match selection prediction. It was explained we would take a profiling "test" and then that would be used to construct the model. Having been involved, outside the Service in some similar work, I questioned the use of just one test (16PF as I recall), and suggested that a battery of tests, probably 3 or 4, would be required to construct a usable model, and that for it to work at all there needed to e a comparison or performance against prediction. My comments were dismissed as "ridiculous" and the test went ahead. I wrote a memo to the Head of the Sec Training School, was interviewed on my comments, and the whole idea subsequently disappeared in a puff of smoke. Pity, because done properly it had promise.