A selection test is inherently a predictive process which tries to predict which candidates will perform best at or after some point in the future.
It does not measure future performance, it predicts it. If you are able to rank the performance of the candidates you accept at some future point, you get the feedback you need to improve your selection process.
Let's say your selection process starts out taking, for example, university graduates with either history degrees or engineering degrees, and ranks them on their academic records.
Now lets say your feedback process 2 or 5 or 10 years down the line tells you that actually history graduates are consistently slightly outperforming engineering graduates as officers.
At this point, you'd go back and weight their academic results slightly higher. You'd say to yourself "my old judgement about who will do best is consistently slightly off, so despite what I think, I need to tweak it.
There is no shame in accepting that when you try your best to use what you hope is an objective selection process to get the best candidates, you consistently discount (bias against, consciously or otherwise) certain classes of candidate, and choosing to identify membership of one of those classes as an unrecognised factor to be added to your selection process.
If your feedback process says that race or gender seems to result in bias, you'd do well to insert race into your selection process.
But you sure as hell don't decide that because a balanced selection process accepts approximately the same proportion of Scottish ancestry as of Irish ancestry (insert other examples if you like) as you see in qualified potential applicants, that when the proportion of Scots has been a bit low for a while, it is time to accept ONLY Scots. Past selection errors don't justify swinging too far the opposite direction today.
Meeting a particular proportion of different races in the entire organisation as a whole by pumping in lots of minorities today should not be the objective; recruiting a particular proportion (with inevitable scatter due to small sample sizes) in each intake should be the objective.
Originally Posted by
RAFEngO74to09
Remember in WWII,
the things that were rarely ever damaged on returning bombers were in fact the weak points?
Similarly, the people you don't succeeded in recruiting despite them appearing to be good candidates are the people your advertising needs to appeal to.
So you go out and find smart cookies outside the RAF who are a year or two older than your usual candidates and you survey them to find out why they didn't consider (or rejected) the RAF as a career, and then your address those issues in your advertising.
Getting people interested in planes is easy, getting them to see past their own misconceptions about joining the military is much more important.
Advertising is the first step of your recruiting selection process, you don't need to inspire every Tom, Dick and Harry to apply, you need to persuade as many of the very best as you can that the RAF offers them a better future than their high performing mates will get elsewhere.