[thechat] predictive indexing - is it evil ?

Martin Burns martin at easyweb.co.uk
Tue Sep 3 11:47:00 CDT 2002

On Tue, 3 Sep 2002, Madhu Menon wrote:

> At 08:04 PM 03-09-02, Ashok at MagicalKenya.com wrote:
> >my company's HR dept. has got hold of a 'predictive-performance-index'
> >consultant
> >- this bloke is gonna 'profile' individual employees.
> My arse!
> It's voodoo psychology man. Being the skeptic that I am, I can tell you
> that I could fake my answers to give them any "profile" I want. All those
> Myers-Briggs, Keirsey, etc. tests are one big bunch of bull.

Depends. Depends on what the aim is and how well that's communicated to
the people taking the test.

If it's perceived as "Something to work out whether you get fired or a pay
rise" people are going to fake it. If it's perceived as "Help us all
understand how to work more effectively with each other" then it's going
to be more honestly taken and more useful anyway.

Tests such as MBTI or Belbin all come with massive health warnings about
them not being useful to appraise how *good* an employee is, but what
their working style is (of course, if you're in a role which requires you
to be a Belbin Completer/Finisher and you're rubbish at it, that may be
different, but if that's the case then you're going to be obviously
rubbish at it). They also come with health warnings (usefully ignored in
the dictionary entry you cited about not being the whole story, but simply
the *preferred* mode of operation. (Interesting how selectivity is one of
the charges laid by the dictionary - skeptic, question theyself).

But yeah - *any* test performed badly isn't going to be accurate.

OTOH, if it's an ability-based test, which has been properly normalised
for your role and organisation (ie "People who do well at these
capabilities (eg abstract reasoning, or analytical thinking) do well at
this job here"), then that's also valid.

Some tests (particularly Belbin) also don't give an "You are X" answer,
but more of a "You're strongest in X and Y and weaker at Z" result. And of
course this varies by context. Duh.

> I'll simply point you to the Skeptic's dictionary entry for the
> Myers-Briggs test: http://www.skepdic.com/myersb.html
> The problem with any of such profiling questions is that situations are
> never black and white. Without context, you have no way of predicting
> behaviour. If you asked me, "do you get along with people", I'd say "yes".
> Does that mean I'll get along with everyone? Of course not. Another such
> question is "do you prefer to work in a team or by yourself". Again, no
> reliable answer to this one. My answer would be "who's in the team and what
> are they doing?". There have been situations in my work where I've got
> things done faster by doing it solo, but there are other times when I've
> worked with great people to produce great work. Unfortunately, the
> "profiling" test don't allow "it depends" as an answer. You're either here
> or there.

...which is why tests tend to
a) have "Hell yeah" to "No way" spread answers over a spectrum
b) Assess binary answers ("Do you usually do (a0) or (b)") against several
alternatives ("Do you usually do (a) or (c)")


"Names, once they are in common use, quickly
 become mere sounds, their etymology being
 buried, like so many of the earth's marvels,
 beneath the dust of habit." - Salman Rushdie

More information about the thechat mailing list