Home | Statistics and
Research Methods | Psychometrics | Human-Computer Interaction
| Organisational Psychology
| Psychology for
Work | Commercial
sites
Psychometrics
Psychometric
tests are widely used for Personnel Selection and Development. They are
designed to give information relating to ability or personality
characteristics. The information gained can then be used as a starting
point for a Selection or Development process. Different tests assess
different elements of performance, personality, or interpersonal
skills, largely determined by the company who produce the tests.
The original premise for such tests was that they removed any effects
of 'personal bias' in the selection process, and represented therefore,
a more 'scientific' method of selecting personnel. This approach has
been developed to include Development applications, and arguably,
Psychometrics are more successful in this application than in
traditional selection procedures (whose success is highly dependent on
the skill of the person delivering the test, and the manner in which
the information is used).
There are, however, several things that are important about
Psychometric tests:
First and foremost, it is preferable to use a Psychometric test
developed by someone with some expertise in Psychology and
Psychometrics. This is crucial as this may be the only indicator you
have that the test is useful. Secondly, a test which is accredited by
the British Psychological Society can be guaranteed to have been
through a rigorous procedure including Statistical and practical
validation.
The test should be valid: the test should measure what it says
it measures. In the past it has been easy to tell exactly what the test
measures, as the distinction was often between ability (such as IQ) or
personality (such as Extraversion or Introversion). More recent
tests are more varied in their purported purpose (e.g.
"Interpersonal Effectiveness"), and so you need to rely on other
measures to assess the efficacy of the test.
Other measures would include the relationship between the test in
question, and other tests measuring the same thing: so for example, a
test might include a measure of Introversion and Extraversion: in this
case the measure should bear some relation to other tests assessing
these constructs (and the test manufacturer should be able to provide
this information).
A second way of assessing the efficacy of a test is to look at the
Statistics. All reputable Psychometric tests will be subjected to
statistical analysis, this will include examination of constructs as
well as individual test items (questions). Of course not everyone will
understand what the statistical output means (see hints and tips for
some basic pointers), but the statistics should demonstrate a
significantly better performance than chance performance, and that the
test items relate together as predicted.
The test should be reliable:
this means, at the very least, that the test should perform in the same
way for the same person over time. This can be reduced to producing the
same results for the same person when tested on different occasions.
You would expect this to be true, if everything else is equal (but
beware, tests on people are often not as reliable as we would like
-people can be unreliable things!) The point is though, that the test should be reliable, and thus
if there are differences in performance, these should reflect
genuine differences, rather than be artifacts of the test. Again,
Statistics can be produced to demonstrate the reliability of the test (see
hints and tips
for some basic pointers).