Chapter 3: Aspects of
concentrates on the different aspects of personality questionnaire.
Specifically, focus will be on the choice, evaluation and issues related
to the use of personality questionnaires.
Choice of Questionnaire
There is a wide array of published trait
questionnaires that are potentially relevant to certain human resource
needs. Personality questionnaires range from those that aim to assess
general qualities, such as measures of the Big Five, to those that measure
more narrowly drawn traits that may be critical in certain situations.
There are no definite rules for choosing between the different
questionnaires. However, it is essential that the user of these
personality questionnaires have a clear objective as to what the test
should assess or measure. The goal of using personality tests helps in
limiting the choices of questionnaire the user can apply and ensure that
useful results are obtained.
Evaluation of questionnaires
There are some well-established benchmarks
that may be used for evaluating the questionnaires. One of which is
reliability, referring to whether repeated measurements will give similar
questionnaire scores. However, a questionnaire may be reliable for the
wrong reasons. The user must ensure that questionnaire items do not
increase the tendency for biases. The validity of the questionnaire is
also an important factor for assessing the quality of a personality test.
In evaluating personality tests it is important that the questionnaire
possesses good reliability, stability and validity. If it has subscales,
their differentiation should be supported by factor analysis. Evaluation
of internal consistency (i.e., reliability) and stability over time is
straightforward. Generally, researchers take a reliability value of 0.7 as
the minimum for research use, although 0.8 or more is preferable.
Individual assessment requires a reliability of 0.9 or better. Determining
factor structure may raise technical issues such as the nature of the
factor structure to be used, although, if the factor structure is robust,
choice of analytic method should have minor effects only.
Assessment of validity may be a little
more complex, as I will now discuss. As previously discussed, the key
element of validity is criterion validity–the ability of the questionnaire
to predict meaningful criteria such as emotional states, abnormal
behaviours and job performance. Criterion validity has two main aspects.
These include the concurrent (present) and predictive (future) validity.
Both aspect can be useful; for instance, the clinician may want an index
of current behavioural disturbance, while the personnel manager needs to
predict future job performance, following training. In any case, the
validity coefficient expresses how strongly the trait predicts the
criterion; the trait may not be of much practical use if the coefficient
is too low. It is also important to establish whether the validity
coefficient generalises across different contexts; it is dangerous to
assume that a single study establishes validity, even if the coefficient
Face validity is the least important of
the remaining aspects of validity, although lack of face validity may
sometimes alienate respondents. Content validity is especially important
in the early stages of research, before the development of a detailed
nomological network that demonstrates the meaning of the construct from
its relationships to other indices and behavioural outcomes. Convergent
and divergent validity are usually considered together. For example, an
extraversion–introversion scale should correlate moderately high with
related constructs such as sociability and assertiveness (convergent
validity). If it fails to do so, the scale is probably not measuring
extraversion. It should also show small correlations with other constructs
that are known to be distinct from extraversion, such as neuroticism and
intelligence (divergent validity).
Establishing divergent validity is
especially important in developing scales for new constructs, which, all
too often, turn out to be similar to existing ones. Incremental validity
is related to divergent validity. It refers to tests of whether the scale
predicts criteria if other constructs correlated with both the scale and
the criterion are statistically controlled, typically using partial
correlation or multiple regression. If I had a new scale for stress
vulnerability, incremental validity would be demonstrated if the scale
predicted anxiety symptoms with neuroticism and extraversion controlled,
for example. Finally, construct validity refers to the often elusive
theoretical basis for the trait, and its psychological meaning. The
relevance of theory to the practitioner varies according to the nature of
the practical problem. Sometimes, prediction proceeds on an actuarial
basis. That is, if it is known that a battery of scales predicts
performance on some job (with good validity generalisation), the scales
can be used for personnel selection without too much concern about theory.
However, this approach is often negated by the existence of moderator
variables, that is, additional variables that influence the association
between the trait scale and the criterion. For example, correlations
between traits and job performance depend critically on factors such as
the nature of the work, the stressfulness of the work environment, and the
level of stimulation or arousal it affords. Although the influence of
moderator variables can be mapped out empirically, prediction is enhanced
when the user can utilize theory to determine when a trait is or is not
likely to be predictive.
The user of psychometric tests is, of
course, bound by the same ethical principles as any other psychologist.
Lanyon and Goodstein (1997) discuss some misuses of tests, which would be
contraindicated by the APA code. Naturally, it is unethical to use
professionally a test whose validity has not been established. Even if the
test has been systematically developed, problems may arise when there is
no clear criterion for the construct that is assessed, and when tests are
interpreted on the basis of common sense or the tester's personal
insights. Such problems are often more acute for projective tests than for
trait measures. More subtly, tests that are valid for one purpose may be
misused in a different context. Lanyon and Goodstein pointed out that
tests developed for use in psychiatric settings such as the Rorschach and
MMPI may not be suitable as selection devices in industry, especially when
administered by people with no clinical training. Several countries,
including the UK, have formal systems for accrediting test users to
counter such problems.
Ethical obligations are discharged within
a legal framework, which, of course, differs from nation to nation, and,
in the USA, from state to state. Laws typically deal with issues such as
confidentiality and data protection, protection of privacy, and fairness
in occupational selection. Naturally, the practitioner requires
familiarity with such laws, especially in an increasingly litigious
society. If a trait assessment is a factor in a job applicant not being
hired or promoted, the psychologist may have to justify the relevance of
the trait in court. Occasionally, legal decisions may seem capricious. In
1996, the police force of New London, Connecticut, obtained some notoriety
for refusing employment to an applicant whose mental ability was deemed
too high (corresponding to an IQ of about125). The police department
successfully argued in court that applicants who score too high could get
bored with police work and leave soon after receiving costly training.
Arthur and colleagues (2001) reviewed some
legal implications of organisational personality assessment in the USA.
They pointed out that personality measures may be less vulnerable than
mental ability tests to the perception that they are unfair to minority
applicants. Indeed, they quote a statement made by Hogan et al. (1996, p.
“… we want to suggest in the strongest
possible terms that the use of well-constructed measures of normal
personality in pre-employment screening will be a force for equal
employment opportunity, social justice, and increased productivity.”
Arthur and associates (2001) pointed out
two unresolved problems in line with the use of personality questionnaire
within the occupational field. First, although fakers may be identified as
having very high scores on desired traits, rejecting a job applicant
because they score too highly might be difficult to justify legally.
Second, the well-replicated sex differences in some personality traits
lead to conflict between legal and scientific principles. The use of score
adjustments or differential cut-offs in the use of employment-related
tests, on the basis of race, colour, religion, sex or national origin was
made unlawful by the US Civil Rights Act. This is implemented primarily to
prevent racial discrimination. However, it is normal and scientifically
justified practice in personality assessment to use separate norms for men
and women, a procedure that in fact promotes fairness in occupational
selection. It remains to be seen how this issue will play out in future
Read the rest of the thesis
[ Back ] [ Next ]
Read what our customers
say about our work