RELIABILITY AND VALIDITY IN ASSESSMENT:
inter-rater reliability: refers to the degree to which two independent observers or judges agree.
test-retest reliability: measures the extent to which people being observed twice or taking the same test twice score in
generally the same way. Makes sense only when the theory assumes that people will not change greatly between testings
on the variable being measured.
alternate-form reliability: using two forms of tests rather than giving the same test twice, perhaps when there is
concern that people will remember their answers from the first test and aim merely to be consistent. The extent to which
scores on the two forms of the test are consistent.
internal consistency reliability: assesses whether the items on a test are related to one another.
In each of these types of reliability, a correlation is calculated between raters or sets of items. The higher the
correlation, the better the reliability.
Validity is generally related to whether a measure fulfills its intended purpose.
content validity: refers to whether a measure adequately samples the domain of interest.
criterion validity: is evaluated by determining whether a measure is associated in an expected way with some other
measure (the criterion).
- concurrent validity: both variables are measured at the same point in time, the relationships are concurrent.
- predictive validity: criterion validity can be assessed by evaluating the measure’s ability to predict some other
variable that is measured in the future (eg; IQ tests used to predict future school performance).
construct validity: is relelvant when we want to interpret a test as a measure of some characteristic or construct that is
not simply defined.
- A construct is an inferred attribute, such as anxiousness, that a test is trying to measure.
interview: any interpersonal encounter, conversational in style, in which one person, the interviewer uses language as
the principal means of finding out about another, the interviewee.
Characteristics of clinical interviews:
clinical interview: is different from a casual conversation or a poll by the attention the interviewer pays to how the
respondent answers, or does not answer, questions.
The paradigm within which an interviewer operates influences the type of information sought, how it is obtained and
how it is interpreted. A psychoanalytically trained clinician can be expected to inquire about the person’s childhood
history because the analytic paradigm holds that the most significant aspects of a disturbed or normal person’s
developmental history are repressed into the unconscious.
Creating a rapport is important, as well as building trust. It doesn’t always suffice to be an ‘authority’ figure such as a
doctor to have a patient open up. Empathising, repeating what they said to show you’re listening,
Interviews vary in the degree to which they are structured. In practice, most clinicians operate from only the vaguest
outlines. Exactly how information is collected is left largely up to the particular interviewer and depends too on the
responsiveness and responses of the interviewee.
Because of the overwhelming majority of clinical interviews are conducted within confidential relationships, it has not
been possible to establish either their reliability or their validity through systematic research.
At times, mental health professionals need to collect standardized information, particularly for making diagnostic
judgements based on the DSM.
structured interview: one in which the questions are set out in a prescribed fashion for the interviewer. structured clinical interview diagnosis (SCID) for axis 1 of DSM-IV: a branching interview; that is the client’s response to
one question determines the next question that is asked. It also contains detailed instructions to the interviewer
concerning when and how to probe in detail and when to go on to questions bearing on another diagnosis (See p.103,
Structured interviews have also been developed for diagnosing personality disorders and more specific disorders, such
as the anxiety disorders.
psychological tests: standardized procedures designed to measure a person’s performance on a particular task or to
assess his or her personality or thoughts, feelings, and behaviour. If the results of a diagnostic test are inconclusive,
psychological tests can provide information that can be used in a supplementary way to arrive at a diagnosis.
standardization: the same test is administered to many people at different times, and the responses are analysed to
indicate how certain kinds of people tend to respond. Statistical norms for the test can thereby be established as
sufficient data have been collected.
There are three basic types of psychological tests; self-report personality inventories, projective personality tests and
tests of intelligence.
(1) Personality Inventories:
o Personality inventory: the person is asked to complete a self-report questionnaire indicating whether
statements assessing habitual tendencies apply to him/her.
o The best known is MMPI, intended to serve as an inexpensive means of detecting psychopathology. Is
multiphasic because it is designed to detect a number of psych. problems at a time.
In designing the test, first, many clinicians provided statements that they considered indicative of
various mental illnesses. Second, these were rated by patients already diagnosed with particular
disorders and by a large group of individuals considered normal.
o The MMPI-2 has several noteworthy cha