Читать книгу Abnormal Psychology - William J. Ray - Страница 164

Reliability asks the question of whether the instrument is consistent. We would expect, for example, that the odometer in our car would reflect that we drove a mile each time we drove 5,280 feet. We would also expect our bathroom scale to show the same reading if our weight had not changed. Researchers interested in questions of measurement discuss a number of types of reliability:

 Internal reliability—Internal reliability assesses whether different questions on an instrument relate to one another. If we were seeking a general measure of depression, for example, we would want to use questions that relate to one another. Questions related to feeling sad, not having energy, and wanting to stay in bed would be expected to show internal reliability.

 Test–retest reliability—Test–retest reliability determines whether two measurement opportunities result in similar scores. A key consideration with test–retest reliability is the nature of the underlying construct. Constructs seen as stable, such as intelligence or hypnotizability, would be expected to show similar scores if the same instrument was given on more than one occasion. In psychopathological research, measures of long-term depression or trait anxiety would be expected to show a higher index of test–retest reliability than measures that reflect momentary feelings of mood.

 Alternate-form reliability—As the name implies, alternate-form reliability asks whether different forms of an instrument give similar results. If you were giving an IQ test, for example, you would not want to ask the same question each time, since the individual could learn the answers from taking the test. Thus, it would be important to create alternate forms that reflect the same underlying construct.

 Inter-rater reliability—Inter-rater reliability asks how similar two or more individuals are when they observe and rate specific behaviors. Psychopathology researchers often rate the emotional responses of children as they engage in various activities. An index of inter-rater reliability would measure how consistent different observers would be in rating the same situation. Historically, one of the motivating factors for developing the DSM classification system was the discovery that different clinicians in different locations watched a film of a person with a mental health disorder and diagnosed it in different ways.

We expect our bathroom scale to show the same reading if our weight has not changed. Likewise, researchers are concerned with the reliability, or consistency of measurement, by assessment instruments.

© iStockphoto.com/tetmc