Page 61 - PhD GT

Problems of bias are addressed in the design of self completion instruments used to obtain clinical data. Often measures are validated for a specific population and may need further validation in different populations. Cultural neutrality can be relevant to different age groups as well as to those more commonly applied criteria for cultural diversity. For example, MacEwan (personal communication) found that a questionnaire measuring readiness to change in problem drinkers (Rollnick et al. 1992) that had been validated in Australia could not be understood by a population of Maori in New Zealand due to the wording of the questions.
3.2.4 Retest reliability of self-report
Babor et al. (1987b) described 17 studies which evaluated the test-retest reliability of self-report using a variety of measures of self-reported drinking and consequences in different population groups including inpatients, pregnant women, college students, homeless people and adolescents. Acceptable levels of reliability were reported and these varied between the different populations and for different measures targeting different behaviours. In general in the studies reviewed, reports of frequency of drinking were more reliable than reports of quantity, it being easier to recall on how many days alcohol was consumed as opposed to how much alcohol was consumed. In the present main study, quantity of alcohol or heroin consumed on the heaviest day in the past week was asked about as a day that may have stood out above the rest. Number of drinking or drug taking days out of the past seven constituted the frequency question.
The conclusion offered by Maisto et al. (1990) for the accuracy of the self-report of drug users is rather less enthusiastic. In a review of fourteen studies of the reliability and accuracy of drug users’ self-report of their drug use, the authors concluded that while their analysis of the data provided “some evidence for self-report accuracy...the degree of accuracy is not high” (Maisto et al. 1990 p. 120). Nonetheless these authors conclude that the way forward is to enhance the reliability and validity of self-report rather than simply dismiss the method as being too inaccurate.
Reliability and validity are affected by the same variables, classified below as task variables and respondent variables. With reference to task variables, the reliability and validity of the instruments chosen for assessment is an essential prerequisite.
3.3 The validity of self-report
Chermack et al. (1998) tested the validity of self-report in the context of screening for alcoholism in medical settings. They compared self-report with collateral report in 581 pairs of medical patient and collateral responses. Patient and collateral reports of alcohol consumption and
49