Page 127 - PhD GT

that a given dose will have a much more predictable effect and duration than alcohol, the availability and the cost of the drug. Interviews with respondents in the validation sample revealed that, despite these potential differences, people using heroin did find meaningful the questions about limiting their intake. They described limiting the amount used within a single session as well as limiting the number of times they used the drug throughout the day or the week. Thus while the idea of having no limit might be problematic with heroin users, the idea of limiting use was found to mean much the same as it does with reference to alcohol use. It would appear to be the case that the setting of absolute limits was determined by different criteria, i.e. those spelt out above, than the criteria used by those drinking alcohol.
Examination of the adapted Impaired Control Scale revealed similar findings to those of Heather et al. in both validation studies of the original scale (Heather et al. 1993, 1998). Internal consistency, content validity, construct validity and concurrent validity have been demonstrated. In the main, factor analysis lent support to the intention to measure separately attempts to control and ability to control use, whether experienced or expected. Principal components analysis of the original Impaired Control Scale (Heather et al.1993) was conducted for the three parts combined and yielded one component on which all items in Parts 2 and 3 loaded and a different factor on which Part 1 items all loaded. Items from Part 1 were reported to load only weakly on the first factor. In the rotated solution, three factors were found, corresponding to Parts 1, 2 and 3 respectively, with low correlations between Parts 1 and 2 (.01) and between Parts 1 and 3 (.14). Correlation between Parts 2 and 3 was much higher (.48), leading the authors to confirm their a priori assumption that Part 1 measured the conscious attempt to control drinking, while Parts 2 and 3 measured the more related questions of whether control had been successfully maintained in the past and whether the respondent believed they were able to maintain control in the present. Factor loadings for items belonging respectively to Parts 2 and 3 in the present study were not distinguishable; items from these parts of the scale showed highest loadings on the same first factor.
Nor was it possible to identify separate factors for items associated more clearly with impaired control than with inability to abstain, along the lines discussed in the earlier part of this chapter. Unlike the original scale, Parts 2 and 3 did not appear to form separate factors in this study.
More worryingly, factor analysis revealed that the attempt to avoid the response bias that is response set (the tendency to tick the same responses for all items) that was built into the scale may have resulted simply in the identification of this source of bias. The fact that the negatively worded items consistently formed a separate factor for both parts of the scale which contained such items (namely Parts 2 and 3), that these items were distinguished from the rest only by the fact that they
121