Page 230 - PhD GT

the effects of conditioning on responses. In the present study, the reference periods varied for the different self-completion scales used in order to remain faithful to the original scale in its validated form and to preserve consistency with the meaning of the construct. The Leeds Dependence Questionnaire enquires about the past week. The General Health Questionnaire enquires about the past month, a time frame retained from the original validation of the scale. The Social Satisfaction Questionnaire enquires directly about present satisfaction. Instructions for the Coping Behaviours Inventory were also in line with the original scale, enquiring about which method the respondent has used and implying ever. Adhering to the validated form of each scale was considered to be of paramount importance, with one exception. In its validated form the Impaired Control Scale (ICS) enquired about the past six months. As this time scale was twice as long as the first follow-up period in the present study, and because the ICS was thought to measure a component of dependence which was anticipated to be capable of change over a much shorter time scale than this, a change in the time reference was proposed. In a personal communication with N. Heather, the first author of this scale, it was agreed to reduce the six month time reference of the original to three months for the first two parts of the scale (the third part enquires about beliefs about what would happen now). Although it was suggested that it would have been preferable to enquire about the past week in line with the LDQ, Heather felt that three months would be more consistent with the instructions given during the validation studies. A researcher was available to offer assistance during questionnaire completion when requested, and no queries were reported regarding the time-scales in the different questionnaires.
With reference to the complexity of the instructions or the concepts contained in the individual scale items, a small number of respondents (fewer than ten) queried the meaning of items 5 and 8, as was reported in the validation study (Raistrick et al. 1994) and thought at that time not to constitute sufficient grounds for altering this item. Item total correlations for these items in the present study were slightly lower than those for other items in the scale (.44 and .39 compared with .49 for the next lowest correlation and .72 for the highest correlation) but the effect of their removal in raising the alpha coefficient was only marginally greater than the effect of the removal of other items. Readability of the LDQ, at the very least achieving scores consistent with ‘fairly easy’, and ‘standard writing’, was reported in Chapter 4.
Conditioning did not appear to be a problem given the considerable differences between the individual scales and the three and nine month time lags between completion of these scales. However, a response set was detected in responses to the Impaired Control Scale and discussed in Chapter 5. Such a finding raises the question of response sets in other scales where it was not
225