Page 83 - PhD GT

alcohol and heroin users. Furthermore, it was found to measure dependence in a student sample and in a small general practice sample. Principal components analysis produced a single factor solution which accounted for 58.3% of the variance and all of the ten items had loadings of 0.67 or more on this factor. The lowest loadings were for items 5 and 8 (0.67 and 0.68 respectively). The remainder of the item loadings were 0.75 and above. Cronbach’s alpha coefficient of reliability was reported to be 0.94. High positive correlations were reported for total score and for individual items in test re-test reliability in a sample of 33 individuals who completed the questionnaire twice over an interval of 2-5 days.
Content validity was refined through the eight pilot versions of the scale; respondents were reported to have found the items emotionally neutral and easy to understand, with the exception of three individuals who reported difficulty in understanding items 5 and 8. This was considered an insufficient number to justify changing the wording of these items.
In order to examine the concurrent validity of the LDQ, scores were compared with two separate instruments, the SADQ and the SODQ, both described in previous sections of this chapter. Though these scales do not measure an identical construct but rather measure the alcohol and the drug dependence syndromes respectively, they were thought by the authors to have been “sufficiently close to be useful” (Raistrick et al. 1994 p. 566). Significant positive correlations were found between mean scores on the LDQ and the SADQ for the alcohol sub-sample and between mean scores for the LDQ and the SODQ for the opiate sub-sample.
The scale was shown to discriminate a clinic population from a student population and a general practice population with significantly different mean dependence scores for the three samples. No significant differences were found in the mean dependence scores for the two substance groups, though the mean dependence score for the opiate sub-sample was higher at 20.1 (SD = 6.8) than for the alcohol sub-sample which was 16.3 (SD = 8.9).
The authors concluded that there was evidence to support the use of the LDQ as a valid and reliable instrument capable of measuring dependence as a unitary construct.
Further validation study of the LDQ was reported by Heather et al. (submitted) following its administration to a total of 1681 clients attending two agencies (Leeds and Newcastle) for treatment of substance dependence and misuse. Data were collected over an 18 month period from 1994 to 1996. The LDQ was administered at first appointment at the agency as part of an evaluation package which contained the 12-item General Health Questionnaire (GHQ) (Goldberg 1972) and a Social Satisfaction Questionnaire, designed specifically for this evaluation package. Both these questionnaires are discussed in Chapter 7. Respondents were classified into three drug groups,
71