Page 187 - PhD GT

total dependence scores at intake and three months and between total dependence scores at three months and twelve months for the whole follow-up sample. The correlation between dependence scores at three months and twelve months was smaller than that between intake and three months. No such significant correlation was found between dependence scores at intake and at twelve months for the whole sample and the heroin sub-sample. Nor was a significant correlation found between dependence at three months and at twelve months for the heroin sub-sample. The largest correlation was found between dependence scores at intake and three months for the alcohol sub-sample and a smaller but significant positive correlation (correlation coefficient .29; p<.05, 1-tailed) was found between intake and twelve month dependence scores for the alcohol sub-sample suggesting that dependence at baseline may be more predictive for alcohol dependence than it is for heroin dependence. Correlation coefficients and their levels of significance for Sample 6c and the constituent substance groups are shown in Table 9.1.
Table 9.1 Correlations between total dependence scores at t1 (LDQ t1) and t2 (LDQ t2), t2 and t3 (LDQ t3) and t1 and t3: Spearman’s rho correlation coefficients and levels of significance: Sample 6c (n = 101, data missing for 2 participants)
Sample 6c (n=99)
LDQ t1
LDQ t2
heroin sub-sample
(n=59)
LDQ t1
LDQ t2
alcohol sub-sample
(n=40)
LDQ t1
LDQ t2
*p < .05 (1-tailed) **p < .01 (1-tailed)
LDQ t2
.46**
.3*
.7**
LDQ t3
.08 .21*
-.08 .03
.29* .48**
One-tailed tests of significance for the difference between these correlations (Howell 1997) showed that the correlation between total dependence score at intake (LDQ t1) and at three months (LDQ t2) for the heroin sub-sample (correlation coefficient .3) was significantly different (z=2.633;
182