Page 124 - PhD GT

had more control. Nunnally used the term prediction in the validation of scales to refer to:
“functional relations between an instrument and events occurring before, during or after the instrument is applied” (Nunnally 1978 p. 88)
What is being suggested here is the examination of correlations between scores on measures that would be expected to covary with impaired control because of a hypothesised relationship between the phenomena. Nunnally referred to criterion-related validity or predictive validity, while elsewhere, for example in the validation studies of the impaired control scale already referred to, the term concurrent validity is used.
In the present validation study, relationships with levels and patterns of use would be expected to hold when examining the sub-scale scores: more attempts to control would be associated with lower levels of use, more failed control with higher levels of use and more expected failed control would also be associated with higher levels of use. Measures of substance use were therefore chosen to examine the criterion-related validity of the adapted Impaired Control Scale. Correlations between ICSdrg, patterns of use and quantities used are shown in Table 5.3.
Table 5.3
Correlations between ICSdrg part scores, number of days using and quantity used on heaviest day: Sample 4 (n=118)
DAYS QF
-.26*** -.29** .47* .45* .37* .41*
.75*
ICSAC
ICSFC
ICSPC
DAYS
* p= .000; **p=.002; *** p=.005
Correlations were all significant; as shown in Table 5.2, attempts to control were negatively correlated with failed control and perceived control. They were also negatively correlated with number of days used in the past week and with quantity of heroin used on the heaviest day as shown in Table 5.3. Significant positive correlations were found between failed control and perceived (failed) control as well as between both of these and the number of days and quantity of heroin used
119