Page 72 - PhD GT

item version, the authors report high correlation between the 29 and the 25 item versions, suggesting that findings for the former apply equally to the latter (Skinner and Horn 1984). Reported internal consistency of items was high and “the majority of items had a moderate to substantial correlation to the total score.” (p. 202). Principal component analysis revealed a first factor which accounted for 31.0% of the variance. Tests of concurrent validity demonstrated that higher alcohol dependence scores correlated positively with greater daily consumption of alcohol, social consequences of drinking and with scores on the Michigan Alcoholism Screening Test (Selzer 1971). Predictive validity for the attendance rate for out-patient treatment was demonstrated by a statistically significant declining rate of attendance with increasing levels of alcohol dependence. In their clinical sample, Alcohol Dependence scores “conformed quite closely to a normal distribution providing empirical support for a quantitative interpretation of the syndrome as existing in degrees of severity” (p. 206).
The development of this scale highlights differences in approaches to the measurement of a construct. The Alcohol Use Inventory was developed in 1974 (Horn et al. 1974) with 16 primary scales covering symptoms and consequences of excessive drinking, drinking styles and perceived benefits from drinking. Skinner described that, having heard about the Alcohol Dependence Syndrome in 1979, he then examined a factor analysis of the Alcohol Use Inventory and found that the first factor “definitely resembled the concept of alcohol dependence syndrome” (Skinner 1986 p. 193). This may be the reason why the scale includes items like consumption levels and the consequences of extreme intoxication which, by no stretch of the imagination were ever described as elements of the syndrome. It may also explain the response bias imputed in the validation study when individuals scoring high on the Social Desirability Scale taken from the Personality Research Form (Jackson 1974, reported by Skinner and Allen 1982). When an instrument enquires about socially deviant and undesirable (by most standards) behaviour it is more likely to be fraught with response bias than is a questionnaire testing beliefs about other behaviours. Skinner and Allen (1982 p. 205) claim that their scale is “ more closely linked to an underlying theoretical concept” than, for example the MAST (Selzer 1971) but the way Skinner (1986) tells it, the sequence of events sounded more like a pre-existing ‘everything you wanted to know about a person’s drinking’ scale looking for theories to determine what its various sub-scales might be or mean.
3.4.3 The Short-form Alcohol Dependence Data (SADD)
In their development of this scale, Raistrick et al. (1983) aimed to measure the presence of the alcohol dependence syndrome and its severity and in order to do so, they operationalised each of
60