Page 229 - PhD GT

10.3.2 Social desirability bias
The problem of social desirability as a source of bias in questionnaire and interview response was raised in Chapter 3. The under-reporting of drinking in all study participants would be an anticipated result of this source of bias, with perhaps more under-reporting of drinking by women than men, if the suggested greater stigma attached to female excessive drinking compared with male excessive drinking is correct. In her study of barriers to treatment for women, Thom (1986) did not confirm the hypothesis that social stigma was one such barrier. An interesting and unusual finding in the present study was that women reported drinking on more days of the week than men (a mean of 5.8 days for women compared with 4.3 days for men), a mean that is so high as to render the possibility of under-reporting to be unlikely. The suggestion of over-reporting, made in Chapter 3 with reference to heroin users would not have the same rationale for problem drinkers.
The finding of greater frequency of drinking in women is at odds with the epidemiological findings regarding gender differences in patterns of alcohol consumption. Fillmore et al. (1997) have noted a universal gender difference which transcends culture and social context (p. 27) wherein “women tend to drink less frequently and in smaller quantities than men”. Could it be that the response bias resulting from social desirability operated for men and resulted in an under-reporting of their drinking, or might this be a true finding? A possible explanation for this finding is that women are also found to enter treatment later in their drinking career and it has been suggested that this too may be due to the greater stigma thought to be attached to problem drinking in women compared to that in men (Walitzer and Connors 1997); although these authors assert that gender differences are insufficiently researched in treatment populations due to the under-representation of women, it may be that population gender differences are cancelled out in clinical populations because women seek help later and therefore when their problems have become more severe.
10.3.3 Reliability of the instruments
In order to enhance the quality of self-report data, various recommendations that were discussed in Chapter 3 were adopted in the present study. Some of these were relevant to the instruments used and some were relevant both to the method of conducting the short interview and to its content.
Midanik (1988) proposed attention to three key areas: the effects of bounded and unbounded reference points on the temporal judgements of respondents, the effects of complex concepts and
224