AIDS knowledge and attitudes among injection drug users: the issue of reliability

Abstract

Among injection drug users (IDUs), AIDS-related knowledge and attitudes have not consistently predicted AIDS risk behavior. This may be due in part to the limited reliability of indexes used to measure drug users' AIDS knowledge and attitudes. In addition, the substantive interpretation of findings is confounded if index reliability is lower for particular demographic groups (e.g., ethnic populations and women). This report is based on 8 measures of AIDS-related knowledge and attitudes in a sample of 332 injection drug users in Los Angeles. The reliability of knowledge and attitude indexes for the overall sample is generally acceptable for the purpose of group comparison (average alpha = .60). But reliability is consistently lower for respondents who are Hispanic (average alpha = .49) and respondents with less formal education (alpha = .56). The reliability of 2 measures of sex-related attitudes is lower for female respondents. It is therefore important that the reliability of knowledge and attitude indexes be assessed not just for drug-user samples as a whole, but also within demographic groups of substantive interest.

PIP: This study pertains to a survey of 365 injection drug users (IDUs) in Los Angeles County from methadone maintenance/detoxification or residential drug free programs and earlier studies of AIDS risk reduction. The sample was stratified by ethnicity and gender and appeared to represent the local IDU population with treatment experience. There were 129 whites, 118 Hispanics, and 85 blacks; 174 were women and 158 men. 189 had completed high school and 143 had not. The purpose of the study was to investigate the reliability of indexes measuring AIDS-related knowledge and attitudes. Background information is provided on AIDS risk demographics and measurement, AIDS knowledge and attitudes, and summary of prior research. The instrument was newly designed to measure general knowledge about AIDS, perceived susceptibility to HIV infection and to self-efficacy regarding drug, and sex related risk reduction techniques and response efficacy of those techniques, and drug and sex-related risk reduction norms. 45 questions were asked by trained interviewers and answered by respondents in English in a Likert agree/disagree format. The results showed that knowledge, susceptibility, self-efficacy/drugs, self efficacy/sex, and norms/sex had the highest alphas ranging from .64 to .78. The minimum acceptable level is .50. The response norm/drugs is barely acceptable at .56. In the demographic analysis, alphas are the lowest for Hispanics in 7 out of 8 indexes, and with statistically significant differences in 4 indexes: response efficacy/drugs, response efficacy/sex, self-efficacy/sex, and norms/drugs. Response efficacy/sex and norms/drugs reach the acceptable minimum of .50 for blacks and whites only. Self-efficacy/sex for total does not reach a minimum acceptable level, however, it does for white at .58 and almost reaches acceptability for blacks at .43. There are inconsistent patterns by sex. In general, men are less reliable on the susceptibility index and women on the response efficacy/sex and self-efficacy/sex indices. Respondents with a high school education had alphas of .61 compared with .56 for low education respondents. There is lower reliability among lower educated respondents on self-efficacy/drugs and norms/drugs. The implications are that subgroup reliability is related to responses, such that low education and Hispanics score lower. Inconsistency of response may be due to methodological problems, which may be corrected statistically or by asking open-ended questions, and affect the effectiveness of interventions.