Hays R D, Revetto J P
RAND Corporation, Santa Monica, CA 90406-2138.
J Drug Educ. 1990;20(3):191-8. doi: 10.2190/65RP-7LNE-91LT-7J4L.
Peer cluster theory hypothesizes that peer drug association has a direct effect on adolescent drug use. In turn, peer drug associations are influenced by familial factors (sanctions and strength) and individual variables (religious identification, school adjustment). Oetting and Beauvais evaluated peer cluster theory in a cross-sectional survey of 415 high school students [1]. We hypothesized, evaluated, and found support for an alternative model in which poorer school adjustment was specified as a consequence of drug use, peer drug associations, lack of family sanctions against drug use, low religious identification, and absence of family strength. This reanalysis illustrates that interpretations of structural equation modeling results from cross-sectional data are especially problematic.
同伴群体理论假设同伴间的药物关联对青少年药物使用有直接影响。反过来，同伴间的药物关联又受到家庭因素（制裁和家庭影响力）和个体变量（宗教认同、学校适应情况）的影响。奥廷和博韦在对415名高中生进行的横断面调查中评估了同伴群体理论[1]。我们提出假设、进行评估，并发现了一个替代模型的支持证据，在该模型中，较差的学校适应情况被确定为药物使用、同伴间的药物关联、缺乏家庭对药物使用的制裁、宗教认同度低以及家庭影响力缺失的结果。这种重新分析表明，对横断面数据的结构方程建模结果的解释尤其存在问题。
