Adlaf E M, Kohn P M
Br J Addict. 1989 Jul;84(7):749-57. doi: 10.1111/j.1360-0443.1989.tb03054.x.
Re-analysis employing covariance-structural models was conducted on Strickland's (1983) survey data on 772 drinking students from Grades 7, 9 and 11. These data bear on the relations among alcohol consumption, alcohol abuse, association with drinking peers and exposure to televised alcohol advertising. Whereas Strickland used a just-identified model which, therefore, could not be tested for goodness of fit, our re-analysis tested several alternative models, which could be contradicted by the data. One model did fit his data particularly well. Its major implications are as follows: (1) Symptomatic consumption, negative consequences and self-rated severity of alcohol-related problems apparently reflect a common underlying factor, namely alcohol abuse. (2) Use of alcohol to relieve distress and frequency of intoxication, however, appear not to reflect abuse, although frequent intoxication contributes substantially to it. (3). Alcohol advertising affects consumption directly and abuse indirectly, although peer association has far greater impact on both consumption and abuse. These findings are interpreted as lending little support to further restrictions on advertising.
运用协方差结构模型对斯特里克兰（1983年）关于772名7年级、9年级和11年级饮酒学生的调查数据进行了重新分析。这些数据涉及酒精消费、酒精滥用、与饮酒同伴的交往以及接触电视酒精广告之间的关系。斯特里克兰使用的是一个恰好识别的模型，因此无法检验其拟合优度，而我们的重新分析检验了几个可被数据反驳的替代模型。有一个模型与他的数据拟合得特别好。其主要结论如下：（1）有症状的饮酒行为、负面后果以及与酒精相关问题的自评严重程度显然反映了一个共同的潜在因素，即酒精滥用。（2）然而，为缓解痛苦而饮酒以及醉酒频率似乎并不反映滥用，尽管频繁醉酒对其有很大影响。（3）酒精广告直接影响消费，间接影响滥用，尽管同伴交往对消费和滥用的影响要大得多。这些发现被解释为几乎没有支持进一步限制广告的依据。
