Mäkelä K, Simpura J
Drug Alcohol Depend. 1985 Aug;15(4):389-404. doi: 10.1016/0376-8716(85)90018-3.
On data from a representative mail survey of the adult population in Finland, generalized linear modelling was applied to express the probability of consequences of drinking as a function of annual intake of alcohol. Most experiences related to drinking could be predicted as a simple function of annual intake. Annual intake therefore should be retained as a basic variable describing involvement with alcohol. The slopes of the models indicated that particular behavioral concomitants of drinking and direct causal consequences of drinking grow less quickly with increasing consumption than the probability of social control reactions to drinking. Important differences between age groups in the incidence of hazardous drinking behaviours and of causal consequences of drinking prevailed, but, with the exception of drunken driving, sex differences practically vanished when annual intake was held constant. Contrary to the common view that women's drinking is more closely controlled than male drinking, men on each intake level more frequently reported control reactions by significant others.
根据芬兰成年人口具有代表性的邮件调查数据，运用广义线性模型来表示饮酒后果的概率是酒精年摄入量的函数。大多数与饮酒相关的经历都可以作为年摄入量的简单函数进行预测。因此，年摄入量应作为描述饮酒情况的一个基本变量保留下来。模型的斜率表明，随着饮酒量增加，饮酒的特定行为伴随情况和饮酒的直接因果后果的增长速度，比社会对饮酒控制反应的概率增长得慢。危险饮酒行为的发生率以及饮酒的因果后果在不同年龄组之间存在重要差异，但是，除了酒后驾车外，当年摄入量保持恒定时，性别差异实际上消失了。与普遍认为女性饮酒比男性饮酒受到更严格控制的观点相反，在每个摄入量水平上，男性更频繁地报告有重要他人的控制反应。
