Non-response bias in alcohol and drug population surveys
Non-response bias in alcohol and drug population surveys
复制标题
DOI：
10.1111/j.1465-3362.2009.00077.x
复制
发表时间：
2009-11-01
影响因子：
3.8
通讯作者：
MacDonald, Scott
中科院分区：
文献类型：
--
作者：
Zhao, Jinhui;Stockwell, Tim;MacDonald, Scott
引言和目的。这项拟议的研究旨在评估2004年加拿大成瘾调查(CAS)中的无反应偏差。设计与方法。两种方法被用来评估CAS中的无反应偏差，其应答率仅为47%。首先，将CAS样本特征与2002年加拿大社区健康调查(CCHS，应答率77%)和2001年加拿大人口普查数据进行比较。其次，比较了早、晚被调查者的特点。结果。与人口普查相比，收入最低、高中教育程度低的人和从未结婚的人在CAS中的比例较低，但与CCHS相似。物质使用在CAS样本中比CCHS样本中更普遍，但CAS和CCHS的大多数估计值不超过+/-3%。晚些时候的受访者也明显更有可能是男性、年轻人、高学历、吸毒者、高收入、居住在不同省份并报告使用药物。多因素Logistic回归分析发现，终生、过去12个月、慢性危险、急性危险和每月大量饮酒、终生和过去一年吸食大麻、终生使用迷幻剂、终生和过去一年任何非法药物使用均存在显著的无反应偏倚。对无反应偏差的调整大大增加了患病率估计。例如，对终身和过去12个月非法药物使用的估计分别增加了5.22%和10.34%。讨论和结论。结论是，在答复率较低的物质使用调查中，无回答偏差是一个严重的问题，但可以做出一些调整来弥补。[赵静，斯托克韦尔，麦克唐纳.酒精和毒品人群调查中的无反应偏向.毒品酒精2009年修订版]。
Introduction and Aims. This proposed study was to assess non-response bias in the 2004 Canadian Addictions Survey (CAS). Design and Methods. Two approaches were used to assess non-response bias in the CAS which had a response rate of only 47%. First, the CAS sample characteristics were compared with the 2002 Canadian Community Health Survey (CCHS, response rate 77%) and the 2001 Canada Census data. Second, characteristics of early and late respondents were compared. Results. People with lowest income and less than high-school education and those who never married were under-represented in the CAS compared with the Census, but similar to the CCHS. Substance use was more prevalent in the CAS than the CCHS sample, but most of the CAS and CCHS estimates did not exceed +/- 3% points. Late respondents were also significantly more likely to be male, young adult, highly educated, used, have high income, live in different provinces and report substance use. Multivariate logistic regression found significant non-response bias for lifetime, past 12 months, chronic risky, acute risky and heavy monthly alcohol use, lifetime and past year cannabis use, lifetime hallucinogen use, any illicit drug uses of lifetime and past year. Adjustment for non-response bias substantially increased prevalence estimates. For example, the estimates for lifetime and past 12 month illicit drug use increased by 5.22% and 10.34%. Discussion and Conclusions. It is concluded that non-response bias is a significant problem in substance use surveys with low response rates but that some adjustments can be made to compensate. [Zhao J, Stockwell T, Macdonald S. Non-response bias in alcohol and drug population surveys. Drug Alcohol Rev 2009].