目的　介绍群随机试验的统计方法并与传统方法比较。方法　以青少年戒烟干预试验为例 ,介绍群随机试验的假设检验及样本含量估计方法。结果　用传统统计方法分析群随机试验会产生虚假的统计显著性结论 ,增大了Ⅰ型错误 ;用传统样本含量计算公式将低估群随机试验所需的样本含量 ,增大了Ⅱ型错误。结论　忽视群效应将产生严重的统计分析偏性 Objective To introduce statistical methods for group randomized trials and compare them with traditional methods. Methods Adolescent smoking cessation intervention trials were used as an example to introduce hypothesis testing and sample content estimation methods for group randomized trials. Results The use of traditional statistical methods to analyze group randomized trials yielded false statistically significant conclusions that increased Type I errors. The use of traditional sample content calculation formulas would underestimate the sample size required for group randomized trials and increase Type II errors. Conclusion Ignoring group effects will produce serious statistical analysis bias