Burns K C
Aviat Space Environ Med. 1984 Jun;55(6):521-7.
A statistical mixture model is used to fit time-to-emesis data. The Weibull probability distribution is shown to provide a good fit for those subjects who either become sick or withdraw from the experiment within 2 h. The second part of the mixture accounts for those subjects who neither quit nor vomit within 2 h. The log-normal probability model is shown to give a poorer fit to the data and figures showing the relative fits of the estimated Weibull and log-normal distributions are provided. A nonparametric test is used to compare the five motion conditions of the Correlation Study. That test shows that there are significant differences in severity among the conditions.
使用统计混合模型来拟合呕吐发生时间数据。结果表明，威布尔概率分布能很好地拟合那些在2小时内出现恶心或退出实验的受试者的数据。混合模型的第二部分用于描述那些在2小时内既未退出也未呕吐的受试者。结果表明，对数正态概率模型对数据的拟合效果较差，并提供了显示估计的威布尔分布和对数正态分布相对拟合情况的图表。使用非参数检验来比较相关性研究的五种运动条件。该检验表明，各条件之间在严重程度上存在显著差异。
