※ 引述《binchung (ideal)》之铭言： : ※ 引述《gsuper (绿色苏打心)》之铭言： : : 我刚刚一直再想这个问题 : : 也是觉得只有 rank 才能解决 : : 可是 他的data最少最少也要画分成3个categories : : "满意" "普通" "不满意" : : 若是rank下来 : : (Oi-Ei)^2 : : sigma ___________ = test statistic = chi-square : : Ei : : 不论 满意度 上升或是下降都会趋向显着 : : 重点就是 满意和不满意 : : 两个catagories在他的命题条件之下 : : 是相冲突的 : : 这到底要怎麽解决阿? : : 难到是要 rank 成 1 2 0 -1 -2 这样吗? I was not clear in my previous post, so I made clear in this one. To be precise, first consider each subject has a rank. For example, the first categories have 30+45=75 subjects, so the ranks are 1,2,...,75. Then you can calculate the midrank for the first group. : Yes, consider the data are the ranks with lots of ties. : Check the Wilcoxon midrank test, then you will see : how to do it. You don't need to worry about how many categories the data set : has. --

※ 发信站: 批踢踢实业坊(ptt.cc) ◆ From: 208.125.17.166