Question

in Questions 6-7, we have four numerical samples, each of (sune) size n, denoted (Group 1) Y1,1, . . . . У (Group 3) Yi.3, Y,,a, and (Group 4) Yǐ,4, x,4. ї, (Group 2) Yī.2, .. .. Y., Question 6. Write down the Kruskal-Wallis (KW) test statistic (stripped of useless constants). Specify how the ranks are computed Question 7. (continued from Q6) What is te hypothesis that is most appropriate for this test? Explain why.

Answer 1

Dala Yn.2 → Group 2 er eup3 botoa m YOW aueshon-6 krus kal ais test statistie is deined a - 11)4 为 n: sample she for each grou 闪kin 2 2 tv ド1 camtu Y r(si): rank of yj e at first e have to miz tua hole dote and ordun i in com bined ordan. mag in increasing

s.Imi lrtly cal cudate Pe, 3 24 uestion; Kruskal- au i test isequivalent to RNoy A the ﹃uautg 여 meers tor diffe7 ent tent . tuAince oe have ne dictribushonal means ix ne- In non-partemetric set assumpHơw tanto e' fest for equality doen not exish). But median exist aloaja. Hence, it is v approhinte to test alit of medMank. 1,2 ,3/u se u 부ㅈ it is boncauy o each 1uantiu. tu. specia