Відео

Dear Yakir, You should know that in academic and scientific arguments you can’t simply say someone is wrong (and get away with it :) ) without providing evidence! I am just worried that newbies to statistics who want to learn the technique may get misguided because of your unsupported claims!! Additionally, I would love to help you understand the technique :)  First, Tukey’s post hoc is based on computing the honestly significant difference value (HSD) as follows: HSD=q*sqrt of MSwithin / n (you said in your comment that I used Tukey’s, but you did not provide any equations or explain how). Second, the F statistic you calculate in ANOVA informs you about any statically significant difference among all means of the treatments. ANOVA F statistic doesn’t help you identify if a particular treatment’s mean is significantly different form the other observed means. That is why you conduct a post hoc test (For example, ANOVA tells you if there is a statically significant difference among the means of treatments A, B, and C. However, it does not tell if treatment A’s mean is significantly different from B’s and C’s and so on). While calculating Tukey’s test depends on computing the statistically significant minimum difference between two means , the Scheffe post hoc depends on computing the F- ratio as I explained in the video (the critical value of the Scheffe's F -ration is used just like that of the overall ANOVA to evaluate differences and their significance among treatments' means)! I recommend this book to my graduate students: Statistics for the Behavioral Sciences by Frederick Gravetter and Larry Wallnau. If you can get the seventh edition, read pp. 416-419. This link might also help: statisticslectures.com/topics/posthoconewayanova/ I wish that helped and thank you for watching the video!

Thank you for watching!