Thank You Soo Much! I have and online "DATA ANALY FOR BUSINESS APPL" class and I was confused with the homework assignment. I found this video and followed your video step by step and...... A+ On My Homework!!! "Thumbs Up"
Dude, thank you for explaining this in a very digestible way! I spent a solid hour trying to find a straight forward breakdown of this concept and stumbled across this just as I had given up finding anything useful. Cheers baud!!
Excellent presentation on Tukey-Kramer multiple comparison. The presentation is really valuable and informative. Thank you. Mohd Younus Associate Professor Ph.D Research Scholar Hyderabad, India.
thank you so much for your video. this is really useful and makes me more understanding. about the significance, how do we how much our result is significant, p
lordandladyg The formulas change somewhat when you have an 'unbalanced' study, i.e. what you are describing. I have it on my list of tutorials to make, thanks for the comment.
mistfam sorry haven't made it yet, I have been making a series of playlists on Prediction Techniques using the R statistical computing language and environment
+lordandladyg Here is what you put in under the square root in case of unequal number of observations: 1/2 x pooled variance x (1/n1+1/n2). Btw the pooled variance is the same as mean square within from the ANOVA.
The explanation is very clear, thanks a lot. I just want to ask: in scientific literature, some specific value is used to express statistical significance for the Tukey test (like p-value in ANOVA). But here you only use the difference between the critical range and the absolute difference to express the significance. I mean, we could have also expressed the significance in ANOVA via F minus Fcrit, but it's way more convenient to use a p-value :) Could you please tell, how to calculate such value for the Tukey test? Thank you. Marina.
Thanks! It was very useful for me. I just was wondering how n0 can be considered when we are calculating critical range for the groups with different number of observation.
great video! quick question: what would the n_dot value be if the observations were different between two groups (i.e: group a = 10 obs. and group b = 8 obs.)
Good question. It is important to understand that these data arose from a balanced design where one-factor anova was first utilized. When you have an unbalance design in this situation the formulas change a little and the n. is no longer a useful notation in those formulas. They are not difficult to understand and very closely related but I would have to make another series of videos for unbalanced design anova and multiple comparison.
Really great tutorial. I love the methodical step by step approach. Thank you. Is there a way of getting the level of significant difference rather than significant or not significant following the Tukey post test? Also, could I peform a Dunnett's post test instead?
What is n. if you don't have an equal number in each group. Say in group 1 i have 4 observations but in group 2 i have 3 observations. If i'm comparing 1 to 2 what should i use as n.?
Is the Tukey Num_df calculated differently from the Anova Num_df? The Anova one is 3 but the Tukey one is 4. Just curious if it's a mistake or not. Thanks!!
Thank you for posting this video. I am doing a Tukey test for the first time and this explanation is very clear! Could I ask a question? For the formula, my professor uses the mean square (within) from the ANOVA table instead of the pooled standard deviation. Is this the same thing? Thanks again!
I’m not 100% sure, but I believe in that case you would use the harmonic mean of all the sample sizes, which in this case is the harmonic mean of (17, 24, 24, 24, 24, 24). This is equal to 6/(1/17 + 5/24), or about 22.459.
Quite nicely explained!! one doubt regarding "Numerator degree of freedom =4" (7:50). Could you please explain, how is it 4? should it be 3?? please confirm. Thank you sir.
thank you very much !! i am rescued thanks to you :D but the critical range doesn't work for !! could it b beauce i use a french keyboard or something ?? can u tell me why do we use the F4 and if i can do it manually because itseems that i have a problem with that ! and thank u very much
just add $ sign after the column, row or both that you wish to lock. For example A$1 to just lock column A. A1$ to just lock row 1. A$1$ to lock both column A and row 1.
