Great series! Do you have a suggestion on how to combine ordinal-based items (motor skills classified as 1, 2, 3, 4, where 1 is less skilled and 4 most skilled) to form a composite score, say total motor skill competence? This is would be my IV. Then, compare the composite scores between 2 groups (1= normal weight and 2= overweight). Do you have a suggestion on the type of analysis? Any help would be greatly appreciated.
Hello, do you have a literure reference for me where I can find the statements "The median test is more robust to violations of normality and homogeneity of variances in comparison to the Kruskal-Wallis test and Mann-Whitney U test " and "median-test is not powerful enough"? Because I would need literature references for these statements for my masterthesis and I can't find such. That would be really kind :)
If I'm understanding this right. If my variance isn't homogenous I can't use the kruskul-wallis test but I can use the median test. Right? Are there any other options when my variance isn't homogenous? best regards.
Hi there, Thanks very much for your useful videos. I watched so many of them already. I hope you are still actively playing around with your youtube channel. I have one question that is: How can we conduct an "1 sample t-test" for a non-normally distributed interval/ratio variable? Many thanks!
4:10 is where the explanation starts
Great video! I was losing hope I will finish my Masters degree :)
Great series!
Do you have a suggestion on how to combine ordinal-based items (motor skills classified as 1, 2, 3, 4, where 1 is less skilled and 4 most skilled) to form a composite score, say total motor skill competence? This is would be my IV.
Then, compare the composite scores between 2 groups (1= normal weight and 2= overweight).
Do you have a suggestion on the type of analysis? Any help would be greatly appreciated.
is the median test only for three and more groups? i.e. can you use it as equivalent to the Mann-Whit when you dont have homg. of variance?
Hello, do you have a literure reference for me where I can find the statements "The median test is more robust to violations of normality and homogeneity of variances in comparison to the Kruskal-Wallis test and Mann-Whitney U test " and "median-test is not powerful enough"?
Because I would need literature references for these statements for my masterthesis and I can't find such.
That would be really kind :)
Go into Google Books and type: "median test" is more robust than Kruskal-Wallis
You'll find several references, without even having to go into the books!
@@how2stats thank you very very much for all your information :) It helped me a lot!!!
Just sum across items to create the DV. Then, probably do an independent groups t-test if assumptions are satisfied.
If I'm understanding this right. If my variance isn't homogenous I can't use the kruskul-wallis test but I can use the median test. Right? Are there any other options when my variance isn't homogenous?
best regards.
Jason Már Bergsteinsson Welch's t-test; Brown-Forsythe F test.
is it possible to use it with one dependent and one independent varible?
That's the example in this video: group is the independent variable and drinks is the dependent variable.
Great test! You have three groups (locations) with 15 in each. Can the groups be different sizes?
Sure.
Sure, median test can be used with just two groups.
Hi there, Thanks very much for your useful videos. I watched so many of them already. I hope you are still actively playing around with your youtube channel. I have one question that is: How can we conduct an "1 sample t-test" for a non-normally distributed interval/ratio variable? Many thanks!