Nonparametic statistics
10 important questions on Nonparametic statistics
How do we call the category of methods for comparing treatments without assuming a normal distribution for the response variable in the population?
How do you get numbers to perform a wilcoxon nonparametric test for comparing groups?
- take two independent groups
- create a table of all possible ranking combinations between participants
- take the mean ranking for each group on each combination
- for each combination, take the mean difference between groups
- create a sampling distribution with x-axis mean differences per combination, y-axis is probability of occurring
- then make the actual ranking of the participants and take the mean ranking dofference between groups.
How do you get to a conclusion using the wilcoxon nonparamtric test for comparing groups?
- make assumptions
- H0: identical population distributions, so equal expected values for sample mean differences
- H1: different values for sample mean differences, or higher sample mean differences
- test statistic: difference in sample mean ranking.
- p-value: sum of prob. of difference in sample means and all differnces in sample means above that.
- conclusion.
- Higher grades + faster learning
- Never study anything twice
- 100% sure, 100% understanding
How can you use the wilcoxon nonparametric test with sums of ranks?
- create the same sampling distribution, so with the prob. on y-axis
- only this time the x axis will consist of all possible sum of rankings for only one group.
- then calculate p value with the observed rank sum.
What do we do if two participants tend to have equally good ranks?
How do you get a two sided p value from a single z score?
When is it advisable to summarize a group by median instead of mean?
What is the assumption for using the median to compare groups?
When would you use the kruskal-wallis test instead of anova f?
How do you perform a kruskal wallis nonparametric test for comparing several groups?
- assumptions
- H0: identical population distributions for g groups.
- H1: population distributions not identical
- use kruskal wallis test statistic
- p value from chi squared with df=g-1
- conclusion
The question on the page originate from the summary of the following study material:
- A unique study and practice tool
- Never study anything twice again
- Get the grades you hope for
- 100% sure, 100% understanding