Mann-Whitney U test Flashcards
What 2 tests investigate whether there is evidence for a significant difference between scores in 2 conditions based on ranks?
1) The Mann-Whitney U test
2) Wilcoxon signed-rank test
What is the Mann-Whitney U test?
A non-parametric alternative for the independent t-test
A non-parametric alternative for the independent t-test
This is known as…?
Mann-Whitney U test
What is Wilcoxon signed rank test?
A non parametric alternative for the paired/related t-test
A non parametric alternative for the paired/related t-test
This is known as…?
Wilcoxon signed rank test (Wilcoxon)
How do you find the U value using the Mann-Whitney U test?
1) List the data from lowest to highest, regardless of which condition/groups they’re from
2) Rank the data
3) Calculate the sum of ranks in each condition
4) Calculate the smallest sum of ranks that could be for both conditions
5) Calculate how far the overall sum of ranks is above the smallest sum of ranks for both conditions
6) Mann Whitney U is the smaller value between the two conditions
What does Mann Whitney U tell us?
How far away is the sum of ranks for the group which has the smallest sum of ranks
The smallest the sum of ranks could be depends on…?
The size of the sample (how many people you have in each condition)
What is the p-value assuming that the sample size is large?
Asymp. Sig (2-tailed)
What is the p-value associated with Mann-Whitney U?
Exact Sig. [2*(1-tailed Sig.)]
Exact Sig. [2*(1-tailed Sig.)]
What is this p-value associated with?
Mann-Whitney U
Asymp. Sig (2-tailed)
What is this p-value associated with?
Assumes that the sample size is large
You want to test the idea that sad vs happy facial expressions impact on likeability ratings.
You formulate a research hypothesis stating that there will be a difference in the likeability ratings (on a 20-point scale) given for sad vs happy face pictures.
14 participants are assigned at random to either the happy or sad face group and the following data are obtained:
Happy ratings: {7, 15, 14, 12, 17, 12} (N = 6)
Sad ratings: {4, 6, 11, 7, 9, 4, 7, 18} (N = 8)
You wish to conduct a Mann-Whitney U test on this data to test your hypothesis.
First, calculate the ranks overall scores across both groups (taking ties into account).
a) The rank sum for the happy group is:
b) The rank sum for the sad group is:
The minimum possible sums of ranks for the happy and sad groups are 21 and 36 respectively.
c) U1, i.e. how much higher the rank sum is than the minimum rank sum for the happy group, is:
d) U2, i.e. how much higher the rank sum is than the minimum rank sum for the happy group, is:
e) The Mann-Whitney U statistic is:
a) 60
b) 45
c) 39
d) 9
e) 9
You want to test the idea that smart vs casual clothing impacts on employability ratings.
You formulate a research hypothesis stating that there will be a difference in the employability rating (on a 20-point scale) given for a picture of a person wearing either smart or casual clothes.
16 participants are assigned at random to either the smart or casual picture group and the following data are obtained:
Smart ratings: {7, 15, 14, 12, 17, 12, 11, 11} (N = 8)
Casual ratings: {4, 4, 12, 7, 9, 12, 9, 18} (N = 8)
You wish to conduct a Mann-Whitney U test on this data to test your hypothesis.
First, calculate the ranks overall scores across both groups (taking ties into account).
a) The rank sum for the smart group is:
b) The rank sum for the casual group is:
Note that the minimum possible sum of ranks for both the smart and casual groups is 36.
c) U1, i.e. how much higher the rank sum is than the minimum rank sum for the smart group, is:
d) U2, i.e. how much higher the rank sum is than the minimum rank sum for the casual group, is:
e) The Mann-Whitney U statistic is:
a) 81.5
b) 54.5
c) 45.5
d) 18.5
e) 18.5
