Non parametric Tests

?
How do you know if the distribution is not normal?
When the bars are different sizes for example a few 10 and a few 20
1 of 48
Where is the most data recorded?
On a ratio or interval scale are normally distributed
2 of 48
When data is heavily skewed, the step size between points is what?
on the scale is probably constant
3 of 48
What should happen at the simplest level?
One should probably be using a non parametric test for ordinal data
4 of 48
What is examples of non parametric tests?
Spearman's rank, Wilcoxon T, Mann Whitney U, Friedman's, Kruskal Wallis
5 of 48
What is a Mann Whitney U?
Designed to look at difference between treatments in a between groups design
6 of 48
When is it used?
When there are only two treatment levels (1 IV with exactly 2 levels
7 of 48
What does it require?
DV to be on at least an ordinal scale
8 of 48
What is the rational?
If we arrange all the data into an ascending sequence, when the null hypothesis is true, we would expect the scores of the two groups to be randomly distributed
9 of 48
What happens when the null hypothesis is false?
We would expect the scores of the two groups to be clustered at either end of the sequence
10 of 48
Mann Whitney U test example?
Experiment: compare two groups of subjects learning lists of words and pictures, rank all scores, if the null hypothesis is true the rank totals should be the same, if the null hypothesis is false the rank totals should be different
11 of 48
What happens to calculate U?
U=N1N2 + Nx(Nx+1)/2 - Tx
12 of 48
What are all the components of this equation?
N1-number of subjects in group 1, N2 - number of subjects in group 2, Nx - number of subjects in larger rank total, Tx - total rank (the larger one)
13 of 48
When is the null hypothesis accepted or rejected according to Mann-Whitney U?
We then compare the observed value of the test statistic, U, to a critical value, We reject the null hypothesis when : Uobserved
14 of 48
Where is SPSS found?
Independent samples, Mann Whitney U, Define Groups
15 of 48
What is a Wilcoxon related samples Test?
Designed to look at differences between treatments in a within subjects design
16 of 48
What is it used for?
When there are only two treatment levels
17 of 48
What is the rationale?
Test is based on the difference between the 2 scores for each subject, it uses both direction (which score is greater) and magnitude of the differences (how much greater)
18 of 48
If H0 is true then differences in one direction will be as large as the differences in the other then?
that is what the test statistic will measure – … but it will use ranks of magnitudes rather than actual magnitudes so it can be used on skewed data
19 of 48
The test statistic we are interested in is what?
Smaller of the T values (whether thats T+ or T-
20 of 48
What do we compare
The observed value of the test statistic, W to a critical value
21 of 48
When do we reject the null hypothesis?
Wobserved
22 of 48
What happens when there are tied ranks?
We take the average of the range of ranks that ties cover and allocate this to other ties
23 of 48
What happens if there is a zero difference score?
ALl those subjects essentially get removed from analysis, Their score does not contribute to the W statistic, simply reduce N by the number of subjects where this happens
24 of 48
Where is Wilcoxon found in SPSS?
Analyse/Nonparametrictests/2related samples
25 of 48
What is the Kruskal Wallis H test?
Designed to look at differences between treatments in between group tests, used when there are multiple levels, requires DV to be an ordinal scale
26 of 48
What is the rationale?
When the null hypothesis is true we expect a random distribution of ranks across groups: average ranks of levels should be equal, however when the null hypothesis is false we expect a systematic distribution of ranks across groups:different average
27 of 48
For example?
we want to test the effects on learning of different types of reinforcement given to a child
28 of 48
What are the three conditions?
No reinforcement, positive, negative
29 of 48
What do you do next?
work out H
30 of 48
using what
An equation where where N is the total number of scores Ri is the sum of ranks for each level of the variable ni is the number of scores in each level of the variable
31 of 48
Where can you find it in SPSS?
Analyse/Non parametric tests/ k independent samples
32 of 48
In SPSS what does it convert it to?
Chi squared
33 of 48
How can we test the difference?
Kruskal- Wallis only tells us that the 3 conditions differ but not which ones
34 of 48
What can we use?
multiple Mann-Whitney tests and apply the Bonferroni correction (multiply the p value by the possible number of tests
35 of 48
Why use Bonferroni ?
TO reduce the chances of obtaining a false positive
36 of 48
What is Friedman's test?
Designed for paired samples in both within subject and matched subject design, used when there are multiple levels
37 of 48
What does it require
The Dv to be an ordinal scale
38 of 48
What is the friedman's test rationale?
Each individuals scores are ranked:
39 of 48
When the null hypothesis is true what do we expect?
random distribution of ranks across the subjects: rank totals of d.v. for the different levels of the i.v. will be about equal
40 of 48
What happens when the null hypothesis is false?
we expect a systematic distribution of ranks across the subjects: rank totals of d.v. for the different levels of the i.v. will be different
41 of 48
For example what is the Friedman test/
Same as before but run over several days in a within subject design manner
42 of 48
What happens next
Rank within each subject
43 of 48
What is the equation?
We calculate X2 R
44 of 48
Where is it found in SPSS?
Analyse|NonparametricTests|k Related Samples
45 of 48
As with Kruskall Wallis, the Friedman test does not tell us which condition differs?
is praise is better than no reinforcement? – is criticism worse than no reinforcement? – is praise better than criticism?
46 of 48
What can we use?
Multiple Bonferroni corrected wilcoxon tests
47 of 48
What is a real world example?
Repeated exposure to social stress alters the development of agonistic behavior in male golden hamsters.
48 of 48

Other cards in this set

Card 2

Front

Where is the most data recorded?

Back

On a ratio or interval scale are normally distributed

Card 3

Front

When data is heavily skewed, the step size between points is what?

Back

Preview of the front of card 3

Card 4

Front

What should happen at the simplest level?

Back

Preview of the front of card 4

Card 5

Front

What is examples of non parametric tests?

Back

Preview of the front of card 5
View more cards

Comments

No comments have yet been made

Similar Psychology resources:

See all Psychology resources »See all Statistics resources »