Non-Parametric Models - From Textbook Flashcards
What are the 4 most common non-parametric procedures?
Mann-Whitney test
Wilcoxon signed-rank test
Friedman’s test
Kruskal-Wallis test
What do the 4 most common non-parametric procedures overcome?
Distributional problems
How do the 4 most common non-parametric procedures overcome distributional problems?
By ranking the data
The model is then fitted to the ranks and NOT the raw scores.
Why are the 4 most common non-parametric procedures fitted to the ranks and NOT the raw scores?
To eliminate the effect of outliers + irons out problems w/ skew.
i.e you have 20 data points + the 2 highest are 30 and 60, the scores become ranks of 19 + 20.
-ive to non-parametric tests when fitting the mode to ranks and NOT the raw scores
Lose info about magnitude of differences between scores
What is meant by statistical power?
Ability of a test to find an effect that genuinely exists.
If there’s a genuine effects = non-parametric test is less likely to detect it than a parametric one.
What do you do on SPSS when comparing groups containing different entities
Select analyse
Non-parametric tests
Independent samples
What do you do on SPSS when comparing scores that were taken from the same entities under difference conditions
Select analyse
Non-parametric tests
Related samples
SPSS NON-PARAMETRIC TESTS
What do you select in objectives tab
Customise analysis
What are the options to compare the distribution in 2 conditions containing scores from different entities?
Mann-Whitney test
Wilcoxon rank-sum test
Rank data - ignoring the group to which a person belonged - from lowest to highest.
What should you do if there’s no difference between the groups?
Find a similar no. of high + low ranks in each groups.
If you added up the ranks, you would expect summed total of ranks in each group to be ~ same.
When groups have unequal no. of participants in them.
What is the Test statistic (Ws) for the Wilcoxon rank-sum test?
Sum of ranks in the group containing the fewer people
When group sizes are equal its the value of the smaller summed rank.
How to run a Mann-Whitney test using SPSS
Select analyse
Non-parametric tests
Independent samples
Fields tab
Custom field assignments
Specify model by dragging variables into test fields
Calculation for effect size
r = z score / (square root of size of study)
What must you write for the results from the Mann-Whitney test
Test statistic + its significance
