Non parametric tests Flashcards
What are meant by non-parametric tests?
Non-parametric tests are tests where there are no assumptions about the distribution
Do non-parametric tests use a random sample?
Yeah baby
Comment of the sensitivity of non-parametric tests to outliers
not sensitive to outliers
Do non parametric tests work with large or small data sets?
Both
What distribution are non-parametric tests useful for?
All
How do non-parametric tests give values to scores?
Rankings
What is the test statistic for non-parametric tests?
the test statistic of non-parametric tests is the difference between the mean ranking or sum ranking.
What is the p-value for non-parametric tests?
the difference between the mean ranking or sum ranking is standardised and the p-value for non-parametric tests is based on this standardisation.
Why are parametric tests preferred over non-parametric tests when possible?
because parametric tests have more power than non-parametric tests and not every parametric test has a non-parametric alternative.
What is the equivalent of the independent samples t test for non parametric? What does it test?
The Wilcoxon rank-sum test, two independent samples
What does the Wilcoxon test use?
It uses the sum of the ranks per group and uses a z-score. The difference in group rank is used.
What test does the same as the Wilcoxon test?
The Mann-Whitney test
What non parametric test is used for two dependent samples?
The Wilcoxon signed-rank test
What does The Wilcoxon signed-rank test use as a test statistic?
It assigned a ‘+’ or ‘-‘ to the difference between two repeated measures. The ranks for the positive group are summed and standardised.
What is the Kruskal-Wallis test used for?
The Kruskal-Wallis test uses more than two independent samples.
What does the Kruskal-Wallis test do?
It subtracts the expected mean ranking from the observed mean ranking and makes use of a Chi-square distribution.
What test utilises more than two dependent samples?
Friedman’s ANOVA
What does Ri denote?
the sum of ranks for each group.
How is the effect size for the Wilcoxon rank-sum test and the Mann-Whitney test calculated?
r= z/ sqrt(n)
When is a Kolmogorov-Smirnov Z test utilised?
This tests whether two groups have been drawn from the same population and has more power than the Mann-Whitney test when the sample sizes are smaller than 25 per group.
When is the Moses Extreme reaction test utilised?
This tests the variability of scores across the two groups and is a non-parametric equivalent of Levene’s test of equal variances.
What does the Wald-Wolfowitz runs test do?
This tests clusters of scores in order to determine whether groups differ.
What should happen if there is no difference in the Wald-Wolfowitz run test?
The ranks should be randomly interspersed if there is no difference.
What does the sign test do?
This does the same as the Wilcoxon-signed rank test but is only based on the direction of the difference in ranks. The magnitude of change is neglected.
Comment on the power of the sign test
This test lacks power unless the sample size is very small.
What does McNemar’s test compare?
This test compares the number of people who changed their response to one direction to those who changed it in the opposite direction
What data does McNemar’s test use?
uses nominal, rather than ordinal data.
When is McNemar’s test useful?
It is useful when looking for changes in people’s scores.
Name an extension to McNemar’s test and what it is similar to
Marginal homogeneity is an extension of McNemar’s test and is similar to the Wilcoxon test.
Name the non-parametric equivelant of an ANOVA to compare two groups and comment on its power
Friedman’s 2-way ANOVA by ranks (k samples) is the non-parametric equivalent of an ANOVA to compare two groups but it has low power compared to the Wilcoxon signed-rank test.
What does the median test test?
This tests whether the samples are drawn from a population with the same median.
What does the Jonckheere-Terstra test for?
an ordered pattern of the medians of the group
How does the Jonckheere-Terstra test differ from the Kruskal-Wallis test
It does the same as the Kruskal- Wallis test but incorporates the order of the groups.
When is the Jonckheere-Terpstra useful?
It is useful to test for trends in the data and should be used when a meaningful order of medians is expected.
What does Kendall’s W test?
the agreement between raters and ranges between 0 and 1.
What does Cochran’s Q test do?
This is Friedman’s test on dichotomous data.