14. Nonparametric Tests 1 Flashcards
When are nonparametric tests used?
when assumptions of a parametric test has been violated
How do nonparametric tests work?
by ranking data
- lowest score ranked 1
- next highest score ranked 2
- etc..
Nonparametric tests are powerful as long as
the assumptions of a parametric test have been violated
What measure of central tendency do nonparametric tests use?
median (as opposed to mean in parametric)
What type of graph do you generate before running nonparametric tests?
box and whisker
What types of things can you compare for nonparametric tests?
- 2 independent groups
- 2 related groups
- 3+ independent groups
- 3+ related groups
compare 2 independent groups
Mann-Whitney U test
compare 2 related scores/samples
Wilcoxin rank-sum (W) test
compare 3+ independent groups
Kruskall-Wallis ANOVA by ranks
compare 3+ related groups
Friedman 2 way ANOVA by ranks
Mann Whitney test (U) is the equivalent to
independent/unpaired t-test
How do nonparametric tests work?
use the values of the ranks to assess differences between groups
Which two tests can be used for comparing two independent groups?
Mann-Whitney (U)
Wilcoxin rank-sum (W)
Wilcoxin rank-sum (W) is the equivalent of
dependent/paired t-test
The Kruskall-Wallis test is the equivalent to
1-way ANOVA
What is the statistic used for the Kruskall-Wallis test (or rather, the nonparametric ANOVA)
chi square statistic
Why a chi square stat?
the data creates a chi square distribution (special distribution type)
When must post-hoc tests be performed for a Kruskall-Wallis test?
if H stat is less than .05
What post hoc test would be performed if H-stat is less than .05?
Mann-Whitney test for each comparison with a Bonferroni-corrected alpha
*carefully choose to avoid Type II error
The Friedman’s 2-way ANOVA is the equivalent of this test
repeated measures anova
What test statistic is used for the Friedman’s test?
chi-square
When must you perform post-hoc test for Friedman’s?
x^2 stat is less than .05