Non- parametric ANOVAs

Question 1

What are the post hoc tests used for between and within participants design ANOVAs?

Answer 1

between = Kruskall Wallis
within = Freidmans

Question 2

Differences between parametric and non-parametric?

Answer 2

• uses the ranks of scores rather than the scores
• tests whether there are differences between the ranks of scores across more than 2 groups
• if outcome is significant = follow up tests

Question 3

Kruskal-Wallis test

Post-hoc used?

Answer 3

look at mean ranks for the 3 groups
chi-square = tells us association between the 3 levels
p value= tells us the probability of this effect being observed
X^2(df) = ,p =
-significant = means there are differences between the ranks of scores across more than 2 groups = follow up tests needed with bonferroni correction

Post-hoc used for between p’s = Mann-Whitney U
-report ‘U’ value, correct ‘p’ value with bonferroni

Question 4

Freidmans

post-hoc test used?

Answer 4

look at mean ranks for each condition
chi square = tells us association between levels
p value = tells us probability of this effect being observed
X^2(df)= , p=
if significant = post-hoc test

post-hoc used for within p’s = Wilcoxon

Question 5

Graphing data

what do we use?

Answer 5

box and whisker plot to look at where medians lie and at the spread of data
(error bar graph not appropriate as it uses means and SE’s which are only appropriate for parametric)