Non-parametric 1 way ANOVA

Question 1

what are the assumptions we should not violate for parametric analyses?

Answer 1

equal sphericity (within ppts), equal variance (between ppts), no outliers, normal distribution, continuous DV

Question 2

if you have 2 histograms that are normally distributed and 1 that is not do you run parametric or non parametric analyses?

Answer 2

non parametric

Question 3

if one group has a significantly wider set of data should we run a non parametric test?

Answer 3

yes (this bigger spread will be caused by an outlier) so this results in violation of assumtions of equal variance/sphericity

Question 4

if it is only the normal distribution that is violated can you run a parametric ANOVA?

Answer 4

yes (parametric ANOVAs are robust to violations of normality if there are equal sample sizes)

Question 5

what is the non parametric test when there is a violation of assumptions for BETWEEN participants (independent t test) for 2 GROUPS?

Answer 5

Mann-Whitney

Question 6

what what is the non parametric test when there is a violation of assumptions for BETWEEN participants (independent t test) for MORE THAN 2 GROUPS?

Answer 6

Kruskall-Wallis

Question 7

what is the non parametric test when there is a violation of assumptions for WITHIN participants (paired t test) for 2 GROUPS?

Answer 7

Wilcoxon

Question 8

what is the non parametric test when there is a violation of assumptions for WITHIN participants (paired t test) for MORE THAN 2 GROUPS?

Answer 8

Friedman Test

Question 9

do non-parametric ANOVAs use actual or rank scores?

Answer 9

rank

Question 10

non parametric ANOVAs rank scores for the whole sample = which design?
non parametric ANOVAs rank scores across conditions = which design?

Answer 10

whole sample = between ppts, across conditions = within ppts

Question 11

why are error bar graphs with mean and SE not appropriate for non-parametric data?

Answer 11

because we report the median values, so we use BAR GRAPHS based on median values

Question 12

provide an account of how non-parametric ANOVAs handle violations of parametric ANOVAs

Answer 12

they RANK scores instead of using the raw scores. This manages violations of distribution, outliers and sphercity/variances. Outliers are reduced as ranking scores brings them closer to the majority of data. Differences in the spread of data is therefore also reduced which reduces the variability in the scores and brings the range of data closer to the median scores