Non-Parametric Between-Subjects Tests IV - that changes DV - measures Level 1 and level 2 - different groups How to measure the DV? E.g. questionnaire Levine scale would produce ordinal data Hypothesis testing: Null hypothesis (HOti - there will be no difference between the two groups Experimental hypothesis (Hati - One tailed/two tailed Significance levels for hypothesis testing - p values (less than 0.05 = significantti P values: The probability of obtaining a result given the null hypothesis is true P < . 05 = significant - the result is highly unlikely to be due to chance (reject the nullti P > .05 = not significant - difference is due to random factors (confoundingti Parametric assumptions: · Normal distribution · Homogeneity of variance · Independence of data points - data from one participant does not affect another What do you do if data violates parametric assumptions? 1. Check for outliers - re-examine 2. Transform the data 3. Run alternative tests that do not rely on parametric assumptions Otherwise it will compromise the precision of estimates (inaccurate statsti and lead to inaccurate predictions of behaviour (wrong conclusionsti Example of a non-parametric between-subjects test: 1. Mann-Whitney U Test · Testing differences · When data does not met assumptions of parametric data · Between-subjects design · 2 groups Used to test between the conditions to identify where the actual differences lie Effect sizes: (r) You must always report effect sizes with significant effects r = z VN
2. Kruskal Wallis Test It is similar to Mann-Whitney · Between-subjects · Non-parametric · But more than 2 groups Test tells us if our IV has some significant effect across all the groups Post hoc comparisons: Problem of familywise error - Every statistical test produces a chance of finding a result as significant when actually the null hypothesis is true (type 1 errorti Error rate = 1 - (1-ati n (a is the alpha/p level and n is the number of tests runti