Integrating analyses

Question 1

Non-parametric

Answer 1

data = not normally distributed
use ranks
less powerful
(median =most appropriate descriptive stats to use)

Question 2

If independent isn’t normally distributed use:

Answer 2

Mann Whitney (between subjects)

Question 3

Mann Whitney

Normality - Shapiro Wilk

Answer 3

If in the shapiro wilk test for normality the result is significant = not normally distributed = USE NON-PARAMETRIC
- mean ranks calculated - sig difference between them?
in table look at:
Mann Whitney u value= value of test statistic
Asymp. Sig = p value
(less than 0.05 = significant)
To report results:
first condition(median), second condition (median). State whether this was found to be sig or not MWU value and p value.

Question 4

If paired t test isn’t normally distributed use:

Answer 4

Wilcoxon(comparing ranks of the ps who scored higher on condition 1 to ps who scored higher on condition 2)

Question 5

Wilcoxon

Answer 5

find differences for before and after - find mean ranks for positive and negative pairs
–significant difference ?
Z score = value of the test statistic
Asymp sig, (2-tailed) = p value
Reporting the results:
sig or non-sig difference between 2 conditions before(median) and after(median; Z; P) This suggests…

Question 6

If pearsons is not normally distributed use:

Answer 6

Spearmans’s Rho (using ranked scores)

Question 7

Spearman’s Rho

Answer 7

Scatterplot used
Output box look at only top right corner for:
Correlation coefficent = spearmans cc (r)
Sig (2-tailed) = probability that 2 variables aren’t correlated (p)
need ‘r’ and ‘p’ value
Reporting the results:
say whether strong/weak positive/negative, then significant or not rs=, p< , what this tells us