Statistical tests

When would you use a Mann Whitney U test?

• Looking for a difference

• Data is at least ordinal

• Independent measures design

Steps for Mann Whitney (and relationship to critical value)

• Rank whole data as one set and add the total ranks for each option (lowest score gets rank 1)

• Plug into the formula to find Ua and Ub

• Whichever value of U gives the smaller value take as your observed value

• Observed value must be equal to or less than critical value

When to use a Wilcoxon test

• Looking for a difference

• Data is at least ordinal

• Repeated measures design

Method for Wilcoxon test (and relationship to the observed value)

• Calculate the difference between each score given in the data pairs A and B

• Rank the differences as a whole list, ignoring the + or - signs. (1 is the smallest difference)

• Add positive ranks in one column and negative ranks in the other

• Whichever is the smallest value is the T value

• Observed value must be equal to or less than the critical value

When to use a Spearman’s test

• Looking for a relationship

• Data is at least ordinal

Method of Spearman’s

• Rank each data separately

• Calculate the differences between these ranks, ignoring + or -

• Calculate the differences squared (then add the total of d squared)

• Plug into formula

• Observed value must be equal to or greater than the critical value

When to use a Chi squared test

• Looking for a difference

• Nominal data

Method for Chi squared test

• Calculate the total number of expected choices. This is the total of results divided by the number of options

• Calculate the degree of freedom - R-1 X C-1

• Do the formula in a table form: observed value minus expected; square all of the results; divide each answer by expected value

• Observed value must be equal to or greater than critical value