Data Techniques Revision

Null Hypothesis: This hypothesis states that the relationship is by chance and not by a non-random cause. You use these statistical analysis to prove that actually the relationship is not down to luck.

Spearman's Rank

How?

•        Formulate a null hypothesis.

•        Individually rank the values of each variable. 1 = highest value.

•        Find the difference between the two.

•        Square the differences and sum the values.

•        Input into the formula.

Appropriateness:

•        Appropriate for data with 10-30 values with 2 variables that are believed to be related.

Pros:

•        Indicates the statistical significance of a result - rules out chance.

•        Gives numerical value to the strength and direction of a correlation.

Cons:

•        Does not show if there is a casual link

•        Too many tied ranks affect the validity of the test.

•

Mann Whitney-U

How:

•        Select null hypothesis.

•        Rank the data sets across the two columns. 1 = lowest value.

•        Treat as two separate columns. Add ranks in first column to get your R1 value then add ranks in the second column to get your R2 value.

•        Input into the formula.

•        Choose the smaller U value of either U1 or U2.

•        Compare to the critical values table: less than the critical value means you should reject the null hypothesis at 95…