Lies, damned lies, and statistics

During the last half term we've covered four statistical tests in some detail, and if you've completed all the work set then you will be fully prepared for any questions on stats in the unit 4 exam. You won't be asked to complete a test from start to finish starting with raw data; past questions have focused on one section of the process eg comparing obtained vs critical values or looking or critical values or ranking data.

A common question on the paper is to ask which test should be used for a particular set of data. This sheet gives a summary of each test and when to use them. This sort of question is often worth 2 or 3 marks - meaning that it isn't enough to state that the data is correlational (or independent groups or whatever) - you also need to match the type of data to the test. This sheet is a good summary of different data types (nominal, ordinal, ratio & interval).

Resources by test:

Spearman's rho:

• Used for correlational data, when the two sets of data are pairs of scores from one person, and when the data are ordinal or interval.
• Intro ppt
• Step-by-step worksheet
• Worksheet 2

Chi-squared test:

• Used when the hypothesis predicts a difference or an association - is membership of one category associated with  membership of another category? Used for categorical data and for comparing frequencies. The data must be unrelated - you can't be in more than one category.
• Intro ppt
• Step-by-step worksheet
• Worksheet 2

Mann-Whitney test:

• Used for independent groups design, when the hypothesis predicts a difference between two sets of data. The data must be interval or ordinal. 
• Intro ppt
• Step-by-step worksheet
• Worksheet 2

Wilcoxon T test:

• Used in repeated measures or matched pairs design, for ordinal or interval data,
• Intro ppt
• Step-by-step worksheet
• Worksheet 2