# testing for differences

basics of statisical testing and notes on what the mann whitney u test and the chi squared test are useful for

- Created by: Thegirlwhoknewtoomuch - Team GR
- Created on: 17-10-13 13:23

First 297 words of the document:

Geographical skills theory

Testing for differences

These statistical tests tell you if there is a difference between two sets of data

as well as how much of a difference there is.

Statistical testing

1. State the null hypothesis (no significant difference between data 1 and data

2)

2. State the alternate hypothesis, this one must be true if the null hypothesis

is wrong (there is a significant difference between data 1 and data 2)

3. Select a statistical test. This should be done on the basis of whether the

data fits the assumptions made by the test.

4. Decide on a significance level- this is the level of confidence you must have

to know that the null hypothesis has been correctly rejected.

5. Work out the critical value for the test result beyond which the null

hypothesis will be rejected.

6. Do the test

7. Decide the significance of the data on the basis of the result.

Mann Whitney U test

Test the difference between the medians of two sets of data and can be used when

the data is in an awkward form such as samples of very different sizes. The test

starts out be assuming both sets of data are similar and then establishes whether

that assumption should be rejected or not. It uses ranks of data so it can be used

for any data that can be put into order.

Chi-squared test

This looks at the frequencies that were actually observed and compares them with

the frequencies expected. As you decide the expected, you must be clear on what

you expect and why. Usually the expected value is that which would have occurred

due to chance.

## Comments

No comments have yet been made