# Statistical Tests

HideShow resource information
• Created by: mollyb
• Created on: 13-06-13 14:59
Mann Whitney U
A test for differences. Compares medians and ranks data to see if it differs.
1 of 11
Shows the median between two sets of data, good with dealing with skewed data so doesnt need to be normally distributed, you can decide the boundaries of the 2 groups and only needs one variable in a set of data.
2 of 11
more appropriate when the sets of data are independent of each other or when both dat sets have the same shape distribution, have to have equale sample sizes and become less accurate when the sample size is below 5 or above 20
3 of 11
Chi-Squared (Difference)
when used to test differences it compares observed and expected frequencies
4 of 11
uses a complex formula, calculation and uses a significance table making it hard to use. Cant use percentages for this test and data must be numerical
5 of 11
Chi-squared (association)
compares the observed and the freqency expected in a given hypothesis
6 of 11
can test asssociations between variables
7 of 11
can't use percentages, data must be numerical, categories of 2 can not be compared, number of observations must be 20+, test becomes invalid if any of the expected values are below 5, quiet complicated to get right and has a difficult formula
8 of 11
Spearman's Rank
Measures a correlation or association and measures the strength of a relationship bwteen two ranked data sets
9 of 11
shows the significance of the data, proves/disporves a correlation, allows for further analysis and doesn't assume normal distribution
10 of 11
can be difficult to work out, quiet a complicated formula, can be misinterpreted, need two sets of variable data so the test can be performed.
11 of 11

## Other cards in this set

### Card 2

#### Front

Shows the median between two sets of data, good with dealing with skewed data so doesnt need to be normally distributed, you can decide the boundaries of the 2 groups and only needs one variable in a set of data.

### Card 3

#### Front

more appropriate when the sets of data are independent of each other or when both dat sets have the same shape distribution, have to have equale sample sizes and become less accurate when the sample size is below 5 or above 20

### Card 4

#### Front

when used to test differences it compares observed and expected frequencies

### Card 5

#### Front

uses a complex formula, calculation and uses a significance table making it hard to use. Cant use percentages for this test and data must be numerical