Choosing a statistical test Flashcards
What are the 3 key questions when choosing a statistical test?
Testing for difference or correlation, related or unrelated design and nominal, ordinal or interval level data
What do the hypothesis for testing for difference predict?
The group/condition scores will differ
How are the results of a test for difference displayed?
A bar chart
What does the hypothesis for a test of a correlation predict?
There will be a positive or negative correlation
How are the results of a test for correlation displayed?
A scattergram
What are the types of tests for difference?
A related or unrelated design
Where does related data come from?
The same participants (repeated measures) and matched pairs
Where does unrelated data come from?
Independent groups - including a comparison of 2 different things
What are the types of data?
Nominal, ordinal and interval
What is nominal data?
When particpants are put in a category - yes or no / attachment type
What is ordinal and interval data?
A participant has a numerical score for each particpant
What is interval data?
An agreed way or measuring the variable, no judgment, intervals between points on scale are fixed and can make mathematical comparison
What are examples of interval data
Time taken and height
What is ordinal data?
Different possible ways of measuring variable involving a degree of judgment where intervals between points on the scale are not ‘fixed’ so can’t make mathematical comparison
What are examples of ordinal data?
Strength of attachment and how aggressive someone is
What is the main difference between interval and ordinal data?
Interval is objective and ordinal is subjective
What is the pneumonic for remembering statistical tests?
Carrots Should Come Mashed With Swede Under Roast Potatoes
What is the rule of rs?
If a test name has an r in it, the calculated value should be equal to or more than the critical value
What measure of central tendency should you use with interval data?
The mean
What measure of central tendency should you use with ordinal data?
The median
What are the types of tests?
A one tailed or two tailed test
When do you use a one tailed test?
When there is previous research suggesting a direction
What do the results for a one tailed test need to be to be significant?
In the most extreme 5% at one specified end of distribution
When do you use a two-tailed test?
When there is no previous research or findings are insignificant
What must the results of a two tailed test be to be significant?
In the most extreme 2.5% at one of the 2 ends of a distribution to be significant
What are the types of errors?
Type 1 and type 2
What is a type 1 error?
A false positive when you think there is a difference but there isn’t
What is a type 2 error?
A false negative, thinking there is no difference when there is