Please enable JavaScript.
Coggle requires JavaScript to display documents.
CHOOSING A STATISTICAL TEST - Coggle Diagram
CHOOSING A STATISTICAL TEST
Tests are used to determine whether a significant difference or correlation exists
This lets you know whether you accept the null or experimental hypothesis
3 STEPS
Is the experiment looking for a difference or a correlation?
What is the experimental design?
What is the level of measurement?
STEP 1:
Is the experiment looking for a difference or a correlation?
This should be clear from the hypothesis/methodology of the experiment
STEP 2:
Independent groups: participants complete one condition
Repeated measures/matched pairs: participants complete both conditions (these two designs are considered the same when choosing a statistical test)
What is the experimental design?
STEP 3:
Level of measurement
Ordinal
Nominal
Interval
Ordinal data
Data can be ordered/ranked
Can count and order data
House number, swimming times
Scale of 1(dislike) - 5 (like)
Nominal data
Sex, eye colour
Male, female
Can count but can't order the data
Usually use a code e.g. Male = 0, female - 1
Occurs in frequencies
Interval data
units of equal measurements (a scale with equal intervals) are used e.g. minutes, kilograms, number of words recalled in a memory test or percentage score in an exam
Most precise
Based on numerical scales e.g. time, temperature, weighing scales
Tests you must remember:
Wilcoxon
Spearman's rho
Mann Whitney
Unrelated t-test
Sign test
Related t-test
Chi-squared
Pearson's R