Statistical tests Flashcards
Why do we use statistical tests?
To test if results obtained are significant
What are the two types of test and when are they used?
Parametric- When data is normally distributed
Non parametric- When data is non evenly distributed
What do we do if results are significant?
Reject null hypothesis and accept alternative hypothesis
What is the criteria for parametric tests?
Normally distributed data
Homogeneity of variance
No extreme outliers
Data from participants should not be correlated
What are the types of parametric tests?
Correlation- measures the relationship between two variables
Paired t-test- compares mean value of two variables obtained from same ppts
Unpaired t-test- Compares mean value of a variable measured from two different groups
ANOVA- analyses differences between means of more than two groups; one way uses one IV, two way uses two IVs
How do you calculate the sign test?
1) Work out the difference between the two sets of data
2) Add up how many are positive and how many are negative, the S-value is the less frequent signs, so if there are 2 positive and 5 negative, the S-value is 2
3)Find the N value (the total number where the difference isn’t 0)
4)Use the N value and significance level to find the critical value
5)If the s-value is equal or less than the critical value then results are significant
What are the two types of errors?
Type 1- rejecting the null hypothesis when its true, i.e researching suggesting significance when actually insignificant
Type 2- Mistakenly accepting null hypothesis when it should be rejected
When should you use the Pearson’s r?
Parametric
Correlational data
When should you use t-tests?
Parametric
Non correlational
When should you use Chi-squared?
Non parametric
Nominal data
Independent groups design used
When should you use sign test?
Non parametric
Nominal data
Repeated measures
When should you use Spearman’s Rho?
Non parametric
Not nominal
Correlational
When should you use Mann Whitney?
Non parametric
Not nominal
Not correlational
Independent groups
When should you use Wilcoxon?
Non parametric
Not nominal
Not correlational
Repeated measures
