Non Parametric Methods Flashcards
What is the usage for a Mann-Whitney U test ?
To compare two independent samples when outcome is interval scale but assumptions for t-test (Normality and equality of variance) are not satisfied or when outcome is ordinal
What is the Mann-Whitney test ?
Mann-Whitney U test is a non-parametric test, meaning it does not assume a specific distribution (like the normal distribution) for the outcome variable. It is often used to compare differences between two independent groups when the dependent variable is ordinal or continuous, but not necessarily normally distributed
How do you complete a Mann-Whitney U test on SPSS ?
Analyse>Nonparametric tests> Legacy dialogs>2 independent samples then enter the interval scale variable as the Test Variable and Treat as the Grouping variable
What is the usage of a Wilcoxon Signed Rank Test ?
To compare two paired samples when outcome is interval scale but assumptions for paired samples t-test (i.e. Normality of within pair difference) is not satisfied or when outcome is ordinal
How do you conduct a Wilcoxon Signed Rank test in SPSS ?
Use Analyse>Nonparametric tests> Legacy dialogs>2 related samples
Enter the data so there are 2 interval scale variables
What is the correlation coefficient ?
Used to measure the the degree of association between two interval scale variables (but assumptions for Pearson’s correlation coefficient are not met), or when at least one of the variables is ordinal
Best assessed using a scatter diagram to check association
How do you conduct a Spearman’s rank correlation coefficient test in SPSS ?
Use Analyse>Correlate>Bivariate
To obtain a scatter diagram use Graphs>Scatter, enter control variable on y axis and dependent variable on x
What are the disadvantages with non-parametric test ?
Tests typically have less power than their parametric counterparts if the relevant assumptions hold
They focus on hypothesis testing (deriving a P value) rather than on estimation (obtaining a confidence interval)
They are less flexible (they cannot be subsumed into more complex forms of analysis in the way that is possible for parametric methods
When and Why to Use Non-parametric Methods ?
Sensitive to non-normality when sample sizes are small
less affected by extreme values
handle ordinal scales better
What are the non-parametric methods for parametric methods relations ?
Independent samples t-test = use Mann-Whitney U test
Paired samples t-test = use Wilcoxon Signed-Rank Test
One-way analysis of variance = use Kruskal-Wallis one-way analysis of variance
Pearson’s correlation coefficient = use Spearman’s rank correlation coefficient
What do you look for on SPSS when completing Mann-Whitney U test ?
The relevant P-value is shown under Asymp. Sig. (2-tailed)
What are the scale values and their meaning in a correlation coefficient ?
-1 = perfect inverse relationship
0 = the two variables are independent
1 = perfect direct relationship
What is the usage for Spearman’s Rank Correlation Coefficient ?
To measure the degree of association between two interval scale variables but assumptions for Pearson’s correlation coefficient are not met (the joint distribution of the variables is bivariate Normal – i.e. both variables must be Normally distributed) or when at least one of the variables is ordinal
