DATA ANALYSIS 2 Flashcards
T test is used to…
Compare two population means
Can a t-test be used for either parametric or non-parametric data
The t-test (unpaired or paired) is only appropriate for analysing parametric data because the method by
which the p-value is calculated assumes the data adheres to a normal distribution.
For non-parametric data, that relationship
doesn’t exist, therefore any test that assumes the relationship exists will generate an inaccurate p-value.
Why can’t we just conduct multiple t-tests?
Performing multiple comparisons within a single study/analysis
increases the possibility of false positives occurring (“a type 1 error”).
Anova
Analysis of variance
The ANOVA generates a p-value that is compared to an alpha value (e.g. 0.05) determined prior to the study (similar to a t-test).
One way anova
Comparing one thing with multiple things
studies investigating the effect of 3 or more varying
conditions (i.e. 3 or more groups) on a single continuous variable (e.g. the effect of varying
treatments on blood pressure)
effectively test the null hypothesis that there are no statistically significant differences between any of the groups.
Two way anova
Comparing two things with multiple things
where studies also investigate whether there is an interaction
between two categorical variables on a single continuous variable (e.g. whether there is an
interaction between the sex of patients and the effect of varying treatments on blood
pressure).
Two types of 2 way ANOVA
With replication
Without replication
One way anova between groups
Tests to see if there is a difference between two groups
Like a T-test
Two way anova without replication
Used if you have one group that you are double testing
Testing the same set of individuals before and after treatment
Two way anova with replication
The 2 groups and their members are doing more than one thing
- two different hospitals and 2 different therapies
Limitations of the one way anova
Highlights that there is a difference between the groups
But it won’t tell you which groups there are differences between
How do you resolve for a false positive from multiple comparisons
a correction is made that reduces the alpha (or increases the p-value) in proportion to the number of comparisons being made, hence compensating for the otherwise increased probability of type-1 errors occurring.
Bonferroni correction
• Bonferroni correction is used when the researcher wishes to select specific pairwise comparisons that do not a have particular pattern to them.
Dunnett’s test
• Dunnett’s test is used when there are multiple groups to be compared to a control group
Tukey’s test
• Tukey’s test is used when a study requires pairwise comparison of every possible combination of groups.
