Study for class test Flashcards
One continuous variable, compare its mean to a pre-selected value
One sample t-test
One sample t-test assumptions
Normally distributed continuous variable
One sample t-test H0
Mean of variable is equal to the population (or pre selected value)
2 measurements of the same continuous variable at different time points
Paired samples t-test
Paired samples t-test H0
Mean of variable at follow up is the same as the mean at baseline in population
Paired samples t-test assumptions
Difference in the variable between baseline and follow up should be normally distributed
Assumptions not met for paired samples t-test
Wilcoxon test
One continuous variable and a categorical variable with more than 2 groups, to compare mean of continuous variable between the groups
ANOVA
ANOVA H0
Mean of variable is the same across all 3 groups
ANOVA assumptions
Independent groups
Normal distribution
Equal variance of continuous variable between groups - use Levenes test (p>0.05 for variance)
ANOVA show which groups differ
Post-hoc test -> Scheffe test
Assumptions for ANOVA not met
Kruskal-Wallis test
Compare continuous variable between 2 binary variables
Independent samples t-test
Independent samples t-test assumptions
Independent groups
Normal distribution
Equal variance between groups - Levenes test
Assumptions not met independent samples t-test
Mann-Whitney test
Independent samples t-test H0
Mean continuous variable in one group is the same as the mean in other group
Krustak-Wallis instead of
ANOVA
Man-Whitney instead of
Independent samples t-test
Wilcoxone instead of
Paired t-test
When to use non-parametric tests for continous variable
- assumptions for parametric counterpart not met
- OR data is ordinal (ranked)
- OR small sample size
What do non-parametric tests conclude
Compares the distribution of the variable between groups by using ranks
H0 non parametric tests
Distribution of continuous variable between groups is the same
Chi squared assumptions
No expected cell counts <1
At least 80% of expected cell counts >5
What does chi squared look at
Proportion of people in each group
2 independent categorical (nominal) variables and assumptions met
Pearsons chi squared test
2 independent categorical (ordinal) variables and assumptions met
Chi squared test for trend
Categorical variables - 2x2 table, assumptions met
Chi squared test with continuity correction
Categorical variable - 2x2 table, assumptions not met
Fisher test
Categorical variable - 2x2 table, variables are paired (not independent)
McNemar test
Check if observed frequencies within each cell is what is expected (categorical data)
Chi squared goodness of fit test
