Lecture 3 Comparing groups Flashcards
what is inference?
Process of making generalisation about a population from a sample.
What is a hypothesis?
statement about the value of a population parameter or association
logically easier to disprove rather than prove a hypothesis
first step in hypothesis testing is to define the null hypothesis
what is the null hypothesis?
often expresses the idea of ‘no difference’ or ‘no association’
The symbol used to denote a null hypothesis is
H0
No difference between groups we are interested in – easier to prove this is incorrect rather than correct
what is the alternative hypothesis?
opposite of null hypothesis
expresses the idea of ‘some difference’ or ‘some association’
Symbol used is H1 (or Ha in some books)
A parametric test relies upon the assumption that the data you want to test is (or approximately is) normally distributed
Hypothesis tests (Parametric):
what is a Independent t-test?
compares the mean value of a continuous variable between two independent groups
A parametric test relies upon the assumption that the data you want to test is (or approximately is) normally distributed
Hypothesis tests (Parametric):
what is a •One-way Analysis of variance (ANOVA)?
compares the mean value of a continuous variable between more than two independent groups
A parametric test relies upon the assumption that the data you want to test is (or approximately is) normally distributed
Hypothesis tests (Parametric):
what is a Paired t-test?
compares the mean value of a continuous variable between two related groups
These three make no distributional assumptions (hence non-parametric)
Hypothesis tests (Non-parametric):
what is a Mann Whitney test?
compares the distribution of a continuous/ordinal variable between two independent groups
Hypothesis tests (Non-parametric):
what is a Kruskal Wallis test?
compares the distribution of a continuous/ordinal variable between more than two independent groups
Hypothesis tests (Non-parametric):
what is a Wilcoxon signed rank test?
compares the distribution of a continuous/ordinal variable between two related groups
what is a p-value?
probability of seeing the observed value or more extreme observed values if the null hypothesis is true
p < 0.05 is regarded as statistically significant
A significance level of 5% is the risk we take in rejecting the null hypothesis, in favour of the alternative, when in fact the null hypothesis is true (Type I error)
