Statistical Hypothesis Testing (Statistics) Flashcards
Define Statistical Hypothesis
A form of statistical inference for assessing the strength of evidence supporting the null hyporthesis
Null hypothesis is defined prior to collection of data
In research the null hypothesis is that there is no difference between intervention and control groups
Define P value
A theoretical concept
Strength of evidence in support of null hypothesis
A percentage likelihood of obtaining the sample estimate assuming that the null hypothesis is true
0.05 is the level of significance most frequently studied used in clinical studies to reject the null hypothesis
Criticism of the us of P-values
Level of significance is a measure of statistical significance NOT clinical significance
Process of statistical hypothesis testing
Determine null hypothesis and alternative hypothesis. E.g. There is no difference between the control group and intervention - 0
For testing exposures assumes risk ratio of 1.
Set critical region (0.05)
Sample from population
Compare observed to hypothesised value
Probability of an outcome as extreme or more extreme that the one you got given that the hypothesis is true
High value=higher probability
Low value =lower probability
Less than 0.05 sufficient evidence to reject null hypothesis
What is a Type 1 error?
Rejection of null hypothesis when null hypothesis is true in population
Probability of this determines the power of the study
Type 2 error?
Failure to reject null hypothesis when null hypothesis is false in population
Determines power of the study
Which error is worse?
Dependent on cost, side effects, alternatives
Type 1 error may conclude effectiveness of an ineffective treatment. May have side effects, be costly, or cause patients to avoid other more effective treatments
Type 2 error may have put patients through a trial for nothing, deprive patients of an effective treatment
