Lec 78: Evidence-Appraisal: Hypothesis testing Flashcards
What are the paired study designs?
Pre-post comparison study
Cross-over study
Matched case-control study
Twin study
What are the unpaired study designs?
Traditional randomized control trial
Epidemiology studies without matching
Describe pre-post comparison study
- Single intervention
* Each pt observed before and after intervention (same person measured twice)
Describe cross-over study
- Each person receives sequence of interventions
- Pt randomized to sequence of interventions
- Advantages: good for short and/or acute treatments and treatments measuring palliation not cure
- Limitations: not feasible for certain outcomes/treatments (i.e. deaths)
Describe matched case-control study
Identify study participants then match with similar control participants
Describe twin study
Can be used to distinguish between genetics and environment
Provide fundamental concepts of hypothesis testing
Hypothesis tests: Estimate the probability of obtaining the observed result, or one more extreme, if there is no true difference. Use info to make decisions about the population
- Process for choosing between competing hypotheses
- One-sided test: greater than or less than for null and alternative hypotheses
- Two-sided test: equal and does not equal for null and alternative hypotheses
- Compare P-value (determined from test statistic) to alpha to determine statistical significance
•General form
Test statistic = (observed value - hypothesized value) / Std Error of observed value
Provide fundamental concepts of confidence intervals
Confidence intervals: Estimate a range of values that is likely to include the true (population) value
•Provides range of plausible values
•General form
CI = estimate plus/minus critical value X standard error
How are clinical and statistical significance different
Statistical significance provides the answer to a hypothesis test which removes chance from a study
Clinical significance considers the statistical significance and other factors in order to determine if a study is relevant