unit 6 pt 1

# unit 6 pt 1 - Unit 6 Overview Evaluating Valid Statistical...

This preview shows pages 1–3. Sign up to view the full content.

Evaluating Valid Statistical Associations: Statistical Associations: Chance Dr Borenstein Dr. Borenstein NO AUDIO ON THIS SLIDE. CLICK NEXT TO CONTINUE >> Unit 6: Overview ± Precision/validity of an association ± Internal vs. external validity ± Generalizability ± Chance: population vs. samples ± Hypothesis testing ( p -values to reject or accept Ho) ± Estimation: Confidence intervals (95%, 99%) ± Computation and interpretation of a 95% confidence interval ± Statistical power: review of Type I and Type II errors ± Four features to power ± Example of a sample size and a power calculation Od l ± Overpowered samples Unit 6 Evaluating valid statistical associations: Random error: Precision and ± Chance measurement error (non-differential misclassification – Unit 7) Internal validity ± Bias – Unit 7 Confounding Unit 8 ± Confounding – Unit 8 Testing an epidemiologic hypothesis involves consideration of the concept of an association ± An association refers to: ² Relation between an exposure and an outcome ² Statistical dependence between two variables (e.g., correlation) ² Degree to which the rate of the outcome in persons with a specific exposure is higher or lower than the rate of the outcome among persons without the exposure ² Degree to which the odds of exposure in persons with D are higher than the odds of exposure in persons without D If we observe an association, we If we observe an association, we must consider: 1. Is the association precise ? (Could the study findings be due to chance alone?) (Unit 6 findings be due to chance alone?) (Unit 6) 2. Is the association valid ? (Do the study findings reflect the true relation between E and D?) (Units 6-8) 3. Is the association causal ? (Is there sufficient evidence to infer that a causal association exists between E and D?) (Unit 10) In any epidemiologic study, there are at least four possible explanations for the observed results: ± Chance ± Bias ± Confounding (a type of bias, since its presence distorts the true E-D association) ± The findings are true The findings are true. The first three explanations are not mutually exclusive th b t i th t d -- more than one can be present in the same study (and usually are)

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
A “precise and valid” statistical A precise and valid statistical association implies “internal validity” ± Definition of validity: the extent to which the observation measures the true phenomenon in nature Two kinds nature. Two kinds: ² Internal validity : The results from the study measures the phenomenon we are trying to measure in the study sample . Are the study results valid = unbiased? ² External validity: The results from the study measure the phenomenon we are trying to measure in the world - outside the study sample - to all other populations ± Generalizability. E.g., Nun Study: applicable to other women?
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

### Page1 / 9

unit 6 pt 1 - Unit 6 Overview Evaluating Valid Statistical...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online