Statistics Flashcards Preview

STEP > Statistics > Flashcards

Flashcards in Statistics Deck (14)
Loading flashcards...

Effect modification

o Definition = Effect of an exposure on an outcome is modified by another variable. NOT a bias.
o Example: A study showing a new drug increases the risk of DVT in treated women who smoke compared to untreated women who smoke, with an RR of 1.7. In nonsmokers, no increased risk of DVT is evident with use of the drug.
o Other notes: You can resolve this problem with STRATIFICATION. Ie in an analysis of primary school students, age can be a confounder that muddles the association between shoe size and intelligence. Children with bigger shoe sizes may appear to be more intelligent on initial analysis. When older and younger children are stratified, the association between shoe size and intelligence disappears.


Rare disease bias effect

odds ratio approximates RR


Determining significance of relative risk...

If 1.0 falls within the presented interval, then the relative risk is not significant. If 1.0 does not fall within the interval, then the presented relative risk is significant.


How do you calculate a type II error?

Type 2 error = 1 - Power


Example of a type II error

Indicating a drug does not work when it really does.


Type 1 (alpha error) statistical definition

Null hypothesis is rejected when it is really true.


Type 2 (beta error) statistical definition

You fail to reject the null hypothesis when it is really false. (indicating a drug doesn't work when it really does)


How do you determine if null hypothesis is rejected?

If the p value is greater than the alpha criterion (usually 0.05), null hypothesis is not rejected. If it's less than 0.05, the null hypothesis is rejected and statistical significance is proven.


state confidence interval equation



How to approach problems about changing threshold...

Think about how test will work in new population at large.


Affect of lowering a threshold...

*Increase NPV.
*Increase sensitivity.


Case series design

Collects detailed information about people who are all believed to have the same disease or condition. Intended to describe common elements defining the disease.


Cross sectional study assesses...



statistics tool for cross-sectional study

Chi-square analysis