# Probability and Statistical significance Flashcards

What are 2 ways studies can screw up?

random error and bias/systematic error

Does random error bias a study?

No. It may be wrong, but the study isn’t biased.

How do you deal with random error?

Statistical Inference.

T/F: a study who’s results are deemed statistically significant is also clinically relevant/meaningful

FALSE. You can have small measures of association be statistically significant, that are too small to matter.

What are 2 values that we can use to estimate how much random variation there is in our study?

confidence intervals and P values.

____ samples have ____ CIs, and _____ samples have ____ CIs

A. large samples, large CIs; small samples small CIs

B. small samples, large CIs, large samples, large CIs

C. small samples, large CIs, large samples, small CIs

C. small samples, large CIs, large samples, small CIs.

What does the CI need to include or not in order for the odds ratio to be statistically significant?

The CI for the OR should not include 1, for the OR to be statistically significant.

Same rules apply to relative risk and prevalence ratios.

What are P-values?

they essentially tell you the same thing as a CI, but CIs also tell you more. P-values only estimate whether a measured association was likely to have been caused by chance.

-no info about where true value is, or size of sample.

What does the P-value need to be, for it to be statistically significant?

It needs to be less than 0.05

T/F: you can prove that something is true.

False. You can only prove that something is not true.

What is the null hypothesis?

The hypothesis of no association…there is no association between exposure and dz

What is the alternative hypothesis?

That there is an association between exposure and dz.

If your P-value is less than 0.05, do you accept or reject the null hypothesis?

reject it. There is an association, so you accept the alternative hypothesis.

What is type I error?

False positive. You reject the null hypothesis when it is not false (so you are saying there is an association when there isn’t)

What is a Type II error?

False negative. You are failing to reject the null hypothesis, when it is actually false.