Interpreting Associations - Chance

Question 1

Your measure of association indicates a causal relationship - what else do you need to do (besides satisfying Bradford-Hill guidelines) before inferring causality?

Answer 1

1. Critically appraise the evidence available for the relationship
2. Eliminate other potential reasons for measuring an association

IE - rule out CHANCE, BIAS, CONFOUNDING in your calculation.

Question 2

What is chance?

Definition & explain relationship between P-Value and chance

Answer 2

Chance is a sampling variation caused by random error.

Rule this out by stats - P Value will tell you what probability that a measure from a sample occured by chance (and does not exist in the sample population the measures were taken from)

P- Value = 0.05 = 5% probability that the observation occurred by chance.
Anything below 0/05 indicated observation is not due to chance.

Question 3

What happens if you calculate a P-value of 0.04 or 0.06 (ie borderline)

Answer 3

This needs further investigation and should be stated in the findings that the measurements were borderline stat significant.

Question 4

How can we reduce chance?

Answer 4

By increasing the sample size

Question 5

Explain the relationship between chance and confidence intervals?

Answer 5

CI tells us how confident we are that the true population mean lies between A&B.

The narrower the confidence interval, the more precise the measure.

If we are 95% sure the pop value lies between A&B, there is a 5% prob that the value would be observed outside of that range and therfore 5% chance there is no association.

Question 6

What does it mean when 2 95% CI’s do not overlap?

Answer 6

There is a statisitically significant difference at the P=0.05 level