What is the 95% confidence interval?
The range of valves you can be 95% sure the the mean of the population lies within
When is the 95% confidence interval larger?
when more variety in population values and smaller samples
why is 95% confidence level used?
to estimate precision of observed values and determine statistical significance of a difference between two differnt observations
Describe how data should be set up in a case control study analysis
- top row is expose ppl bottom row is of unexposed ppl
- first column is the cases (number of outcomes) of exposed and unexposed
- 2nd column is of number of controls (no outcome)
How is odds ratio calculated?
odds of outcome in exposed ppl / odds of outcome in unexposed
Can also be calculated by: a X d/ c X b or (a/b)/(c/d)
how is error factor calculated?
1) 1/a+ 1/b+ 1/c +1/d = ans
2) square of ans, X 2= ans2
3) e^ans2= error factor
How are confidence limits calculated from error factor and odds ratio?
OR/ error factor is lower limit
OR X error factor is upper limit
What is the null hypothesis?
that there is no significant difference between the two variables and so the odds ratio is one
1 lies within the confidence interval, is the null hypothesis accepted or rejected?
null hypothesis is accepted as no significant difference
Can you ever fully reject a null hypothesis?
no as there is always a possibility it occurred by chance, but 5% chance is acceptable for significance of differences
Is odds ratio relative or absolute risk?
relative risk- it is only comparing risk as compared to the a different population, not compared to the whole population
How is incidence rate calculated?
of new events/ (# of persons x # of years)
What is tendency?
the true value (observation is out best estimate of the tendency)
Why are case control studies good?
- theyre quick
- theyre cheap
- theyre good for rare diseases as you can select based on outcome
What issues are there with case control studies?
- selection bias (need to select control from similar population to the cases)
- information bias (make sure you do the same to both groups)
- confounding
how can confounding be reduced in case control studies?
match cases with similar people who dont have disease (controls)
adjust for it in analysis by standardisation
What are disadvantages of cohort studies?
- big scale
- expensive
- take long time
- survivor bias (high number of losses to follow up, healthy ppl tend to drop out)
- not good for rare outcomes
- unknown confounders cannot be accounted for
What ratios can be used to describe relative risk?
- risk ratio (prevelance proportions at particular time (100 per thousand))
- rate ratio (incidence rates after a particular time (50 per 1,000 person years))
- odds ratio (odds between groups at end of time period (#dead: #alive)
What measurements of relative risk can be applied to case control and cohort studies?
cohort: risk ratio, indcience rate ratio, odds ratio
case- control: odds ratio
How would you describe the relative risk if the odds ratio for getting cancer if you smoke vs if you dont smoke is 3?
you are 3 times as likely to get cancer if you smoke than if you didnt
