UTF-8 U+6211 U+662F wrote:Breast Cancer and Calcium Tablets:
Cases Controls
Exposed 70 25
Unexposed 30 75
Odds of exposure of cases = 70/30 = 2.3
Odds of exposure of non-cases = 25/75 = .33
Odds Ratio = Odds of exposure of cases/Odds of exposure of non-cases= 2.3/.33 = 6.97
The odds of exposure to calcium tablets of cases-patients was 6.97 times those of controls, therefore, case-patients were 5.97 times more likely to have used calcium tablets than controls (not sure how accurate that is).
Odds ratio = OR = ad/bc, calculates correlation, not direct causation (I'm not trying to criticize anyone here).
Relative risk = RR = AR for exposed/AR for unexposed = [a/(a+b)]/[c/(c+d)], measures the relative risk of a specific exposure
Hope that helped!
Exactly right on the odds ratio, but another caution: you can't go from odds ratio to "times more likely" as easily as you can with risk ratio. Remember that if you do odds ratio on the roll of two dice, where one or two count as hits for the first but only one is a hit for the second, your odds ratio will actually be 2.5 (2:4 over 1:5). In reality, though, the event is only twice as likely, not 2.5 times as likely.