Skip navigation

Comparing proportions

Learning outcomes

On watching this video, students should be able to

  1. Interpret measures of association or effect measure for proportions, specifically, risk differences, risk ratios and odds ratios and understand their null value (the value if there is no association).
  2. Explain the difference between relative measures of effect such as odds ratios and risk ratios and absolute measures of effect such as risk differences and identify situations when we prefer to report each or both. 
  3. State that an approximate 95% confidence interval (based on normal theory) for measures of effects for proportions may include nonsensical negative values in its range and that this is an indication that the normal model wasn't a good fit to the data. In these situations we need to use another method to calculate a confidence interval. 
  4. Recall that these methods rely on assumptions and understand that different approaches should be investigated when these assumptions are not met.


04:38: I should say “pi1-pi0” NOT “pi1-pi2”

16:36: I say “the Odds ratio is always a bigger number”, this is wrong, what I mean is that the OR is always indicating a stronger effect, so if the RR is 0.4 say, then the OR will be less than 0.4