lecture 7 - binary logistic regression Flashcards
what type of variable is y when using OLS and linear regression models?
continuous
when is binary logistic regression needed?
when the outcome variable is binary
what does the binary logistic model predict?
the probability of getting y = 1
what shape graph does the binary logistic model produce?
s shaped
what are the key distinctions between OLS regression and binary logistic regression?
with OLS y can have any value
with BLR y can only fall between 0 and 1 as it predicts probability
what are the steps in BLR including odds ratios
- estimate probability for the case given Xs
- convert the probability to odds (p/(1-p))
- compare the odds of the 2 cases to get odds ratio
how do you work out odds ratio?
odds after one unit increases in X / original odds
*the original odds is the category = 0
odds ratio indicates a change in odds resulting from…?
a unit change in the predictor
what do odds ratios centre on?
1
- > 1 = positive association
- <1 = negative association
odds ratio is also referred to as?
exp (b)
rather than obtaining odds ration you can?
convert odds to log odds which will make it linear
* not recommended as odds ratio are easier to interpret