two way tables
paradox
- variables may alter the results of your distribution - adding more variables can give you different results in 2 way table
separate (marginal) distribution
- 1 variable at a time - totals
joint "and" distribution
- overall percentage in each percentage - looking at two variables at a time - "probability of girl and brown hair"
conditional distribution
- use to find relationships
if conditionals are same/close
they are independent and have no relationship
two way tables examine
categorical data - 2 variables at a time
in a table what kind of distribution do you have?
joint and marginal ONLY
for percentage change, where do you look?
look at subgroups NOT marginals
simpsons paradox
look for third variable that may change results ( but this doesnt always happen)
if conditionals are different
there is a relationship, x & y are dependent