Categorical data Flashcards
What’s Pearsons’ chi for?
one categorical predictor, is the association by chance? compares observed and expected values
Pearsons’ chi formula
Sum of (observed - expected)2 / expected
Expected value formula
(row total x column total) / N
How many ppts needed (PC)
Each category must have 5 ppts, in large samples 20% can be over 5 but all must be over 1
How to report Pearsons’ chi
X2 = (1, N= ??) = ??
Correction of continuity or Yate’s purpose?
does hypothesised value differ from proportion of population?
When Correction of continuity or Yate’s not used?
More than 2 outcomes
When is liklihood ratio used
When samples are small
Maximum liklihood theory
A model that maximises the proabability of obtaining the observed set of data
Correction of continuity or Yate’s formula
Sum of natural log (observed/model)
Calculate standardised residuals
(observed-model) / squrt of model
Loglinear analysis, what is backward elimination method
Removes a term then compares to previous model in order to find the simpliest model
Goodness of fit sig or non sig
Want to be non-significant so term can be left out, no significant different between the two
