Dear Allstaters,
I have a query about assessing the extent to which a model explains the
data. I have a dataset of women giving birth in hospitals. My outcome is
binary (delivery by caesarean section or not) and I am doing a logistic
regression clustered on hospital using Stata 7.0. I have performed two
goodness of fit tests (Pearson's and Hosmer-Lemeshow) but they only give me
a chi-squared statistic and p-value. I want to know to what extent my model
explains the data (ideally an r-squared equivalent) and how good or bad the
fit is. My background is in Epidemiology which means I have studied some
statistics but I am not a statistician. Can anyone advise me on how to
assess the proportion of the data variation explained by my model.
Many thanks
Bernadette
----------------------------------------------------------------
Bernadette Alves
Health Services Researcher
Department of Primary Care and General Practice
ICSM Charing Cross Campus
The Reynolds Building
St Dunstan's Road
London W6 8RP
|