Good afternoon,
We have asked mothers to evaluated their child's height & weight.
We then measured the children.
We classified the children according to BMI USDA CDC-growth curve as being
>=95th percentile or not (obese or not).
We have now an indicator of perceived body weight and one for the measured
(or actual) body weight in which 0=not obese and 1=obese.
We crossed them and we can computed several statistics. Misclassification
can be computed as being "actual obese but not perceived as it".
We are interrested to see which socioeconomic factors that contribute the
most on the misclassification. Which is the best statistical technique to
describe this ?
Also we would like to put all the factors in a "multivariate" analysis ?
Again which stats is the most appropriate ?
Thanks
Manon Girard
|