Why do it with categories? Do you have both height and weight? Why not
use something like Body Mass Index? It is a continuous measure. You
could coarsen it later if you still wished. Do you have definitions of
what is "underweight", "normal" "overweight"? If you google "bmi" and
"standard" you should find info on its limitations. However, just using
3 or 4 categories (yours + "obese") still runs into a problem of
thresholds. One standard is that men are "overweight" with 25% body
fat or more. Suppose that someone is 24.9% at time 1, 25.0% at time 2,
24.9% at time 3, 25.0% at time 4, etc. The categories would change,
but the continuous measure would should constancy.
In SPSS if you have the weight in kilograms and height in centimeters
the syntax would look something like this.
compute bmi = wgt_kg / ((heightcm/100)**2) .
Art
[log in to unmask]
Social Research Consultants
University Park, MD USA
(301) 864-5570
manon girard wrote:
>Hi (and sorry for cross-posting),
>
>We would like to analyze the change children' weight over time, first by looking at the continuous variable (i.e. weight in kilograms) and then by 3-category variable which is underweight, normal and overweight.
>
>We have 5 measurement over time (i.e. mother was asked to tell the weight ans height her child when he was 5, 18, 30, 48 and 60 months).
>
>I think we can model the weight (continuous variable) with latent growth trajectory. I don't think it is possible with the 3-category variable.
>
>Is there someone who did the same thing ? If yes, can you provide me with SAS/SPSS commands ?
>
>Furthermore, we would like to analyzed what variables (socioeconomic, nutrition, etc) contributed to the change (i.e. being normal to overweight ?). Is it possible to do this ?
>
>Thanks
>Manon Girard
>
>
>
|