Phil,
Stata will assign a high value to a variable with a missing value.
You should decide whether you will use some form of imputation to
replace the missing value (mean, mean + random error, median, or none).
After possible replacement of some missing values,
you might issue a modified command:
generate BMI = weight1998/(height1998^2) if weight1998 < .
Regards,
RAY
Robert A. Yaffee, Ph.D.
Research Professor
Shirley M. Ehrenkranz
School of Social Work
New York University
home address:
Apt 19-W
2100 Linwood Ave.
Fort Lee, NJ
07024-3171
Phone: 201-242-3824
Fax: 201-242-3825
[email protected]
----- Original Message -----
From: Phil Schumm <[email protected]>
Date: Thursday, May 11, 2006 8:40 pm
Subject: Re: st: A quick question
> On May 11, 2006, at 5:34 PM, Albert Newton wrote:
> > I have two varibles: weight and height of respondents in 1989.
> >
> > weight1998 float %9.0g
> > height1998 float %9.0g
> >
> > I want to create a new variable:
> >
> > gen BMI = weight1998/(height1998^2)
> >
> > My little problem is that: BMI takes strange values if
> weight1998
> > is missing. Could anyone tell me why? Thanks a lot!
> >
> > height1998 weight1998 BMI1989
> > 1.5748 290 53.04115
> > 1.4732 256 53.50351
> > 1.5494 285 53.84973
> > 1.5494 306 57.8176
> > 1.6764 375 60.52597
> > 1.905 . 124.4902
> > 1.778 . 142.9097
> > 1.778 . 142.9097
> > 1.6764 . 160.757
> > 1.6764 . 160.757
> > 1.6764 . 160.757
>
>
> It is clear that the variable you are listing (BMI1989) was not
> calculated using the expression above. For example, if we take the
>
> first observation:
>
>
> . di 290/(1.5748^2)
> 116.93572
>
>
> which is not equal to 53.04115. We can help you a lot better if
> you
> include part of the actual log. For example:
>
>
> . des
>
> Contains data
> obs: 11
> vars: 2
> size: 110 (99.9% of memory free)
> --------------------------------------------------------------------
> ----
> -------
> storage display value
> variable name type format label variable label
> --------------------------------------------------------------------
> ----
> -------
> weight1998 int %8.0g
> height1998 float %9.0g
> --------------------------------------------------------------------
> ----
> -------
> Sorted by:
> Note: dataset has changed since last saved
>
> . gen BMI = weight1998/(height1998^2)
> (6 missing values generated)
>
> . li
>
> +--------------------------------+
> | wei~1998 hei~1998 BMI |
> |--------------------------------|
> 1. | 290 1.5748 116.9357 |
> 2. | 256 1.4732 117.9551 |
> 3. | 285 1.5494 118.7183 |
> 4. | 306 1.5494 127.466 |
> 5. | 375 1.6764 133.4369 |
> |--------------------------------|
> 6. | . 1.905 . |
> 7. | . 1.778 . |
> 8. | . 1.778 . |
> 9. | . 1.6764 . |
> 10. | . 1.6764 . |
> |--------------------------------|
> 11. | . 1.6764 . |
> +--------------------------------+
>
>
> BTW, BMI is typically calculated as:
>
> BMI = weight (kg) / height (m^2)
>
> Your heights look about right (for meters), but your weights are
> off
> by about a factor of 4.
>
>
> -- Phil
>
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
