Open SPSS. Paste the syntax below into a new syntax window.
run it.
See if this does what you want.
* this is the data from your spreadsheet.
* there are other ways to do this but this way should be clear.
*first create a table of cuts.
new file.
data list list/gender (a4) AGE (f3) cutforpctile5
cutforpctile85 cutforpctile95(3f4.1).
begin data.
girl 5 13.1 17 18.3
girl 6 13.1 17.2 18.8
girl 7 13.1 17.6 19.4
girl 8 13.3 18.1 20.2
girl 9 13.6 18.8 21.1
girl 10 13.9 19.5 22.1
girl 11 14.4 20.4 23.2
girl 12 14.9 21.4 24.3
girl 13 15.5 22.4 25.4
girl 14 16 23.3 26.3
girl 15 16.5 24 26.9
girl 16 16.8 24.5 27.4
girl 17 17 24.8 27.6
girl 18 17.1 25 27.8
boy 5 13.4 16.7 17.8
boy 6 13.4 16.9 18.2
boy 7 13.5 17.3 18.8
boy 8 13.7 17.7 19.4
boy 9 13.9 18.3 20.2
boy 10 14.1 18.9 21
boy 11 14.5 19.6 22
boy 12 14.9 20.5 23.1
boy 13 15.4 21.4 24.1
boy 14 16 22.4 25.1
boy 15 16.5 23.2 26.1
boy 16 17.1 24 26.8
boy 17 17.5 24.7 27.5
boy 18 17.9 25.3 28
end data.
dataset name cuttable.
* make up some child data.
data list list /id (f2) gender (a4) age(f2) rawbmi (f4.1).
begin data
1 girl 5 10
2 girl 5 15
3 girl 5 18
4 girl 5 20
5 boy 5 10
6 boy 5 15
7 boy 5 17
8 boy 5 20
9 boy 3 15
10 girl 5 x
end data.
dataset name children.
*make sure files are in correct order.
dataset activate cuttable.
sort cases by gender age.
dataset activate children.
sort cases by gender age.
match files file=children /table=cuttable
/by gender age.
dataset name combined.
*check the logical operators to see if they do what you want when a raw
bmi is equal to the cut.
numeric bmi_category (f1).
do if missing(rawbmi).
compute bmi_category = 0.
else if rawbmi lt cutforpctile5.
compute bmi_category = 1.
else if rawbmi ge cutforpctile5 and rawbmi lt cutforpctile85.
compute bmi_category = 2.
else if rawbmi ge cutforpctile85 and rawbmi lt cutforpctile95.
compute bmi_category = 3.
else if rawbmi ge cutforpctile95.
compute bmi_category = 4.
ELSE .
compute bmi_category = 5.
end if.
missing values bmi_category (0,5).
value labels bmi_category
0 'missing bmi'
1 'underweight'
2 'normal'
3 'overweight'
4 'obese'
5 'oops check data and logic'.
list.
Art Kendall
Social Research Consultants
Jennifer Ryan wrote:
Art,
Thanks for your reply! Sorry I didn't get back to you earlier -- this
is not my preferred email address.
I am using the standardized age and sex-specific percentiles from 2007
WHO standards. I have attached the percentile sheets I am referencing -
one for girls and one for boys. I am also attaching an excel
spreadsheet I made, based on these two percentile sheets. As you will
see, the ranges up to and between the 5th, 85th, and 95th percentiles
are the significant cut-offs for my study. So, for example, for Girl
Age 5, I took percentile numbers for 5y/1m, 5y/6m, and 5y/11m at 5th,
85th, and 95th respectively. Using my excel sheet for the cut-offs for
SPSS, I figured that would be a reasonable solution for accurate
numbers.
I hope to hear from you again.
Thank you,
Jennifer
On Thu, Nov 5, 2009 at 1:00 PM, Art Kendall
<Art@...>
wrote:
Please
tell us more about the data you have available. Do you need to get the
percentiles from your own data or from pre-existing tables?
Art Kendall
Social Research Consultants
Jennifer Ryan wrote:
Hi,
I'm a grad student and am working with SPSS for the first time for a
sociology paper, which is about inequalities and obesity in China. I
have learned some basics, but am running into a problem trying to
create weight ranges and categories for children of different ages.
Basically, I have the data on the child weight and height, which is
what you need to calculate BMI. However, calculating BMI for children
is a bit tricky because child BMI is age and sex-specific, meaning
weight categories are defined by where the child falls in the
percentile for his or her age and gender. (For example, over 95th
percentile is considered obese).
I need to know how to create ranges for each gender at each age (5-18
yrs.), so I can then place them into the right weight categories
(underweight, healthy weight, overweight, obese).
PLEASE HELP!! I would be forever grateful!!
Jennifer
=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
