Hello I am doing a research looking at maternal perception of the child's weight. I downloaded the macro of the WHO to obtain the BMI z-scores of the child however, the majority of the studies I've read use the percentiles instead. I know I can convert the z-scores to percentiles if the sample has normal distribution which it has, does anyone has converted the z-scores to percentiles? How do you do it? I am using STATA and SPSSS.
My answer is not based on any Software. We usually denote by z-alpha the z-score that cuts off the area (or probability) alpha to the right tail of the z-distribution. The z-scores representing the percentiles are consecutively -z0.99, -z0.98, -z0.97, ..., z0.50, ..., z0.97, z0.98, z0.99 with z0.50 equal to zero. A z-score w can be converted to a percentile by calculating using the z-table the probability P(Z
Have you seen this document?
http://www.who.int/childgrowth/standards/Chap_7.pdf
If your BMI values are entered in SPSS as numbers (kg/m2) , you can calculate percentiles through Data- Rank Cases- rank types- proportion estimates-rankit. If your BMI values are entered just as z values, you can calculate BMI value for each individual with Transform- compute variable. By doing calculations you must change the equation for z value from z=(x-averageBMI)/st.dev bmi to x indivi. BMI val.=z value*st.dev.bmi+averageBMI. As result you will get individual BMI values that can be then calculated with rank cases into percentiles values for each individual.
I hope my advice will solve your problem!
Best regards,
Tim.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS