I am new to this but maybe you can help me with the following issue.

I'm using the body mass index from the Indian survey of 2015 as my outcome variable.

However, in my sample selection it seems to range from 1204 to 5606, which is pretty unlikely to be true.

Has some of you encountered a similar problem? What do you think the problem can be: should there be a decimal point?

]]>

Many DHS variables are multiplied by some factor just to get rid of a decimal point. The BMI, like the HAZ, WAZ, and WHZ, is multiplied by 100. To get to the correct scale, you need to divide by 100. There may be some extreme values of the BMI that are implausible or even impossible, reflecting possible measurement or coding errors in height or weight. v445 is coded 9998 for flagged values but there are long tails on both the low and high ends of the distribution and you may want to set more conservative boundaries for the plausible range.

]]>

Thank-you for the clarification. Much appreciated.]]>