Calculating BMI when ht and wt have been recorded with different units

[Rene McCrae]

Hello,

I am trying to create a new variable for BMI and I have a dataset with continuous variable "height" and another variable "htunits" recorded as cm or inches, and continuous variable "weight" with variable "wtunits" recorded as kg or lbs. When I attempt to convert all height values to m and weight to kg using the gen and replace commands I am getting an error "type mismatch"

Could anyone suggest syntax to do this properly?

Thanks in advance

Rene

[Weiwen Ng]

You are likely referring to a string variable as if it were numeric. Any string variable has to have quotes when its value is referenced, e.g.

gen height_inches = height * 2.54 if htunits == "inches"

Or perhaps vice versa. If you show your code (use the code delimiters like I did, see my signature), we can help you further.

[Rene McCrae]

I checked and the variables height, weight, htunit and wtunit are storage type "double", meaning the unit variables are incorrectly stored as numerical

When I use the following, as you suggested

gen weight_kg= weight/2.205 if wtunit=="lbs"

I get a type mismatch error. Could you explain 1) how I would convert my unit variables into the proper format to use the code? and 2) what command would I use to carry over the weight data that is inputted in kg already?

Here is a sample of my data

* Example generated by -dataex-. To install: ssc install dataex
clear
input double(weight wtunit height htunit)
 57 1 157 1
 50 1  61 2
 64 1 163 1
 75 1 170 1
 74 1 165 1
 93 1 175 1
107 1 189 1
143 1 185 1
132 2  60 2
120 1 183 1
 66 1 162 1
 59 1 164 1
 93 1 173 1
 78 1 174 1
119 2  60 2
 72 1 168 1
 43 1 147 1
 95 1 175 1
 80 1 167 1
109 1 169 1
 70 1 175 1
 86 1  72 2
 61 1 168 1
160 2  67 2
111 1 178 1
114 1 155 1
 88 1 177 1
 57 1 150 1
 87 1 178 1
 89 1 160 1
 57 1 157 1
 75 1 182 1
 99 1 182 1
 48 1 161 1
 77 1 165 1
 77 1 170 1
 59 1 165 1
146 2 155 1
 65 1 170 1
 66 1 167 1
110 1 178 1
 91 1 168 1
130 2  62 2
 64 1 160 1
 92 1 185 1
 43 1 157 1
 58 1 149 1
 85 1 171 1
102 1 187 1
114 1 176 1
 82 1 180 1
 93 1 175 1
 94 1 177 1
 90 1 167 1
 87 1 165 1
 77 1 169 1
 91 1 171 1
 45 1 157 1
 70 1 178 1
 75 1 172 1
 50 1 156 1
 67 1 160 1
 63 1 172 1
 94 1 172 1
 75 1 162 1
106 1 168 1
 62 1 157 1
220 2  68 2
 96 1 170 1
150 2  69 2
 82 1 152 1
 70 1 178 1
 75 1 158 1
 67 1 158 1
 75 1 155 1
109 1 165 1
 64 1 170 1
 92 1 183 1
 63 1 157 1
 79 1 180 1
 88 1 173 1
 90 1 183 1
 80 1 160 1
 75 1 156 1
102 1 170 1
 89 1 190 1
 73 1 165 1
 48 1 150 1
 96 1 162 1
 68 1 170 1
 93 1 170 1
 84 1 175 1
162 2  64 2
 99 1 180 1
 45 1 155 1
 93 1 172 1
 74 1 168 1
 64 1 150 1
 59 1 157 1
 53 1 168 1
end
label values wtunit wtunit
label def wtunit 1 "kg", modify
label def wtunit 2 "lbs", modify
label values htunit htunit
label def htunit 1 "cm", modify
label def htunit 2 "in", modify

[Marcos Almeida]

I gather you’ll need to - generate - and - replace- two variables (say, height_m and weight_kg) under the ‘if’ clause. After that, you may proceed with the generation of the BMI variable.

[Rene McCrae]

Thank you Marcos! I was incorrectly using the label in my code rather than the number 1 or 2. It works now

Rene

[Nick Cox]

gen BMI = cond(wtunit == 1, weight, weight * 0.453592) / (cond(htunit == 1, height/100, 2.54 * height/100)^2)

[Laurita Norkute]

I have a related question. How would you convert a variable that is coded as "stones and pounds" to kilograms?
Look forward to hearing from you.
Best wishes,
Laurita

[Marcos Almeida]

Please present data as shown in #3. Since it is your first post, please read the FAQ, particularly the topics about sharing data/command/output. You may wish to start a new thread as well.