How to classify handgrip?

[Vivian Le]

Dear all,

I'm currently have a research about handgrip strength in highschool student. I need to classify it into 3 categories: Under, Nomal, Over, based on the table here:

Age	Gender	Under	Normal	Over
15	Male	<34	>=34	<40.9
15	Female	<24.5	>=24.5	>28.5
16	Male	<36.9	>=36.9	>43.2
16	Female	<26	>=26	>29

And I have some data:

Age	Gender	Handgrip (kg)
15	Male	40.8
15	Female	19.2
15	Female	30.1
16	Male	27.7
16	Female	23.3
16	Female	17

Please forgive my poor English.
Thank you so much in advance for any help!

Best regards,
Vivian

[George Ford]

clear
set obs 1000

g male = runiform()>0.5
g age = 15
replace age = 16 if runiform()>0.5
g handgrip = runiform(20,80)
g age16 = age==16

g bound1 = 34*(male & !age16) + 36.9*(male & age16) + 24.5*(!male & !age16) + 26*(!male & age16) 
g bound2 = 40.9*(male & !age16) + 43.2*(male & age16) + 28.5*(!male & !age16) + 29*(!male & age16) 

g griptype = 2
replace griptype = 1*(handgrip < bound1) + 3*(handgrip > bound2)
label var griptype "Grip Category"
label define glab 1 "Under" 2 "Normal" 3 "Over"
label values griptype glab

[Nick Cox]

There are many ways to do this. Here's another: assume students are 2 for normal unless there is evidence to the contrary. This assumes an indicator female which is 1 for female and 0 for male.

gen class = 2 if !missing(female, age, handgrip) 
replace class = 1 if  (female == 0 & age == 15 & handgrip < 34) | (female == 0 & age == 16 & handgrip < 36.9) 
replace class = 1 if  (female == 1 & age == 15 & handgrip < 24.5) | (female == 1 & age == 16 & handgrip < 26)

There are two more statements for class 3. Watch out for missing values.

[Vivian Le]

clear
set obs 1000

g male = runiform()>0.5
g age = 15
replace age = 16 if runiform()>0.5
g handgrip = runiform(20,80)
g age16 = age==16

g bound1 = 34*(male & !age16) + 36.9*(male & age16) + 24.5*(!male & !age16) + 26*(!male & age16)
g bound2 = 40.9*(male & !age16) + 43.2*(male & age16) + 28.5*(!male & !age16) + 29*(!male & age16)

g griptype = 2
replace griptype = 1*(handgrip < bound1) + 3*(handgrip > bound2)
label var griptype "Grip Category"
label define glab 1 "Under" 2 "Normal" 3 "Over"
label values griptype glab

Thank you both of you.

But what happens if I have more than two value of age. For example,

Age	Gender	Under	Normal	Over
15	Male	<34	>=34	>40.9
15	Female	<24.5	>=24.5	>28.5
16	Male	<36.9	>=36.9	>43.2
16	Female	<26	>=26	>29
17	Male	<39.6	>=39.6	>46.2
17	Female	<26.3	>=26.3	>30.3
18	Male	<40.7	>=40.7	>47.2
18	Female	<26.5	>=26.5	>31.5

[Nick Cox]

You need more code to handle extra possibilities.

[George Ford]

I think perhaps the simplest way is to create a "threshold.dta" file with the info in the table from #4, with Age, Gender, and the values for Under and Over (ignore the Normal, since you're just going to use < and >, with the rest being 2).

Then,

joinby Age Gender using threshold

and you'll get two variables Under and Over that represent the thresholds specific to each age and gender.

Then use the command

g griptype = 2 
 replace griptype = 1*(handgrip < Under) + 3*(handgrip > Over)