P-value for between group difference by row in a 4 by 2 frequency table

[JLang055]

Hello,

I have a 4 by 2 frequency table:

BMI groups	cat 1	cat 2
underweight	1.4	1.6
normal	67.2	71.2
overweight	23.7	21.6
obese	7.7	5.6

Is there a way to obtain a p-value for significant difference for each BMI group when comparing cat 1 vs cat 2?

Thank you,

Justin

[StatDave]

See this note. 

[Ksharp]

Your data is not COUNT data. I suppose just to use ANOVA ?


data have;
infile cards expandtabs;
input BMI  : $40. cat1 cat2;
cards;
underweight 1.4 1.6
normal 67.2 71.2
overweight 23.7 21.6
obese 7.7 5.6
;
data temp;
 set have;
 cat='cat1';v=cat1;output;
 cat='cat2';v=cat2;output;
 drop cat1 cat2;
run;
proc glm data=temp;
class bmi cat;
model v=bmi cat/solution;
quit;