03-27-2018 04:39 PM
I have a model like this if I run the regression:
y = cons + b1*x1 + b2*x2+ b3*x3 + se
y = dependent variable
cons = intercept
se = error term
x1,x2,x3 = independent variables
b1,b2,b3 = coefficient of each independent variables
1. How can I capture the b1,b2,b3 in a new variable?
2. How can I capture b1/b2, b2/b3?
3. How can I calculate the estimator and standard error of b1/b2, b2/b3? Does it need bootstrap?
03-27-2018 04:51 PM
All of the SAS regression procedures have at least one if not two ways of getting parameters into a data set such as OUTEST= somesetname on Proc Reg statement or ODS OUTPUT ParameterEstimates = somesetname statement. Which regression procedure are you using? That would address question 1.
With the values in a data set then a data step would calculate those ratios in question 2
I'm not familiar with anything resembling standard errors or the ratio of parameter estimates and not sure what the interpretation would be.