Post-estimation, predicting the dependent variable with a change to one or more independent variables

[Jonah Busch]

Dear Stata community,

I'm attempting to predict the value of a dependent variable if one of the independent variables is changed.

For example, let's suppose I've modeled testscore as a function of independent variables height, weight, and cupsofcoffee:

regress testscore height weight cupsofcoffee

I can generate predictions with
predict testscorehat

But now I want to predict what everyone's testscore would have been if they'd drank twice as many cups of coffee. What command do I use? Predict? Adjust? Something else? And with what syntax?

If I wanted to predict what would happen if cupsofcoffee had a single value for everyone, e.g. if everyone had one cup of coffee, i.e. I could run:
adjust cupsofcoffee=1
(https://www.stata.com/support/faqs/s...ct-and-adjust/)

But if I try to double everyone's coffee intake by running
generate doublecoffee=cupsofcoffee*2
adjust cupsofcoffee, by(doublecoffee)

I get the error message 'too many values', which seems to mean that the the vector doublecoffee has too many individual values.

It seems that I might be able to make an out-of-sample prediction on a "new" dataset that is the same as the old data set but for the doubled coffee. (https://www.stata.com/manuals13/u20.pdf)
But this seems like a lot of work for what I would think would be a simple command.

I'm sure this must be a routine command or syntax, but I'm having a difficult time finding it.

Thanks,

Jonah

[William Lisowski]

Welcome to Statalist.

It seems to me that what you ask for can be accomplished without much effort by

generate savecoffee = cupsofcoffee
replace cupsofcoffee = 2*cupsofcoffee
predict ...
replace cupsofcoffee = savecoffee

[Jonah Busch]

Hi William,

Thanks for your quick response. That works perfectly. Exactly what I was looking for.

Best,

Jonah