Constant interpretation - multiple regression with log dependent variable

[Sebastian Cernadas]

Dear all,

Could anyone please explain me how to interpret the constant in a multiple regression when the dependent variable is used with log? (see in the image attached)

Command:

reg logpayamount loggamehour i.gend-ag

Variables used:

logpayamount -> log of the amount paid by participants for video game consumption
loggamehour -> log of the number of hours participants play video games per week
gend -> gender of the participant
ag -> age category of the participant

Thank you upfront for your help!

[Joro Kolev]

From

E(log Y) = a + b*X

we see that the constant is the average of logpayamount when all your regressors are set to 0.

[Sebastian Cernadas]

Thank you!

[Phil Bromiley]

I would normally describe the constant as the predicted value of the dv when all regressors are set to 0. It may not be close to the actual mean of the dv for any observations that happen to have all zero values. I don't think it even has to be within the range of the dv.