0

I am currently trying to run a t test on a variable and determine if it's statistically significantly different from 1. Here is the code I am using:

ttest dm1=1

And it is spitting out this output:

enter image description here

I don't want my null hypothesis to be that mean=1, I want it to be that dm1=1. When I do regular calculations ({Beta(dm1)-1}/SE(Beta(dm1))) on the ttest, I get that the new t statistic should be around -48.89. What is the code to determine if the coefficient is statistically different than one, if this is not the proper way? Also, here is an image of the regression model for reference: enter image description here

Nick Cox
  • 35,529
  • 6
  • 31
  • 47
  • This is more a problem in understanding what each command does. Statalist is a better forum for this kind of question. There isn't really a programming problem here. – Nick Cox Apr 04 '16 at 07:25

1 Answers1

2

The first t-test syntax is for testing that the null that the mean of dm1 is 1. It has nothing to do with the regression coefficients at all.

If I understand what you are asking, you want a Wald test:

sysuse auto
reg price mpg weight i.foreign 
test mpg=1
dimitriy
  • 9,077
  • 2
  • 25
  • 50