one sided t test for dummy variables

[rob86]

Hi

I posted a question a while ago about testing whether the coeffiects of a number of dummy variables follow a certain order, i.e c1 > c2 > c3, and the answer I got was to perform a one sided t test. I am at the stage now where I need to perform the test and am wondering now to perform it on eviews? Had a look in the manual but couldnt the answer. Also is there a a way to test whether the dummy variables are significantly different from each other and if so how would I do it?

Thanks

Rob

[startz]

1. You perform a one-sided test the same way you perform a two-sided test, except you have to look up the appropriate critical value in a table.

2. (1) holds for comparing two coefficients. If you really need to compare three, as in your example, that's going to be very hard.

3. Hypothesis tests on the coefficients of dummy variables are no different than any other test. You can use the Wald test in the equation view menu.

[rob86]

thanks again startz, i can forget about the first part then. Is the correct way to test whether the coefficents of the dummy variables are significantly different from each other using a wald test c(1) = c(2) = c(3) = c(4)? is so how would I interpret an F-stat of .21? I have attached a the wald test results. thanks

[startz]

Take a look at the p-value that EViews supplies.

[rob86]

ok but is c(1) = c(2) = c(3) = c(4) the correct thing to enter into the wald test to test are the coefficents significantly different from each other? will check out how to interpret results when im in college and have accesss to the help manual.

[startz]

Yes, that's the right specification.