Which test is better to use with categorical variables?

[Bakhytzhan Shakhmanov]

Good afternoon!

I regress the model:

GDP = β₀ + β₁ERR₁ + β₂ERR₂ + ơ₁Pop+ ơ₂Surf + ơ₃ExGS + ơ₄ExpEd + ơ₅RIR + u

My main focus is ERR variable which is categorical (exchange rate regime of the country). The base level is floating regime, ERR₁ is pegged regime and ERR₂ is residual regime.
Thus, I need to test significance of these variable (ERR). I try to use command ttest, but Stata shows me mistake.

ttest GDP, by(ERR)

more than 2 groups found, only 2 allowed

I tried to understand more appropriate test on https://stats.idre.ucla.edu/other/mult-pkg/whatstat/, but the table is not clear.

Could you recommend which test to use for whole model to asses it and test separately for categorical variable?
I need to understand significance of the model and of the variables. So I more lean to F test for whole model and t and p tests for variables. Thank you!

[Felix Bittmann]

As long as GDP is a continuous variable, just use an OLS regression.

regress GDP i.ERR

This is a very simple model, you can easily recreate the equation from above using regress and get the results under control of all independent variables in the model.

[Bakhytzhan Shakhmanov]

Than you, Felix! But my question is not about how to run regression. I did it.
I need to test whole model and variables.

[Phil Bromiley]

You need to read the documentation for regress much more carefully. You also probably need to read an introduction to regression. The regression F tests the hypothesis that all the parameters are zero. The regression output has T statistics and probabilities beside each parameter.

If you need to test the equality of specific parameters with in the regression, then just use the test command.

[Bakhytzhan Shakhmanov]

Thank you for your answer, Phil!
Yes, I understand these moments of regression analysis.
But now I have found answer.
Thank you again!