error distributin test
Posted: Sun Sep 28, 2008 2:04 pm
by tg128
I have selected the "normal" option for the error distibution when estimate ARCH, however, I cannot find any results regarding to normal distribution from the equation window, how can I tell if the residuals from the my model is normally distributed or not please?
Re: error distributin test
Posted: Sun Sep 28, 2008 2:41 pm
by startz
I have selected the "normal" option for the error distibution when estimate ARCH, however, I cannot find any results regarding to normal distribution from the equation window, how can I tell if the residuals from the my model is normally distributed or not please?
Look at the Jarque-Bera statistic to the right of the histogram.
Re: error distributin test
Posted: Sun Sep 28, 2008 3:56 pm
by tg128
I have selected the "normal" option for the error distibution when estimate ARCH, however, I cannot find any results regarding to normal distribution from the equation window, how can I tell if the residuals from the my model is normally distributed or not please?
Look at the Jarque-Bera statistic to the right of the histogram.
but that's not from the ARCH test window itself, you have to conduct another test by using 'view' ... if I selected t-distribution, i will have a row of t-distribution statistic results shown, I find nothing when I use 'normal distribution' option...
Re: error distributin test
Posted: Sun Sep 28, 2008 5:55 pm
by startz
I have selected the "normal" option for the error distibution when estimate ARCH, however, I cannot find any results regarding to normal distribution from the equation window, how can I tell if the residuals from the my model is normally distributed or not please?
Look at the Jarque-Bera statistic to the right of the histogram.
but that's not from the ARCH test window itself, you have to conduct another test by using 'view' ... if I selected t-distribution, i will have a row of t-distribution statistic results shown, I find nothing when I use 'normal distribution' option...
You are choosing the distributional assumption for the error. I believe the row you are seeing is the estimated degrees of freedom. It's not a test of the distributional assumption.