Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Normally distributed error term & testing normality of residuals


From   "JVerkuilen (Gmail)" <jvverkuilen@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Normally distributed error term & testing normality of residuals
Date   Sat, 13 Oct 2012 09:42:38 -0400

On Sat, Oct 13, 2012 at 9:27 AM, Ebru Ozturk <ebru_0512@hotmail.com> wrote:
> Thanks. I checked it too. But I just wanted to learn is there a way of testing normality apart from a formal test like histogram, scatter plots?

Ah, yes, I get that point. Well the trick is that the observed data
are what they are but we don't know the censored data.

I can think of a few "devices" but I'm not sure what good properties
they would have. Cameron and Trivedi mention (but do not show) the
notion of comparing the predicted values for non-censored cases to the
observed values. The distribution won't be normal for them and the
relationship won't be perfectly linear, but you shouldn't see gross
differences in a good fitting model.

Example:

sysuse auto
gen wgt = weight/1000
tobit mpg wgt, ll(17)
predict predicted, xb
predict seforecast, stdf
gen censored = 0
replace censored = 1 if mpg <= 17
summarize mpg predicted if censored == 0, detail
cor mpg predicted
qqplot predicted mpg if censored == 0
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index