Re: st: re: tsset with non-integers

[Michael Hanson <mshanson@mac.com>]

On May 7, 2009, at 1:23 PM, Kit Baum wrote:

> Calculating a D-W statistic on, e.g., a cross-section is asinine,  
> and a program that lets you do it is a pretty lousy piece of  
> software. As the sort order of a cross section is arbitrary, I can  
> change the "serial correlation" between successive observations by  
> just sorting the data on some other key.


On May 7, 2009, at 2:06 PM, Nick Cox wrote:

> While we're chuckling at what appears to be poor practice elsewhere,  
> let's remember that all statistical programs basically assume at  
> some level that users take responsibility for what they do.


On May 8, 2009, at 1:30 AM, John Antonakis wrote:

> I have seen this wrong suggestion all over the place (based on a  
> quick scan of the net).


Back when I used to teach intro econometrics, my department decided to  
standardize on EViews -- version 4 or 5, I believe -- because of  
purported "ease of use".  It would always report a Durbin-Watson  
statistic with every OLS estimation, whether the data were cross- 
sectional or time series.  I would repeatedly explain that the D-W  
statistic was only applicable to time series questions -- and, even  
then, only under certain rather restrictive conditions.  Nonetheless,  
a small percentage of students each year would insist on reporting the  
value of the D-W statistic for cross-sectional regressions in their  
final course project, and proudly explain that a value near 2 meant  
their regressions did not suffer from serial correlation.  Such papers  
helped significantly with determining the grade distribution.

In the advanced econometrics course, I allowed students to use EViews  
if they insisted, but taught the course (and assigned problem sets)  
using Stata.  The mantra of that class was: "Never push a button or  
type a command you do not fully understand."  Most (but not all) the  
students got the message the second time around.

-- Mike
 
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/