Page 1 of 1

hard to reproduce bug

Posted: Tue Sep 21, 2010 4:58 pm
by startz
I had been looking at some stats or a historgram or something in a panel data set, but had closed all windows except the workfile window. I then changed the sample such that all the remaining observations were from a single date and all equaled zero. When I double-clicked on the series I got the "zero variance" alert. After I picked the spreadsheet view everything was okay.

Re: hard to reproduce bug

Posted: Wed Sep 22, 2010 11:23 am
by EViews Pamela
I can't reproduce this ... let me know if you come up with any more info.

Re: hard to reproduce bug

Posted: Wed Sep 22, 2010 1:01 pm
by startz
I can't reproduce this ... let me know if you come up with any more info.
Well, the problem is probably that EViews likes you and doesn't like me. This favoritism has to stop :evil:

In fact, I've been able to reproduce it. First I did Histogram and stats. Then I closed the window. Then I reset the sample such that all observations in the sample equalled zero. When I reopen the series window I get the error message.

If you can't make this happen, I can ship you the file. But it's 10 MB.

Re: hard to reproduce bug

Posted: Wed Sep 22, 2010 1:46 pm
by EViews Pamela
There are two different ways of getting a histogram in EViews: via the Histogram & Stats menu item, and by selecting the Histogram graph type in the main graph dialog. I was using the latter, which doesn't show the error. However, the first method requires that the variance be non-zero. So, I would suggest using the main graph dialog to get your histogram in this case. (Even though, unfortunately, you won't get the stats.)