Date: Thu, 16 Mar 2006 09:24:55 -0500
From: arne <arne
To: Walt Cooney <waltc
References: <000c01c648a4$07a2bd20$6401a8c0

Walt Cooney wrote:
> In my opinion, what you guys suggest and do is really the best way to do it.
> That is a much better measure of the error than the reciprocal of S/N.  However,
> I'd still argue that we only ask observers for the reciprocal S/N for their
> error.  Some of the reasons why I would argue that are
> 
> will do it
> work
> 
> because you only shoot one image or must stack many faint ones to get one with
> decent S/N.
> 
> 
> users of the data are familiar with and can use.
> 
> I wholeheartedly agree that seeing the scatter in multiple data points is a much
> better measure of error but I'm not sure if it is what we as AAVSO should be
> demanding when a much easier estimate gets us most of the way there.
> 
That is why we generally request reciprocal S/N right now; simplicity.
The true error analysis is quite complex, and even taking multiple
observations only tells you about the "flux" noise and not any systematic
well.  We will be addressing this issue in more detail in the future.
Arne
_______________________________________________