Archived posting to the Leica Users Group, 1997/09/02

[Author Prev] [Author Next] [Thread Prev] [Thread Next] [Author Index] [Topic Index] [Home] [Search]

Subject: Re: M6 problem "survey"
From: "Patrick G. Sobalvarro" <pgs@sobalvarro.org>
Date: Tue, 02 Sep 1997 23:42:19 -0700

At 09:28 PM 9/2/97 -0500, Eric Welch wrote:
>At 01:48 PM 9/2/97 EDT, you wrote:
>>you want to make claims based on statistics, get meaningful ones --
>>survey thousands of owners, or try to get the facts on repair rates
>>from Leica service.
>
>Actually, for most populations, a sample of 400 is adequate. :-)
>
Actually, it depends on the overall frequency in the population of what
you're trying to measure.  If you're trying to measure something that has
an overall frequency of about a half, then 400 samples give you pretty
tight error bounds.  If you're trying to measure something that has an
overall frequency of 1/1000, then 400 samples give you pretty poor error
bounds... but that wasn't the problem with this survey.  Personally, I
think that surveys like this one of M6 owners are worse than useless --
they're misleading because they're more likely to produce incorrect results
than correct ones.  Unhappy people are more likely to respond to such
surveys with a complaint than satisfied people with a message saying "mine
works fine."