The AFU and Urban Legend Archive
Animals
 falling cats more




Date: 26 Mar 1996 10:07:07 -0500
From: "Ian A. York" <iayork@panix.com>
Newsgroups: alt.folklore.urban
Subject: Re: Falling cats

In article <4j9021$648@universe.digex.net>, Elizabeth Jones <ebeth@universe.digex.net> wrote: >
>It seems to me that 132 cats is a rather small sample size from which
>to draw conclusions, especially when several of the more important
>conclusions are based on 1 sample "1 of 22", "1 among 13". Is this
>considered a statistically significant sample size in medical research?

Nope (as I noted in my comment, this "may be a statistical anomaly in a relatively small sample), and you're quite right that these conclusions are not 'statistically significant'. On the other hand, to their credit (maybe) these guys don't make the claim that these are statistically significant findings.

There's a fundamental difference between clinical observation and medical research. This paper represents the former. Even though there aren't statistics involved, the conclusions can still be valuable to a clinician. If, for example, you're faced with a phone call from a distraught owner who has just seen their cat falling from the 20th floor, you can point out that death is not inevitable - that's clear from these results. You'd be out of line to say that their cat has a 90% chance of survival, because there's no confidence to the numbers.

So this is a moderately useful paper in that sense. Should it not have been published until they got large enough sample sizes for statistical analysis? I don't think it's reasonable to wait that long; this is the result of several years of data collection, and there's nowhere in the US (or probably the world) that gets more cases than this place.

I don't want to defend this too strongly - as I pointed out in my introduction to the paper, the 'conclusion' that's often repeated from this paper is no more than a speculation - and you may have noticed that I do medical research, not clinical observations; but I think it's unfair to ignore the paper because they don't have stats.

Ian


http://tafkac.org/