|
![](/i/fill.gif) |
On Tue, 11 May 2010 22:42:30 +0100, Orchid XP v8 wrote:
>>> What the heck did you search for?! It's kind of a rather specific
>>> question...
>>
>> I searched for "newton apple myth". First hit took me to a page that
>> linked to this.
>
> And that WORKED??
Absolutely. Try it.
>>>> You might refute it, a straw poll might be a start, but a poll of 20
>>>> people isn't a particularly statistically valid poll.
>>> Well it would be more valid than a straw poll of *one* wouldn't it?
>>> :-P
>>
>> I don't think statistically it would be. A sample size that's too
>> small is too small.
>
> I don't think stats works like that. You get a number which indicates
> how accurate your estimate is. You them apply some arbitrary threshold
> to decide what is "valid" and "invalid". Choose the right threshold and
> anything can be "valid". My point is that a larger sample size will give
> a better result (assuming reasonable sampling).
Below a certain threshold, a sample holds no statistical validity. At
work we use statistical analysis on exam items, and if we have less than
50 samples or so to work with, we can't get any useful statistics; 1
result is no more statistically valid than 20 in such a case, because in
order to do the analysis, a sample size of 50 or more gives us what we
need for the analysis.
Statisticians I've talked to over the years have confirmed that - and I
work with a few on a somewhat regular basis in my work (analysing survey
results and the like, as well as indirectly working with people who are
professional psychometricians).
>> Still, you might give it a go anyways, if anything it'll get you
>> talking to people in meatspace. ;-)
>
> In other news, I just spent an hour and a half in a pub.
Also not a bad thing - humans are - by our very nature - generally social
creatures.
>>> Still, I guess this is going to be one of those things where no matter
>>> how much evidence I produce that nobody has heard of these people,
>>> everybody will continue to assert that my statistics are just wrong...
>>
>> That's because you don't *have* statistics. You have a guess. You say
>> "nobody", but to prove that, you have to prove that *everybody* hasn't
>> heard of them. That's pretty easy to disprove.
>
> Clearly I meant "nobody" in the sense of "a very small fraction of the
> population" rather than "zero people in the entire world". :-P
Ah, hyperbole. ;-) Still, you don't have statistics, you have a logical
argument that starts with what I believe is a false premise.
>>> Quite a few of the names look hard to pronounce, but we'll see...
>>
>> Such as?
>
> Sagan?
Just like it's spelt: Say-gan
> Rechecking the list, it doesn't look so bad...
I didn't think it was, hence the question.
Jim
Post a reply to this message
|
![](/i/fill.gif) |