In article <firstname.lastname@example.org>,
email@example.com (Seth Breidbart) wrote:
> In article <firstname.lastname@example.org>, Robert Bonomi
> <email@example.com> wrote:
>> In article <firstname.lastname@example.org>, Thor Lancelot Simon
>> <email@example.com> wrote:
>>> In article <firstname.lastname@example.org>, Dave Garland
>>> <email@example.com> wrote:
>>>> The exercise revealed numerous errors in both encyclopaedias, but
>>>> among 42 entries tested, the difference in accuracy was not
>>>> particularly great: the average science entry in Wikipedia contained
>>>> around four inaccuracies; Britannica, about three ...
>>> I'm astonished that a 25% difference is considered "not particularly
>> I'm astonished that something that can be explained by "jitter" of
>> "plus/minus one count" in 'ordinal' numeric data, would be considered
>> anything _other_ than "not particularly great". Well, unless they do
>> not really understand statistical analysis, that is.
>> 3 vs 4 is jitter.
> 126 vs. 168 is a bigger difference, though it's the same 25%.
> (Unless you believe that there are a lot of off-by-one errors, _all_
> in the same direction.)
Except that these numbers were averages, not actual counts. But then
they rounded them off for the article. It's possible that around four
is 3.7, and about 3 is 3.4, so they're actually much closer; but they
could also be 2.8 and 4.3, a 35% difference.
But what they also didn't include in the article was information about
the distribution, standard deviation, etc. If most of the articles in
Wikipedia have 3-5 innacuracies, while most of the Brittanica articles
have 2-4, that's a significant overlap. On the other hand, if 2/3 of
Wikipedia articles have no errors, and the other third have 10-14,
while Brittanica is 90% clean with the other 10% having around 30
errors, that's quite different.
Barry Margolin, firstname.lastname@example.org
*** PLEASE post questions in newsgroups, not directly to me ***
*** PLEASE don't copy me on replies, I'll read them in the group ***