Subject:
|
Re: On the veracity of statistics in general
|
Newsgroups:
|
lugnet.off-topic.debate
|
Date:
|
Fri, 19 Oct 2001 03:08:03 GMT
|
Viewed:
|
165 times
|
| |
|
|
Larry Pieniazek wrote:
>
> I was doing some Googletrolling with various search keywords, looking for
> some scholarly work on the accuracy of UN statistics. It's a relatively
> tough search...
>
> I ran across this tidbit:
>
> http://www.security-policy.org/papers/1997/97-P154at1.html
>
> Now, this is anecdotal of course, but there is a lot of difference between
> the 100-114M figure bandied about by the US state department, the UN and a
> bunch of other people, and the 10M figure the author claims
>
> But I do think it highlights how difficult it is to come up with accurate
> statistics on things that aren't easy to measure.
>
> I have no position on the relative merits of the author's position on the
> subject he was writing about (land mines), I just found it in my search, so
> don't focus on that bit, just the statistic on numbers that he is
> challenging. (or if you want to, go start a new thread, this one is about
> statistics)
>
> Here's my thinking. Statistics have a tendency to self perpetuate. We all
> know that to be true, Someone quotes them and gets cited and before you know
> it they are enshrined as gospel. Statistics also have a tendency to get
> cited out of context in order to make various points. Take the
> healthcare/crime statistics about various crimes/causes of deaths/injuries
> that are being bandied about here. Different factions here are citing the
> same statistics to prove different things.
>
> Many organisations are highly political. How likely is it that they always
> rigorously validate every statistic they hear, especially if it happens to
> fit their preconceived notions? How EASY is it to rigorously validate a
> statistic anyway, if it's a statistic about a closed society?
>
> That's why I take UN statistics, in particular, with larger grains of salt
> than most, because the UN is more highly political than just about any
> organization ever in existance. I'd say people that don't use similar
> scepticism are naiive or foolish or deliberately misconstruing things for
> their own ends.
>
> The number of times a statistic is cited in the media, while an interesting
> statistic in itself, has no bearing on how accurate the statistic itself is
> unless the fact checking process behind it is rigorous. And we KNOW the
> media are far from rigorous at checking facts. Unfortunately, it DOES have
> bearing on how believable a statistic is because things that are repeated
> tend to be remembered as true by a large fraction of people, who tend not to
> think critically.
>
> Do we have any statisticians, or polling experts or econometric experts in
> the house that can comment authoritatively?
Good summary Larry,
I've been thinking about these issues a lot lately and agree with a
lot of what you have said in your post.
I think some of this is covered in the book called "how to lie with
Statistics" by
Darrell Huff. I've never read this book... it's on my "to read" list
though.
And as far as being a statistician, I pretty much do some sort of
statistical analysis most days of the week (1), but I don't think I know
enough to qualify as an expert.
-chris
1. it's going to be a statistics weekend. Mini conferance in 8 days
and I signed up to present two posters and none of my data is
analyzed... :(
|
|
Message is in Reply To:
 | | On the veracity of statistics in general
|
| I was doing some Googletrolling with various search keywords, looking for some scholarly work on the accuracy of UN statistics. It's a relatively tough search... I ran across this tidbit: (URL) this is anecdotal of course, but there is a lot of (...) (23 years ago, 18-Oct-01, to lugnet.off-topic.debate)
|
26 Messages in This Thread:
 
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
