View Full Version : Lies, Damned Lies, and Statistics!
druid
04-26-2004, 01:56 PM
Here is
a book I found on amazon.com, How to Lie With Statistics. I think I might pick this one
up.
http://www.amazon.com/exec/obidos/tg/detail/-/0393310728/ref=cm_cr_dp_2_1/1
04-5484362-5481516?v=glance&s=books&vi=customer-reviews (\"http://www.amazon.com/exec/obidos/tg/detail/-/0393310728/ref=cm_cr_dp_2_1/104-5484362-5481516?v=glanc
e&s=books&vi=customer-reviews\")
koolking1
04-26-2004, 02:06 PM
I read that book many years ago and it was an \"eye-opener\". If this kind of thing interests you, you might
also want to read some books on \"marketing\" - you\'d be mighty surprised how things are done to make you want
to buy something or believe in something.
DrSmellThis
04-26-2004, 09:28 PM
Nice post. You can get much more value out of statistics and appreciate it more by learning how it can be abused.
The are a million ways to run a crappy study or write it up in a crappy way. I wish everybody had the quality of
intelligent skepticism contained in that book.
One of the most important skeptical principles to keep in
mind when you see statistics reported somewhere is that just because two things are correlated (meaning that the
level of one changes with the other in a consistent way) does not mean that one of them causes the other to
change. For example, even if every person who ate twinkies got cancer (a perfect correlation); we could not show
that eating twinkies causes cancer, based on that fact. The fallacy that correlation somehow implies
causality appears every day in the papers, even today. Journalists aren\'t taught about it, unfortunately.
Even some \"scientists\" (bad ones) fall for this BS.
Another important principle of research-consumer
skepticism is that we should review every aspect of how researchers choose to design their studies. Why? Because the
design of a study determines the kinds of results it is possible to end up with. What is being compared in the
study? How and why? How could the comparisons be different? If the design is sloppy or irresponsible, the researcher
could be seeing only what he or she wants to see. (This is called \"confirmation bias\")
On the other hand,
stats are just a neutral tool of reasoning; a natural part of reasoning. You have to appreciate stats to be able to
understand what we know about the world. Some believe that statistics are \"bad\". That belief is every bit as
dangerous as \"believing every number you read\"! That would be like saying logic (another tool of reasoning) is
bad just because logic can be misused.
To be sure, we need statistics to be able to understand physics,
mathematics, genetics, psychology, and many other fields of knowledge. Why? Because nature behaves in statistical
ways, at just about every level; wherever and whenever there are repetitions, trends, numbers, possibilities or
probabilities. That\'s just about everywhere and always.
We have no choice but to turn to statistics --
albeit carefully -- if we want to understand our world.
</font><blockquote><font class=\"small\">Quote:</font><hr />
To be sure, we need statistics to be able to
understand physics, mathematics, genetics, psychology, and many other fields of knowledge. Why? Because nature
behaves in statistical ways, at just about every level; wherever and whenever there are repetitions, trends,
numbers, possibilities or probabilities. That\'s just about everywhere and always.
We have no choice but to
turn to statistics -- albeit carefully -- if we want to understand our world.
<hr /></blockquote><font
class=\"post\">
Man, I HATE statistics. P.S., try my new BF #7, Pi:NPA:SOE:WAGG:AE:PF 14:9:7:4:2:1
DrSmellThis
04-27-2004, 08:32 AM
You are starting to sound like watcher, posting your great mixes (sounds like a good one, BTW).
/ubbthreads/images/graemlins/smile.gif
Don\'t blame you one bit! I always hated stats too, with a
passion (I was cruelly forced to take many courses in it! /ubbthreads/images/graemlins/tongue.gif), until
I finally got a great, brilliant stats teacher who focused on the simple ideas behind it rather than the number
crunching. Then I was able to just put it together in one place under my understanding of science and research. I
still hate the number crunching, but computers do all that now anyway.
/ubbthreads/images/graemlins/smile.gif
Sexyredhead
04-27-2004, 08:33 AM
Yes. I give thanks daily for SAS. /ubbthreads/images/graemlins/smirk.gif
</font><blockquote><font class=\"small\">Quote:</font><hr />
You are starting to sound like watcher, posting
your great mixes (sounds like a good one, BTW). /ubbthreads/images/graemlins/smile.gif
Don\'t
blame you one bit! I always hated stats too, with a passion (I was cruelly forced to take many courses in it!
/ubbthreads/images/graemlins/tongue.gif), until I finally got a great, brilliant stats teacher who
focused on the simple ideas behind it rather than the number crunching. Then I was able to just put it together in
one place under my understanding of science and research. I still hate the number crunching, but computers do all
that now anyway. /ubbthreads/images/graemlins/smile.gif
<hr /></blockquote><font
class=\"post\">
DST, I was trying to touch on the irony of it (as we use ratios in the mixes). I actually like
stats. Anyway, what you said about stats being in nature and the way it works, I agree, and I think part of that
could be in our pheromone profile and the relationship between the pheromones we produce.
When you said what you
did I was reminded of your Fiboccini Numbers trials, as you were trying to find ratios in natures and apply that to
pheromone ratio that might be percieved as beautiful.
DrSmellThis
04-27-2004, 09:20 AM
</font><blockquote><font class=\"small\">Quote:</font><hr />
Yes. I give thanks daily for SAS.
/ubbthreads/images/graemlins/smirk.gif
<hr /></blockquote><font class=\"post\">
So is that the
best stuff now? I remember that and SPSSX from grad school. But it seems like it might be simpler now.
Sexyredhead
04-27-2004, 09:35 AM
SAS is evil, until you understand it, then you can chain it up and make it do your dirty work for you.
/ubbthreads/images/graemlins/grin.gif
DrSmellThis
04-27-2004, 10:05 AM
...wouldn\'t mind having a girlfriend like SAS! /ubbthreads/images/graemlins/tongue.gif
/ubbthreads/images/graemlins/blush.gif /ubbthreads/images/graemlins/grin.gif
/ubbthreads/images/graemlins/laugh.gif
Mtnjim
04-27-2004, 10:07 AM
</font><blockquote><font class=\"small\">Quote:</font><hr />
</font><blockquote><font
class=\"small\">Quote:</font><hr />
Yes. I give thanks daily for SAS.
/ubbthreads/images/graemlins/smirk.gif
<hr /></blockquote><font class=\"post\">
So is that the
best stuff now? I remember that and SPSSX from grad school. But it seems like it might be simpler now.
<hr
/></blockquote><font class=\"post\">
Slide Rules dominate /ubbthreads/images/graemlins/grin.gif
/ubbthreads/images/graemlins/grin.gif
Pancho1188
04-27-2004, 10:13 AM
</font><blockquote><font class=\"small\">Quote:</font><hr />
...wouldn\'t mind having a girlfriend like SAS!
/ubbthreads/images/graemlins/tongue.gif /ubbthreads/images/graemlins/blush.gif
/ubbthreads/images/graemlins/grin.gif /ubbthreads/images/graemlins/laugh.gif
<hr
/></blockquote><font class=\"post\">
...wouldn\'t mind having a girlfriend like SRH! Oh, wait...we\'re
talking about statistics...
Well, in that case...
30% of women have made love to a man while he was
asleep
The female orgasm involves 87 muscle contractions per second
...but who\'s counting?
Sexyredhead
04-27-2004, 10:50 AM
</font><blockquote><font class=\"small\">Quote:</font><hr />
</font><blockquote><font
class=\"small\">Quote:</font><hr />
...wouldn\'t mind having a girlfriend like SAS!
/ubbthreads/images/graemlins/tongue.gif /ubbthreads/images/graemlins/blush.gif
/ubbthreads/images/graemlins/grin.gif /ubbthreads/images/graemlins/laugh.gif
<hr
/></blockquote><font class=\"post\">
...wouldn\'t mind having a girlfriend like SRH! Oh, wait...we\'re
talking about statistics...
Well, in that case...
30% of women have made love to a man while he was
asleep
The female orgasm involves 87 muscle contractions per second
...but who\'s counting?
<hr
/></blockquote><font class=\"post\">
I think I only counted 85.
/ubbthreads/images/graemlins/confused.gif Oh welll, I\'ll just have to count again!
/ubbthreads/images/graemlins/blush.gif /ubbthreads/images/graemlins/grin.gif
koolking1
04-27-2004, 01:28 PM
no wonder I haven\'t been sleeping so well!!!!
Powered by vBulletin® Version 4.2.2 Copyright © 2024 vBulletin Solutions, Inc. All rights reserved.