Just read the blurb and there are two problems .. one is comparing
retrospective % with prospective (see below) and the other is inferring
causation. But it's probably the reporters who blotched this; they probably
cherry picked % from a lengthy interview and the result makes no sense.
I sometimes spent class time discussing pro- versus retrospective studies, and
also time doing 2 by 2 boxes with different base rates. If you take their "69%
of Candy eaters were violent" and "42% of non-violent children ate candy" and
construct a table, and arbitrarily assign "100" to represent the # of candy
eaters, certain features of the box are fixed, namely
Violent Nonviolent
Candy 69 31 100
No Candy __________________42___
73
You can make things come out differently depending on the base rate of candy
eating. If only 33%, it follows that:
Violent Nonviolent
Candy 69 31 100
No Candy _____158__________42__ 200
227 73 300
It's still true that 69% of candy eaters are violent, 42% (31/73) of nonviolent
kids ate candy, but, 79% of non-candy eaters (158/200) became violent, as
compared to 69% of candy eaters become violent. But you can reverse the
apparent effect by increasing the base rate of candy eating to 50% say, and
having 100 candy eaters. (hopefully I did not mess this up, but I think it's
right).
--------------------------
John W. Kulig
Professor of Psychology
Plymouth State University
Plymouth NH 03264
--------------------------
----- Original Message -----
From: "Dr. Martin Bourgeois" <[email protected]>
To: "Teaching in the Psychological Sciences (TIPS)" <[email protected]>
Sent: Thursday, October 1, 2009 12:21:53 PM GMT -05:00 US/Canada Eastern
Subject: RE: [tips] Correlation example
Good one. Here's what I find puzzling:
British experts studied more than 17,000 children born in 1970 for about four
decades. Of the children who ate candies or chocolates daily at age 10, 69
percent were later arrested for a violent offense by the age of 34. Of those
who didn't have any violent clashes, 42 percent ate sweets daily.
Am I missing something, or does this comparison make no sense? And 69% of the
kids in the study who ate sweets were later arrested for a violent offense! And
I thought we were a violent society on this side of the pond.
From: Beth Benoit [[email protected]]
Sent: Thursday, October 01, 2009 12:04 PM
To: Teaching in the Psychological Sciences (TIPS)
Subject: [tips] Correlation example
Here's an almost laughable example of "correlation is not causation" that some
might find a good example for class. (Well, aren't they almost always
laughable??)
It's about a study that found that children who eat lots of candy are more
likely to be arrested for violent behavior as adults. In all fairness, one
researcher did try to encourage people to dig a little deeper:
"Previous studies have found better nutrition leads to better behavior, in both
children and adults.
Moore said his results were not strong enough to recommend parents stop giving
their children candies and chocolates. "This is an incredibly complex area," he
said. "It's not fair to blame it on the candy." "
But in my morning newspaper, neither that conclusion was posted, nor was the
journal cited. Only the term "British researchers" was used.
Here's the story:
http://www.cbsnews.com/blogs/2009/10/01/crimesider/entry5355367.shtml
Beth Benoit
Granite State College
Plymouth State University
New Hampshire ---
To make changes to your subscription contact:
Bill Southerly ([email protected])
---
To make changes to your subscription contact:
Bill Southerly ([email protected])
---
To make changes to your subscription contact:
Bill Southerly ([email protected])
