September 2003
Sun Mon Tue Wed Thu Fri Sat
  1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30        
Aug   Oct

Other blogs I visit

Subscribe to this blog in Radio:
Subscribe to "Cathemeral Thinking" in Radio UserLand.

Click to see the XML version of this web page.

E-mail this blog's author, David Harris:
Click here to send an email to the editor of this weblog.

Monday, September 15, 2003

My meme experiment

By now, you've probably read the "Aoccdrnig to rsereach..." meme floating about the web. I talked about it last Friday. Since then, the meme has exploded on the web, and further propagated by Slashdot today.

What I didn't mention the other day is that I was using this as a very informal experiment. I originally received the paragraph by email on Friday. But then I changed it to post to my site. The original version I received was:

Aoccdrnig to a rscheearch at an Elingsh uinervtisy, it deosn't mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht frist and lsat ltteer is at the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit porbelm. Tihs is bcuseae we do not raed ervey lteter by it slef but the wrod as a wlohe. ceehiro.

But then I changed on Friday to:

Aoccdrnig to rscheearch at an Elingsh uinervtisy, it deosn’t mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht frist and lsat ltteer is at the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit a porbelm. Tihs is bcuseae we do not raed ervey lteter by it slef but the wrod as a wlohe.

On Monday morning I changed it in my post to:

Aoccdrnig to rsereach at an Elingsh uinervtisy, it deosn't mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht frist and lsat ltteres are in the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit a porbelm. Tihs is bcuseae we do not raed ervey lteter by istlef, but the wrod as a wlohe.

Apart from the fact that the original contained some grammatical errors and there were misspellings, I wanted to see how this paragraph propagated in blogspace in different versions.

Before I posted my version, I checked that there were no versions on the web that began precisely with "Aoccdrnig to rscheearch". I also checked on my second rewrite to confirm it didn't yet exist. So now I can use Google to see how far my versions have propagated.

At time of writing, my Friday version is found by Google on 106 sites. The Monday version is not yet found anywhere else. So basically, this shows that my Friday version spread pretty quickly through blogspace when it wasn't yet floating all over the web but the latest version which has only had a day to propagate in an already saturated web, hasn't made any impact, presumably because everybody already knows about it or people are just linking to the entry without copying the text now.

Like I said, it's not a scientific experiment, but still an interesting one to see how memes spread.

[Update: Tuesday morning - google now finds 206 copies of my Friday version, 8 copies of the Monday version.]

[Update: Wednesday evening - original version: 569, my Friday version: 396, my Monday version: 35]

[Update: Saturday morning - original version: 1750, my Friday version: 1010, my Monday version: 42]

[Update: Saturday evening- original version: 3450, my Friday version: 1420, my Monday version: 44]

Fish diet prevents teen violence - a critical look

Yesterday I read a disturbing news report that begins:

Feeding children a diet rich in fish could prevent violent and anti-social behaviour in their teens, according to research to be announced this week which suggests the root causes of crime may be biological rather than social.

Why is this disturbing? Primarily because this is an extraordinary claim and so it needs extraordinary evidence - evidence that does not appear to exist. There are two major problems that I'm going to discuss. First, whether the newspaper report is an accurate representation of the scientific paper and, second, whether the claims in the scientific paper are justified. I'm going to talk about both as I go because they are interwoven.

Here are the resources you might find interesting if you want to see the original reports:
Guardian newspaper article
Abstract of research paper

From the abstract:

Eighty-three children were assigned to an experimental enrichment program from ages 3 to 5 years and matched on temperament, nutritional, cognitive, autonomic, and demographic variables with 355 children who experienced usual community conditions (control group).

The children studied in this group all live in Mauritius. This brings us to the first interpretational problem. Criminality is extremely culture-dependent. That means that the study here has little generalizability. (i.e. we can't extrapolate from these results and think that they would be similar in other cultures, such as the US or the UK, both very different from Mauritius.)  Lack of generalizability is not a problem as such, as long as it is always kept in mind. Seeing as the study is done in Mauritius, there is a chance it could be of great benefit to Mauritiuan communities. However, the newspaper article does not even mention Mauritius until the 5th paragraph and implies that the results are generalizable to other countries. The abstract of the research paper doesn't mention Mauritius at all.

Something else in the abstract set off warning bells in my mind - the idea of matching on temperament. I have spoken with a number of criminologists and psychologists who say that the idea of being able to quantify or classify three-year old children on the basis of "temperament" is ludicrous and impossible. But we'll let that slide for the moment. We'll also let slide the strange use of 83 experimental subjects and 355 control subjects. Well-known theorems in statistics suggest this is a silly way of doing things - most of the control subjects are wasted and the results would be nearly as precise with a far smaller number. It would have been better use of resources to do this differently.

The main conclusion of the paper is that children who went through the enrichment program between ages 3 and 5 ended up being less likely to show schizotypal personalities and criminal behaviors at ages 17 and 23. The abstract states:

The beneficial effects of the intervention were greater for children who showed signs of malnutrition at age 3 years, particularly with respect to outcomes for schizotypy at ages 17 and 23 and for antisocial behavior at age 17.

Notice here that the connection between malnutrition and criminality is not the most significant of the findings - which seems at odds with the implications of the newspaper story.

But really, these points are relatively minor compared with something else that is apparent from the newspaper article and abstract. The experimental group went through an "intensive programme of enriched diet, exercise and cognitive stimulation, which included being read to and involved in conversation," according to the newspaper article.

We already know that there is a strong link between good socialization and lowered criminality. We also know there is a strong link between higher education levels and lower criminality. So why would you claim that the main reason for lowered criminality in the study was the fish?

The situation is a little like hearing "listening to Bach makes you a better runner" and then finding out that the experimental group were provided with "Bach listening sessions and running coaching".

Now it is possible that the fish played a role. That is because there is a small amount of evidence suggesting that eating fish helps learning, etc. So perhaps eating fish made the children better able to learn and therefore less likely to be criminal. But we expect this only to be really significant where malnutrition is a problem, and so we confront the generalizability problem again.

From reading the abstract, the authors do not make strong claims that it is the fish that is key, however the following quote appears in the newspaper article from one of the authors of the research paper:

'Could it be the exposure to increased omega 3 fatty acids, which we know are the building blocks of cell membranes, leads to better brain function which we did discern at age 11 - and better outcomes at 23?' said Raine.

The way the newspaper article is written implies that the author is suggesting the fish diet is responsible for the reduced criminality. I am not convinced that the researchers would support such an implication. What this does tell us is that Raine is probably aware of the intermediate step in this process from fish diet to better learning to lower criminality. That being the case, why design an experiment that appears to cut out exploration of the middle step and confound the matter so heavily?

One reponse to this whole issue is to just say that the study doesn't really say much that is interesting that we didn't already know. However, the way it is reported has very serious ramifications.

Notice that the newspaper article is written by a political reporter rather than the science reporter. This already suggests something about the way this research is seen in the wider community. Then look at the subtitle of the article: "New study reveals that the root cause of crime may be biological, not social."

Does this study really support the idea that the cause of crime is biological? Absolutely not, except in the sense that amount of learning, etc. is biological. It certainly doesn't support the fairly widely discredited theories that criminality is dependant on intrinsic biological properties such as genetics, but those are the types of ideas that people usually associate with "biological causes of crime".

The topic is important because if biological causes of crime become strongly accepted, we suddenly have a situation where defendants claim that they have some biological imperative to be criminals and so the crime is, in some sense, not their fault. This argument is already doing the rounds in US courts and is generally dismissed, although not sufficiently often (remember the "Twinkie defense"?). It also gives excuses to avoid dealing with crime and criminality in other ways (ways that tend to be politically difficult or expensive). If we are going to start claiming there is a biological imperative for crime and making policy based on those claims, we had better have some good evidence. This paper is not that evidence.

All in all, the newspaper report does not seem to fairly represent what was involved in the research (but when did a political correspondent ever get a science story right?) and the science itself does not seem to make any supportable extraordinary claims that are new.

To conclude, perhaps it worth repeating what one criminologist commented to me about this paper and news report: "It is this sort of thing that gives psychology a bad name."

New Virtual Occoquan

The latest issue of the Virtual Occoquan is out and includes something of mine...

© Copyright 2006 David Harris. Click here to send an email to the editor of this weblog.
Last update: 1/25/2006; 2:49:16 PM.
Powered by