Science abuse: Omega-6 drives homicide rate

I am currently reading the Perfect Health Diet book, and whilst I do overall like it, I am taken aback by some of the “science” that is cited. Which is the shame – there is a lot of science cited in there, and most of it sounds reasonable to the untrained eye, so bloopers like this one are really not necessary, because they undermine the confidence. For the record, my impression so far of the book is very good. I do perceive a tendency though to overplay his hand and in particular to present his 20/15/65 ratio as the only way to go. I would think that there is a bit more variation…What I am referring to is the following figure

With the caption: Homicide mortality versus consumption of omega-6 linoleic acid, plotted for 1961, 1970, 1980, 1990, and 2000 in five countries. Within each country, homicide rates go up as linoleic acid consumption goes up.

I am not sure how an analysis like this can get past any peer review, but to be fair, maybe the message is a bit taken out of context and maybe the underlying analysis by Joseph Hibbeln, Levi Nieminin, and William Lands was a bit more differentiated on the findings, and/or had more data to back it up.

Now before I go on, what I am refuting is the implicit statement “Omega-6 drives homicide rate, and this graph supports this analysis in a meaningful way”. I am not refuting the first part of the statement – I have no idea – and I am not saying that the graph is unrelated. Of course the explanation could be the one given by the statement implied. But so could be hundreds of incompatible one’s, all of equal likelihood and credibility.

They key point of course is that since the 60’s homicide rates have essentially gone up in a reasonably steady manner in all the countries I have checked (UK, Argentina, US) and I am venturing a wild guess that it is the case for the other countries in this analysis too (as for most other countries in the world – homicide rates have just generally gone up). And of course Omega-6 consumption has gone up worldwide. So if two quantities go up at the same time, they will be correlated.

I could go on forever, but just a couple of points why this graph just does not make sense

  • You can take ANY other parameter that has increase over the years, and you get the same correlation (try speed of microprocessors, number of computers per household, average income in nominal terms, income disparity, whatever); there will be a bonus of course if the number you choose is higher in the US to start with
  • It is ludicrous to do any kind of analysis with homicide rates and to not control for availability of firearms
  • The graph is extremely driven by the fact that US is high on both, and the other one’s arent, and so is the whole analysis; this makes it even more ludicrous: on this basis, samba drives homicide rates (if you look at South American countries you’ll probably find that Brazil is high)
  • Having seen in the point before that cross-country data does not really add much, we should look at time series at one country. In the US, why is there a rate of 6 and of 10 associated with the same level of Omega-6
Advertisements

3 thoughts on “Science abuse: Omega-6 drives homicide rate

  1. Well actually, the data proves it. I have done some more research on it, and apparently the correlation between homicide rates and samba is high in Brazil even during the year, ie when people are dancing more Samba (for some reason, samba dancing peaks every year, but on a unpredictable schedule) then homicide rates move up at the same time.

    helau!

  2. Pingback: Perfect Health Diet – a quick review | Thor Falk

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s