I tried to stay away from Conservapedia, but I just can’t. Despite what I said in my previous post, nobody on the web does stupid like Conservapedia. And I’m not talking about your everday, vanilla kind of stupid. No, this is spicy, arrogant stupid. It’s the stupid that comes from people who are too stupid to know they’re stupid, and my latest Conservapedia discovery serves as a perfect example of this arrogant ignorance. Stumbling around the site earlier, I eventually came across a link to the Conservapedia article on Richard Lenski, so I decided to take a peek and see if they’d found any new bullshit to throw at the man or his work. Well, I was pleasantly surprised to see that they had.

The bullshit in question has been hurled at the statistical analysis used in Lenski’s inaugural PNAS paper, and it’s particularly ripe.  It opens with a heaping pile of fail:

Blount, Borland, and Lenski^[1] claimed that a key evolutionary innovation was observed during a laboratory experiment. That claim is false. The claim was based on incorrect measurements of statistical significance.

The person who wrote those words understands approximately shit about Lenski’s paper.  The claim of an “evolutionary innovation”, specifically the evolution of aerobic citrate utilization in a population of E. coli, was not based on any statistical argument.  It was based on the direct observation of the Cit+ phenotype.  Statistical methods were used in the paper to evaluate the hypothesis of evolutionary contingency, not to support the observation of aerobic citrate metabolism.  So right off the bat it’s clearer than Crystal Pepsi that the dildo who wrote this exercise in creationist self-delusion doesn’t know the first thing about the paper he/she claims to be critiquing.

As we enter the body of the article, things get even worse.  The author tells us that, “The null hypothesis from the paper was constant mutation rate over all generations”, which is actually true-ish.  The null hypothesis for Lenski’s Monte Carlo resampling test was that the emergence of the Cit+ phenotype was due to a single, rare mutation.  Under this hypothesis, Cit+ was equally likely to emerge in any given generation.  But how in the hell does this statement square with the article’s introduction, which implies that Lenski failed to demonstrate the evolution of aerobic citrate metabolism in E. coli?  I mean, if the introduction is correct, then there are no experimental data and, therefore, no need for the statistical analysis that Conservapedia is trying to rebut.  I’ve tried to figure out what the Conservapedia author thinks the statistical analysis was used for in Lenki’s paper, but I can’t.  Whatever it is, though, it’s clearly wrong as hell.

Anyway, let’s take a look at Conservapedia’s criticism of Lenski’s statistics.  The claim is that Monte Carlo resampling was the wrong statistical tool to use, one that gave artificially low p values.  They base this claim on the fact that a chi-square test for independence gives very large p values, indicating that Lenski’s data is not statistically significant.  That is, quite simply, one of the dumbest things I’ve ever seen in print.  The chi-square test that Conservapedia uses is the chitest function of Excel.  Chitest uses the following equation to determine chi-square values:

Here A_ij is the experimental value in the i^th row, j^th column, and E_ij is the expected value in the i^th row, j^th column.  Since this is a summation, it is clear that the order of the rows of data doesn’t matter.  As an example, here is the Conservapedia presentation of the data from one of Lenski’s replay experiments:

As Conservapedia says, chitest gives a p value of 0.19 for this data.  But what if we rearrange the order of the experimental data?

Here, chitest gives us the exact same p value of 0.19, despite the fact that this arrangement of the data more clearly supports the hypothesis that later generations are more likely to produce Cit+ mutants than earlier generations.  If the chi-square test is an appropriate tool in this situation, then it should return, at the least, a smaller p value for the second table than it does for the first.  But it doesn’t.  In fact, as we would expect from the equation, it returns the same p value for any possible arrangement of the experimental data.  It is totally insensitive to when in time the mutations occur, making it an absolutely useless method for determining if Lenski’s data significantly support the historical contingency hypothesis over the hypothesis of a single, rare mutation in the origin of the Cit+ phenotype.  This would be clear to anyone with i) a functioning cerebral cortex, and ii) a trivial understanding of Lenski’s paper.  Unfortunately, our Conservapedian likely lacks both of those.  And, like everyone else who views Conservapedia in any sort of favorable light, they also lack the self-awareness necessary to properly recognize their own shortcomings, which is why they are writing about things that they don’t understand.  So congratulations, Conservapedia.  Once again you’ve failed spectacularly, and once again you are too stupid to realize it.  You truly are THE preeminent repository of stupid on the internet.  All others…take heed.

This entry was posted on Tuesday, April 28th, 2009 at 8:59 pm and is filed under Asshats, Creationism. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.