I am always suspicious when activists present simple figures to confirm their bias and fool their audience. I think anti-fluoride activists do this a lot. Here is an example in Paul Connett’s presentation to the recent Sydney anti-fluoride conference.
Connett uses data from Xiang et al (2003) and some of Xiang’s other papers and presentation to push his claim that fluoridation is bad for your IQ. Apparently he and Bill Hirzy (currently described as Fluoride Action Networks “chemist in residence”) are working on a paper attempting to justify a case that the maximum permissible level of fluoride in drinking water should be reduced to practically zero! They use a simplification of data from Xiang’s paper for this.
First of all the figure shows what Xiang’s data is like. It compares IQ with urine fluoride concentration – unfortunately he did not give a similar figure for fluoride concentration in drinking water. However, this is well correlated with urine fluoride.
There is certainly a lot of scatter, but Xiang (2003) reports a “Pearson correlation coefficient of 0.174, p=0.003.” So a statistically significant relationship (helped by having a large number of samples) but it still only explains about 3% of the variance!
This is important because, although Xiang did consider some confounding factors he could well have missed a factor which explains more of the variance which, when considered, may make the relationship with serum fluoride concentration non-significant. For example, I would be interested to see a statistical analysis which included incidence of moderate and severe dental fluorosis as this may be more important than the drinking water fluoride concentration itself).
But have a look at how Paul Connett present this data (or the equivalent data for drinking water fluoride concentration) in his Sydney anti-fluoride conference presentation.
The “trick” has been to divide the data into “categories” based on inclusion in a separate water fluoride concentration ranges and then presenting only the averages within each category. I can see the point of sometimes using such categories, but this figure conveys a very misleading message.
The Sydney audience could have been excused for thinking that Xiang’s data showed a very strong connection between IQ and drinking water fluoride – a relationship explaining almost all the variance. Completely misleading as this relationship probably only explains only about 3% of the variance in the original data.
Paul Connett and William Hirzy are currently campaigning to make IQ the key factor for determining the maximum permissible levels of fluoride in drinking water. They might confuse a few politicians with these sort of distortions but hopefully the real decision-makers will be awake to such tricks.It really is
It really is a matter of “the reader beware.” Never take on trust what these political activists are saying. Always go to the original sources and consider them critically and intelligently.