Last year I picked up on a story about alleged poor water quality in the United States. This centred around a report about hexavalent chromium levels in tap water. I wasn’t particularly impressed when I saw how low the figures being quoted were. But I did point out that there might be an issue because of the variability of the data.
What do I mean? Well it is often the case that the variability of data is telling you something that the absolute values aren’t. If you want are buying clothes to stock a clothes shop for example, you need to know not just how big people are but also how wide the variation is. But it goes a lot further than that. High variability in data is something that needs to be explained – or at the very least tells you something about what is going on.
Well it turns out that there was an explanation for the variability in the data. The people doing the survey had mixed up treated and untreated water. Not surprisingly, this gave a bigger range of data than they would have got if they had just tested tap water.
As I originally thought, the variability did indeed point to a hidden problem. The problem being that the pressure group who created the story are half wits. Their name is the Environmental Working Group and as a not for profit organisation they don’t pay any taxes and rely on contributions. There aren’t any regulations governing what groups like this are allowed to say, and this particular group seem to take this as a green light to say anything that gets them a quick headline.
I had originally linked to a site that gave the full details of EWG’s poor sampling of hexavalent chromium. I’m afraid the link has now vanished, but basically it was a water engineer explaining patiently that you can’t work out what is in drinking water from samples taken from water before it had been treated.