Fracking and Arsenic

Katie Colaneri recently blogged on the NPR website with the headline New Study Examines Link Between Fracking and Arsenic Contamination.

Her post begins with commentary on a ProPublica piece New Study Finds High Levels of Arsenic in Groundwater Near Fracking Sites by Theodoric Meyer.

These articles refers to a paper with the lengthy title An evaluation of water quality in private drinking water wells near natural gas extraction sites in the Barnett Shale Formation by researchers at the University of Texas at Arlington.

The paper's abstract claims:

Lower levels of arsenic ... were detected at reference sites outside the Barnett Shale region as well as sites within the Barnett Shale region located more than 3 km from active natural gas wells.

Which is not surprising, because concentrations of heavy metals in groundwater tend to differ between geological regions. Not to mention, these reference sites are on top of a completely different aquifer (Nacatocha) than the fracking-adjacent wells (Trinity and Woodbine), as described on page 5.

Also, there were only 5 reference sites measured in the study. Statistical significance can't be satisfied with such a small sample size. The authors even admit this:

...our sample size is too small to make definitive conclusion. page 11

Anyway, by comparing apples to oranges (i.e. not controlling for geology/aquifer), the authors fail to address the:

...potential impact of natural gas extraction activities on groundwater quality in aquifers overlying the Barnett Shale ... page 6

Which they claim is the goal of the paper! The observed variation in arsenic (and other heavy metals) is easily explained by the lack of a control for geology and aquifer

The authors should have looked for changes in each well over time. In fact, this information is available from the Texas Water Development Board. I know the authors are aware of this data source, because they used it in further, poorer analysis:

Concentrations were significantly higher in active extraction areas compared to reference samples and historical samples. page 9

Hey, they're comparing apples to oranges again! They have averaged data from all wells in all counties. Then, they compared this to averaged data from fracking-adjacent wells. That's stupid! Why not find the changes on a per-well basis, and see if there is a statistically significant result?

My guess is, they tried that, and it didn't work, so they resorted to unjustified grouping by country. Maye they even did a bit of hand-picking in order to come up with their desired outcome?

But here's the most astounding part of this study. Take a look at this graph:

(Red marks are mine.)

This graph is accompanied by no statistical analysis, whatsoever! This is how you know they're liars. They went to the trouble of graphing distance vs. arsenic concentration, but buried it on page 26 (of 28), and didn't perform a correlation analysis, which is necessary to achieve one of the three goals of the paper:

... evaluating the relationship between water quality and geographic proximity to natural gas extraction activities. page 6

The data didn't fit their preconceptions, so they just skipped the statistical analysis. Wow.

The mountain of problems like bogus historical data, faulty methods, cherry picking facts, hiding statistics, lack of proper controls, and small sample sizes force me to conclude one thing.

This paper is junk science.

If the Katie Colaneri's headline were true, we'd see these red-circled points lying somewhere along the red line. So, as it happens, no such link exists between arsenic levels and proximity to fracking wells!

The high levels of arsenic in these people's wells is real, but natural, caused by the local geology.

At least Theodoric Meyer's headline is only misleading. It implies some link between "high levels of arsenic" and the nearby fracking sites, without coming out and saying it.

Katie Colaneri's headline is just a lie.