Flint Water: A Political Football. Part 10

I looked at Marc Edwards' numbers and did some calculatin'

Source

Here is what the Virginia Tech Research Team tell us:

Over the weekend, we analyzed all samples shipped to Virginia Tech from Flint to date. Flint residents have already returned an astonishing 84% of the sample kits we sent out (252 out of 300 samples). 

The protocol they used can be found here:

Students and scientists (see About Us) at Virginia Tech assembled and shipped 300 kits for Flint residents to sample their water and get it analyzed for lead. The kits comprise of three bottles (1L, 500 mL and 250 mL ones) and an instruction sheet for sampling tap water. All 300 kits were sent out August 14th (Friday) and reached Flint on August 19th (Wednesday). 

The Excel file I downloaded,  Flint-Samples-FINAL.xlsx, contains 300 sample points (I am kind of confused as to why this number does not match with the statement above) from which three samples were collected at each point:

• Pb Bottle 1 (ppb) - First Draw
• Pb Bottle 2 (ppb) - 45 secs flushing
• Pb Bottle 3 (ppb) - 2 mins flushing

The average lead concentration reported on the sheet for the first draw is 10.646 ppb.  I took all these sample results and recalculated using Excel's statistical function and I got the same number.

I then looked to see what the largest number reported was for each of the three sample columns. With this many sample points you can see a trend.  Since these samples were collected by the homeowners, the possibility of human error in sampling bottle identification is high.

That being the case, I felt that three samples were incorrectly identified, so I removed them from the rest of the data points.  This left 297 sample points from which to run some calculations.

What I wanted to see was if there was a difference between the first draw and a flush.  The problem with this data as presented was that sample 97 reported results in this order:

1. 7.244
2. 1051
3. 1.328

It is possible that 1051 ppb is a legit number, but it appears doubtful that the 45 second flush would produce such a high number when the first draw was only 7.2. Additionally, this one numbers skews the results for the 45 second draw too much to make a comparison between first draw and flushes appropriate.

Removing samples from the data is a judgement call, and I think I can justify removing them in an effort to get a truer picture of what is going on.  I also removed samples 258 and 24 because I think they are in the wrong order - their removal changes the average but still shows a drop in lead when a flush was used.

So here are my results.  For 297 sample points:

As you can see, there is a reduction in the amount of lead found in the 45 second sample and even less in the 2 minute flush sample.  This supports the contention that the flushing requirement Flint was having the homeowners perform to meet the Lead and Copper Rule was most likely under reporting the lead.

What's important here is to not get caught up in the averages. This data came from a mix of homes in Flint.

Flint’s non-profit group Water You Fighting For? is helping coordinate distribution of these kits to volunteering homes 

There are, as the Williams' sample results showed - when compared to their neighbors -, houses within Flint that are more impacted then others.  

There is overwhelming evidence that there is lead present in Flint's water and it is higher in lead than water produced by Detroit.

Response:  We coordinated a citizen sampling event of homes in Detroit during 2010, and confirmed that Detroit water met Federal Standards—Flint’s water does not. 

What is more telling as to the extent of the problem is the spike in elevated blood-lead levels that I wrote about in my last post.  Exposure is problematic, uptake is worst when you know it has happened, but proof of an elevated level found children under five...well that's pretty damning evidence that you have a problem.

So what can we take away from all of this?


Next post: Flint Water: A Political Football.  Part 11