iWitness Fraud

02 12, 2013 by LMOGA

The Louisiana Bucket Brigade released its January iWitness Map Monthly Report last week, presenting a host of figures it claims accurately characterizes the frequency of oil and natural gas industry-related incidents across the state. The report presents data from two sources: the National Response Center (NRC), a federal communications center that records and responds to reports of hazardous spills; and the Bucket Brigade’s own iWitness Pollution map, which is the group’s homemade, self-populated, unverifiable database for soliciting and tracking “eyewitnesses reports of accidents at oil refineries and chemical plants.”

Each report claims to derive its numbers from the accident reports included in one of those two databases. According to this latest report, the total for January 2013 is 273 accident reports, 183 of which the Bucket Brigade identifies through its own analysis of the NRC database -- and 90 of which were reported via the “Submit a Report” function of the iWitness Pollution Map.

Answers to a number of very basic questions – such as: how “average citizens” can be “eye witnesses” to so-called accidents occurring inside a refining facility – are unfortunately beyond the scope of these reports. But the red flags don’t end there.

For example: Tallying the individual figures listed under “All Reports” yields a total of 185. But the figure listed as the total number at the top of the report is 273. So what’s with this discrepancy? Is one wrong to assume that “All Reports” and “Total Reports” are the same thing? Regrettably, it’s impossible to tell.

Giving the group the benefit of the doubt, 185 is closer to the NRC Report total of 183, so perhaps those two data sets should be synonymous. Even so, the numbers still don’t add up.

What about utilizing the group’s own search function on the online iWitness Pollution Map, setting a date of Jan. 1, 2013-Jan. 31, 2013, and searching for the total number of reports? That search generates 278 reports, not 273. How is one version of the January 2013 report undercounting by five accidents, or vice versa? Again, because no methodology is given, it remains a mystery.

iwitness 1
Unfortunately, things go downhill even from there, particularly when focusing on the individual categories of “Citizen Reports.” Leaving aside unanswered questions about what is needed for a citizen report to be considered “verified” as opposed to “unverified,” and whether or not text messages and email reports, for example, are filtered to eliminate double counting -- the more glaring unanswered question is; How can the figures in the individual categories add up to 94 reports when the top of the document indicates that there were 90?
Moving down the columns sorting-out the facilities suspected of being the sources of several accidents in January doesn’t provide much relief. The Bucket Brigade doesn’t share how it arrived at any of its figures, including the total of 183 NRC reports. So we dug up our own data from the NRC site (found HERE) and determined that, while the NRC reports for Norco and Shreveport are accurate (one and zero reports, respectively), the reports for both the ExxonMobil Baton Rouge and Chalmette refineries are inflated by at least a factor of two in both instances. Take a look for yourself:

You don’t need a PhD in statistics or access to high-powered data processing software to see that virtually none of the Bucket Brigade’s figures actually add up. As we’ve shown above, all you need is a calculator.

Of course, if the Bucket Brigade were genuinely interested in informing the discussion with facts and science and valid insights, they’d share their process methodology and commit themselves to being just as transparent as they routinely demand of the companies they target. Unfortunately, as the group continues to show us: the ends (doing whatever they can to attack job-creators all around the state) always seem to justify the means (distorting the data, and sometimes just plain making it up).