Beyond Cows: Another Look at PolitiFact Wisconsin

Earlier I looked at allegations of so-called conservative bias at PolitiFact Wisconsin by one Jed Lounsbury of the website Uppity Wisconsin; now I’ve just happened upon a blogger here on Blogspot called “XbeyondX” who has posted some very detailed, cogent arguments almost worthy of somebody’s “Sith Blogroll” on several of PolitiFact Wisconsin’s rulings, including one that Jed Lounsbury expounded upon, again alleging conservative bias. Only this time, this blogger may really be onto something.

For “proof” of the Milwaukee Journal Sentinel’s source of right-wing bias, “XbeyondX” provides some interesting information:

“…I am starting to wonder if we can add Milwaukee's media monopoly Journal Communications to the growing list of blatant conservative ideologues. Yes, this is the corporate parent of a radio station that interrupts its constant barrage of conservative talking points only for occasional sports.”

XbeyondX had just rated a PolitiFact (PF) Wisconsin fact check with a “Mostly True” on Wisconsin’s Republican Attorney General J. B. Van Hollen, who stated that “more than 22,000 IP addresses in Wisconsin have downloaded child pornography.” XBeyondX’s argument was that PF’s methodology—by extrapolating an estimate of the number by looking at other states—was flawed because:

To merely multiply the number of IP addresses identified during a 1 month period to conclude the number would be 9 times larger if the analysis continued for a 9 month period completely ignores the fact that one IP address may be downloading child pornography throughout that period.

The key word here--which XBeyondX appears to ignore or doesn’t see--is unique: Van Hollen’s statement as reported by PolitiFact Wisconsin (as written by Tom Kertscher) was “Over 22,000 unique Internet Protocol addresses -- in Wisconsin alone -- have downloaded child pornography." The statement in the “deck,” however, does not include the word unique (which happens frequently, as noted by one of my conservative blog counterparts). The investigation from which Van Hollen drew the information for his statement, from the Wisconsin Department of Justice in November of 2008, uses the word unique as well, and adds that this is a conservative estimate: (emphasis added)

During this time frame there were 22,304 unique IP addresses that were identified as attempting to download child pornography. This is believed to be a conservative estimate that only shows the minimum size of the problem. For example, the map does not include IP addresses that were identified but could not be mapped. It is possible that some IP addresses may come back to the same offender.

However, there is a problem in Kertscher’s attempts to extrapolate this number when comparing to the Indiana and Illinois stats in regard to the “unique” qualification: only Indiana uses the “unique” descriptor for the IP addresses it uncovered downloading porn. The national numbers seem to be “non”-unique as well. So Indiana can be considered more reliable comparison-wise than Illinois or nationally.



"Unique" is there, but numbers is not spelled numebrs.

I’m not a computer hack, but based on my limited understanding (when I did database analysis, it meant "no duplicates"), if the IP addresses are unique, the occurrences of that address being used again have been taken into account for the 9.5-month period of time which Van Hollen said he was referring to when he made his statement. Kertscher may have erred by making a comparison of the number of addresses found with Illinois, since there’s no verification of whether the IP addresses it turned up were unique. XBeyondX’s other comment about it being the same offender is not relevant here, although it might have been well for Kertscher to address that point.

So what we have here is that XBeyondX made a good argument that didn’t sound like a partisan hack but missed a crucial descriptor (“unique”) that would have made the difference in what he was arguing. PolitiFact’s link label used the word “unique” although it is possible XBeyondX overlooked it. But PolitiFact should be “downgraded” for this shoddy deed on the mismatched title/deck as well as not determining and reporting whether the comparison states also had unique addresses. Given all this, should it change the ruling of Mostly True? Assuming I am correct about the “unique” unless XBeyondX has some other plausible explanation, I’d still say Mostly True is probably a fair ruling.

That was one of the four PolitiFact Wisconsin rulings examined by XBeyondX and found to be wanting; I might post a review on two of them later, but for this post I must look at one other ruling because it matches one reviewed by Jed Lounsbury: that of a ruling of Pants on Fire on a statement by now-former Wisconsin Senator Russ Feingold that he’s been “been outspent by my opponents every time I've run for U.S. Senate.” A quick review of Jed’s arguments led me to this conclusion:

What PF was faulting Feingold for was stating that he was outspent by ALL his opponents, and totaling them together would make for, as PolitiFact notes, a very convenient argument, because “tallying up spending by Republican primary candidates he never faced to obscure that he vastly outspent the one he did.” In other words, it’s an argument ridiculous on its face. Yet Jed expects us to believe just that. Feingold, in my opinion, may not deserve a Pants on Fire ruling, but it was False nonetheless.

XBeyondX finds the devil in the details as to the reason that what the Republicans spent in the primaries should count for something and should not be considered so ridiculous:

PolitiFact is correct that a reasonable voter would not suspect that Feingold was considering the money spent by individuals who failed to earn the Republican nomination because these candidates were not technically Feingold's opponent in any race. Therefore it would be inappropriate to rate this claim as "True." However, anyone who remembers the nature of the last election will recall that Feingold was a target of each candidate in the Republican primary. Each candidate made attacking Feingold a central part of their primary campaign and thus, to wholly dismiss these candidates as opponents of Feingold's is too simplistic. It is certainly not such an indefensible argument that it should earn the "Pants on Fire" label. This statement contains some element of truth but ignores critical facts that would give a different impression and therefore should have been appropriately rated "Barely True."

Senator Feingold was a major target for unseating by the Republicans in 2010, so XBeyondX’s analogy reasonably fits. (And it obviously worked for the Republicans!) I must commend “X” too, for carefully framing his arguments with PolitiFact’s ruling parameters, that “this statement contains some element of truth but ignores critical facts that would give a different impression.” PolitiFact showed that Feingold was outspent in the previous two races, and as “X” said “therefore Feingold's statement is only 2/3 true.” Yet Feingold was ruled Pants on Fire. This has given me some second thoughts and I find I must renege on the possibly “False” and concur with the "Barely True".

XBarelyX posted a great cartoon in his last post regarding PolitiFact. And he wrote something on PolitiFact’s right-wing bias that I’ve heard similarly somewhere before in the opposite direction:

PolitiFact [Wisconsin] is an embarrassment to the PolitiFact name and a profound disservice to Wisconsin politics and most importantly Wisconsin voters. If there was a McDonald's franchise so regularly dishing out such a crappy product, the corporate parent would immediately intervene and revoke the franchise in an effort to preserve the company's good name. But here, the parent PolitiFact organization seems content to let the Milwaukee Journal Sentinel's shoddy and biased work tarnish what used to be the good name of PolitiFact.

Poor old PolitiFact. Damned if they do and damned if they don’t.

XBeyondX's cartoon....loved it!