Sidebar: It's Gettin' Uglier Down Under

I’ve hardly said anything about a recent “holiday” (vacation) I took with a friend to merry old England and the rest of the British Isles—that is, Wales, Scotland and Ireland. It was one of the top items on my “Bucket List.” Recent events, however, bring to mind something that happened when we were a bus load of people from all over the world stuck together, sharing a common bond for almost two weeks traveling the British/Irish countryside.

The national makeup of our group was mostly Australian, with a few Americans and Canadians thrown in. My friend and I got to calling them the Ugly Australians, like the Ugly Americans. It started with my friend who had an experience where a very large Australian fellow traveler physically pushed then clumsily pulled her out of a lift (elevator) along with her luggage because he wanted to accompany his wife who was already in it. He didn’t ask her, he just did it. He totally ignored her, unapologetic, as she fell over her luggage. For the frequent times he saw us later during the trip, he acted like it never happened. It briefly paralyzed my friend’s enjoyment of the trip in her angst and anger toward this man; as we experienced some other questionable conduct we came to appreciate the appropriateness of that nasty epithet about arrogant American tourists, only this time it wasn’t us. It surprised me because before I retired I worked with Australians in my position and found them to be warm, friendly and self-effacing.

Suddenly now more meaning has been brought to that expression the Ugly Australian with the exposure of the corrupt business practices of Rupert “Fox News” Murdoch:

Reporters at a Murdoch-owned newspaper in Britain hacked into a 13-year-old kidnapping victim's voicemail to get tabloid gossip, even deleting messages to make room for new messages. Their actions led her parents and police to believe she was alive. But she wasn't—she'd been murdered.

It appears Murdoch's reporters regularly hacked into the cell phone accounts of prominent politicians, celebrities, and crime victims. And news just broke that they may have even hacked the phones of 9/11 victims.  (from an e-mail by Daniel Mintz of Moveon.org)

Our experience with the Australian tourists may have just been something anecdotal that could be chalked up to that occasional bad traveling episode. But Rupert? In a way, we see what greed can do through him—and I don’t care what those Ayn-Randian Gordon Gekko disciples say about the “good” of greed—in the greater scheme of things, it’s not just ultimately nasty, sleazy, corrupting and destructive….it’s ugly.

Lil White Lies: Take These Net Jobs and Shove 'em

Jobs statistics are a sticky, complex topic for PolitiFact: as noted by PolitiFact Ohio writer Mark Naymik in this recent PolitiFact (PF) ruling on Joe Biden: “Politifact also frequently points [out] that when framing job figures, slicing and timing make all the difference.”

And wouldn’t you know, PF even knows quite well what selection bias is. Only it calls it for what it really is: cherry picking: (emphasis added)

Take, for example, some other accurate statistics we can create using the same figures.

If job growth continues on pace in 2010, employment still will have decreased by 1.6 million jobs since Obama took office in 2009.

While Obama is on a pace to add 2.356 million jobs in 2010, Bush created 2.36 million jobs in 2005 (a specific year to year comparison).

And from April 2003 to June 2007, while Bush was president, the U.S. economy added almost 7.9 million jobs.

All true. And all in seeming contradiction to Wasserman Schultz's claim.

Why?

Because we cherry-picked numbers to make our point.

But the ruling that is the subject of this Lil White Lies post is not about Joe Biden or Debbie Wasserman Schultz.  It's about a claim by Virginia Governor Bob McDonnell about Texas Governor Rick Perry's "job creation performance" in an op-ed piece he wrote for the Wall Street Journal.

My conservative counterpart is correct in how PolitiFact approaches job numbers claims: “PolitiFact's history of (appropriately) treating job creation claims as dealing with net job creation where the context warrants taking them that way.” So, in that same track, is he also correct when he says that PolitiFact failed to note that Rick Perry “did not mention the term "net" while talking about job gains?

Here is the complete quote from McDonnell:

Look no further than growing and diverse Texas, a state Democrats targeted in 2010, where data from the Bureau of Labor Statistics reveal that under Gov. Rick Perry’s leadership the state has created more jobs over the last decade than the rest of the states combined. That’s a record of job creation the Obama administration can only dream about.

There is no mention of this “net” by McDonnell throughout his op-ed piece extolling the great things all the new Republican governors have been doing for their states. For example, he says about his own, Virginia: “Since February 2010, 67,400 new jobs have been created in Virginia and our unemployment rate has fallen to 6% from 7.2%.” He says about Wisconsin: “In six months, 25,000 private-sector jobs have been created.”

PolitiFact Truth Index Rankings

It’s finally time to look at individual Truth Index ratings and do something that I don’t know is available on PolitiFact’s mobile app: ranking them in order of high to low for the Truth Index scores. A few qualifications are in order, however.

As noted in my Politi-Score description, you can’t do much when politician “Joe Sixpack” has only two rulings from PolitiFact. It means very little. But the ongoing aggregation of rulings is starting to provide a more substantive record for every type of politico. The last time I did this ranking, I set my ruling minimum at 10, because if I set it higher, I would not have had much to list. Now, however, I am pleased to say I can bring that minimum up to 15 rulings or more, and have an end result of a ranked list of 25. (Click here to see a list of those with 10 or more rulings done with accumulated rulings through the end of 2010.)

Certain people were removed with 15 or more rulings, who were those no longer active in politics, such as Charlie Crist and John Edwards. I kept Hillary Clinton, since she is still around as Secretary of State, even though she’s had no rulings in the last six months.

With that said, below is the bar chart which shows the ranking using the Truth Index report card color coding that is on the PolitiFact mobile app. You can see that it’s hard to tell how they compare with the colors, though the ranking gives it more meaning. For example, in terms of the Truth Index, Republican Ohio Governor John Kasich has never said anything that merited a Pants on Fire ruling, yet he is below the zero or Half True marker at Number 13 on the list, below fellow Republican Florida Governor Rick Scott, who hit the Half True on the nose (i.e., he scored exactly zero with the Truth Index—there is an arrow marking there to indicate the “Half True” baseline), and has had three Pants on Fire rulings.

Click to enlarge

This group of 25 individuals/groups ranked represent just over a third of all PolitiFact rulings since they began doing them up through the end of June. The collective Truth Index for this group, FYI, is (9.29).

As in-office politicians go, however, Michele Bachmann bottoms out the list. And yes, according to the Truth Index, Bill Adair is (technically!) correct in saying that she is only exceeded in being the biggest “liar” by chain e-mails. As I’ve noted before the Truth Index is heavily weighted to the False/Pants on Fire side. She needs at least three Trues to offset two Pants on Fire…don’t think that’s going to happen even with continual accusations of selection bias.

Below is the list from which I created the chart, giving the exact numbers.  While not as “pretty to look at” it does give further clarification to the bar chart.

Subjected to the Truth Index

The new PolitiFact (PF) app for mobile devices supposedly has a feature which shows the Truth Index by subject or topic. I started keeping track of the subject or topic in the first quarter of 2011, and published the overall Politi-scores, just to see what subject was being misrepresented the most. I’m assuming that PolitiFact is doing it this way as well; so now I’m going to take it one step further here.

To begin as in my previous post from April, here is the updated breakdown of subjects in a pie chart for the first half of 2011. I decided to do the Top 15, in order to make the chart less confusing.

Instead of showing “The Most Lied-About Subjects by Politicians” as I did before, I’m going to make another division:  “The Most Lied-About Subjects by Political Party.”  I decided to do the Top Ten subjects, since making this division would make it the Top Twenty since I’m reporting by party.  Here they are top "lies" to the most truthful:  Healthcare makes it to the top again, only this time it’s when Republicans are talking about it.  The Democrats made Labor/Unions the second biggest lie, even topping out the Republicans.

The Truth Index base value is zero for Half True, as is No. 16 Education by Democrats.

PolitiFact’s Truth Index by subject on its mobile app may have one important superior feature to mine in that (I’m assuming) the index is calculated as a percent of ALL PolitiFact rulings since they began doing them, which would be 3,774. Mine is only based on 986, the number of rulings so far in 2011. In addition, many of the rulings had multiple subjects (as tags) so it’s possible their numbers are based on the total number of subjects assigned rather than rulings…which would create some overlap if you’re doing a Truth Index, I’d think. It seems to me doing it this way would disconnect it from the rulings.

Next up is an update on individual rulings using the Truth Index.

Visiting More PolitiFact Writers with the Truth Index

About 14 writers do the bulk of PolitiFact’s rulings (roughly 70% of them). Generally there is one “lead” writer at each PolitiFact (PF) state, so I’m assuming the other writers are interns, part time or have other functions. For example, at PF National, Louis Jacobson writes about twice as many rulings as the other two writers, Robert Farley and Angie Drobnic Holan. Holan is the moderator at the Facebook page, so that’s probably one of her other “functions” while Jacobson concentrates on writing rulings. PolitiFact Ohio seems to divide everything somewhat equitably among four writers except for (the fifth) Thomas Feran, who appears to be the lead writer like Jacobson. In Wisconsin, Dave Umhoefer and Tom Kertscher are about in an even partnership with writing rulings, while James Nelson does fewer rulings and handles the Wisconsin Flip-o-meter.

Previously I focused on the three PolitiFact National writers; but here’s a better look at all the writers now that there’s a sufficient number of rulings to do a Politi-Score Truth Index applied to Democrat versus Republican rulings. All of these writers have done 20 rulings or more. The average is 48 rulings: Louis Jacobson of PolitiFact National was at the front with 115; Ryan Kost of PolitiFact Oregon was last with 20. I also wanted to include ALL PolitiFact states which necessitated the minimum of 20 because Kost had the highest number of rulings (20) written for PolitiFact Oregon.

Here are the writers as listed in an Excel spreadsheet from highest variance between Republican and Democrat Truth Index scores, followed by a graphical depiction.

Gene "Jim Morrison" Emery seems to light his fire with Republicans more than Democrats.

In the graph, I’ve placed an arrow to point at the Truth Index scores of James Nelson of PolitiFact Wisconsin: he seems to be the “culprit” as to why I’ve called Wisconsin not just cheesy but sleazy, finding a lot of false statements coming from both parties. He and his co-worker Tom Kertscher also appear to find Republicans more truthful as evidenced by the Truth Index, unlike almost everyone else.

I also notice, until I’m able to get used to it, that’s it’s a bit more difficult to follow/interpret the Truth Index as opposed to my own Politi-Score. To make the graph more understandable, I placed a “Half True” at the zero baseline. It’s kind of two-dimensional with the negatives and the positives….maybe that’s deliberate? I don’t know.

Politi-Score: Lots in Translation, Part 2: States to Date

In early July, just before the PolitiFact Truth Index made its debut, I had compiled my state by state PolitiFact ruling statistics in total through the end of June so I could get a broader reading—this way, I had a lot of rulings to work with.  Here’s how the aggregate Politi-Score looked for Democrats and Republicans:

This seemed to be a “cleaner” and more graphically understandable way to present it, with the horizontal “combined” bars, sort of like the individual ruling spread PolitiFact has on its mobile app, but with just a red for Republican and blue for Democrats color code.

Now that the conversion from Politi-Score to Truth Index has been established, it threw my horizontal bar chart under the bus. Since the Truth Index has Half True as its zero value baseline, the horizontal bars could not be used because now there were a lot of negative scores, which cannot be handled by that type of chart. So I had to return to a vertical bar format, and here is what that looks like:

The total number of rulings (up to 6/30/2011) is in parentheses next to the state's name.

Politi-Score: Lots in Translation

One of the reasons PolitiFact (PF) may have settled on the scoring method it uses in the Truth Index is that it shows a greater range of variation. This may have prompted me, as well, to switch from a student grade type scoring to what I called for a while the “Wide Measure” Politi-Score, back in 2010 when I first started quantifying the Truth-o-Meter. For PolitiFact, the large total range between True and Pants on Fire of 250 points makes for greater "average" variances, especially when calculating them day to day. Although in general I agree with my conservative counterpart’s view of this daily Truth Index reporting as somewhat “nonsensical” based on the fact PF is using only a select number of statements for review and ruling in calculating it, and  certainly don’t reflect all statements made. So as a measure of daily or weekly “truthiness” it is somewhat questionable.

On a month-to-month basis (which, with more rulings in a more meaningful period of time, may be more telling), here is the Politi-Score since January, compared to the Truth Index. While a monthly comparison really smooths out the scores, it is a little too smooth for the Politi-Score at the top, while the Truth Index at the bottom shows clear ups and downs, although not anywhere nearly like the daily scoring of a weeks’ worth of rulings displayed on the mobile app.



The trend is down for the year 2011, but the "bumps" are more evident on the Truth Index.
NOTE:  The number in parentheses (after the month) is the total number of rulings.

Most of us have learned the formula for changing Fahrenheit to Celsius temperatures and vice versa, and translating the Politi-Score to the Truth Index is very similar. To convert my Politi-Score to its equivalent Truth Index, you double it, then subtract 100. To convert the Truth Index to the Politi-Score, you add 100 to it, then divide the result in half. Although it appears that eventually I’m going to have to retire the Politi-Score in favor of the Truth Index, as it becomes the Truth-o-Meter standard for its mathematical interpretation.