While some call it style over substance, even entertainment or novelty, PolitiFact’s introduction of its Truth Index was inevitable. Too many pundits out there, myself included, were applying scoring systems to their Truth-o-Meter rankings to see what trends and statistics could be gleaned from it, especially on an individual basis.
For now, it appears that PolitiFact is avoiding the “individual” basis and is using it to show time period trends, such as, what I pointed out earlier using the Politi-Score, how scores fell (got “less truthful”) during the 2010 Election Cycle or this past February in Wisconsin. It also appears to be avoiding that topic which likely interests people the most: macro comparisons of Democrat versus Republican. But the inevitable may occur: I suppose Adair may be looking for a good reason to justify why Democrats score consistently more truthful than Republicans, on the other hand, the weighted to False scoring system he chose will make it far worse, so it’s going to be difficult to “even it up” between now and then, if that is his intent.
Here is Bill Adair’s description of how it is scored:
* Here's the full formula:
(# True x 20) + (# Mostly X 10) + (# Half X 0) + (# Barely X -10) +
(# False X -20) + (# Pants on Fire X -30)
____________________________________________________________ X 100
(Total # Statements) X 20
Let’s reduce the fraction of this formula down to its lowest common denominators, which takes into account the double-down of the “actual truth/ideal truth” (the divisor, [Total # Statements] x 20) in the calculation. While the Politi-Score starts at assigning 100 points for True and then moves down in 25 point increments, PolitiFact’s Truth Index starts at assigning 100 points for True and then moves down in 50 point increments. This makes Half True the “zero” line for the Truth Index (instead of False, which is the “zero” line on the Politi-Score), and “bottoms out” at -150 for a Pants on Fire ruling, as compared to -25 in the Politi-Score (a whopping 600% higher!).
Here is the full comparison in table format:
| Ruling | Politi-Score Assignment Points | PolitiFact Truth Index Assignment Points. |
| True | 100 | 100 |
| Mostly True | 75 | 50 |
| Half True | 50 | 0 |
| Barely True | 25 | -50 |
| False | 0 | -100 |
| Pants on Fire! | -25 | -150 |
So the Truth Index is a heavily-weighted-to-false index. This has been a common feature in the other scoring methods I’ve reviewed, particularly that of a recent colorful Daily Beast piece. When I started the Politi-Score, I used a student grading system (100, True or A, 90, Mostly True or B, etc.), as I thought this would be more understandable. When I posted it on a PolitiFact Facebook Discussion Board, I received a recommendation that it might be more intuitive to widen the measure so that 100 would be True, down to zero for False, although there was some disagreement on how punishing the Pants on Fire score should be. That way, Half True would be “half of True” at 50, and it would be easy to discern the scores. It was felt by some that -25 points, which I eventually decided to use, might not be a sufficient enough penalty for Pants on Fire. Apparently, when PolitiFact decided to do this, they felt it wasn’t sufficient enough either, since their scoring method gives Pants on Fire -150 points.
With using this scoring system, however, the trends will still show very similarly, since they are high-to-low by equal increments. My only caution to PolitiFact is that, as I’ve already noted, it serves to showcase the bias even more of that which already appears to some to be already biased, as it magnifies the effects of Barely True, False and Pants on Fire rulings on the overall scores. So Bill Adair better stick with “daily average” comparisons (which he is doing for now) and leave out political affiliation or individual comparisons unless he just doesn’t really care.
No comments:
Post a Comment