Monday, April 18, 2016

Politifact's LAME HATCHET JOB on Ellen Troxclair....


"having a good conscience, that when they defame you as evildoers, those who revile your good conduct in Christ may be ashamed."
1 Peter 3:16

Politifact is trying to call Ellen Troxclair over something she said at a Prop. 1 forum a little over a week ago; at best, proved she cited a statistic that was incomplete.

At issue was Troxclair's claim that "When you look at the number of taxi drivers and the number of TNC drivers and the number of complaints that APD...has received over that time, you are about nine times more likely to be assaulted by a taxi driver."  They then compared Troxclair's interpretation of the data with the opinion of a couple local statistics professors.  All of which leads them to shriek "Pants on Fire!"

[Sidenote: Both sides are throwing out doomsday claims about sexual assault related to Taxis vs. TNC's.  We realized while writing this post that there are very few claims against either one.  However you feel about the underlying substance, your odds of being assaulted by either a Taxi driver or a TNC driver are minuscule.]

Troxclair aide Michael Searle explained the statistic to Politifact:
To our request for Troxclair's backup information, her aide Michael Searle emailed a document he said Troxclair received from the Austin Police Department listing all reported complaints to the police about assaults in taxi cabs or ride-hailed vehicles since 2014.

....

The document lists 37 reported incidents from Feb. 26, 2014, through Jan. 21, 2016, breaking out to 14 complaints about cabs, 21 complaints tied to vehicles driven for Uber or Lyft and two complaints related to "independent" ride-hailed vehicles. Each date-and-time entry indicates a crime ranging from assault "sexual nature" to sexual assault to rape.

....

Searle wrote: "If the estimate is that there are 15,000 TNC drivers, subject to name-based background checks, and there are 913 cab permits, subject to fingerprint background check, then you are 9 times more likely to be assaulted by a cab driver than a TNC driver in Austin."

....

Searle said Troxclair reached her "nine times" conclusion by dividing the 14 cab-connected complaints since June 2014 by the 913 cab permits (getting 0.015) and comparing that to what you get from juxtaposing the 23 ride-hail complaints versus 15,000 ride-hail drivers in Austin (or 0.0015) -- which actually suggests a 10-fold difference.
Given that taxi drivers and TNC drivers are subject to different screening procedures, it would be helpful to know how many assaults are committed by drivers subject to each.  Considering that the whole point of the anti-TNC ordinance was to mandate one of those procedures for everyone, it's actually very useful to know.  This isn't controversial.

But local statistics professors took issue with Ellen's characterization:
Rachelle Wilkinson, an adjunct professor of statistics at Austin Community College, said the calculations offered by Troxclair aren’t valid ways to gauge the relative incidences of sexual assaults. It’s "comparing apples to oranges (number of sexual assault reports to number of drivers)," Wilkinson said by email. "A much more valid way to look at the data would be to compare the sexual assault reports to the number of rides given---NOT the number of drivers. Taxi drivers tend to drive as their profession or job whereas ride-hailing drivers often do it on the side."

That is, the number of rides "given by taxi drivers is likely much higher (per driver) than the number of rides given by ride-hailing drivers (per driver)," Wilkinson wrote, adding that ride counts didn’t appear to be part of Troxclair’s backup. Also, Wilkinson suggested, the probability of such assaults appears to be extremely small whether riding in a taxi or a ride-hailed car, she wrote.

Carol Gee, a math professor at Austin’s St. Edward’s University, similarly said by email that from the perspective of a passenger seeking a safe ride, "it is likely more appropriate to measure safety by the number of incidents per trip, rather than the number of incidents per driver."

[Author's Note: Emphasis added.]
We don't disagree.  From the perspective of a passenger, assaults per trip are a more important consideration.  But the passenger's perspective isn't the only one relevant; in a discussion of whether or not to fingerprint drivers, the number of assaults per driver is very relevant.

So, at worst, you can cite Troxclair for making an incomplete argument, but that's not good enough for politifact:
Troxclair said statistics show "you are about nine times more likely to be assaulted by a taxi driver" in Austin than a driver for a ride-hailing service.

This claim shakes out to dividing oranges into apples and getting grapefruits--a ridiculous notion. Vital data, at the least about the number and length of trips by taxi cabs or ride-hailed vehicles, remains to be seen, making it impossible to definitively say which type of driver or ride more likely risks an assault. It's also worth mention that the chances of assault in either seems very low.

This leaves the "nine times" statement incorrect and ridiculous. Pants on Fire!
Not exactly Politifact; as we explained above:

  • Assaults per driver is a very relevant piece of data when examining the efficacy of fingerprint based background checks.
  • Assaults per trip is what's relevant to a passenger.
  • Your odds of being assaulted in either a taxi or a TNC driver are minuscule.
Bottom Line: Why give your readers a complete picture of the issue when there's a lazy narrative to push?!?

3 comments:

  1. I live just outside Austin, but have been barraged with Prop 1 political ads on local TV.

    ReplyDelete
  2. I wonder if the odds of assault are higher with a relatively-anonymous cab driver than with a Lyft or Uber driver whose name and license-plate number you have on your phone, and how many of the assaults attributed to Lyft or Uber were committed by non-drivers. I have heard of intoxicated riders getting into a car believing it to be an Uber, only to find out later they got into the wrong car. This is hardly the fault of Uber or Lyft.

    ReplyDelete
  3. Well, whatever the case there was some truth to what she said...
    discount concert tickets
    buy concert tickets

    ReplyDelete