Posts Tagged ‘lying with statistics’

“The data also shows 6 of every 1,000 students are victims.”

That’s National Public Radio, summarizing a new report from the federal Bureau of Justice Statistics that tells a very different story from the “1 in 5” campus sexual assault slogan heard from the White House on down. Earlier here, etc.

P.S. Cato has now posted my Commentary article from last year on federal pressure for universities to reduce the procedural rights of accused students and faculty.

An in-faux-graphic on rape statistics

In recent days media outlets, including respectable ones like Washington Post “WonkBlog”, have circulated an infographic on rape incidence claiming (among other things) that false accusations of sexual assault are a vanishingly rare phenomenon. The chart claims to be sourced to official statistics, but Mark Bennett digs in a bit and finds a pile of at best strained speculation, at worst made-up nonsense. [Defending People]

P.S. This supposedly corrective piece at Slate is if anything worse than the chart it purports to correct, straining to minimize false accusation as even rarer than portrayed. (It’s worth remembering that its author, Amanda Marcotte, has a bit of a history herself when it comes to credulity on this subject.) Bennett again provides a needed corrective: “Forensic DNA typing laboratories — as numerous commentators have noted — encounter rates of exclusion of suspected attackers in close to 25 percent of cases.” (& Greenfield; and an informative followup from Bennett regarding the incidence of false accusation.) Yet more: Washington Post ombudsman says mistakes were made.

Lies, Damned Lies and Court Statistics…

Hartford, CT likes to count Mark Twain as one of its native sons; for those planning your next vacation to Hartford, his old house is designed to evoke a steamship. Among the quotes attributed to him (but actually cribbed from Disraeli) is: “There are three kinds of lies: lies, damned lies and statistics.”

I wonder what he’d think of the latest statistics about case filings and dispositions in our federal courts. [Update: The statistics were just released on August 20th.]  We just got our printed version in the office on Friday (and the stats are now available here online). Could the days of “Overlawyered” be numbered with an actual decline in certain types of lawsuits? Nah, but some of the numbers sure are curious.

In Connecticut, for example, new lawsuits are down over 20 percent in the last ten years or so. Employment discrimination lawsuits are down almost 25 percent nationwide since 2000.

But as Twain hinted, stats aren’t always what they seem. While certain areas have seen decreases, others have seen increases. Wage & hour claims are up 25 percent since 2000 and claims filed in parts of Florida have skyrocketed over the last year or so.

So, is litigation up or down in federal courts? Yes and no. It just depends how you crunch the numbers.

How trial lawyer urban legends get started

Public Citizen wrote a report about New York medical malpractice that said:

Physicians who made three or more malpractice payments between 1990 and 2006 – accounting for no more than 4 percent of New York’s doctors – were responsible for nearly half (49.6 percent) of medical malpractice dollars paid out on behalf of doctors in the time period.

This is technically true, but wildly misleading; we previously refuted this precise statistic as a natural statistical consequence of any randomly distributed set of payouts–and given that doctors in high-risk professions such as neurosurgery or ob/gyn are far more likely to be sued than dermatologists or gerontologists, the random concentration effect is going to be even more pronounced, so the Public Citizen statistic is meaningless without a showing of speciality-adjusted correlation between time periods–something no study has ever found.

But note how blogger Eric Turkewitz writes an op-ed in a small-town New York newspaper that isn’t even satisfied with simply misleading the public, and says something that is out-and-out false:

4 percent of the state’s doctors contribut[e] to half of the malpractice suits [emphasis added]

Not remotely true. “Nearly half of payments” has been turned into “half of malpractice suits.” Justinian Lane, who knows or should know that the latter statistic isn’t true, because his blog posted about the original statistic, then repeats the lie either thoughtlessly or deliberately:

Maybe doctors should discipline the four percent of doctors that make up half of all malpractice claims.

Will either of them retract the false claim with the same fanfare that they made it? Stay tuned. (They certainly won’t explain that there’s nothing damning about the accurate statistic–though I have been refuting this for over three years, Public Citizen and trial lawyers and their fans continue to regurgitate the data as if it means something.)

“4% of doctors responsible for 50% of payouts”

Trial lawyers like to repeat statistics similar to this (Bizarro-Overlawyered just did so this week) as an argument for medical malpractice being a problem of the doctors, rather than the lawyers. The problem is, as I noted three years ago, that the statistic is fallacious.

Some small X% of doctors responsible for large Y% of payouts is always going to be true simply by random chance. It’s going to be true over any time period: the problem is that if you take that time period and divide by two, the X% in the first half of the time period are going to be almost entirely different than the X% in the second half of the time period. Even if you were to fire every single one of those doctors in the tail in the first time period, all you have is X% fewer doctors; the very next year, it’s going to be a different small A% of doctors responsible for large B% of payouts, and you’ve solved nothing. With very rare exceptions medical malpractice payouts have absolutely nothing to do with the quality of the doctor, and everything to do with the risk profile of their practice.

It’s worth noting Eugene Volokh’s excellent explication of the issue:

Read On…