The search for data

The New York Yankees' right fielder, Aaron Judge, was the rookie of the year in the American league this past season.  I know a lot about him.  I know that he absolutely killed the Baltimore Orioles:  he hit .426, and clubbed 11 homers in only 61 at-bats against them.  That works out to one home run every 5.5 at-bats.  Babe Ruth hit a home run every 14.8 at-bats.

But against the Boston Red Sox, Judge was a bum.  He managed only 11 hits in 73 at-bats, batting an abysmal .151, with only two home runs.

There's all kinds of data out there about baseball players.  And football, basketball, and hockey players, too.  Data is not merely useful in sports, it is indispensable.  It informs virtually every decision a manager or a coach makes.

Data is very useful in the law, too.  Or would be, if we had it.  

How many appeals have you had where some damaging and inadmissible piece of evidence came in at trial, and the judge instructed the jury to disregard it?  In how many of those appeals has the court affirmed on the basis that jurors are presumed to follow the instructions?

There is no empirical support for that statement.  In fact, most judges and lawyers think it's a complete fiction.  But we don't know for sure, because we never get to interview the jurors to see if they really did ignore the inadmissible evidence.

When the Ohio legislature passed SB 2, the sentencing reform package, in 1996, it also created an agency which was supposed to collect data about sentencing.  Consistency was one of the goals of sentencing, and data on sentencing certainly would be helpful in achieving that.

We don't have that data; the agency was never funded.

One thing we do know, though:  sentencing on child pornography is wildly out of whack.

Every now and then, the local Internet Crimes Against Children Task Force up here does a bust of about fifteen or twenty defendants who ICAC has determined are downloading child porn.  The defendants are virtually identical in character traits:  white males in their thirties or forties with no prior criminal record.  The cases are virtually identical:  a defendant will be charged with 60 or 70 counts, with an equal sprinkling of felony twos and felony fours, and will enter a plea to about two-thirds of the charges.

Depending upon which judge he drew in the arraignment room, the defendant will get anything from probation to well more than a decade in prison.

It's actually worse in the Federal system.  The Bureau of Justice Statistics just released a study of sentencing in child porn cases from 2004 to 2013.  The number of such cases has doubled in that time.  (That's largely the result of better detection methods, not more criminal activity.)  So has the average prison sentence, from 70 to 139 months, "among the longest in the Federal system."

Some of that is due to the peculiarities of the Federal sentencing scheme.  The recommended sentence is derived from a matrix which cross-indexes the "offense level" with the defendant's criminal history and spits out a range of months.  But that offense level is enhanced if the defendant used a computer -- imagine that, a computer - or if there are more than 10 images (a video file is counted as 75) or if there was "distribution," meaning that the defendant left the file in the default folder of his peer-to-peer network program, where other people might access it. 

In both the Federal and state systems, the basic rationale for the Draconian punishment for end users is that they drive the market for child pornography:  if users dry up, then nobody will make child porn anymore.  A similar argument was made with regard to drug users, and we all know how stunningly effective the Drug War has been.  Again, we have an assumption for which there is no data which shows that; as the Sentencing Commission admitted in a 2012 report to Congress, there is no empirical support for the "market forces" theory.

And although Ohio law specifies that "a sentence imposed for a felony shall be ... consistent with sentences imposed for similar crimes committed by similar offenders," there's little empirical support to show that's being done, either, especially in child porn cases.  Much of that is due to the appellate review process.

There are two problems here.  The first is sample bias.  Nobody is appealing probation or a two- or three-year sentence for child porn.  They're appealing the ten-, fifteen-, or twenty-year sentence, so that's all the appellate court sees.  According to the "data" they have, that's the norm.

The second problem is that the appellate courts are disinterested in policing inconsistent sentencing.  A few years back, one of the top lawyers on sex cases here picked up one of the defendants rounded up in a child porn bust.  He took the time to review the cases of the other fifteen defendants who'd been charged, and produced a nice chart, showing what sentences those other defendants had received.  The others had gotten anywhere from probation to five years.  His client had gotten fifteen.

The appellate panel's response?  That was immaterial, because consistency in sentencing is achieved not by looking at the sentences of comparable defendants, but by application of the sentencing factors.  In short, if the judge applied the factors - and all the judge has to do is say that he did, and sometimes he doesn't even have to do that - that makes the sentence "consistent."  That's sort of like saying that somebody can determine if they're obese by comparing their own weight to their own weight. 

I've been harping on child porn cases, because that's where sentencing has become completely unmoored from reality, but it applies to other cases as well.  It's a common practice here for a lawyer who's drawn a particularly tough judge to complain that he "lost the case in the arraignment room." 

That's a problem.  The reason some consistency in sentencing is a worthy goal is that promotes respect for the justice system.  If two similarly situated defendants wind up with wildly disparate sentences, people are more inclined to believe that the system is rigged:  that's it not what you did or who you are, but who you know, how good your lawyer is, and who your lawyer knows.  

Search