| |
 |
 |
|
Len Watkins
Joshua Genser
|
 |
| |
|
 |
Offices
in Point Richmond and Emeryville
125 Park Place, Suite 210
Point Richmond, CA 94801
Phone: 510-237-6916
Fax: 510-236-9851
2200 Powell Street, Suite 890
The Watergate Office Towers
Emeryville, CA 94608
Phone: 510-237-6916
Fax: 510-236-9851
|
|
|
 |
| |
| Statistics,
Probability and the Law: What You Don't Know Can Lose
the Case |
|
My
brother-in-law, Russ, was contemplating whether to
try to hit his golf ball through the tree directly
between him and the green, or whether to play it safe
but lose a stroke by getting out from behind the tree.
Bill, another of our foursome, advised Russ to go for
it. "After all," he said, "the tree is ninety percent air." Russ’s shot ricocheted off a branch and deeper into the woods, costing him two
strokes.
Russ’s decision to try to hit the ball through the tree was based
on his judgment that the probability of hitting a tree branch was
low, based on the statistic that ninety percent of the volume of
the tree was air. His decision was flawed, not only because of his
high handicap, but because the ten percent of the tree that was solid
was evenly distributed throughout its volume, leaving very few clear
paths for the golf ball to travel through the tree. In other words,
as I pointed out to Russ as I collected my winnings, a window screen
is also ninety percent air, but I wouldn’t bet on hitting a golf
ball through one.
Every day, every person makes decisions based on probabilities and
statistics, usually unconscious of that fact. Often, however, ignorance
of some of the most fundamental rules of probability and statistics
dooms those decisions to be bad ones. Judges and justices of courts
of appeal suffer from this affliction as much as anyone, but their
flawed decisions have major public policy repercussions. Three recent
decisions of Courts of Appeal illustrate the idea.
In Sharon P. v. Arman, Ltd., Division Three of the Second Appellate
District of California ruled that a commercial parking garage was
inherently dangerous and, therefore, that owners of such structures
could be liable for criminal acts committed on the premises even
in the absence of any previous incidents.
Sharon P. is only the latest in a lengthy history of decisions of
California Courts of Appeal deciding whether property owners owe
duties to prevent criminal assaults on their property. Prior decisions
had focused on the question of whether it was reasonably foreseeable
that such criminal assaults were likely to occur in the absence of
reasonable precautionary measures. Such foreseeability had been found
in prior cases only where there had been previous criminal acts.
The question in Sharon P. was whether previous criminal acts were
necessary to establish such foreseeability.
The plaintiff, Sharon P., had been sexually assaulted at gunpoint
in the subterranean parking garage of the office building in which
she worked. She sued the owners of the building, but the trial court
granted summary judgment in favor of the defendants. The trial court
noted that, other than robberies of a bank located in the building
above the parking garage, there was no evidence of prior criminal
activity in the vicinity.
The Court of Appeal reversed, finding that prior criminal acts were
not the only factor to be used in determining the foreseeability
of criminal assaults. However, the Court went beyond that, ruling,
also, that commercial parking garages are so inherently dangerous
that criminal assaults are, as a matter of law, always foreseeable.
The Sharon P. decision, that commercial parking garages were inherently
dangerous, was based purely on the justices’ logic: that dark areas
out of sight of the public and the police are likely places for a
criminal to lie in wait for a victim. The opinion cites not one iota
of empirical evidence to support this proposition, nor was any such
evidence presented to the trial court in opposition to the Motion
for Summary Judgment.
While it does sound logical that parking garages might suffer from
more criminal assaults than other places, what if it is not, as a
matter of fact, true? What if rapists simply haven’t figured out
the apparent vulnerability of women who park in commercial garages?
Perhaps the discomfort of lurking in a garage outweighs the apparent
opportunities to find victims. Perhaps criminals fear being trapped
and unable to escape if detected committing a crime in a garage.
Even if commercial parking garages are places where many criminal
assaults occur, how much more frequent need they be for them to be
declared inherently dangerous? To what are they being compared?
To put these questions into the language of probability and statistics:
Is the probability of a sexual assault, or, indeed, any sort of crime
against a person, significantly greater in a commercial parking garage
than in other places? If so, the public policy question would then
be: Is the cost of imposing additional risk of liability on owners
of commercial parking garages justified by the costs of the increased
risks of assault to users of those garages?
The Sharon P. opinion provides answers to none of these questions,
yet it obviously has profound public policy implications. The cost
of insurance for commercial parking garages has probably increased.
Owners of commercial parking garages must now take additional security
measures, further increasing their costs of doing business. Despite
such security measures, one can be confident that the number of lawsuits
against owners and operators of commercial parking garages will increase,
further crowding courts. Of course, those costs will be ultimately
borne by consumers.
The legislature is the body that ought to be responsible for deciding
how the risks of civil liability for criminal acts should be allocated.
However, Courts have a long history of stepping in when legislatures
abdicate their responsibility to make such decisions. The Court of
Appeal in this case, though, should not have made this decision without
empirical proof that the probability of a criminal assault in a garage
was significantly higher than on other types of commercial premises.
The proper result would have been to remand the case to the Superior
Court for a trial on the factual issue of whether commercial parking
garages are especially dangerous. Ideally, the Court of Appeal should
have established for the trial court an objective standard for the
determination of what increased level of risk is necessary in order
to support a finding of inherent danger, or it should have instructed
the trial court to develop such a standard based on expert testimony
adduced at the trial on remand.
Another pair of disturbing decisions are those of the Ninth Circuit
Court of Appeals in United States v. Bighead, and of the California
Court of Appeal in People v. Roberts. In both cases, the courts upheld
the admissibility of expert testimony on statistical issues without
any foundation for determining whether the statistical opinions rendered
were valid.
In Bighead, the defendant was accused of sexual molestation of a
child. The prosecution offered the testimony of a sexual abuse counselor
who opined that victims of sexual abuse commonly delay reporting
abuse and blend events together in their memories. This testimony
converted what would ordinarily have been profound doubts about the
truth of the alleged victim’s story into bolstered credibility. That
is, the alleged victim’s delay in reporting the alleged assault and
the victim’s inability to remember accurately the details of the
incident would ordinarily severely damage that witness’ credibility.
However, after the testimony of the sexual abuse counselor, those
very defects became evidence of the truth of the accusations.
The expert’s opinion was based on interviews of more than 1,300 people
who said they had been abused as children. Thus, the sexual abuse
counselor’s opinion was based on statistics and probability. That
is, based on statistics derived from a study of 1,300 sexual abuse
victims, this person had concluded that the probability of delay
in reporting assaults and the probability of confusing different
events were higher in victims of sexual abuse than in victims of
other types of crimes.
The problem is that this sexual abuse counselor had no valid basis
for coming to that conclusion. The 1,300 persons who had been interviewed
all claimed to have been victims of sexual abuse, but there was no
evidence of how many had actually been victims of childhood sexual
molestation. Thus, the sample population tested could, in fact, have
been evidence of the opposite of her testimony: that those who make
false claims of sexual abuse are more likely than average to delay
making such claims and to confuse the facts of their false claims.
Even if all of the 1,300 persons in the sample had been genuine victims
of sexual abuse, the expert could not testify to how many of those
1,300 persons had delayed reporting the alleged molestations, nor
how many tended to blend events together in their memories. In other
words, the expert made a statistical conclusion without having gathered
any of the statistics upon which she based that conclusion.In fact,
because the sexual abuse counselor had not interviewed a similar
sample of victims of other crimes with regard to their tendencies
to delay reporting the crimes or to be confused in their memories
of the details, the expert didn’t really know if her group of sexual
assault victims was different from anyone else.
The fault is not the expert’s. This witness wasn’t trying to determine
scientifically whether her observations about victims of sexual abuse
were statistically valid, she was trying to counsel those who needed
it. The fault lies with the trial court judge and the justices of
the Court of Appeal who permitted such testimony without regard for
its flaws.
In Roberts, the defendant was accused of robbery. The defendant,
who was not a member of a gang, presented the testimony of a gang
member who was already convicted of the same robbery, who said that
the defendant was not his accomplice and was not present at the scene
of the crime. The prosecution, in rebuttal, offered the testimony
of a police officer as an expert witness, who opined that gang members,
such as the defense witness, will lie to protect residents of their
neighborhoods, even when those residents are not fellow gang members.
Again, this is entirely a statistical statement. That is, in order
for the police officer’s testimony to have had any meaning at all,
he must have been saying that this gang member is more likely to
be lying than would be a non-gang member.
The basis for the police officer’s opinion is, of course, his years
on the force and numerous contacts with gang members. However, like
the sexual assault counselor in Bighead, the police officer could
not testify to what proportion of those gang members had lied to
him, nor even how he knew which ones had lied and which ones had
told the truth. The police officer certainly didn’t know as a factual
matter how likely were other neighborhood residents who were not
gang members to lie for a neighbor, so he could not make a valid
comparison with gang members.
Both the sexual abuse counselor and the police officer also could
not account for any number of other factors that must go into an
opinion such as they gave. For example, one would expect that age,
culture, race and gender would make significant differences among
the behavior of both victims of sexual abuse and gang members. That
is, Asian gang members in Los Angeles might be more willing to lie
than African-American gang members in San Francisco, and impoverished
White children might be more reluctant to report sexual molestation
than middle-class Latino children. While ignorant of the importance
of such other factors, and ignorant of the distribution of these
other important factors among the populations of gang members and
alleged victims of sexual molestation, one cannot legitimately opine
about the truthfulness of the testimony of a member of that group.
Judges are not trained in probability and statistics, but perhaps
they should be. As the gate keepers of expert testimony, only they
can keep invalid evidence from a jury that is certain to be even
more ignorant, and, thus, more easily swayed by invalid scientific
opinion, than the judge. Judges should, at least, be aware of their
ignorance, and insist that the parties educate them before they make
factual or public policy pronouncements without any empirical support.
Most importantly, litigators should be cognizant of the most basic
principles of probability and statistics so that they can object
to factual findings made without proof. |
| |
| |
|
|
|
|
