At last! The prosecution is currently presenting its “computer forensics” witness—the Ford Motor Company engineer who analyzed Trooper Robert Higbee’s vehicle black box. And he’s committing all the rhetorical sins a technical witness can commit in front of a jury (IMHO).

Expert Error No. 1

I’ve said it before, but the worst rhetorical error a technical witness can make is to confuse the jury. In this case, the poor explanation of the black-box data may be the fault of the prosecutor (his choice and sequence of questions, for example), or it may simply be that the expert witness doesn’t have much experience testifying in criminal trials (as defense attorney Subin appears to have pointed out).

In any case, the testimony is belaboring a sub-second by sub-second record of the ups and downs of acceleration and deceleration before what the witness calls “the impact.” I guess this witness didn’t take Freshman Comp 101 in college, so he never learned the value of a thesis statement. Listening to his testimony is like reading the footnotes of an essay before you read the opening paragraph.

Expert Error No. 2

This expert witness is also making a mistake I’ve noticed in other trials, such as the expert in garbage bags in the Melanie McGuire murder trial (also NJ—pure coincidence, I assume). This expert is trying to support the prosecution’s side of the case rather than his own data—and it’s painfully obvious. Attorney Subin is doing a nice job of jumping all over the witness’s biased word choice, such as adjectives that suggest the black box could tell how much pressure the trooper was applying to the brakes. I believe most jurors appreciate an objective expert witness, regardless of the side for which he's testifying.

Expert Error No. 3 (or is it trick no. 1?)

The expert witness is also irritating me by his loose ranges of numbers. Statistics drawn from data are always subject to a range of “plus or minus,” because statistics are only estimates. We all know this. But the range has to be within a “margin of error” that is reasonable.

There’s always at least a 1% variance because chance is 50:50. Most valid statistical predictions (that is accurate statistics) assume a margin of error of plus or minus 2%. For example, a well-calibrated black box might record a car as traveling 100 miles an hour when it was really traveling 98 miles an hour or 102 miles an hour. At 65 miles an hour the 2% error rate is plus or minus 1.3 mph or 63.7 to 66.3 mph. In fact, don’t all drivers “know” they can speed 5 miles over the limit without being pulled over, because the margin of error for radar devices is about plus or minus 2?

But if I heard correctly, the expert testified that the black box in question had an error rate of minus 2 mph to minus 8 mph at 65 mph. Where’s the plus? Is he saying the black box is calibrated to under-report the vehicle’s speed? And the amount it can be in error varies up to 400%. What good is it, then? And clearly the rate of error lies far outside accepted statistical error rates—especially in a homicide trial.

But this isn’t the greatest rhetorical trick being employed here. The greatest trick is stating that the data representing the “final 20 feet” of the accident are missing, because black-box data is recorded in 2/10ths of a second increments and, at a speed of 100 feet per second, that 2/10ths of a second represents 20 feet.

OK, even this fairly mathematically illiterate former juror can see from the expert’s charts that the car was not traveling at 100 mph when the final data point was collected. That means it’s less than 20 feet—closer to 12 feet, it seems to me.

The Point of Impact

Not to mention that the expert witness identified “the point of impact” in his data as being the point at which the black box ceased to function. That makes no sense. The black box must have operated until the motor (electrical power) cut off. Of course, maybe I missed testimony from mechanics or accident reconstructionists who testified that the power went off on impact. If so, then ignore the rest of what I have to say.

Why should the expert assume that the trooper’s last application of the brakes was the first impact? Didn’t the cars bounce around a bit? What if he applied the brakes before the first impact (as appears to have happened in the data) and then reapplied them during the collision in an attempt to halt the forward motion of the car? Or what if he applied the breaks briefly several feet from the intersection and then again when he first saw the white van to his right?

My point about “the point of impact” is that this expert witness is making many assumptions without considering alternatives. And he seems to me to be doing it because he thinks that’s his job as a prosecution witness.

Prosecution Exhibit s40

The prosecution is presenting a chart to the jury this afternoon which divides the incident into five phases, each with a duration and a distance. Unfortunately, here too there’s an enormous margin of error. For example, the “4th phase” or “braking” phase, is said to have occurred from minus 2.6 seconds to minus 1.6 seconds before impact (again, that mushy deadline) and to have covered from 276 to 166 feet. (I’m assuming the range of feet is the distance the car traveled, that is 110 feet; but the rhetoric is so confusing that it could mean it’s possible the car traveled 276 feet in distance or 166 feet in distance and the 110 feet is variance. Hmmm. Besides, if his calculations are based on 20 feet per second, how could the car have traveled more than 100 feet in a second if it was going under 80 miles per hours?)

In any case, 110 feet is a long way, isn’t it? I thought the “Rules of the Road” said a car should begin braking at 50 feet before a stop sign. If Trooper Higbee hit the brakes 110 feet before the impact, then he was braking well before the stop sign.

CSI: Confusing Scientific Information

If were on the Higbee jury, I’d be hoping one of my fellow jurors understood calculus enough to check this witness’s figures.