The jury was out on the Sean Bell case before the trial even started. Unwilling to gamble on sympathy from Queens jurors for three NYPD detectives accused of killing an unarmed man just hours before his wedding, defense attorneys chose to take their chances on a judge, not a jury.

Before that decision, attorneys for both sides fought a battle of pollsters.

The three cops—Michael Oliver, Gescard Isnora, and Marc Cooper—used their poll's results to plead for a change of venue because "the well of justice in New York City has already been incurably poisoned." And what was the poison? "An enormous amount of highly prejudicial local media publicity," according to the cops' lawyers. In the November 2006 incident, Bell was killed, and two of his friends (also unarmed) were wounded, in a hail of 50 bullets fired by cops.

When the change-of-venue request was denied, the defense team opted for a bench trial before Justice Arthur Cooperman. Opening statements were February 25.

Oliver and Isnora are charged with first- and second-degree manslaughter, assault, and reckless endangerment. Cooper is charged with reckless endangerment. The judge is generally considered pro-cop, although he's best known for presiding over the stun-gun case in which two Queens cops were convicted of assaulting a prisoner in 1986.

Queens has the highest conviction rate, by far, of any of the five boroughs, indicating that a pool of jurors there would in general be pro–law enforcement. But that's not what the defense team's poll supposedly revealed. Its report, "Juror Attitudes in Queens"—marked "highly confidential" on each page but included in the defense team's filings—was nothing but bad news for the defendants. Conducted by the firm Luntz, Maslansky (founder Frank Luntz was a strategist for Newt Gingrich), the survey indicated that even a majority of white people in Queens thinks there's too much bad behavior by cops.

In November 2007, the firm polled 600 Queens residents. Though Queens has a large Latino population, the poll concerned itself only with whites and blacks—perhaps because Sean Bell was just the latest in a long line of black men shot by cops.

The poll's results were curious. Responding to the statement "I trust police," 83.5 percent of whites agreed and so did 58.3 percent of blacks. But in response to the statement "There is too much police misconduct in the city," 71.2 percent of blacks agreed—and so did 52.5 percent of whites.

Even in response to this statement—"The NYPD is more likely to use excessive force with a black suspect than a white suspect"—51.9 percent of whites agreed. (So did 79.4 percent of blacks.)

Although the cops' lawyers made a big deal about the allegedly slanted media coverage of the Bell shootings, their own poll indicated that there was no difference between blacks and whites in how they viewed the media coverage—each group was just about evenly split on whether the coverage was slanted for or against the cops.

The prosecutors pulled out their own calculators. The Luntz poll claimed a margin of error of plus or minus four percentage points, but the prosecutors pointed out that the margin of error was actually closer to seven. And not to be outdone, they produced their own poll, conducted by the Siena Research Institute. The Siena poll, according to court filings, showed that only 35.5 percent of respondents had formed an opinion about the case—and that only 43.4 percent of those "believe that the police officers used excessive force and are guilty of a serious crime."

In any case, the jury pool has been reduced to one: the judge himself. The trial is expected to last a couple of weeks.