Blacks have as much right as whites to serve on a jury, according to a U.S. Supreme Court ruling that's more than a century old. You might not know it from looking in an Allegheny County jury room.

The recent Tribune-Review story "A jury of peers?" used surveying and GIS analysis to show how a system that's supposedly designed to randomly pick a fair cross-section of the community for jury duty consistently overlooks blacks.

Residents of the county's overwhelmingly white neighborhoods are twice as likely to get called for jury duty as residents of black neighborhoods, the story showed. While blacks make up 11 percent of the county's adult population, they make up only 4 percent of those serving on jury panels.

As a result, black defendants - who outnumber white defendants in our courts - often have their fates decided by all-white juries.

It's possible that courtrooms in your area have similar problems. Just take a look inside jury duty rooms and see whether the racial makeup of the jurors matches the demographics from your county's census numbers. Here's how you can use CAR to answer some key questions.

Surveying for race

Pick a week or two and go to the jury room every morning before the cases start. Ask the court administrators how many jurors they have that day. Then count the number of black jurors. (When I was unsure, I briefly interviewed the person and asked his race.)

When I did this over two weeks I had counted 1,031 jurors, 4 percent of them black. That gave me a very small margin of error, plus or minus 1.2 percent at a 95 percent confidence interval.

Pittsburgh has almost no Hispanics, so I did not count them. If you want to look at Hispanics in the jury room, you have to ask everyone his or her ethnicity. Or you can substitute mapping analysis.

Mapping juror lists

Another way to test for fairness is to map the addresses of people called for jury service and see whether residents of minority neighborhoods are as likely to be called for jury duty. You'll need to get the court's jury array - the list of people called to serve each day - along with their addresses and the dates served. Try to get a least a year; I got 18 months of potential jurors in a database.

Even in privacy-mad Pennsylvania (it has the fourth-worst public records laws in the United States, according to the Better Government Association, a nonpartisan watchdog group based in Chicago) court administrators must release the jury arrays.

Since the arrays document everyone called for jury duty, not just the people who showed up, the list is a better measure of how well the court administrators are summoning a "fair cross-section of the community," as the Supreme Court ruling requires.

Once I had the arrays, I needed to remove duplicate names. Our county court jury arrays are made public 30 days before jury duty, so there are always people on it who later are granted a postponement. Those people then show up months later on another array when they're rescheduled for duty. To find them, I sorted by last names and street names, then made a separate column with an "=IF" test to see if a cell repeated the one above it. (For details see sidebar on page 12).

With the duplicates removed from the data, I was ready to geocode those addresses and create a point map showing the residences of the potential jurors. ArcView 3.2, using Census TIGER street files, matched about 91 percent of the addresses off the bat. I got the match rate up to 99 percent by fixing mistakes in addresses, using the StreetMap geocoding extension and manually plotting particularly troublesome addresses.

Once I had created the shapefile of 45,000 juror points, I overlaid it onto a census block group layer with Census 2000 demographics. After that, I opened the attribute table for the potential jurors and block group layers and performed a spatial join to append the demographic data to the juror layer. Then I used the Summarize function to count the potential juror points inside each block group.

Using the Census 2000 race demographics for block groups, I created an Excel spreadsheet with two key numbers for each: the percent of the adult population that is black, and the number of jurors per 1,000 adults population, which I called the "jury service ratio."

The average jury service ratio in block groups that were at least 98 percent white was 53 per 1,000. The average ratio in block groups that were at least 50 percent black was 26 per 1,000. There was my lede: Black neighborhood residents were half as likely to be called for jury duty. I selected 98 percent white to make sure I excluded suburban neighborhoods where a significant number of blacks also live. I calculated similar numbers for the county's 130 municipalities, so I could refer to them in stories and graphics, because nobody knows what block group he lives in.

I calculated them for most of Pittsburgh's neighborhoods. Like many cities, Pittsburgh has several black and white enclaves. Instead of saying that in Pittsburgh 32 of 1,000 adults got called for jury duty, it was more telling to say that 20 did in (black) Larimer, and 57 did in (white) Regent Square.

If you have clearly defined Hispanic neighborhoods, you might be able to use GIS analysis to make up for difficulties in using jury room surveys to identify Latinos.

Explaining patterns

If you find the similar results, the next step is to determine why. This was the most difficult part of the story, and I only was able to suggest some possibilities.

Most jury pools are picked by computer from voter rolls and driver's licenses lists. Many people say, "Aha! Blacks don't vote as often as whites." Maybe. But according to a Census Bureau study of the 2000 election, blacks register at almost the same rate as whites. "Motor Voter" laws prevent election authorities in Pennsylvania from regularly purging voter rolls, so the lists are useless in trying to see if black neighborhoods have fewer registered voters.

License records in Pennsylvania lack a racial designation for drivers. And the state Department of Transportation is hyper-secretive about those records so it is impossible to check if there are fewer licenses per adult population in black neighborhoods.

I found a very likely culprit for the jury pool discrepancies by looking elsewhere. The jury commission sends "qualifying questionnaires" to the people its computer picks to make sure they will be able to serve. A third of those never come back. The commission ignores the non-responses, even though at least half of them are returned because of a wrong address - in other words, the commission fails keep its address records current.

We know from the census that blacks in the county are more likely to rent. Renters move more often. Ergo, blacks are more likely to be missed by the jury commission because of a recent move.

The commission does an even worse job finding people in the neighborhood around the University of Pittsburgh. Residents there are mainly students, and they move every year. Only 10 adults per 1,000 in the Pitt neighborhood got called for jury duty.

Is that another story? Not likely. The Supreme Court has ruled that people can't challenge a jury room for being disproportionately old.

Contact Mark Houser by e-mail at mhouser@tribweb.com.