Are you a wonk? Even if you're not, here's a mind-bender for you:

Q: Can a place have a negative population?

A: 1.) No, no place can have fewer than zero people in it.

Result: Wrong! Try again.

A: 2.) Yes, if that place is a cemetery.

Result: Wrong! Try again.

A: 3.) Yes, if that place is being used to draw legislative districts in New York.

Result: Right! Go on to the next paragraph.

Democrats in the New York State Senate have been angered for years that Republicans were using downstaters in upstate prisons -- many of them members of minority groups -- to help gerrymander Senate districts. After they finally briefly got control of the body, they added a ban to this practice in a 2010 budget bill.

Republicans recaptured power later that year. They wanted to go back to their usual practice, so they challenged that ban in court. They were unsuccessful. So the Legislative Task Force on Demographic Research and Reapportionment (LATFOR) had to design districts that counted prisoners at their last address before incarceration. But these prisoners were counted by the U.S. census as residents, well, where they live.

Data from the State Department of Corrections and Community Supervision (DOCCS) was therefore used by LATFOR to adjust 2010 census results for the purpose of State Legislature redistricting. DOCSS totals for prisoners were subtracted from totals for census blocks in which prisons were located and added back in at their last addresses before incarceration; detail on population composition (e.g. race, ethnicity) for affected census blocks was adjusted as well.

Because the numbers were from the two discrete and independent counts, done for different purposes, some results were anomalous. After the math was done,LATFOR reported 71 census blocks containing a negative number in one or more fields. This means that DOCSS was reporting more people in a category -- for example, people of Hispanic decent -- than the census reported as residing there. Even more interesting, LATFOR's total adjusted population for three census blocks showed that a negative number of people resided in them. That is, the DOCCS total for each of these blocks was greater than the census total.

The entire "negative population" in the statewide database released by LATFOR is 4,498. This is equal to 3.5 percent of the population required for one Assembly district, or 1.5 percent of that required for one Senate district. Of these "negative people" about half -- 2,130 (47.4 percent) -- are Hispanic.

At the Sept. 17, 2011, LATFOR hearing, Roman Hedges, one of four task force members, said: "Well, you can't end up with a minus number. You can't have negative population. So let's call it zero and say that there is a small error." But in fact,LATFOR did not call it zero. These negative numbers are still there, in the published database.

Elementary school children learn that, when added together, negative numbers offset positive ones. So one consequence of LATFOR's "negative people" in some blocks is to offset "real people," that is, to effectively remove them from the count. Here's one example: There are negative population numbers for voting-age Hispanics in the three census blocks in Franklin County that contain state prisons. The effect is that 44 percent of the voting-age Hispanic population in Franklin County is not counted in the demographic data there.

When asked, Hedges made two additional points to minimize the importance of LATFOR's negative population numbers. First, he said, most of these did not affect the block totals that were actually used for redistricting. They occurred in those demographically descriptive subcategories of the data that are not used for this purpose. And second, he argued, since the New York State constitution requires adhering to town lines in designing state legislative districts, any small errors introduced by the few cases of overall negative totals for blocks were of no consequence for the actual outcomes of the process.

These arguments are not convincing for at least three reasons:

*Negative population numbers (as we suggest at the outset of this essay) strain credulity. LATFOR has said, in effect: We are publishing these numbers because these are what we got when we did -- as best we could -- what the court told us to do, and we don't want anyone to think we are monkeying with the results.

But negative numbers in the database in and of itself brings an already contentious, highly partisan undertaking into question for still another reason: There is no such thing as "negative people."

*The point of one-person-one-vote is to make sure every person is counted, not every person except some of those who are members of minority groups living in mostly rural places where a court decision requires adjustment of federal census data.

*The demographic subcategories of the data do have policy importance. Redistricting must meet the requirements of the Federal Voting Rights Act. LATFOR's "negative people" are clearly disproportionately members of protected minority groups under that act.

Census data is not pristine. The possibility and potential disproportionate impact of an "undercount" is a decennial issue. People self-select into categories (or decline to) that are germane to redistricting. But the census is not purposefully manipulated or adjusted to provide one party or another with political advantage. What state courts now require is adjustment of data underlying a highly politicized process, for a political reason, by the agency widely understood to be using that data for purposes of a bipartisan gerrymander.

We have found no impropriety byLATFOR. But it is not an accusation of impropriety to suggest that this is not a good idea. And it would be an even worse idea if a time comes where there is no divided partisan control of the Legislature.

Relocating where prisoners are counted for the purpose of redistricting -- not for voting, because imprisoned convicted felons lack the franchise -- was advocated by Democrats because it potentially gave them an advantage. But of course, Republicans are not immune from such advocacy. Partisan definition of the database to gain an advantage in redistricting is a hoary idea in New York. Republicans benefited for many years from a state constitutional provision, repealed in 1931, that state legislative districts be based upon a separate state census count of "inhabitants, excluding aliens."

All of this adds one more argument for the need for as neutral a process as possible, and one beyond legislative control, for redistricting the State Legislature.


Josh Simons is a research associate at the Center for Research Regional Education and Outreach (CRREO) at SUNY New Paltz. He was a judge for Redistrict New York, a statewide university-based fair redistricting competition in which students sought to draw the best districting plans for the Senate, Assembly and congressional districts. Gerald Benjamin is a distinguished professor of political science at SUNY New Paltz and director of the CRREO.