Tuesday, March 15, 2011

The “De Minimis” Map

Cross-posted at PolitickerNJ

The Legislative Apportionment Commission has decided on a Democratic map. The 11th member, Alan Rosenthal, made it all but official at his first public appearance with the commission last week.

OK, I’m being a tad flippant. Dr. Rosenthal only laid out his standards for judging which map he would settle on. The list of priorities he announced publicly was less detailed than the confidential memo he reportedly distributed to the other commission members. However, if he sticks to the general principles in his public remarks, I do not see any conceivable way that the new legislative map can have fewer than 22 safe Democratic districts.

Here’s why. Rosenthal’s standards allow for very little change to the current map. The first standard is his definition of equal population, which he sets out as no more than a 5% deviation. That means that all district populations should be between 214,300 and 225,300.

The 2001 map had an approximate 7.5% deviation. The competitive map I drew last week had a 7.7% maximum range. Setting the maximum range at 5% severely limits the possibility of drawing more competitive districts. To do this, you would need to peel off some border towns from Republican districts – thus making them smaller than the ideal – and add them to Democratic towns to make the district competitive – and thus larger than the ideal.

Rosenthal’s next three priorities are adhering to the New Jersey Constitution’s provision on municipal splits (although he said nothing about the identical provision regarding counties), contiguity, and compactness – which, barring using an actual mathematical formula, is simply in the eye of the beholder.

The fifth priority on Rosenthal’s list is maintaining communities of interest. This is an acknowledged, but rather amorphous, concept in redistricting. Communities of interest can be defined in myriad ways, including people who have gotten used to voting in the same district. Rosenthal did not detail what he means by this term.

This brings us to the money card in Rosenthal’s standards – continuity of representation. Rosenthal defines this as incumbents facing a familiar electorate. In other words, incumbents should be drawn into districts where the majority of voters are already represented by them. You could also call this the de minimis rule – any change should not be consequential to the current system as a whole. Anyone who has worked with Dr. Rosenthal or read his published works on state legislatures will not be surprised by how much he values this type of continuity.

Lower on Rosenthal’s priority list is competitiveness, which he defines rather weakly as “absolutely no reduction” in the number of currently competitive districts and “perhaps increase [them] a bit.” Considering that most political observers can’t identify more than one competitive district in the current map, this standard is meaningless as stated.

Those who value continuity see competitiveness as akin to volatility. It doesn’t have to be that way. The competitive map I proposed would likely lead to a stable Democratic majority. The only difference is that Democrats would have earn that majority in each election.

Rounding out Rosenthal’s standards are minority opportunity and partisan fairness. The fact that these two factors are at the bottom of the list indicates that Rosenthal has roundly rejected the Republicans’ two main arguments behind their redistricting map.

The minority opportunity argument would require revisiting the 2001 process, which dispersed minority groups across more districts. Recent court rulings on this issue indicate New Jersey is not required to create new minority majority districts, and so Rosenthal is probably on firm legal ground. In fact, I would presume that any 11th member would have been unlikely to support such a “corrective” measure.

The Republican’s other point of contention – partisan fairness – is also rather weak. They argue that the partisan composition of the legislature should reflect the statewide partisan voting pattern of any single election. This assumption is patently inconsistent with our form of government. It may make sense if we lived in a parliamentary system where legislative elections determined who the governor is. But we don’t. Furthermore, our districts are drawn based on equal population, not equal participation. As such, district by district turnout can vary by as much as 50,000 voters in any given election. It just so happens that Republicans tend to represent higher turnout districts and so garner more votes statewide.

Now, I actually attempted to draw a map using Rosenthal’s stated priorities. I did this with an eye toward maximizing Republican gains given the limitations of these standards.

I was able to create 39 districts that ranged from 214,293 to 225,452 residents – just a couple hundred outside the range of Rosenthal’s 5% equal population definition. The remaining district has just over 227,000 residents. The map also maintains the same basic distribution of African-American, Latino and Asian residents as in the current map.

The real key to the map is its emphasis on continuity of representation. As such, only 107 out of 566 municipalities shifted into different districts. That also meant that only 23 of 120 incumbents would have to challenge other incumbents for available slots. In line with maximizing Republican gains in this exercise, individual Democratic legislators would lose out. Six of them would end up facing six Republicans in safe GOP districts. Five Democrats would have no incumbent opponents, but would find themselves in decidedly less friendly districts. Another 11 Democrats would have to compete against each other for seven available seats in Democratic districts. So, a grand total of 10 to 15 Democratic incumbents would be on the hot seat. The remaining 105 incumbent legislators would be completely safe!

This also means that the best the GOP can probably do is to come up with a map that creates 22 safe Democratic districts and 18 safe Republican districts (with an argument that one or two of these districts could be competitive). Perhaps just as importantly, all 120 legislators elected in 2011 will have a de facto 10 year term (barring death or scandal, of course)! So the map that emerges from these stated standards is de minimus indeed.

From the start of the process, the Republicans tried to sell the line that they were prepared for redistricting. Not exactly – the two parties just organized themselves differently. The Democrats were better prepared to negotiate with whomever became the 11th member, while the Republicans spent their resources girding for a legal battle. The fact they even put Alan Rosenthal on their list of potential nominees, knowing his long held preference for continuity, is a pretty good indication they were looking past the mediation process to the courts. And since Rosenthal’s standards are well-grounded in legal precedent, they’ll likely lose there as well.

Wednesday, March 9, 2011

A Competitive Map

Cross-posted at PolitickerNJ

What makes a legislative map competitive? Definitions vary, but in my view it includes the following characteristics: either party has a chance of winning a majority and the number of districts that could conceivably change hands in each election is maximized. This is not to say that legislative control would, or even should, change hands every two years – just that the possibility exists. Such a map would keep both parties on their toes, increase media coverage of legislative elections, and hopefully lead to greater voter participation.

I define legislative competitiveness rather simply as the raw vote difference between the Democratic and Republican candidates for legislature. For the map I propose here, I utilized a two-cycle (2009 and 2007) average in order to include both Assembly and Senate races as well as turnout in gubernatorial and off-year elections.

This is a fairly straightforward metric. I purposely didn’t include results for governor or federal offices as the vote choice dynamic is different. Indeed, the 2009 election witnessed a decent amount of ticket splitting between Governor and Assembly. Since my objective is to heighten legislative competitiveness, it makes sense to rely on legislative results.

This approach is not without limitations. It does not take into account incumbency. Some current districts may appear to be more competitive because a particular incumbent is able to overcome voters’ partisan preferences (LD14 and LD7 come to mind).

Another limitation is that past voting patterns can mask depressed turnout among one party’s voters when the other party “owns” that district. One potential adjustment to this competitiveness metric would be to merge vote data with partisan registration in each town. However, I’m inclined to think that it would not change the decisions behind the map proposed here.

Basically, I defined district competitiveness in the following way: solidly safe partisan districts have a two-cycle victory margin of 10,000 votes or more; likely safe partisan districts average 2,500 to 10,000 vote margins; leaning partisan districts average between 1,000 and 2,500 vote victories; and competitive districts have average victory margins under 1,000 votes.

In the current 2001 map, 26 districts are solidly safe bets (11 Democrat and 15 Republican) and another 13 have very likely partisan outcomes (12 Democrat and 1 Republican). There are no leaning districts and only one competitive district, which is very nearly Leaning Democrat.

I used past legislative voting behavior (disaggregated to the municipal level) to assign hypothetical margins for the districts in the proposed “competitive” map. Based on this metric, the proposed map has only 17 solid districts (7 D and 10 R) and 10 likely districts (8 D and 2 R), with another 2 leaning Democratic.

That leaves 11 districts with the potential to be truly competitive. (On the map, they are numbered 1, 6, 8, 14, 21, 22, 27, 34, 37, 39, and 40).

[Click the following links for: (1) graphic version of the proposed map, (2) list of municipalities in each district, (3) description of district characteristics in the proposed map; and (4) description of district characteristics in the current map.]

For the record, the districts in this proposed map are contiguous and meet the federal equal population guidelines. The largest district has 228,224 (LD36) residents and the smallest has 211,281 (LD38). The difference between these two is 7.7% of the ideal district size of 219,797 – below the 10% federal limit and similar to the 2001 map when it was first drawn. As to compactness, that is in the eye of the beholder – or the judge who will ultimately review the lawsuit that arises from any map. But it certainly doesn’t look any more contorted than the current map, and in some areas may even be an improvement.

Now, on to the consequences of this map. In order to draw more competitive districts, smaller districts skew Republican while the larger ones tend to be Democratic or competitive. There is a pretty even partisan distribution among districts with 216,000 to 223,000 residents.

These skews are unavoidable when drawing a competitive map in a state where Democrats hold a 33% to 20% partisan voter registration advantage. (The remaining voters are unaffiliated.) That also means Republicans are at a disadvantage in terms of winning control of the legislature, even with a competitive map.

The GOP needs to win nine of the 13 competitive or leaning districts in order to claim a majority in either chamber. Democrats on the other hand, only need to hold on to their two leaning districts and take just four competitive districts to retain legislative control.

Given the state’s voting trends in all elections, it is probably the best Republicans can hope for and fair to both parties. It may also be the fairest map for the state’s more than two million unaffiliated voters, who would have a real stake in the outcomes of one-quarter to one-third of the state’s legislative races.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Other notes on the map.

I did not use minority representation as a factor in drawing this map, but I am aware of the consequences.

The proposed map retains one Hispanic majority district. The number of districts with a 40% to 49% Hispanic population would increase from three to four, while the number between 30% and 39% Hispanic would decrease from three to two. A total of 13 districts would have a Hispanic population of 20% or greater, which is identical to the number in the current map.

Black representation would shift in a few areas in the proposed map. There would continue to be one African-American majority district (LD28), but with a larger majority than it currently has. This can be ameliorated by drawing the Newark dividing line between districts 28 and 29 a little more creatively than I did. LD20, centered on Elizabeth, would also see an increase in black population, while the redrawn districts 5, 7, 17, 22, 31, and 35 would have similar proportions of black residents as in the current map. The Trenton district (LD14 in the proposed map) would also maintain similar black representation. However, the need to pair Democratic towns with Republican towns to make more competitive districts would totally change the face of two districts – 27 and 34.

The number of districts with an Asian population over 20% would decrease by one, to four in the proposed map.

Strange bedfellows. There could be some intriguing match-ups brought about by this new map. From south to north:

Republican Assemblyman Domenick DeCicco has all but announced his candidacy for Senate. But rather than take on Fred Madden in the 4th, the proposed map has him facing off against Jeff Van Drew in a very competitive LD1. Madden ends up sharing LD4 with fellow Democrat Senate President Steve Sweeney. Madden has been rumored to be considering retirement, so this primary match-up may not be a deal-breaker.

The newly competitive LD6 pits incumbents Jim Beach (D) and Dawn Addiego (R) for the Senate seat and Budget Committee chairman Lou Greenwald (D) and Scott Rudder (R) for the Assembly. This means that competitive district LD8 has only two incumbents for the six available seats – Democrat Pamela Lampitt and Republican Patrick Delany.

District 13 may be a hard sell for Republicans, as it includes two high-profile Senate incumbents – Joe Kyrillos and Jen Beck. However, since districts 11, 12 and 13 are all GOP strongholds, a few municipality swaps can ensconce the two in separate districts without affecting the overall competitiveness of the map.

An even more competitive LD14, which includes Trenton and parts of Burlington County, would involve a showdown between Joe Malone on the Republican side and Wayne DeAngelo and Dan Benson for the Democrats. With no Senate incumbents here, the bigger question is who would try to move into the upper chamber. This also means that Shirley Turner and newly-minted Senator Linda Greenstein would have to draw straws in LD15.

LD40, which has been relocated to Middlesex County in the proposed map and should be quite competitive, has only two incumbents – state Democratic chairman John Wisniewski and county Republican chairman Sam Thompson.

Democrat leaning LD18 presents what may be marquee match-ups of this map. It pits Senate Majority Leader Barbara Buono against Minority Leader Tom Kean. The undercard would include Peter Barnes (D), Linda Stender (D), and Jon Bramnick (R).

One also wonders whether the Union or Middlesex Democratic organizations would win out in the proposed LD19, which would be home to both Joe Vitale and Nick Scutari.

LD21, which is now a toss-up district, has only two incumbents – Assembly Majority Leader Joe Cryan (D) and Nancy Munoz (R). LD22 is also a competitive district and includes Democratic incumbents from three different districts, Senator Bob Smith and Assemblymen Patrick Diegnan and Jerry Green.

The proposed LD25 is teeming with Republican incumbents. Tony Bucco and Joe Pennacchio on the Senate side, and Minority Leader Alex DeCroce, Jay Webber, and Anthony Bucco (the younger) in the Assembly.

LD26 has only two incumbents – John Girgenti and Nellie Pou, who are transplants from the current 35th. The only problem for these two Democrats is they would find themselves in a heavily Republican district.

LD27 would become a competitive, even slightly Republican, district defended by former Governor and Senate President Dick Codey. The Assembly face-off could include John McKeon (D) and Michael Patrick Carroll (R), unless the latter’s judicial nomination is confirmed.

In the Newark districts, Ron Rice (LD28) and Teresa Ruiz (LD29) would still be in pole position for the two Senate seats. However, Assembly Speaker Sheila Oliver and Mila Jasey would find themselves in LD28 while Ralph Caputo and Albert Coutinho would land together in LD29. The big problem for Essex Democrats (as well as those concerned with the number of female African-American legislators) is that Cleopatra Tucker and Grace Spencer would also share these districts (either 28 or 29 depending on how the Newark dividing line is drawn).

LD34 would be a competitive district pitting Nia Gill (D) against Kevin O’Toole (R) for the Senate seat, and Tom Giblin (D) and Kevin Ryan (D) against Scott Rumana (R) for the Assembly.

LD37 would also be competitive. Loretta Weinberg (D) is the lone Senate incumbent, while Connie Wagner (D) would battle it out with Bob Schroeder (R) and David Russo (R) for the two Assembly seats.

The far northeast corner of the state (LD39), should also be competitive in the proposed map. Gerry Cardinale (R) is the lone Senate incumbent, while the Democratic trio of Gordon Johnson, Valerie Vainieri Huttle, and Joan Voss would have to be whittled down to two for a face-off against Charlotte Vandervalk and a Republican to be named later.

Finally, the major shifts in these northern districts leaves two Democratic districts with only one incumbent each – Assemblyman Gary Schaer in LD36 and Senator Paul Sarlo in LD38.

Monday, March 7, 2011

A "Constitutional" Map

Cross-posted at PolitickerNJ

Now the work begins. With the appointment of Alan Rosenthal as the 11th member of the New Jersey Apportionment Commission, the work on drawing up a new legislative map commences in earnest.

A number of interested parties have already weighed in at public hearings, asking for a map that would bolster their group’s representation in the legislature. However, I have yet to see draft maps that cover two particular takes on the redistricting process – the constitutional guideline approach and the competitive election approach.

This column examines one possible “constitutional” map. (I’ll tackle a competitive map in a few days.) There are many ways to define a constitutional map. For my purposes, I looked at a few key provisions.

First and foremost are the federal guidelines for legislative districts: equal population (with no more than a 10% variance between the largest and smallest district), contiguous, and compact. The map must also conform to certain provisions of the Voting Rights Act.

The New Jersey Constitution also provides guidelines. Legislative districts must be “as nearly compact and equal in the number of their inhabitants as possible,” and “no county or municipality shall be divided among a number of Assembly districts larger than one plus the whole number obtained by dividing the number of inhabitants in the county or municipality by one-fortieth of the total number of inhabitants of the State.”

In other words, to determine the maximum number of districts that a town or county should have, divide the town or county’s population by 219,797 (one-fortieth of 8,791,894) and round up. For example, Bergen County’s population of 905,116 yields five allowable districts (i.e. 4.1 rounded up to 5).

The municipal/county split provision can be disregarded if it is necessary to meet other federal and state constitutional mandates (i.e. population equality, contiguity, compactness). However, the legislative map drawn up in 2001 appears to have violated this provision more than was statistically necessary. In addition to splitting Newark and Jersey into three districts, 15 of 21 counties have more district “splits” than the formula allows.

My objective in drawing this map was simply to see whether it is possible to draw a map that strictly adheres to the state constitution. Bottom line: it’s not possible. But you can come close. I was able to reduce the number of counties with over-splits to seven – and only one of those counties included more than one “extra” district (for a total of 8 county over-splits, compared to 24 in the current map).

[Click for the following: a list of municipalities in each district, the district characteristics, and the 2001 map characteristics.]

I purposely drew this map totally blind to a number of other important concerns that the commission may take into account, such as include minority group representation, current party control of districts, partisan voting patterns, and incumbent hometowns. That doesn’t mean I am unaware of the consequences that such a map would bring. I’ll discuss those below, but first the map itself.

This so-called constitutional map (larger version linked here) meets standards of contiguity and is relatively compact. [Note: The map is not the most compact possible, but “compactness” has been very broadly defined by the courts.] It also meets guidelines for population equality. The largest district, LD15, has 231,723 people, while the smallest, LD12, has 210,245. The difference between the two (21,478) is just under the 10% federal limit – 9.8%, to be exact.

Reducing the number of districts in Newark and Jersey City from three to two was easier in the former city than the latter. Because of the size and layout of the Hudson County municipalities, the existing majority-minority district shifted from 33 to 32. However, this change is somewhat semantic, since the West New York and Guttenberg moved from 33 to 32 as well.

The need to keep small (i.e. below 220,000 population) counties whole shifted some Cumberland County towns into LD3 and put more of Atlantic County in LD1. This necessitated some wholesale municipality shifts in LD4 through LD7. Similarly, Hunterdon and Warren counties needed to be placed into different districts to preserve their county boundaries.

Because of their shrinking population share, the Bergen/Essex/Passaic districts were basically redrawn, with one district disappearing entirely. That district, LD34, reappears in southern Somerset County. The middle of the state also saw a total makeover of LD14.

Now, on to the consequences. What would this new map do in terms of minority representation? Answer: probably not much. (See my prior column for a discussion of this issue.)

There would still be one Hispanic majority district. The number of districts with a 40% to 49% Hispanic population would increase from three to five, while the number with between 30% and 39% would decrease from three to one. A total of 12 districts – as compared to the current 13 – would have a Hispanic population of 20% or greater.

There would also continue to be one Black majority district, but the number of districts with a 30% to 49% Black population would decrease from five to one. However, a total of 13 districts – as compared to the current 12 – would have a Black population of 20% or greater.

Also, the number of districts with an Asian population over 20% would remain stable at five.

Next question: What would this new map do in terms of partisan control of the legislature? Answer: probably not much, but it would make for some interesting match-ups this June and November.

Based on voting patterns over the past few election cycles, 23 districts in the current map are in the Democratic domain and 16 are Republican. The remaining one, LD14, has been a split district for most of the past decade and has seen a number of close elections. Looking at the current map another way, 26 districts are “sure-fire” partisan – defined as an average victory margin of 10,000 votes or more for one party or the other. Another 13 are strongly partisan, with average victory margins of 3,000+ votes for a single party.

The new map doesn’t change that much: 23 districts are slam-dunks for one party and 12 are strong likelihoods. The remaining five districts lean toward one party by at least a 1,000 vote margin on average (four Democratic and one Republican). The bottom line gives us 23 Democrat trending districts and 17 Republican ones.

The real fireworks come in the incumbent face-offs that would be a byproduct of this map. Keep in mind that I was purposely blind to who lived where when I drew the map.

There are some Senate match-ups that will ensure this map is nixed by both parties. On the Democratic side, Senate President Steve Sweeney and Donald Norcross both land in LD5 and Nick Scutari and Ray Lesniak share LD20. On the Republican side, we have Joe Kyrillos and Jen Beck in LD13 and Tony Bucco and Joe Pennacchio in LD26.

Cross-party incumbent face-offs are led by Minority Leader Tom Kean and Bob Smith in LD22, where the Republican would have a slight advantage based on recent voting patterns. GOP stalwart Gerald Cardinale squares off with Bob Gordon in LD39, which leans to the Democratic side. The most interesting race would involve a primary between Nia Gill and former Governor/Senate President Dick Codey in LD27, with the winner likely defeating Kevin O’Toole in the general election.

On the Assembly side, Rible, Angelini, and O’Scanlon would fight it out for the two GOP spots in LD11 and Schroeder, Russo, and Vandervalk would do the same in LD40. Three Democrats, Johnson, Vaineri-Huttle, and Voss would draw straws in LD37.

A GOP leaning LD22 would pit Democrats Green and Stender against Republican Braminck, LD21 teams Majority Leader Cryan and Jasey in a likely win over Munoz, and LD4 includes Democrats Moriarty and Burzichelli along with Republican DeCicco, who will likely take on Fred madden in the Senate contest. LD30, which leans slightly Democratic but could be a toss-up, has three Democrats – Conaway, Benson, and DeAngelo and one Republican – Malone. Somerset County Assembly members Chivukula (D) and Coyle (R) are the sole legislative incumbents in the new LD34 and could face each other for the Senate seat in what is likely to be a Republican victory.

This map appears to meet all the basic federal and state Constitutional criteria for drawing legislative districts. However, I’m not necessarily advocating for its adoption. It may not be the best map for the state – in fact, it probably isn’t. But it does show what happens if you try to stick to certain rules over other considerations.

Next up – my take on a more competitive map.