FL: Rereredistricting

[jimrtex]

How does one distinguish between a racially-motivated gerrymander and an effort to promote minority interests when considering the cross-bay district?

After they drew the Hillsborough-district, they scored it, and then asked the legislative counsel. In the part that I have listened to, they haven't heard back yet.

The lawyers for the senate and the house said that blacks and Hispanics were not cohesive in the primary and so the map-drawers should go into another county to maintain a black opportunity-to-elect district.

This could prove interesting in the upcoming trial. In the congressional case, the lawyers for the house and senate failed to cross-examine the plaintiff's expert witness, and whether he had considered primary results in his analysis.

When considering the possibility of a Hispanic opportunity seat in the Orlando area, the lawyers for the legislature were of the opinion that there was not sufficiently racially polarized voting to warrant creation of a district.

In the senate case, the plaintiffs presented their Hillsborough-only district as an alternative, and then formally withdrew it. They must have got push back from somewhere.

It appears that the House has not presented a remedial plan.

There's a lot of advice about performing minority districts from legislative attorneys. Is their advice disputed by any of the other parties?

We'll see at the trial. Also the briefs on problems with the opposition proposals were due by last Wednesday, but have not been posted yet.

2012 CA 002842: THE LEAGUE OF WOMEN VOTERS OF FLORIDA vs KENNETH W DETZNER AS FLORIDA SECRETARY OF STATE

The legislature attorneys got smoked by the plaintiffs in the congressional trial (other than the Romo plaintiffs who were trying to argue that the pairing of two Democratic representatives was politically motivated).

There were 3 expert witnesses. One was from the University of Utah, and had a heavy (Chinese) accent, which the plaintiff's attorney was able to exploit, and make it appear as if he didn't know anything. He had done an analysis on election results in South Florida. He had specified the criteria for elections, which had then been supplied to him. But in questioning, the cross-examining attorney makes it sound like they cherry-picked the elections.

The second was a political scientist/Republican consultant form Miami, and apparently of Cuban descent. He was having a lot of fun fencing with the plaintiff's attorney, but came off as being very anecdotal. Whereas the Utah professor knew nothing about Florida politics, he probably knew too much, and could recall particular elections. He had also been part of the preclearance submission in 2012, when he said that a South Florida district with almost the identical racial composition of the SD-26 proposed by the plaintiffs was a performing Hispanic opportunity district.

The plaintiff's expert was Allan Lichtman. Because he had a scheduled flight, they agreed to accept his expert report rather than have direct examination by the plaintiff attorney. The legislature lawyers then passed on cross-examining him. In their post-trial briefs, they said that he had not included primary results in his functional analysis. The judge said that they could have cross-examined him.

[Nyvin]

So after the trial on December 14th,  will the maps be official and we'll know what's decided on?

[jimrtex]

So after the trial on December 14th,  will the maps be official and we'll know what's decided on?

I suspect it will be appealed to the SCOFLA.

The Supreme Court has not ruled on the congressional district map, even though they had the hearing three weeks ago.

[Miles]

You must be logged in to read this quote.

[Nyvin]

You must be logged in to read this quote.

Awesome news!

http://www.firstcoastnews.com/story/news/local/florida/2015/12/02/florida-supreme-court-approves-new-congressional-map/76672702/

[/]

Sabato's new ratings:



Whoever redistricted these is kind of an a**hole; they basically just shoved out two incumbents entirely. As much as I dislike Webster, this is just cruel. Especially with Graham.

[Adam Griffin]

RIP GWEN GRAHAM

[Nyvin]

Sabato's new ratings:



Whoever redistricted these is kind of an a**hole; they basically just shoved out two incumbents entirely. As much as I dislike Webster, this is just cruel. Especially with Graham.

Graham really did get screwed, and that does suck since she's an awesome rep.

Webster doesn't actually have it all that bad,  I'm thinking he's gonna move to FL-11 since Nugent is retiring.  

Graham really needs to convince Murphy to drop out of the Senate race so she can run in it.    That way he can stay in FL-18 and she has an election she can win.

[jimrtex]

So after the trial on December 14th,  will the maps be official and we'll know what's decided on?

I suspect it will be appealed to the SCOFLA.

The Supreme Court has not ruled on the congressional district map, even though they had the hearing three weeks ago.

They issued their opinion. SCOFLA Opinion (PDF)

Take a look at the map on page 71 and, then the one on page 67 and see if you notice anything.

It is of interest that they said (page 67, footnote 11) the diminishment standard (AKA Section 5) does not apply if the minority is not cohesive. This suggests that at least some parts of the Gingles test are applicable to retrogression claims.

But it would also imply that Florida could be required to satisfy a Section 2 challenge based on the overall Hispanic population in Miami-Dade County.

The senate district that crosses Tampa Bay would be illegal under the current Florida Constitution, and not required under Section 2 of the VRA, because it is not compact. But can the non-retrogression standard be used to maintain the existing district?

[Nyvin]

So after the trial on December 14th,  will the maps be official and we'll know what's decided on?

I suspect it will be appealed to the SCOFLA.

The Supreme Court has not ruled on the congressional district map, even though they had the hearing three weeks ago.

They issued their opinion. SCOFLA Opinion (PDF)

Take a look at the map on page 71 and, then the one on page 67 and see if you notice anything.

It is of interest that they said (page 67, footnote 11) the diminishment standard (AKA Section 5) does not apply if the minority is not cohesive. This suggests that at least some parts of the Gingles test are applicable to retrogression claims.

But it would also imply that Florida could be required to satisfy a Section 2 challenge based on the overall Hispanic population in Miami-Dade County.

The senate district that crosses Tampa Bay would be illegal under the current Florida Constitution, and not required under Section 2 of the VRA, because it is not compact. But can the non-retrogression standard be used to maintain the existing district?

That's an interesting opinion piece.   I read through a good portion of it.   

They directly brought up the gerrymandering issue on page 88.   

[jimrtex]

So after the trial on December 14th,  will the maps be official and we'll know what's decided on?

I suspect it will be appealed to the SCOFLA.

The Supreme Court has not ruled on the congressional district map, even though they had the hearing three weeks ago.

They issued their opinion. SCOFLA Opinion (PDF)

Take a look at the map on page 71 and, then the one on page 67 and see if you notice anything.

It is of interest that they said (page 67, footnote 11) the diminishment standard (AKA Section 5) does not apply if the minority is not cohesive. This suggests that at least some parts of the Gingles test are applicable to retrogression claims.

But it would also imply that Florida could be required to satisfy a Section 2 challenge based on the overall Hispanic population in Miami-Dade County.

The senate district that crosses Tampa Bay would be illegal under the current Florida Constitution, and not required under Section 2 of the VRA, because it is not compact. But can the non-retrogression standard be used to maintain the existing district?

That's an interesting opinion piece.   I read through a good portion of it.   

They directly brought up the gerrymandering issue on page 88.   

That was a concurring opinion written by one of the two black justices (and concurred in by the other) and was largely unrelated to the case at hand.

[jimrtex]

Is this better? I increased the chop count by 1 to decrease the range from 7.2% to 5.3%. With 2 more chops in Hillsborough and Broward I could reduce it to 3.8%.

This would be much better than what they used.

It would have three splits beyond what was needed to divide counties entitled to more than one district.

How do you decide which is best?

7.2% and two excess splits.
5.3% and three excess splits.
3.8% and five excess splits.

Your final effort would add a Palm Beach-(Broward) district and a (Hillsborough)-Pinellas district with just a tiny nibble of (Broward) and (Hillsborough) in the cross-border district. This would occur within areas of continuous settlement within large urbanized areas.

Election officials would not have a problem handling the extra district, since they already have so many, and the district would be in a single media market, and not require extra district offices.

[jimrtex]

I have just listened to the 10/11 meeting/conference call between the mapdrawers  and the legislature lawyers. The mapdrawers had prepared 10 statewide maps, and three for the southeast. Though they talked a lot about sandboxes, they only clearly used one, the Palm Beach--Broward--Miami-Dade--Monroe area. The rest of the time they spent a lot of time sliding districts around.

This zip file contains the draft maps, and you won't understand what I'm saying without looking at them.

Draft Maps

The recording of the 12:00 meeting would let you hear the actual discussion.

10/11 mapdrawing recording

When they sent the maps to the lawyers, they deliberately did not specify the methodology they had used in preparing them, so as to check whether they had consistently applied a methodology or not.

I still don't understand the difference in methodologies (PDF). If you open up the deviation range, then most splits disappear, and you don't need to have different methodologies. Fewer split counties generally means that there will be fewer splits.

They really weren't using different methods of drawing districts. It was more how they characterized the map after it was drawn.

Draft 37 illustrates a conceptual Hispanic opportunity district in the Orlando area (SD-8), which is 50.1% HVAP. HCVAP is probably a bit higher share of HVAP in Florida because a large share of the Hispanics are Cuban or Puerto Rican. But the 50.1% is not enough to ensure minimal control of a primary, as the share melts away when citizenship, registration, and actual voting are taken into account. The conceptual district is not at all Tier 2 compliant.

The lawyers had already looked at this district in isolation, and indicated that it failed the Gingles test, and wondered why the mapdrawers had even brought it up again.

Draft 38 was similarly dismissed.

Draft 47 had a quite interesting discussion. Notice the boundary between SD-1 and SD-3 in Orange County. The road is popularly known as the Bee Line (and was formerly the Bee Line Expressway). But it ends up with the two districts about 14,000 different in population.

The mapdrawers really liked the boundary, while the lawyers suggested that much deviation might only be permitted if a county line or city boundary were retained. The mapmakers seemed to be of the opinion that any deviation was OK so long as it did not increase the overall deviation range.

There was also concern about SD-8 (Baker to Sumter) even though it was a whole county district. Draft 47 is a methodology 2 map. It has 12 split counties and 29 county parts, and five districts wholly in one county (the map excludes South Florida).

By comparison, my 2nd plan has 10 split counties, 22 county parts, and 12 districts wholly in one county.

Draft 51 The lawyers were concerned about the non compactness of SD-12, SD-24, and SD-25. Even though SD-25 is made up of whole counties. The lawyers guessed this was a methodology 2 map, but the mapmakers thought it was a methodology 1 map.

Draft 52 The lawyers had concerns about SD-25 which cuts into St. Lucie and splits Indian River in a U-shape. In most drafts, the natural district of St.Lucie, Martin, Okeechobee was used. If they had put Martin in SD-24 they would have had a coast-to-coast district. They were attempting to hold the southern Polk-Osceola boundary. The map illustrates the failure of the mapmakers to create apportionment regions. Effectively, the map has a 11-district region stretching from Putnam to Hendry County. This was a methodology 1 map.

Draft 53 The lawyers were concerned about SD-4 (Baker to Sumter). It had been used on other maps, but it had not been in yellow. This is a methodology 2 map.  Draft 53 is similar to Draft 47, with the differences being in the Tampa Bay and Orlando areas. Because they ended up with four methodology 2 maps, and 3 methodology one maps, this was draft was dropped (at least at the time of the meeting with the lawyers, I have more tapes to listen to).

Draft 54 The mapmakers weren't sure whether this was methodology 1 or 2, and were concerned that the lawyers might think it would have legal problems because it had not been drawn with a consistent methodology. The lawyers particularly did not like SD-23 having two arms into Lee County, nor the triple split of Lee and 5-way split of Orange. Because it was of uncertain methodology it was apparently dropped.

Draft 55 The lawyers thought this was a methodology 2 plan, but the mapmakers thought it was methodology 1.

Draft 56 The lawyers liked the look of SD-13 in the Orlando area, but it is only 41.7% HVAP. They disliked SD-26 because of the way it hooked into St.Lucie. The plan has 37 county splits. The lawyers thought it was methodology 1, but the mapmakers methodology 2.

Draft 60 This had the same Lee County split that had been panned before. It also divided Highlands County. A 9-district apportionment region snakes down the east coast from Union County to Martin and then across the state and up to Manatee. The mapmakers apparently were having trouble keeping the deviations within range, which led them to split Highland. This is methodology 2.

The final three drafts are the South Florida drafts.

Draft 57 This is based on the Broward--Miami-Dade district being along the coast, modified to try to make SD-33 a black opportunity district, which it may barely be, by shifting it south into Miami-Dade. They could make it more black, but it would require more city splits. They did not do a functional analysis on SD-38 (53% HVAP, 19% BVAP).

Draft 58 This is based on a black opportunity district crossing the Broward--Miami-Dade line, which leaves SD-2 as possibly a black opportunity district. They did not do a functional analysis on SD-12 (65% HVAP and 14.5% BVAP). This map has the same split of the black population that the congressional map, putting Homestead and West Perrine (southern tip of SD-11) in different districts. The lawyers were sensitive to the double protrusion of SD-6 from Palm Beach into Broward (apparently when crossing a county boundary it should be done in one place. U-shapes are to be avoided, since it may indicate that you are avoiding an area (it also indicates a deficiency in most mathematical compactness metrics since, such a district may not score poorly).

Draft 59 This is probably the most similar to my map, other than Palm beach. The lawyers had a concern about how the Palm Beach-Broward district crossed the county line in two places.

[jimrtex]

This is the senate brief in the senate trial.

OBJECTIONS TO PLAINTIFFS' PROPOSED REMEDIAL PLANS (PDF)

It turns out the map drawn by Strategic Telemetry, the Democratic firm involved in the Mathismander drew the plaintiffs alternative, and consistently overpopulated Republican-leaning districts.

[jimrtex]

This was an experiment in paint by number with QDIS. It shows the apportionment regions I used for my senate map.



I graphically selected the counties in QDIS and then typed in the number of the region. I then displayed the layer based on the region number using a random color.

Ideally, I would want to click on a county and have its region be assigned, and population totals be automatically calculated and displayed.

[jimrtex]

The lawyers had criticized the Baker-Sumter district in Draft 47. The mapmakers combined two districts made up of whole counties, and then split Alachua.



The mapmakers then balanced population across districts.

[jimrtex]

This illustrates the regions used in Draft 61. There are 11 regions (compared to 18 in my plan). In Central Florida particularly, they tended to draw districts until they could match a county boundary, rather than the reverse process of establishing areas and then drawing the district within.



This is somewhat like how a simple redistricting application  might appear. My actual procedure was:

(1) Select counties and assign region numbers within QGIS.
(2) Merge region assignments with population data in Excel. I also calculated the populated-weighted average lat/long for each region. This was saved as a .csv file.
(3) Read the .csv file back into QGIS. The region statistics were rendered as points. They are superimposed over the regions, because of their calculated lat/long.

[jimrtex]

The trial began in the Senate case on Monday (December 14).

There was a motion to intervene by the county election supervisors which was granted. Their interest was to provide information on a necessary time schedule to conduct elections. Their preliminary estimate was March, which would give them time to realign precincts, and conduct filing, etc. for a summer primary.

There was a stipulation that the senate was claiming that the plaintiff's alternatives retrogressed in 4 South Florida districts. It appears that we will get to hear a lot about the Gingles test, and whether if the Gingles conditions are met, whether you actually have to have a minority majority in the district that is drawn.

The Senate's opening statement was next (the House is not a party). They emphasized the neutral process that was used, and that it had all been recorded, with attorney-client privilege waived.

He explained the difference between the two methodologies:

(1) Maximize the number of whole counties (the memo also said to maximize the number of whole districts drawn in counties).
(2) Reduce splits of counties. Here he gave an example of Hillsborough during the special session to draw congressional districts where there was a concern about the number of districts.

He explained the use of sandboxes, which was the term the map drawers used for whole-county areas with whole number of districts, though they were quite careless in their use of them.

He said the plaintiffs did not propose a map until late in the process when it would be too late to incorporate their ideas. Besides retrogressing, he suggested that the plaintiffs had overpopulated Republican districts.

The plaintiff's opening argument claimed that the burden was on the senate to prove their plan was constitutionally compliant, but the challengers did not have such a burden. They also said that the court only had to choose the "best map".

If the interest were to produce the best map, the court would have drawn the map, and opened the process to the public.

He then went into his claim that the senate picked the alternative that was best performing for Republicans, and that it was worse performing in a number of metrics, including population equality, in which he provided numbers on maximum deviation; standard deviation; and average (absolute deviation). The judge interrupted to ask what standard deviation was.

He also emphasized that the SCOFLA had never set a deviation standard but, instead that it would be necessary to justify the deviation. Since my map performs so much better in terms of county splits, the deviation is justifiable.

His claim is that Senator Galvano, the chair of the apportionment committee chose which of the 6 alternatives would go forward. After it had been approved by the senate, he drafted a new version that paired it with a different version of the South Florida map, but that this map was never considered by the senate, but is the one that is using in the court case.

He argues that it was chosen because it pairs the fewest senators.

[jimrtex]

The first witness for the senate was Jay Ferrin who was the senate mapdrawer. Since it was a senate map, I think he actually held the mouse.

Before he started the judge asked for the deviation within each "sandbox" which he referred to as a component (at least that is my understanding), since he referred to the 10 components (my map had 18 components).

Ferrin explained why the senate chose the alternative that it did. It was methodology 2, that had as a goal fewer county splits, and the particular map chosen had the least division in Hillsborough County. It was merged with the South Florida map that was most compact.

The senate lawyer then led him through a bunch of slides that compared the senate map, with the plaintiff map. There were two sets, one that overlaid the two maps, and showed that they were mostly the same, with small areas that hung out. When they differed, the lawyer would ask him to explain why someone might choose a particular t configuration, and it was explained on the basis of Reock score, which places a premium on adding big areas with no people in them.

The other set of slides showed individual district boundaries, and whether they coincided with county boundaries, city boundaries, primary and secondary roads, water features, or something else. In one instance, the senate map had followed a river (in Florida it looks like a lot of rivers have been channelized so they look like canals). The plaintiff's map had followed a straight line, and Perrin explained that was a VTD boundary, and then got to explain what a VTD boundary was.

During his opening statement, the plaintiff attorney had said that the senate had made up some metric about following roads, and said that nobody else used it - but no one else has a provision about following roads in its constitution, plus an interpretation by the state supreme court reinforcing their use (in Apportionment 1 back in 2012, the SCOFLA had rejected the senate map because it hadn't followed roads, etc.).

[Torie]

Is there any merit to factoring in the "reock" score as part of the Muon2 metrics? I presume not, but it might be worth exploring, after we all understand what the F it means. I sort of like the idea of thinking about empty areas, versus populated ones. How this actually works in application, I have no idea. In my Oregon map "published" today, I went for preferring adding empty spaces in Clackamas into the OR-05 chop into it, as part of the process of avoiding locality chops and state highway cuts.

[jimrtex]

Is there any merit to factoring in the "reock" score as part of the Muon2 metrics? I presume not, but it might be worth exploring, after we all understand what the F it means. I sort of like the idea of thinking about empty areas, versus populated ones. How this actually works in application, I have no idea. In my Oregon map "published" today, I went for preferring adding empty spaces in Clackamas into the OR-05 chop into it, as part of the process of avoiding locality chops and state highway cuts.

Compactness measures can be gamed. Florida added several measures to their reports, which then end up with the SCOFLA using them as the measure of "compactness" in the Florida constitution.

The Reock test compares the area of a district with a circumscribed circle. A circular district would score 1.00.  Regular polygons (polygons with equal sides) such as a square score pretty good, compared to irregular shapes.

Florida itself would have a low Reock score because the minimum circumscribed circle would have to be go around the northwest corner (north of Pensacola) to south of Miami and the Florida Keys, leaving vast areas in Gulf of Mexico outside the state, but inside the circle.

The Reock test was used in Ohio, and I got different scores based on what was included in the map. The contest organizers included two dredging spoils in Lake Erie east of Toledo. Lucas County is somewhat elongated in an east-west direction already, and it is obvious to a blind mathematical formula that I had stretched out my district to include the spoils. The state did not include the spoils and my district scored better with them.

But the best district came from Wood County and took in some of Lucas County, south of the Maumee River. Since the spoils were east of Toledo in the channel to the harbor, they weren't so extreme for a district that included Bowling Green.

The plaintiffs gamed the Reock score in the congressional case with districts 20, 26, and 27, and duped the Democratic justices on the SCOFLA.

FL-26 is the district that contains Monroe County, and thus the Everglades, Florida Keys and even Dry Tortugas which are significantly west of Key West, but all the interesting boundaries are in Miami-Dade. There is practically nothing that you could do to the district that would have much effect on the Reock Score. FL-27 is the district to the east, basically coming down from Miami. Much of that area is unpopulated, with places like Islandia with 18 persons, and a nuclear plant. The plaintiff map chopped this area off of FL-27 and made it more compact.

FL-20 is a black opportunity district in Palm Beach and Broward and now Miami-Dade county. The black population is somewhat inland from the beach, and in a sense the most compact district would be a long narrow district, but that tends to force other districts to wrap around the ends.

But it extends out to Lake Okeechobee and into Hendry County. Very few people live out there, but it makes for a mass that improves the compactness scores even though a third tentacle was added to the district. If you were look at a density map you would see three distinct physically separate areas, plus a small number at Belle Glade. Given the population in south Florida, the district is no more compact than FL-5.

When we use whole counties, there is no reason for using the more complex compactness measures which are designed particularly for maps that have intricate borders.  At worst you can have long strings of counties.

In the past, Iowa compared the north-south extent of a district with its east-west extent. This would often result in "L" shaped districts, which might have a bit of the inside part of the angle filled in.  But you could then put other "L" shaped districts inside. This can particularly be seen with the 2000 map.

Iowa congressional district boundaries.

Districts comprised of whole counties can also use the length of the borders as a compactness measurement. A 3 by 4 block of square counties is more compact than 9 counties strung end to end, whether in a long line, or one that snakes.

Erosity is Muon's compactness measure. Boundary length is mine.

[jimrtex]

I had watched the videos of the trial out of order.

The first witness was Jason Poreda, who was one of the two mapdrawers for the House.

On direct, he explained the overall procedure that had been used. In particular, he explained the sandbox concept. This begins around 20 minutes of the 2nd part of December 14. At 28:11 there is sandbox map similar those that Muon2 and I have posted.

To remember the sequence. I had looked at their maps, and said: "UGH lots of split counties", and had posted my weighted voting district map. Muon2 then posted his modification which permitted the weights to be closer to one.

I then began drawing my map. It was only after I started listening to the mapdrawer tapes that I realized they had used sandboxes. They were much less deliberate in use. Their sandboxes to a larger extent were reverse engineered after they had drawn the map.

They make a distinction between a district made up of whole counties, and an area with one or more counties that had multiple districts. I'm not sure why the distinction is made.

Their map had 10 sandboxes and 2 whole-county district. Mine has 18 sandboxes.

Cross of Jason Poreda was interesting.

It turned out that during the special session, that Poreda had been told by the chair of the House redistricting committee to look at the map passed by the senate, and a map proposed by the plaintiffs, and produced a map that was about 1/3 from the plaintiff's map, 1/3 from the senate map, and 1/3 about the same. I think this map may have actually been passed by the House.

If the House were a party to the case, they would likely be advocating for this map.

The cross was a lot about use of various metrics, trying to have Poreda say that one map was "best".

He was also asked about the use of BCVAP to measure whether a district was majority black. Poreda noted that CVAP numbers come from the ACS and are estimates.

[jimrtex]

The 4th Segment of December 14, 2015 is the cross-examination of Jay Ferrin.

The plaintiff's attorney asked a lot of questions in a manner that he might employ for a jury trial. I would be surprised if it had much effect on a judge.

He was asked about pairing of incumbents under the plan chosen by the senate. Ferrin had no knowledge of where senators lived. On re-direct he confirmed that he assumed that individual senators knew where they personally lived, but had no knowledge of whether senators knew where each other lived.

During the special session, Ferrin got a phone call from a senator, who started asking for a change, and Ferrin said to hold on a minute, and then announced that they were "on the record". The senator tried to explain it in abstract terms. but then expressed that two female Republican senators should be unpaired. Apparently he thought that was abstract enough since he didn't actually name them. Ferrin told he couldn't do that, but told him that if asked he could draft a plan that would remove a certain county from a district - it was his responsible to draft plans in response to senator requests. He reported the phone call to the Senate President.

He was also asked about the merging of the 28 districts in north of Florida, and 12 districts in south Florida. Having listed to the tapes, I knew that there had been 6 plans for northern Florida, and 3 plans for south Florida, and they were done independently. They could have produces all 18 combinations of the two areas, but instead they explained that they had inserted each of the southern Florida maps in two of the northern Florida plans. It was quite trivial to do since their software permits districts to be copied from one plan to another. The attorney was all amazed since it could be done in a very short while.

He was also pushing really hard on metrics. Hillsborough County includes the lower part of Tampa Bay all of the way to the Gulf of Mexico. If you include the tail in a Hillsborough-based district, it blows up the Reock score. You would think you would want to hide how you were gaming the metrics rather than badgering a witness about how more compact is better.

But what will do the senate in is that their plan was never offered. The senate had passed one of the draft maps. The chair of the senate committee had a plan drafted, that was never offered during the session, but is now the "senate plan".

The House had taken the plan passed by the senate, and then merged in some parts of a map submitted by the plaintiffs. In conference committee, the head of the senate redistricting committee agreed to take the House map to the senate, which voted it down - and that is where the special session broke down.

[muon2]

He was also pushing really hard on metrics. Hillsborough County includes the lower part of Tampa Bay all of the way to the Gulf of Mexico. If you include the tail in a Hillsborough-based district, it blows up the Reock score. You would think you would want to hide how you were gaming the metrics rather than badgering a witness about how more compact is better.

This was also my observation during the 2010 cycle. These tails on counties are not unique to FL. The artificial distortion of the score led to my notion of erosity based on connections not on shape.

[jimrtex]

December 15, 2015.

The senate presented the judge the information he wanted about the sandboxes/components. It wasn't I had expected, it was the deviation between components, rather than deviation within components.

The senate's expert witness was Baodong Liu. He was the expert witness at the congressional trial. To save time at that trial, the expert's reports had been submitted in lieu of direct examination. Dr. Liu has a heavy Chinese accent, and on cross-examination, Good Old Boy attorney ripped him apart. He had not selected the elections he had used for his analysis - but had specified the criteria, and had been supplied the data, but it made it look like the data had been cherry-picked. Meanwhile, the legislative lawyers did not cross-exam the plaintiff's expert. So the judge more or less decided everything was true.

This time, the senate attorney led him through detailed questions, beginning with his curriculum vitae. I think Dr. Liu may have been a bit embarrassed to be giving his qualifications orally, since it might sound like boasting, so there was a long series of questions about if he had published, where, etc. His expertise is particularly in racial voting.

He explained in general terms Goodman Regression, Double Regression, and Ecological Inference. The key point from the attorney's perspective was that Goodman Regression was old (dating from 1953) and Ecological Inference (EI) was new (1997). As it turns out, the expert for the plaintiff did not use EI.

Dr. Liu's findings were that blacks were cohesive, but Hispanic's were not outside southern Florida (ie Republican-voting Cubans are in the south).

He then displayed some charts, showing Black and Hispanic participation in elections as a function of BVAP% and HVAP%. The key element was that Black participation was higher in districts with higher BVAP.

For example, if BVAP% was 56%, Black participation might be 57%, while if BVAP% were 43%, Black participation might be 41%. It appears that if blacks control the election, they are more likely to actually vote. This led to the suggestion that BVAP must be over 50% for a district to be a performing district.

For Hispanic districts, HVAP% had to be in the 75%-80% range, for Hispanics to form a majority of the electorate. Dr. Liu indicated that HCVAP was not as useful since it is not available for small areas that match election results.

The benchmark (2002), senate, and plaintiff plans were then compared. For black districts, 2 senate districts had over 50% BVAP vs.1 for for the plaintiffs. I had none over 50%, and only one over 40%, compared to 3 for the other plans.

(There are 6 districts with high BVAP, one in Jacksonville, one in Orlando, one in Tampa-St.Petersburg, and three in Broward and Miami-Dade.).

For Hispanic districts, the benchmark had 3 over 75% HVAP, the senate plan 2, and 1 or 2 for the plaintiff plans. I had 2, and even had one over 90%.

The plaintiffs have claimed that they had 4 performing Hispanic districts, but based on this analysis they might only have 1 or 2.

The judge asked some questions about the table showing the number of districts with certain BVAP and HVAP percentages that he at least understood the argument that the senate was making.

Dr. Liu then did a critique of the report by the plaintiff's expert Allen Lichtman. He said that Lichtman had not used EI, had not included the output file from SPSS (which would include the parameters used during the analysis), and had based most of his conclusions on the congressional district races, and the 2010 senate race (Rubio (Hispanic Republican)-Crist (Anglo former Republican, future Democrat, independent)-Meek (black Democrat) threeway). He had also used repeated elections from the same district, which in reality are only a single data point.