Gerrymandering, mathematically explained

[dismal]

Define "good".

I don't know. I am asking you what your point was. Your point was not clear. Can you please explain the point you were making from that graphic? I can't tell if you were supporting or refuting the district, or how you thought it was good or bad if you indeed thought it was good or bad.

I thought my point was pretty damn clear. Because areas with high concentrations of Democrats are near each other you can't mix them with areas of moderate Republican concentration without spaghetti districts. If you define mixing high democrat areas into moderate republican areas to create proportional representation then I imagine you would think creating spaghetti districts to do it is "good". If you think spaghetti districts are the great evil then you'd probably have to live with some disproportional representation, unless there were some series of offsetting errors that happened to make things proportionate overall.

- - - Updated - - -

It's also that they are concentrated in certain states. You can't win more seats in New England or the Pacific coast than the New England and Pacific Coast states have, no matter how much you win those seats by.

However in both 2010 and 2014 it was the Republicans that got a lot more votes. Maybe the Democrats would have better luck focusing on that.

I thought data was shown that in several states the democrats got many more votes but repubs got the seats. Can you show your source for "in both 2010 and 2014 it was the Republicans that got a lot more votes."

Are you sure your head won't explode if facts are introduced?

Click at your own risk:

http://en.wikipedia.org/wiki/United_States_House_of_Representatives_elections,_2014

http://en.wikipedia.org/wiki/United_States_House_of_Representatives_elections,_2010

 

[Jimmy Higgins]

It wasn't that good of an article. Population is population. A population being dense for one party can help lead to gerrymandering, but it isn't the cause of disproportionate representation.

The point is that if you have a neutral algorithm that produces compact districts with no attempt to skew districting in a partisan way it can still produce disprportionate representation due to the way population is distributed. You would either need to actively gerrymander or else give up on single member districts.

Assuming an absolutely naked algorithm, it would be more honest.

In Ohio, the counties with cities are 44% of the states population.

In 2012 the Republicans received 50% of the vote and Democrats 46% of the vote in House of Rep races in the state of Ohio. The Republicans won 12 seats, the Democrats 4. So even with a naked algorithm that only gave Democrats seats where Dems were packed up in bunches, they should be taking home no less than 7. Dems winning that year had at least 70% of the vote.

 

[NobleSavage]

Who is drawing the lines? Isn't it the state legislatures? I remember way back when the Republicans took Congress in '94 with Newt and the "Contract with America". George Will and others were going off on how bad the Democratic gerrymandering was.

 

[Derec]

Who is drawing the lines? Isn't it the state legislatures? I remember way back when the Republicans took Congress in '94 with Newt and the "Contract with America". George Will and others were going off on how bad the Democratic gerrymandering was.

Kind of like how who is outraged by plans to get rid of the filibuster depends on who has 50<x<60 seats in the Senate.

 

[Jimmy Higgins]

Who is drawing the lines? Isn't it the state legislatures? I remember way back when the Republicans took Congress in '94 with Newt and the "Contract with America". George Will and others were going off on how bad the Democratic gerrymandering was.

Kind of like how who is outraged by plans to get rid of the filibuster depends on who has 50<x<60 seats in the Senate.

Moore-Coulter?

Republicans blocked how many nominees for how many positions, how many bills, how many judicial appointees? Democrats filibustered 8 or so under W and you'd swear they blocked thousands.

 

[Blahface]

The best solution is to use the shortest-split line-algorithm to draw districts. In addition, there should be non-partisan primaries with approval voting. Even if you just happen to get bad boundaries drawn due to chance, the second part will still ensure there are competitive elections and a representative that is in the true center for the population.

 

[Rhea]

I don't know. I am asking you what your point was. Your point was not clear. Can you please explain the point you were making from that graphic? I can't tell if you were supporting or refuting the district, or how you thought it was good or bad if you indeed thought it was good or bad.

I thought my point was pretty damn clear.

Nah, it wasn't.

Because areas with high concentrations of Democrats are near each other you can't mix them with areas of moderate Republican concentration without spaghetti districts. If you define mixing high democrat areas into moderate republican areas to create proportional representation then I imagine you would think creating spaghetti districts to do it is "good". If you think spaghetti districts are the great evil then you'd probably have to live with some disproportional representation, unless there were some series of offsetting errors that happened to make things proportionate overall.

Why do we want to mix dems with moderate Rs?

Well, I guess the Teabaggers would want that. But I don't know why a Dem would.

The better config for dems would be conservatives diluted by _either_ moderate Rs _or_ dems.

It's also that they are concentrated in certain states. You can't win more seats in New England or the Pacific coast than the New England and Pacific Coast states have, no matter how much you win those seats by.

However in both 2010 and 2014 it was the Republicans that got a lot more votes. Maybe the Democrats would have better luck focusing on that.

I thought data was shown that in several states the democrats got many more votes but repubs got the seats. Can you show your source for "in both 2010 and 2014 it was the Republicans that got a lot more votes."

Are you sure your head won't explode if facts are introduced?

Click at your own risk:

http://en.wikipedia.org/wiki/United_States_House_of_Representatives_elections,_2014

http://en.wikipedia.org/wiki/United_States_House_of_Representatives_elections,_2010

Nope, no explosions. I didn't have the data and wanted to see it, thank you. I realize that I was thinking 2012 and you were noting 2010 and 2014. I have nothing at all against data, and I often ask for it.

 

[Rhea]

Oh, I see what you did there...

in 2014
Repubs won 56.8% of the seats with 50.9% of the votes
Dems won 43.2% of the seats with 45.3% of the votes

And you say this demonstrates that the system worked properly for representation of the actual population.

 

[Axulus]

Oh, I see what you did there...

in 2014
Repubs won 56.8% of the seats with 50.9% of the votes
Dems won 43.2% of the seats with 45.3% of the votes

And you say this demonstrates that the system worked properly for representation of the actual population.

What do you mean by "worked properly"? People have a representative for a geographic area as well as by population. If Democrat voters tend to cluster together in higher concentrations compared to Republicans (which the data say they do), then you'll see the districts represented by democrats have a higher percent of democrats, on average, compared to the percent of Republicans in republican districts.

Imagine you have a state that is 50% democrat and 50% republican, 10,000,000 population. You have 10 districts. In 4 of those districts (urban) representing 4 million population, you have 60% democrat (2.4 million people) and 40% republican (1.6 million people) on average. Democrats win all 4 districts. In the remaining 6 districts (suburban and rural), you have 57% Republican (3.4 million people) and 43% Democrat (2.6 million people), on average. The Republicans win 6 districts.

The result? 50% of the voters are D and 50% are R, yet the Republicans won 60% of the seats. No gerrymandering required to get this kind of result on occasion.

 

[Jimmy Higgins]

Oh, I see what you did there...

in 2014
Repubs won 56.8% of the seats with 50.9% of the votes
Dems won 43.2% of the seats with 45.3% of the votes

And you say this demonstrates that the system worked properly for representation of the actual population.

What do you mean by "worked properly"? People have a representative for a geographic area as well as by population. If Democrat voters tend to cluster together in higher concentrations compared to Republicans (which the data say they do), then you'll see the districts represented by democrats have a higher percent of democrats, on average, compared to the percent of Republicans in republican districts.

And how does that work? That Republican majority districts don't see the crazy 70%+ vote totals?

Imagine you have a state that is 50% democrat and 50% republican, 10,000,000 population. You have 10 districts. In 4 of those districts (urban) representing 4 million population, you have 60% democrat (2.4 million people) and 40% republican (1.6 million people) on average. Democrats win all 4 districts. In the remaining 6 districts (suburban and rural), you have 57% Republican (3.4 million people) and 43% Democrat (2.6 million people), on average. The Republicans win 6 districts.

The result? 50% of the voters are D and 50% are R, yet the Republicans won 60% of the seats. No gerrymandering required to get this kind of result on occasion.

And in the real world Republicans won 75% of the seats in Ohio with only 50% of the vote.

 

[fromderinside]

And in the real world Republicans won 75% of the seats in Ohio with only 50% of the vote.

Since there are rules on how much districts can vary in size within a state it is necessary for the party in control to gerrymander so that things work out that republicans get move bang for their votes than do democrat, such as 4.2 million votes get only four seats while 3.4 million votes gets six seats. Small but consistent margins are guaranteed for republicans and large consistent margins are forced on democrats. Whodathunk.

 

[Axulus]

What do you mean by "worked properly"? People have a representative for a geographic area as well as by population. If Democrat voters tend to cluster together in higher concentrations compared to Republicans (which the data say they do), then you'll see the districts represented by democrats have a higher percent of democrats, on average, compared to the percent of Republicans in republican districts.

And how does that work? That Republican majority districts don't see the crazy 70%+ vote totals?

Imagine you have a state that is 50% democrat and 50% republican, 10,000,000 population. You have 10 districts. In 4 of those districts (urban) representing 4 million population, you have 60% democrat (2.4 million people) and 40% republican (1.6 million people) on average. Democrats win all 4 districts. In the remaining 6 districts (suburban and rural), you have 57% Republican (3.4 million people) and 43% Democrat (2.6 million people), on average. The Republicans win 6 districts.

The result? 50% of the voters are D and 50% are R, yet the Republicans won 60% of the seats. No gerrymandering required to get this kind of result on occasion.

And in the real world Republicans won 75% of the seats in Ohio with only 50% of the vote.

They got 60% of the vote. Not sure where you are getting your numbers from. With 60% of the vote you can potentially win every seat if all districts were even split 60% R and 40% D (which they aren't). Winning 75% of the seats while having 60% of the votes, on average, is not in any way surprising.

http://en.wikipedia.org/wiki/United_States_House_of_Representatives_elections_in_Ohio,_2014

 

[dismal]

Oh, I see what you did there...

in 2014
Repubs won 56.8% of the seats with 50.9% of the votes
Dems won 43.2% of the seats with 45.3% of the votes

And you say this demonstrates that the system worked properly for representation of the actual population.

No. You could try reading what I actually say.

 

[Jimmy Higgins]

And how does that work? That Republican majority districts don't see the crazy 70%+ vote totals?

Imagine you have a state that is 50% democrat and 50% republican, 10,000,000 population. You have 10 districts. In 4 of those districts (urban) representing 4 million population, you have 60% democrat (2.4 million people) and 40% republican (1.6 million people) on average. Democrats win all 4 districts. In the remaining 6 districts (suburban and rural), you have 57% Republican (3.4 million people) and 43% Democrat (2.6 million people), on average. The Republicans win 6 districts.

The result? 50% of the voters are D and 50% are R, yet the Republicans won 60% of the seats. No gerrymandering required to get this kind of result on occasion.

And in the real world Republicans won 75% of the seats in Ohio with only 50% of the vote.

They got 60% of the vote. Not sure where you are getting your numbers from.

The 2012 election.

Wanna have more fun. Look at 2008 where the Democrats took 51% of the vote and won 10 of the 18 seats. Did Ohio recently turn into Mississippi between '08 and '12?

 

[dismal]

And how does that work? That Republican majority districts don't see the crazy 70%+ vote totals?

Imagine you have a state that is 50% democrat and 50% republican, 10,000,000 population. You have 10 districts. In 4 of those districts (urban) representing 4 million population, you have 60% democrat (2.4 million people) and 40% republican (1.6 million people) on average. Democrats win all 4 districts. In the remaining 6 districts (suburban and rural), you have 57% Republican (3.4 million people) and 43% Democrat (2.6 million people), on average. The Republicans win 6 districts.

The result? 50% of the voters are D and 50% are R, yet the Republicans won 60% of the seats. No gerrymandering required to get this kind of result on occasion.

And in the real world Republicans won 75% of the seats in Ohio with only 50% of the vote.

They got 60% of the vote. Not sure where you are getting your numbers from. With 60% of the vote you can potentially win every seat if all districts were even split 60% R and 40% D (which they aren't). Winning 75% of the seats while having 60% of the votes, on average, is not in any way surprising.

http://en.wikipedia.org/wiki/United_States_House_of_Representatives_elections_in_Ohio,_2014

Let's take Massachusetts 2014. The Democrats won all 9 house seats. Some of the house races were uncontested, but Markey won his senate seat 62-38 while winning a majority in all 9 congressional districts.

http://www.politico.com/2014-election/results/map/senate/massachusetts/#.VPYDiZ3naUk

Assuming the 38% is representative of the Republican share of the vote, is the progressive view (and the Higgins et al argument) that Massachusetts should if necessary draw up spaghetti districts that give the Republicans 3 or 4 of those house seats?

Massachusetts is a supposed to be a progressive state so they really ought to get after that.

 

[Axulus]

And how does that work? That Republican majority districts don't see the crazy 70%+ vote totals?

Imagine you have a state that is 50% democrat and 50% republican, 10,000,000 population. You have 10 districts. In 4 of those districts (urban) representing 4 million population, you have 60% democrat (2.4 million people) and 40% republican (1.6 million people) on average. Democrats win all 4 districts. In the remaining 6 districts (suburban and rural), you have 57% Republican (3.4 million people) and 43% Democrat (2.6 million people), on average. The Republicans win 6 districts.

The result? 50% of the voters are D and 50% are R, yet the Republicans won 60% of the seats. No gerrymandering required to get this kind of result on occasion.

And in the real world Republicans won 75% of the seats in Ohio with only 50% of the vote.

They got 60% of the vote. Not sure where you are getting your numbers from.

The 2012 election.

Wanna have more fun. Look at 2008 where the Democrats took 51% of the vote and won 10 of the 18 seats. Did Ohio recently turn into Mississippi between '08 and '12?

Ok, in 2012 Ds got 47% of the popular vote vs 51% for Rs. If the districting was somehow perfectly "neutral", and there was no clustering in the districts, then wouldn't one potentially expect Rs to win all 16 seats, with their 4 point average advantage? Because there is in fact clustering, 4 seat win doesn't necessarily mean gerrymandering is a deciding factor. And even if there is a bit of gerrymandering going on (which there is), wouldn't this potentially be balanced out with D states engaging in similar tactics?

In an election with 47% D and 51% R, how many seats would you expect Rs to win vs Ds with no gerrymandering? Since 51% of the vote wins the entire seat, please demonstrate why you would no expect all the seats to go to the Rs in such a scenario.

My main point is that the critics haven't made a strong case that the issue makes much of a difference.

 

[Jimmy Higgins]

And how does that work? That Republican majority districts don't see the crazy 70%+ vote totals?

Imagine you have a state that is 50% democrat and 50% republican, 10,000,000 population. You have 10 districts. In 4 of those districts (urban) representing 4 million population, you have 60% democrat (2.4 million people) and 40% republican (1.6 million people) on average. Democrats win all 4 districts. In the remaining 6 districts (suburban and rural), you have 57% Republican (3.4 million people) and 43% Democrat (2.6 million people), on average. The Republicans win 6 districts.

The result? 50% of the voters are D and 50% are R, yet the Republicans won 60% of the seats. No gerrymandering required to get this kind of result on occasion.

And in the real world Republicans won 75% of the seats in Ohio with only 50% of the vote.

They got 60% of the vote. Not sure where you are getting your numbers from.

The 2012 election.

Wanna have more fun. Look at 2008 where the Democrats took 51% of the vote and won 10 of the 18 seats. Did Ohio recently turn into Mississippi between '08 and '12?

Ok, in 2012 Ds got 47% of the popular vote vs 51% for Rs. If the districting was somehow perfectly "neutral", and there was no clustering in the districts, then wouldn't one potentially expect Rs to win all 16 seats, with their 4 point average advantage? Because there is in fact clustering, 4 seat win doesn't necessarily mean gerrymandering is a deciding factor. And even if there is a bit of gerrymandering going on (which there is), wouldn't this potentially be balanced out with D states engaging in similar tactics?

In an election with 47% D and 51% R, how many seats would you expect Rs to win vs Ds with no gerrymandering? Since 51% of the vote wins the entire seat, please demonstrate why you would no expect all the seats to go to the Rs in such a scenario.

My main point is that the critics haven't made a strong case that the issue makes much of a difference.

My main point was merely that when the Democrats won 51% of the vote, they won 10 of the 18 seats. When the Republicans won 50% of the vote, they won 12 of the 16 seats. Now I'm not an engineer or good with numbers, but if I were... I'd say there appears to be a discrepancy here. A pretty damn obvious one.

 

[Crazy Eddie]

It wasn't that good of an article. Population is population. A population being dense for one party can help lead to gerrymandering, but it isn't the cause of disproportionate representation.

The point is that if you have a neutral algorithm that produces compact districts with no attempt to skew districting in a partisan way it can still produce disprportionate representation due to the way population is distributed.

No it cannot, since the whole point of the algorithm is to produce districts with roughly the same population. Districts with higher population densities would automatically be smaller geographically than districts with lower densities.

There's no need to gerrymander at all. Just accept that each district will have X number of people, with X being a portion of the state's overall population. Equal representation, problem solved.

 

[lpetrich]

Geography matters because the heavily democrat areas tend to be next to each other. This "ZOMFG it's population not area" argument does nothing to solve this fact. To get the heavily Democrat areas into a mixture with moderate Republican areas requires spaghetti districts.

View attachment 2417

That is as bad as the original one ( Gerrymandering):

 

[lpetrich]

Let's take Massachusetts 2014. The Democrats won all 9 house seats. Some of the house races were uncontested, but Markey won his senate seat 62-38 while winning a majority in all 9 congressional districts.

http://www.politico.com/2014-election/results/map/senate/massachusetts/#.VPYDiZ3naUk

Assuming the 38% is representative of the Republican share of the vote, is the progressive view (and the Higgins et al argument) that Massachusetts should if necessary draw up spaghetti districts that give the Republicans 3 or 4 of those house seats?

Massachusetts is a supposed to be a progressive state so they really ought to get after that.

Seems like proportional representation would be a MUCH better solution. Everybody gets represented in proportion to their numbers. No worry about how to draw single-member districts to accommodate them.

Proportional representation is widely used in countries that score high on human development (UN:  List of countries by Human Development Index,  List of countries by inequality-adjusted HDI) and democracy (The Economist magazine:  Democracy Index) and low as fragile states (The Fund for Peace). Furthermore, it is often adopted in constitutional reforms over the last century or so ( Table of voting systems by country).

So what's not to like about it? Is it un-American?