Quo Vadis GOP?

[The Brain]

Which country needs fixing next?

[DGuller]

I decided to try my hand at Texas redistricting.  You never know what skills might come in handy one of these days.  I tried to create 38 districts with only geographical compactness as the criteria, as well as making sure that all of them have between 700,000 and 800,000 people.  I haven't figured out how to post images on the forum, however. 

Looking at preliminary results, I think my computer is racist.     My computer is giving Hispanics 16% of the districts instead of 18% that Texas GOP does.  It definitely doesn't pass the most obvious of tests suggested by Zoupa and Minsky, despite focusing merely on creating tight clusters of roughly equal populations.  I must be missing something, Minsky assured me that the math here is simple. 

If there is a funny business afoot, it's with the percentage of White districts.  Texas GOP has it at 60%, while my algorithm so far has it at 37%. 

So far my conclusion is that it is indeed possible for even purely geographic districting algorithm to start with equal populations, more or less, and result in vastly unequal numbers of majority districts.  Therefore, it pays to be careful like Yi was, and ask oneself the question of what should reasonable numbers be in a fair process, instead of assuming that you know what they can and can't be without funny business.  The gap between 37% and 16% is still pretty unintuitive, given that whites and Hispanics are equal in population at around 40% each.  "How come Hispanics have less than half the districts their population entitles them to, when whites are at parity?"

Obviously we know that Texas proposed districts are packed and cracked to the max to benefit the Republicans, and so by proxy they probably benefit whites as well.  That doesn't mean that every argument in support of that notion is automatically valid and complete.  Not examining arguments that lead to a conclusion you're very sure is right is a road to hell, and it doesn't matter that it's paved with good intentions.

[The Minsky Moment]

Who is getting the other 47%?

[DGuller]

Who is getting the other 47%?

Same people who get the other 22% of districts in the original picture, I imagine.  60% + 18% doesn't add up to 100% either. 

The other 47% of the districts have no majority.  They're probably 3-way splits, but it doesn't have to be just a 3-way split, there are other categories not included in Whites, Hispanics, or Blacks.  I'm ignoring them because the original exhibit is ignoring them.

[Razgovory]

See, this is why we had laws against witchcraft.

[The Minsky Moment]

The other 47% of the districts have no majority.  They're probably 3-way splits, but it doesn't have to be just a 3-way split, there are other categories not included in Whites, Hispanics, or Blacks.  I'm ignoring them because the original exhibit is ignoring them.

What do the numbers look like in these districts you are ignoring?

[The Minsky Moment]

To be clear - assuming we accept DG's map as a fair or at least neutral one, and we continue the interpret the other map as before (and we assume no other ethnicity has a majority of a District), then gong from DG to the GOP map has the effect of reducing the number of no ethnic majority districts from 47% to 22% (i.e. the number of such districts are but more than half). 

Putting concrete numbers based on 38 house districts, that means the GOP plan has 8 no majority districts as compared to 18 on the DG plan, or a swing of 10 districts (or 9 depending on the exact nos and the rounding errors). Of those 10 districts, 1 becomes majority Hispanic (after rounding) and 9 become majority white on the GOP plan.

Seems to me this exercise confirms the problem, even before we investigate what is going on in the no absolute majority districts.

[DGuller]

The other 47% of the districts have no majority.  They're probably 3-way splits, but it doesn't have to be just a 3-way split, there are other categories not included in Whites, Hispanics, or Blacks.  I'm ignoring them because the original exhibit is ignoring them.

What do the numbers look like in these districts you are ignoring?

I'm not ignoring any districts, and for that matter I'm not ignoring any people.  All of 29 million Texans are being assigned, it's just that I'm counting the numbers of districts where one of the three mentioned groups have 50%+ majority.  In GOP scheme, 78% of districts have such a majority, whereas in my scheme 53% of districts have such a majority.  It's all just mirroring the choices made in the exhibit, it's not necessarily my preferred choice of going about it.

[DGuller]

To be clear - assuming we accept DG's map as a fair or at least neutral one, and we continue the interpret the other map as before (and we assume no other ethnicity has a majority of a District), then gong from DG to the GOP map has the effect of reducing the number of no ethnic majority districts from 47% to 22% (i.e. the number of such districts are but more than half). 

Putting concrete numbers based on 38 house districts, that means the GOP plan has 8 no majority districts as compared to 18 on the DG plan, or a swing of 10 districts (or 9 depending on the exact nos and the rounding errors). Of those 10 districts, 1 becomes majority Hispanic (after rounding) and 9 become majority white on the GOP plan.

Seems to me this exercise confirms the problem, even before we investigate what is going on in the no absolute majority districts.

I agree that an existence of a problem is confirmed.  It's just that the problem wasn't what seemed obvious.  It seemed obvious to some that the 18% number was the damning one, but actually it's the 60% number that's revealing.  The point that I was trying to make is that we have to be careful and thoughtful, I wasn't claiming that a careful analysis would validate GOP's numbers.

As far as what goes on in no-majority districts, many of them do become Hispanic+Black majority if you merge the two.  Here are my latest numbers (slightly different from earlier due to the algorithm still churning):  39% of districts are majority white, 16% of districts are majority Hispanic, and 0% of districts are majority Black.  If you merge the last two, 47% of districts are majority Hispanic+Black.  Yes, 16% + 0% = 47% here, math is fun sometimes.

What's the percentage of Hispanic+Black districts in the GOP map?  We don't know, we can't tell from the original exhibit, but obviously it can't be greater than 40% given that Whites have 60% of them locked up.

[Valmy]

5.5% of Texans are Asian-Americans as well.

[The Minsky Moment]

Thanks DG and apologies for the less than charitable comment above.

[DGuller]

Thanks DG and apologies for the less than charitable comment above.

[Zoupa]

So after all that condescension and algorithm and whatever number crunching giving you a semi, you conclude that my initial statement of "more than 18%" is accurate.

Sometimes a cigar is just a cigar, a statement is just a statement,and playing devil's advocate is pointless.

[DGuller]

So after all that condescension and algorithm and whatever number crunching giving you a semi, you conclude that my initial statement of "more than 18%" is accurate.

I don't think that is my conclusion at all, I don't think you were reading my posts carefully.  In fact I believe m conclusion was the opposite.  I also don't think discussing the status of my penis is adding anything to the conversation.

[Zoupa]

Let me rephrase to try and bypass your obtuse road block:

Do you feel 18% of districts in Texas being majority black or Latino is an accurate representation of the state's demographics in 2021? Could this number be brought more in line through an impartial, geographic density model instead of what's currently being used to define district borders?