Quo Vadis GOP?

[Razgovory]

They will tell the courts that the map was design to favor Republicans.

[Valmy]

They will tell the courts that the map was design to favor Republicans.

I mean if Hispanic and black voters voted Republican, and whites voted Democratic, it is amazing how progressive Republican maps would suddenly become.

The real problem is having politicians draw those districts.

[Admiral Yi]

I think this is a misleading question, as evidenced by DGs marching about with great sound and fury.

The question should be more toward "Were these districts drawn to exclude minorities to an undue degree?" or even "Was there intent to strengthen a position of whites in the formation of these districts?"  A simple cold and analytical question such as yours seeks to withdraw the intent from the debate.

It is not a misleading question because, as DGuller has been pointing out in the face of much abuse, the chart is trying to demonstrate that intent, and in my and his opinion, doing it in a misleading way.

[The Minsky Moment]

Numbers are not means to an end.  Numbers are powerful tools.

Numbers are not "tools" they are elements of human language, and human language is pragmatic, flexible and contextual.  If someone says "You look like a million bucks" - they are not presenting the results of a formal valuation report on your outfit.  If someone agrees to give 110% effort it doesn't mean they are Siamese twins each contributing 55.  Not every human conversational interaction that includes a numerical expression is a formal thesis.  In context, it wasn't reasonable to read Zoupa's comments as proposing a precise calculation - indeed the point of the comments was a disclaim mathematic precision but to make an emphatic expression that there is real impact here.  And he's right, there is.

[Admiral Yi]

indeed the point of the comments was a disclaim mathematic precision but to make an emphatic expression that there is real impact here.  And he's right, there is.

Are you talking about Zoupa's comment or the MSNBC table?

[The Minsky Moment]

the chart is trying to demonstrate that intent, and in my and his opinion, doing it in a misleading way.

Before making that point, one would have a clear idea of what the chart is saying.  I don't know what they mean by associating "control" of seats in reference to race.

If "control" means the percentage of districts in which persons of the indicated race are a majority within that district, it is arguably misleading - but not for the reasons being argued.  It is misleading because in our two party system, where both Blacks and (so far . . .) most Hispanic groups lean Democrat, the relevant measure is probably the Black+Hispanic vote in the district. 

But putting that aside, what the chart does presumably shows is that the white and Hispanic population are virtually identical in Texas but that whites form the majorities in 60% of districts and Hispanics 18%.  It is extremely difficult to conceive how such a result could occur without very significant and deliberate manipulation.  It is mathematically possible to imagine such a result occurring randomly based on some set of facts known to be untrue?  Yes - but why on earth should we care about that theoretical point?

The chart is a classic case of where there's smoke there's fire.  Those numbers are billowing black smoke coming out the the building. And when we go out in the street and also see people fleeing the building with burns, and masses of fire trucks converging on the scene, and the cops making arson arrests  - - sure it's possible that one doesn't have absolute 100% positive proof of fire.  But it's a pretty silly point to make, no?

[The Minsky Moment]

indeed the point of the comments was a disclaim mathematic precision but to make an emphatic expression that there is real impact here.  And he's right, there is.

Are you talking about Zoupa's comment or the MSNBC table?

zoupa.

[Admiral Yi]

But putting that aside, what the chart does presumably shows is that the white and Hispanic population are virtually identical in Texas but that whites form the majorities in 60% of districts and Hispanics 18%.  It is extremely difficult to conceive how such a result could occur without very significant and deliberate manipulation.  It is mathematically possible to imagine such a result occurring randomly based on some set of facts known to be untrue?  Yes - but why on earth should we care about that theoretical point?

Conceded.

However I think you will agree that the number for blacks is much more problematic for the good guys.

[The Minsky Moment]

However I think you will agree that the number for blacks is much more problematic for the good guys.

Taken on its own sure.  Because if Black neighborhoods are (e.g.) near Hispanic neighborhoods (like West and East Harlem in NYC) then it might be difficult to draw districts where Blacks are an absolute majority.  But one would expect to see a good number of Black + Hispanic majority districts.  Unfortunately the chart doesn't report that, and in that sense it can be argued to be misleading.  We know the total can't be more than 40% (because white majority districts are at 60 - again, assuming that's the proper interpretation of the chart) but the picture would look different if the total were 20 as opposed to 40.

[DGuller]

I realize this won't matter at all, but I decided to get the data myself and see what I can gleam from it.  Luckily Texas makes the census data easily available.

I haven't gone far with it yet, but I'm already seeing a one bad sign for the intellectual honesty of the people who created that table.  I could indeed exactly replicate the figures in the first column, they appear to be taken from the P002* columns on the census data.  The problem?  That's total population.  Yes, Hispanics had a much higher growth rate than other races, but that means that a lot of them are under 18 for now.  If you switch over to the more appropriate P004* counts of 18+ population, the split is now 43% white, 36% Hispanic, and 12% Black.

Does it make a radical difference whether the gap is 1% or 7% between whites and Hispanics?  I don't know yet.  I wish things were as obvious to me as they are to Minsky, but I guess I've just seen too many counterintuitive things in my field to jump to conclusions as readily as he does.  Sometimes small difference in the averages can make for big differences in truncated tail quantities, and the second column appears to be such a quantity.  It seems like the second column is defined as "percent of counties with the given race making up more than 50% of the population", which is definitely the kind of quantity where small differences can get accentuated.

[PDH]

I think this is a misleading question, as evidenced by DGs marching about with great sound and fury.

The question should be more toward "Were these districts drawn to exclude minorities to an undue degree?" or even "Was there intent to strengthen a position of whites in the formation of these districts?"  A simple cold and analytical question such as yours seeks to withdraw the intent from the debate.

It is not a misleading question because, as DGuller has been pointing out in the face of much abuse, the chart is trying to demonstrate that intent, and in my and his opinion, doing it in a misleading way.

Was the "much abuse" needed?  Recently I have caught the whiff of victimhood here by people who feel some sort of existential dread.

[Admiral Yi]

Was the "much abuse" needed?  Recently I have caught the whiff of victimhood here by people who feel some sort of existential dread.

Was "marching with great sound and fury" needed?

[DGuller]

I realize this won't matter at all, but I decided to get the data myself and see what I can gleam from it.  Luckily Texas makes the census data easily available.

I haven't gone far with it yet, but I'm already seeing a one bad sign for the intellectual honesty of the people who created that table.  I could indeed exactly replicate the figures in the first column, they appear to be taken from the P002* columns on the census data.  The problem?  That's total population.  Yes, Hispanics had a much higher growth rate than other races, but that means that a lot of them are under 18 for now.  If you switch over to the more appropriate P004* counts of 18+ population, the split is now 43% white, 36% Hispanic, and 12% Black.

Does it make a radical difference whether the gap is 1% or 7% between whites and Hispanics?  I don't know yet.  I wish things were as obvious to me as they are to Minsky, but I guess I've just seen too many counterintuitive things in my field to jump to conclusions as readily as he does.  Sometimes small difference in the averages can make for big differences in truncated tail quantities, and the second column appears to be such a quantity.  It seems like the second column is defined as "percent of counties with the given race making up more than 50% of the population", which is definitely the kind of quantity where small differences can get accentuated.

Actually, I'm going to hold on this conclusion for now.  The second column appears to be the percentage of US Congress seats where the race is 50%+ of the district.  If that's the case, then I think it should aim to split the total population evenly, not the voter population.

[DGuller]

One factor that works against Hispanics specifically (but not black people) is that they're naturally gerrymandered much like the urban population.  If you look at census tracts, which are purely statistical area definitions unlike congressional districts, Hispanics are much more all-or-nothing than whites.  Out of all the census tracts, 12% have Hispanics making up 80%+ of their entire population, but only 3% of census tracts have whites making up 80%+ of their population. 

Here is another interesting result:  if you look at it by counties, of which there are 254 in Texas, 58% of them are majority white, and 27% of them are majority Hispanic.  That is despite whites and Hispanics being essentially tied on total population.  Now, I'm not going to suggest that this is a fair comparison to Congressional districts, because counties can be much smaller than congressional districts, which makes it easier for them to get lopsided, and Congressional districts are supposed to be equal on population whereas counties can be hugely unequal.  I'm just putting this figure out there to show why one has to be really careful with potential counterintuitive effects; no one defined countries to be gerrymandered by race, and yet they happened to gerrymander themselves almost as much as the Congressional districts.

[PDH]

Was the "much abuse" needed?  Recently I have caught the whiff of victimhood here by people who feel some sort of existential dread.

Was "marching with great sound and fury" needed?

Of course it was, I am not as foul mouthed as Seedy, but I do sometimes like to stir the fire.