30 April 2016

Soylent Green Is Inefficient

Contrary to popular perception, cannibalism is not an efficient use of scarce resources.  If we want to maximize planetary carrying capacity, vegetables are the way to go.

BYU Loves Rapists, Hates Rape Victims

Brigham Young University is a Mormon educational institution. This means that it applies religious values rathe than secular morality.

So, it isn't at all surprising that it routinely persecutes students who report being raped while taking no action to punish the rapists. According to Sarah Westerberg, BYU's Title IX coordinator, "almost all of the reported rapes and assaults at BYU are false reports made by women that feel . . . morally bad after they're having consensual activities." 

What a remarkable official policy.  But, why would anyone expect anything else from a religious institution? 

In the eyes of Mormons, like most Christians, this is what Jesus would do.  And, both the Bible and the Book of Mormon are full of instances of women being treated like shit with God's blessing and endorsement.  So, maybe they're right.

29 April 2016

Republican Presidential Race Nomination Recap

Donald Trump, Ted Cruz and John Kaisch are all continuing to seek the Republican party nomination for President to face presumptive Democratic party nominee Hillary Clinton this year.

Bottom line: I give Trump an 88% chance of winning the nomination (and a 13.2% chance of being elected President), and an 82% chance of a Clinton victory in the general election, mostly because Trump is weak vis-a-vis Hillary Clinton, but also because I think that Cruz is weaker vis-a-vis Hillary Clinton than polling so far would suggest because Cruz's flaws aren't as well known as they might be.

CNN's prediction market affiliate has a 78% chance of a Clinton Presidency and a 19% of a Trump Presidency, which is within reasonable margins of error of my own predictions, which suggests that my views are well within the mainstream of informed opinion on these issues.  And, it isn't impossible that my liberal bias is pulling me to overstate Clinton's chances somewhat.

The Results So Far

There are 2,472 delegates to the National Republican Convention in Cleveland this July.

Trump has 1,002 pledged delegates and he needs 42% of the remaining delegates (271 of 564 including remaining super delegates) to win the Republican Presidential nomination on a first vote at the convention.  After that first vote, the analysis becomes much more complex because pledged delegates cease to be pledged to their candidate at that point.  The magic number at the Republican convention in Cleveland is 1,273 delegates.

It is mathematically impossible for anyone else to win on the first round, even if they are supported by all uncommitted delegates who can vote their preference in the first round.  Ted Cruz has 571 pledged delegates and one super delegate who has endorsed him.  John Kaisch has 157 pledged delegates.  Marco Rubio has 167 pledged delegates. Ben Carson has 9 pledged delegates.  Jeb Bush has 1 pledged delegate.

Rubio and Bush have given Cruz a lukewarm endorsement, and Kaisch has implied that he will join them in that lukewarm endorsement.  Carson has given a lukewarm endorsement to Trump.  But, none of those endorsements releases the delegates for those candidates in the first round or binds those delegates in later rounds.

Key Facts

Unlike the Democratic primary and caucus process, almost all of the remaining GOP races are winner take all, or winner take all partially at the state level and partially at the Congressional district level.

Republican party rule 40(b), which would have to be amended by a supermajority at the convention to change, limits the convention to considering only candidates who have won eight or more states, i.e. to either Donald Trump or Ted Cruz.  Rule 40(b) excludes John Kaisch, Marco Rubio, Jeb Bush, Ben Carson, or some "white knight" candidate who didn't run, from receiving the nomination unless both Trump and Cruz drop out of the race.  (In theory, Kaisch could still win enough of the remaining states to qualify in addition to the one state, Ohio, that he has already won, but in reality this is simply impossible given his past performance and current polling among like Republican voters in the remaining primary states.)

Remaining Races With Predictions

The remaining Republican Presidential nomination races are as follows (with polling aggregated per Real Clear Politics when available):

May 3

Indiana - 57 delegates - Trump 37.5, Cruz 35.2, Kaisch 18.0 - No clear prediction

May 10

Nebraska - 36 delegates - Conventional wisdom favors Cruz (clean sweep) (Trump zero)

West Virginia - 34 delegates - Conventional wisdom favors Trump (almost clean sweep) (Trump 33)

May 17

Oregon - 28 delegates - No clear prediction

May 24

Washington -  44 delegates (proportional) - No clear prediction (At least 15 for Trump)

June 7

California - 172 delegates - Trump 45.7, Cruz 28.2, Kaisch 18.0 - No clear prediction

Montana - 27 delegates - Conventional wisdom favors Cruz (clean sweep) (Trump zero)

New Jersey - 51 delegates - Trump 52.0, Cruz 18.0, Kaisch 24.0 (Trump 51)

New Mexico - 24 delegates (proportional) - No clear prediction (At least 8 for Trump).

South Dakota - 29 delegates - Conventional wisdom favors Cruz (clean sweep) (Trump zero)

Analysis of The Rest Of Primary Season

It is much harder to robustly predict the outcome of the GOP race, because the winner take all nature of the contests (with incomplete Congressional district level data) creates a large swing in close races such as the Indiana race. The three way races also leave open the possibility of tactical moves in which Cruz or Kaisch drops out of contention in a state to give the other non-Trump candidate a shot at depriving Trump of those delegates.

And, the significant share of the remaining delegates that Trump needs leaves much less margin for error.

Analysis of the races give us 5 races to consider further and 84 Trump delegates from easier to predict races, so he would need 187 additional delegates from those 5 races and uncommitted delegates to win in the first round.  If we make a conservative assumption that he will win at least a third of the delegates in the two states with a proportional allocation, his Trump's quota for Indiana, California and Oregon is 164 delegates out of the 257 delegates at stake in those states.

Trump leads in polling in both Indiana and California, which certainly gives him a viable shot at a first round win of the nomination and makes him even more of a front runner than he would have been otherwise.  He not only leads in delegates, but has a path to victory paved with states where he leads in the polls.

But, it is close in Indiana where the close race could make for a wide swing of possibilities. Certainly, Trump gets some Congressional district delegates and Cruz gets other Congressional district delegates in a close race, but a switch of Kaisch voters to Cruz, as the those two candidates suggested briefly, could shift the balance decisively to Cruz in Indiana and deny Trump any delegates from Indiana.

Oregon sits between Trump territory and Cruz territory and I have seen no polling from this mixed winner take all state, so it is hard to predict.

Most California delegates are awarded at the Congressional district level, rather than at the state level where Trump should win if current polling holds up, so neither Trump nor Cruz will get all of California's delegates in all probability, but the polling seems to favor Trump decisively in California overall.

If Trump wins decisively in Indiana, he will probably win the GOP nomination on the first round, as there is little else to dent his momentum before the race goes to California.  But, if Trump loses decisively in Indiana, he has even odds at best of winning a first round vote in the Convention.

At this point, I give Trump a 60% chance of winning on the first round.

Analysis of a Possible Contested Convention

If Trump secures 1,237 pledged delegates backing him in the first round of the convention, or falls a few delegates short but manages to win the backing of the handful of GOP super delegates in the first round, it is all over and Trump is the GOP nominee.

But, if Trump fails, the question is whether the Convention would suspend Rule 40(b) to allow someone other than Trump or Cruz to be considered, and if not, whether Trump or Cruz would win in a second round vote at the Convention.

Trump will undisputedly have secured more delegates by far, more of the popular vote by far, and more states in the Republican primary than Cruz or anyone else.  A Convention vote for Cruz will look like a coup and weaken him, perhaps irremediably with Trump supporters.  In contrast, supporters of other candidates may be disillusioned by a Trump win, but aren't nearly as likely to bolt the party in the end and won't be in a position to claim that Trump's win was illegitimate.

Trump isn't loved by the GOP establishment, but most of the candidates who were eliminated from the race after winning delegates (except Carson) have reluctantly endorsed Cruz, but almost nobody in the GOP establishment likes Cruz.  For example, former Speaker of the House Hastert denounced Cruz as Lucifer incarnate yesterday.  And, Trump has a much greater position to wheel and deal with personal or political favors for delegates than Cruz who has fewer resources and more unbreakable political commitments (and is simply less of a negotiator).

Cruz has tried hard to plant disloyal individuals as Trump delegates, and maybe he's succeeded.  It is hard to know as an outsider to the process.

I give Trump at least a 55% chance of winning in a second round vote against Cruz.  Combined with my prediction that he has a 60% chance of winning on the first round, this gives Trump an 88% chance of winning the GOP nomination in my estimation, with Cruz having perhaps a 10% chance of winning the nomination, and some other white knight having perhaps a 2% chance.

The crystal ball will be much less cloudy on Tuesday night when we have the results from Indiana.

Onto the General Election

In a Trump v. Clinton race, I would give Clinton an 80%-90% chance of winning, call it 85% to split the difference.  Her current margin in national polls over Trump is almost thee standard deviations, but I expect that to fall to quite a bit more than one standard deviation, but less than two standard deviations (95% chance) by the time the general election rolls along as Trump positions himself closer to the center and moderates his publicly expressed views and maybe even apologizes for some of his previous incendiary statement.

In a Cruz v. Clinton race, I would give Clinton a 75% chance of winning which is probably an underestimate given the legitimacy issues that Cruz will face and the fact that lots of Americans haven't yet had time to get to know him and that doing so will reduce his esteem in the eyes of the average voter.  Clinton leads Cruz in national polls by about one standard deviation, but I expect that the gap will grow over time.

In a White Knight v. Clinton race, I would give Clinton a 50% chance of winning (given Kaisch's strong polling against Clinton, but the true legitimacy problem that any such candidate would face).

This gives us a 16.7% chance of a Republican President in the 2016 general election and a 82% chance of a President Clinton and a 1.3% or so chance of a President Sanders (considering less than 2% but more than 1% chance of him being the nominee and the strong likelihood that he would do as well or better than Clinton in a general election under those circumstances).

So the odds regarding the likelihood of various persons being our next President in my estimation are:

* President Hillary Clinton 82%
* President Donald Trump 13.2%
* President Cruz 2.5%
* President Sanders 1.3%
* President Kaisch 1.0%

* Democratic President 5/6th
* Republican President 1/6th

Democratic Presidential Nomination Race Recap

As I will explain below, there is something like a 98% chance that Hillary Clinton will be the Democratic Party's nominee for President of the United States.  You can safely ignore all developments in the remainder of the Democratic Party primary season and not miss anything.

Pay attention to the GOP race between Trump and Cruz instead, where there is at least a ghost of a chance of something interesting happening, even though Trump is overwhelmingly favored there as well.

The Results So Far

Clinton needs 17.4% of the remaining delegates (207 more) to win the Democratic Presidential nomination.

She leads Sanders in pledged delegates 1666 to 1359 (55% of those awarded so far) with 1,016 pledged delegates left to be awarded. A majority of the pledged delegates is 2,021, a threshold which she needs 355 more pledged delegates to surpass (about 35% of the remaining pledged delegates). Sanders needs to win 65% of the remaining pledged delegates to win a majority of the pledged delegates which he needs to make a moral claim that the super delegates should shift their support to him (in the absence of some new scandal impairing Hillary Clinton's prospects that provide an alternative basis to make this claim).

Clinton needs 720 delegates to have an outright majority of all of the delegates in the convention before even considering a single super delegate.  This is only 84 delegates more than she would get in the worst case realistic scenario for Clinton that I outline below, in which she gets at least 636 more pledged delegates, something that could very conceivably happen or at least come very close to happening, if she wins 40%-45% in states where Sanders is favored (instead of the 20% that I've assumed) and 70%+ in states where she is a strong favorite (instead of the 50% that I've assumed) - which would win her about 67 more delegates than I have assumed in a realistic worst case scenario for her.

She leads Sanders in super delegates 510 to 41, with 174 super delegates still undecided at this point. She has endorsements from more than 92% of the already committed super delegates and from 70% of all of the outstanding super delegates.

Key Facts

All Democratic primaries and caucuses in the Presidential race allocate delegates proportionally.

The winner will be decided in the first round vote at the Democratic National Convention, because only two candidates are winning.

Clinton has gotten at least 20% of the vote in every state except Vermont, where she came close.

Remaining Races With Predictions

The remaining Democratic Presidential nomination races are as follows (with polling aggregated per Real Clear Politics when available):

May 3
Indiana - 83 delegates - Clinton 49.6%, Sanders 43% (Clinton expected delegates 41)

May 7
Guam - 7 delegates (Clinton expected delegates at least 2)

May 10
West Virginia - 29 delegates - conventional wisdom favors Clinton in West Virginia (Clinton expected delegates at least 15)

May 17

Kentucky - 55 delegates - conventional wisdom favors Clinton in Kentucky (Clinton expected delegates at least 28)

Oregon - 61 delegates - conventional wisdom favors Sanders in Oregon (Clinton expected delegates at least 12)

June 4

Virgin Islands - 7 delegates (Clinton expected delegates at least 2)

June 5

Puerto Rico - 60 delegates (Clinton expected delegates at least 12)

June 7

California - 475 delegates - Clinton 49.0%, Sanders 42.3% (Clinton expected delegates more than 238)

Montana - 21 delegates - conventional wisdom favors Sanders in Montana (Clinton expected delegates at least 4)

New Jersey - 126 delegates - Clinton 51%, Sanders 42% (Clinton expected delegates more than 63)

New Mexico - 34 delegates (Clinton expected delegates at least 7)

North Dakota - 18 delegates - conventional wisdom favors Sanders in North Dakota (Clinton expected delegates at least 4)

South Dakota - 20 delegates - conventional wisdom favors Sanders in South Dakota (Clinton expected delegates at least 4)

June 14

District of Columbia - 20 delegates (Clinton expected delegates at least 4).


Minimum expected additional Clinton delegates assuming 20% where polling is not available and she is not clearly favored by conventional wisdom, 50% where polling is available and she is favored by conventional wisdom, and per her polling assuming all undecided voters go to Sanders where polling is available: 636

This is basically a worst realistic case scenario.  In it, she is almost sure to win both a majority of the pledged delegates and a majority of all of the delegates by a wide margin.

This prediction is highly robust because the assumptions made are very conservative and because the proportional delegate allocation rules make huge shifts in the popular vote necessary to cause significant shifts in the delegate count, especially when so many of the delegates have already been pledged, buffering any shift in the late polling that could influence the number of delegates awarded.

In this scenario, she can be abandoned by 84% of her super delegates and still win the race.  And, there is no reason to expect that even half of her super delegates will abandon her in favor of Sanders if Sanders does not win even a majority of the pledged delegates, which he will not.

In reality, Hillary Clinton will almost surely win more than 636 pledged delegates in the remaining races, and could afford to lose even more of her super delegates and still win.  Since Clinton's super delegate advantage will be a non-issue, the legitimacy of the nomination process will be untarnished, and Sanders will almost certainly acknowledge defeat and urge his supporters to back Clinton in the general election.

The odds of Sanders winning the Democratic Presidential nomination is much less than 1% in the absence of a major new scandal implicating Hillary Clinton, or a major deterioration in Hillary Clinton's health (e.g. an assassination prior to the Democratic National Convention like the one that killed Robert Kennedy).  The odds of either of those things happening is at most 1-2% or so in my estimation.

The odds the Hillary Clinton will be the Democratic Presidential nominee is on the order of 98%.

The Sanders campaign has basically conceded this mathematical reality, has laid off hundreds of staffers as a result, and is now focusing on putting together a large slate of delegates to influence the party's platform and spreading his message, rather than on winning the Presidential nomination itself.

Secular Parenting Usually Produces Good Kids

Secular Parenting

One of the things that makes me most proud is that my children, who have been raised non-religiously, have each been publicly recognized for their good character both in elementary school and again middle school.  My daughter, in her pitch to become the President of George Washington High School's National Honor Society (spoiler: she won) told her audience that "service is my thing."

My family's experience isn't unique.
More children are “growing up godless” than at any other time in our nation's history. . . 
Far from being dysfunctional, nihilistic and rudderless without the security and rectitude of religion, secular households provide a sound and solid foundation for children, according to Vern Bengston, a USC professor of gerontology and sociology. For nearly 40 years, Bengston has overseen the Longitudinal Study of Generations, which has become the largest study of religion and family life conducted across several generational cohorts in the United States. When Bengston noticed the growth of nonreligious Americans becoming increasingly pronounced, he decided in 2013 to add secular families to his study in an attempt to understand how family life and intergenerational influences play out among the religionless. He was surprised by what he found: High levels of family solidarity and emotional closeness between parents and nonreligious youth, and strong ethical standards and moral values that had been clearly articulated as they were imparted to the next generation.

“Many nonreligious parents were more coherent and passionate about their ethical principles than some of the ‘religious' parents in our study,” Bengston told me. “The vast majority appeared to live goal-filled lives characterized by moral direction and sense of life having a purpose.”

[N]onreligious family life is replete with its own sustaining moral values and enriching ethical precepts. Chief among those: rational problem solving, personal autonomy, independence of thought, avoidance of corporal punishment, a spirit of “questioning everything” and, far above all, empathy. . . .

[S]ecular teenagers are far less likely to care what the “cool kids” think, or express a need to fit in with them, than their religious peers. When these teens mature into “godless” adults, they exhibit less racism than their religious counterparts, according to a 2010 Duke University study. Many psychological studies show that secular grownups tend to be less vengeful, less nationalistic, less militaristic, less authoritarian and more tolerant, on average, than religious adults. . . .

Secular adults are more likely to understand and accept the science concerning global warming, and to support women's equality and gay rights. . . . Atheists were almost absent from our prison population as of the late 1990s, comprising less than half of 1% of those behind bars, according to Federal Bureau of Prisons statistics. This echoes what the criminology field has documented for more than a century — the unaffiliated and the nonreligious engage in far fewer crimes.

Democratic countries with the lowest levels of religious faith and participation today — such as Sweden, Denmark, Japan, Belgium and New Zealand — have among the lowest violent crime rates in the world and enjoy remarkably high levels of societal well-being. If secular people couldn't raise well-functioning, moral children, then a preponderance of them in a given society would spell societal disaster. Yet quite the opposite is the case.
From the Los Angeles Times.

Note also, with respect to the prison population, that the average incarcerated person is younger than the average adult, because most crimes are committed by adolescents and young adults. A random sample of U.S. adults with the same age and ethnic makeup as the population of adults incarcerated in the United States is much more likely to be religiously unaffiliated than that general population of the United States which includes a much larger share older adults who are much more likely to be religious.

Secular Living And Social Class

If your intuition is that this is a case of high socio-economic class values, or ethnic values, transmission from parents to children, you are mostly, but not entirely, wrong.

Non-Hispanic whites and people who are more educated or more affluent, are more likely to be secular.  But, the differences in likelihood of being non-religious along class and ethnic lines are far more modest than most people would naively assume.

For example, people who are religiously "unaffiliated" make up about 20% of whites, 16% of Hispanics and 15% of African-Americans.

The breakdown (from the same source) by education is as follows:

The breakdown by income is similarly, only modestly tilted towards affluence relative to the general public:

An atheist or agnostic has the same likelihood of having "some college" or being in a lower middle income range as members of the U.S. general public, even though an atheist or agnostic is less likely to have no college education and is less likely to be in the lowest income category.

Secular population of the United States is much younger than the religiously affiliated population of the United States, however. "A plurality of atheists and agnostics (42%) are ages 18 to 29, and just 9% are 65 and older. By comparison, about one-fifth of the religiously affiliated (18%) are ages 18 to 29, and a similar portion are 65 and older (19%)."

This is not because secular people tend to convert to a religion when they become adults. It is uncommon for people raised non-religiously to adopt a religion as an adult. President Obama, who was raised non-religiously, but converted to Christianity as an adult, is an outlier in that respect.

Complicating parenting issues, however, there is also a significant gender gap in religious affiliation. Men are much more likely to not have a religious affiliation or to be actively atheist or agnostic, than women.

Secular people age 30+ are also about 10% less likely to be married than religious people in that age bracket (only some of which is attributable to a younger age distribution within that age bracket), and secular people under age 30 are almost half as likely to be married as religious people in that age bracket (again, with only some of that difference attributable to a younger age distribution within that age bracket).

Most secular people were raised in religious households.  Only 4% of households in the 1950s that produced the Baby Boomers were non-religious, and only 11% of the households that produced Generation X were non-religious.  Far larger shares of both of these age groups are non-religious today.  But, given that 30% of people aged 18-29 are religiously unaffiliated, the age people are most likely to be when they first have children, this may not be true for much longer.

What Do Secular People Believe?

About three in ten people who are not religiously affiliated specifically identify as atheist or agnostic.

About one in ten religiously unaffiliated people are looking for a religion that is right for them, but haven't found it yet.

About four in ten religiously unaffiliated people consider themselves to be "spiritual but not religious."

The likelihood that someone without a religious affiliation will have spiritual or supernatural beliefs not commonly viewed as religious in the United States (e.g. believing in astrology, ghosts, or the existence of magical spells) is roughly the same as the likelihood that religiously affiliated people hold such beliefs.

U.S. Teen Birth Rate At All Time Low

Teen birth rates are lower now than they have ever been in the history of the United States, although they remain high by the standard of other developed countries.

According to the Center for Disease Control the trend towards lower teen pregnancy rates has continued with only modest interruption since 1991 and was particularly pronounced in the eight years from 2006 to 2014 (the latest year for which data are available). "In 1991, the birth rate among females age 15 to 19 was 61.8 per 1,000. As of 2014, that number has declined to 24.2 per 1,000."
Birth rates are down a whopping 51 percent among Hispanics age 15 to 19 since 2006, and down 44 percent among black teens, according to a survey of census data by the Centers for Disease Control and Prevention. Teen pregnancy rates among whites also fell by a third.
From here

This has reduced racial and ethnic disparities in teen birth rates.  Hispanic teen birth rates (38.0 per 1,000) are now about 9% more than black teen birth rates (34.9 per 1,000), which are in turn about twice the white teen birth rate (17.3 per 1,000).

Greater use of contraception and especially long term contraception has been an important factor in the change: "use of long-acting contraceptives like IUDs and implants jumped from 1 percent of teens a decade ago to 7 percent in 2014." But, teens are also simply having sex less often.

Abortion rates per teen pregnancy have been more or less constant during this time period (resulting in dramatically fewer teen abortions as a result).