Wednesday, January 5, 2011

Charles Seife's "Proofiness: The Dark Arts of Mathematical Deception"


This is an excellent look at how statistics is being mangled and manipulated by people who know better to achieve results that border on or go beyond the criminal. The first couple of chapters are dedicated to giving you a quick background in math and statistics so that you can see how these tools can be used to manipulate you. The really interesting stuff comes later.

I really sat up and started paying attention in Chapter 4 which looks at how polls are used to manipulate us. He has a lot of interesting background and historical examples of polls that mispredicted elections and he explains why this occurs and why it is bound to occur. He goes into fair depth to explain why a 1936 poll by Literary Digest which got responses from millions of Americans could so badly predict the FDR vs Alf Landon presidential vote. He looks at the 1948 Truman vs. Dewey polling that "predicted" the election of Thomas Dewey.

In Chapter 5 he delves into even more fascinating material by looking at the 2008 Al Franken vs. Norm Coleman senatorial race in Minnesota. I bet you haven't heard of the "lizard people" but you really must read this chapter to understand their crucial role in that election. He also looks at the 2000 "hanging chad" controversy of the 2000 Bush vs. Gore presidential election. You might think you "know" about these elections and what went wrong, but I assure you, until you read Seife's book you won't truly understand them. Just this chapter alone is worth getting the book. Excellent stuff!

Chapter 6 delves into the history of gerrymandering and how that is undermining democracy in the US. Again, this chapter alone justifies getting this book and reading it!
Proofiness has undermined the very foundations of our democracy -- the mechanisms that we use to count our citizens and ensure that they are represented in the Republic. Gerrymandering for political gain is deemed acceptable, even though it clearly dilutes the votes of some of our citizens. Statistical sampling is deemed unacceptable, even though rejecting it forces the government to use numbers that it knows are inaccurate. No matter how many intellectual backflips legislators and judges go through to justify their positions, the fact remains: bad mathematics is being used to deny our citizens -- mostly our minorities -- their rightful vote. In a democracy, there can be no graver sin.
In Chapter 7 he looks at some of the more egregious attempts by legislatures to use laws to redefine reality. In 1897 the Indiana state legislature passed a bill to define the mathematical constant pi to be 3.2. This was speeding toward senate approval and the governor's signature until the Purdue University's math department explained that passing this law wouldn't change mathematical reality but it sure would make the state of Indiana a laughingstock around the world.
Indiana's abortive attempt to rewrite the laws of nature is absurd, but it's not an isolated incident. Our governments misue mathematics in subtler ways all the time, trying to banish facts that are embarrassing and inconvenient. And the most mind-bending denials of mathematical reality come from the branch of government that's supposed to be the guardian of truth: the judiciary.
He then goes on to document incredible judicial decisions -- life and death decisions -- taken by US courts, all the way up to the Surpreme Court, and not just in the 19th century but the 20th and now the 21st century -- that are just as incredibly wrong-headed at that Indiana attempt to redefine pi.

He looks at the infamous 1944 paternity suit brought against Charlie Chaplin. It was a "he said, she said" case until they did a blood test. The result? Chaplin was type O. The child was type B. The Mother type A. There was absolutely no way that Chaplin could be the father of the baby. Did that scientific fact stop the judicial process. Did hard reality force the judge to face facts? No!
It didn't matter. The prosecutor -- and the courts -- rejected the notion that the blood test was conclusive and pressed on with the trials, using such scientific evidence as a "resemblance" test, where jurors gazed intently at the baby to determine whether she looked like Chaplin. The Mann Act charges at first ended in a mistrial and then in an acquittal; however, in his paternity trial, Chaplin was found to be the father. The court's truth -- the legal truth -- was that Chaplin was the father of Joan Berry's baby, even thought the Truth, with nearly absolute mathematical certainty, was otherwise. And the prosecutors and the judges knew it.
Shades of the O.J. Simpson trial. (In fact, the book has a good discussion of that trial as well and the misuse of math by the defense to trick the jury into finding Simpson innocent.)

In Chapter 8 he looks at how the "Right to Life" crowd cynically uses misinformation and untruths to push its ideological agenda. They have seized on a published "research" article that linked abortion with an increased risk of breast cancer:
The link between abortion and breast cancer is randumbness; people saw a pattern when there really wasn't one. In 2003, the National Cancer Institute held a conference on the subject -- at the end of the meeting, the report stated that it was well established, scientifically, that abortions did not increase the risk of breast cancer. (The report was endorsed by all the conference except for one: Joel Brind [the original "researcher"].) The scientific consensus was as clear as it could be; the link was fictional. Yet a few months later, Texas passed the Worman's Right to Know Act. Abortion doctors around the state were forced to put a phony doubt into their patients' minds -- doubt about whether they'll get breast cancer because of an abortion.
This is a blatant misuse of statistics to advance a religious/political viewpoint. Here's the truth behind the phony "research" and its bad math:
According to a 2005 Finnish study, women who terminate their pregnancies are about six times more likely to kill themselves. However, the same study also found that women who have abortions have a higher risk of being murdered -- ten times more likely than those who went through with the birth. They were also 4.4 times more likely to die in an accident. Thse statistics hint at what's really going on here: causuistry. It's not that abortions are causing people to commit suicide, any more than abortions are causing people to get murdered or have accidents. Women who are in a high-risk category are almost certainly more likely to get into a situation where they need an abortion than those in the general population. Mentally ill people, risk-takers, women who run with a wild crowd -- they are more likely to get into trouble of all sorts, including having an unwanted pregnancy. Conversely, women who have a child are more likely to be more stable and responsible, and less likely to take needless risks. Yet antiabortion lawmakers turn that correlation into causation, forcing doctors to scare their patients with the idea that going through with the abortion might cause them to commit suicide later on. The "right to know" laws have nothing to do with knowledge. They are vehicles for whatever bits of proofiness can be deployed to support the lawmakers' opinions.
He finishes with a final chapter on propaganda and points out:
Proofiness is toxic to a democracy, because numbers have a hold on us. They are powerful -- almost mystical. Because we think that numbers represent truth, it's hard for us to imagine that a number can be made to lie. Even the oafish Joe McCarthy knew this; when he declared that 205 communists had infiltrated the State Department, his outrageous falsehood was given the appearance of absolute fact. But proofiness is not merely a tool for propaganda as as it was for McCarthy -- it is much more dangerous than that. Democracy is a system of government based upon numbers, and rotten numbers are eroding the entire edifice from within. Proofiness in America is used to disenfranchise voters and to bias elections. It's used to drain money from our treasury and put it in the pockets of unscrupulous businessmen. Proofiness tilts the scales of justice and helps condemn the innocent. It puts lies in the mouths of reporters. There's no institution in a democracy that's immune.

No comments: