September 1, 2007

What we're really interested in

Science is in the business of making predictions, but the better it gets at predicting anything, the more boring those predictions are for us. For example, I predict that the sun will set at the O'Hare Airport in Chicago today at 7:26 pm CDT. When you think of all the effort that has gone into astronomical observation and prediction over the millennia (for example, Stonehenge), that's an incredible feat the human race has achieved to be able to accurately predict that.

It's also phenomenally boring.

Now, here's a different prediction: Republican nominee Mike Huckabee will outpoll Democratic nominee Bill Richardson 51%-47% in the November 2008 Presidential election. "What an idiot!" you say, "Don't you know that the Clintons will stop at nothing to get back to the White House? Richardson and Huckabee? You don't know anything about the election!" And you're right. I don't. I'm not even sure where Huckabee is from. I think it's that state, you know, the one you drive through to get to that other state.

Now, here are some more predictions. USC will not finish #1 in college football this season. Instead, Rutgers will bring the national title home to Delaware. (Or maybe to Connecticut, depending on where, precisely, Rutgers is located. Assuming it's located somewhere. Maybe it's like the DeVry Institute and is located everywhere. But I digress.) On the other hand, USC will win the NCAA basketball championship next spring behind frosh sensation OJ Mayo.

"What a jerk!" you exclaim, "Everybody knows that USC's linebacking corps is the most devastating in college football since Penn State's back in 1987." Well, I don't know that. In fact, I know barely anything about college football these days.

But the point is that, unlike the sunset forecast, these predictions are interesting, as brainless as they are. The reason that making up nonsense off the top of my head about elections and sports is interesting is because nobody can predict accurately sports and far-off elections with a lot of candidates. Sports, especially, are designed to be hard to predict just so that they will keep our interest. The same with gambling. Randomness isn't natural in the world, at least above the subatomic level. It takes a lot of work to develop gambling devices that are close to random, but a roulette wheel is more interesting than betting when the sun will go down because it's hard to predict.

You often hear that the social sciences aren't real sciences like astronomy because they can't predict anything. But that's not true. Indeed, I'll make a social science prediction for 25 years into the future. I predict that in the year 2032, the students at the schools in Beverly Hills will enjoy higher average scores on statewide and nationwide standardized tests than the students at schools in Compton. Anybody want to bet against me?

I've got a million more predictions like that. For example, in 2032, the children of today's unskilled immigrants will be more of a burden on society than the children of today's skilled immigrants. (That seems like an important use of social science -- to make predictions extremely important for choosing the optimal immigration legislation, right?)

"Well, sure," you say, "Of course. But those predictions are boring. And depressing. In fact, it's in bad taste to mention things that we all sort of know are true but that we really don't want to think about. Who wants to hear predictions like that? Tell us something interesting."

Okay, on December 31, 2032, the Dow Jones Average will stand at 107,391. But just one year later it will have crashed, in the wake of Black Wednesday, all the way to 33,828. But by 2042, during the bubble following a major breakthrough in cold fusion, the Dow will have reached the 201,537 barrier.

"Now that's better! That's the kind of prediction we like: specific and exciting. Of course, you're probably just randomly punching numbers on your keypad, but we forgive you because you're not boring and depressing us anymore."


My published articles are archived at iSteve.com -- Steve Sailer

August 30, 2007

Designer color names

One of the challenges faced by fashion designers is coming up with new names for the same old colors. For example, here is a sandal whose strap color an unfashion-forward individual like myself might describe as "blackish" but a professional designer describes as "Ballistic Anthracite." What the hell is that? It sounds like a weapons system from one of those sci-fi alternate histories of the Civil War in which the War Between the States finally ends in 1887 when Pennsylvania wipes out Virginia's fleet of steam-engine tanks with a salvo of coal-powered missiles.


My published articles are archived at iSteve.com -- Steve Sailer

Carol M. Swain

My new VDARE.com column is a review of the anthology she edited:


Yale Law School Professor Peter H. Schuck observes:


"In a polity in which only 17 percent of the public thinks that immigration levels should be higher and 39 percent thinks they should be lower, one would expect that at least some legal scholars who write about immigration issues would favor restriction. If so, one would be wrong. In over two decades of immersion in immigration scholarship, I have not encountered a single academic specialist on immigration law who favors reducing the number of legal immigrants admitted each year." ...


So, Carol M. Swain, a law and political science professor at Vanderbilt, has done the academic world a service (although one it probably won't appreciate) with her new book Debating Immigration. She brings together 16 chapters from academic and think tank luminaries such as Nathan Glazer, Amitai Etzioni, Douglas S. Massey, and Steven A. Camarota, along with lively essays from journalists Peter Brimelow and Jonathan Tilove.

Swain
is one of the more unusual and admirable scholars in public policy. Growing up black and poor in rural Virginia, one of twelve children, she dropped out of 9th grade and married at 16. In her mid-20s she started back to school. Eventually, she earned tenure at
Princeton as an expert on how Congress operates.

Her views are difficult to categorize politically. I would say she's an advocate of black enlightened self-interest, left of center on economics, right of center on culture. For example, her 2002 book The New White Nationalism sensibly advocated depriving white nationalists such as Jared Taylor of their best issues by restricting immigration and cutting back on affirmative action, especially for immigrants and affluent blacks. Needless to say, that hasn’t happened.
[More]


My published articles are archived at iSteve.com -- Steve Sailer

Top 5 scientists ever?

Anthropologist John Hawks offers some good suggestions:


Don't get me wrong, I like physics as much as anybody. But once your list includes Newton, Einstein, and Maxwell, and then you throw in Galileo, well there's not much room for anything else. None at all if you take Darwin as a given.

So I decided to do something a little different: What five scientists have had the greatest impact on human life? Yes,
Newton was great, but gravity goes on without him.

Many later discoveries stood on his shoulders, but
Newton's achievements were far more intellectual than practical. I'm looking for people whose accomplishments saved lives, prevented wars, stopped hunger, or released people from endless drudgery. This isn't a list of inventors -- if it were, there would be a lot of ancient inventions like the moldboard plow that deserve more attention than anything modern. It's a list of scientists whose impact stretched across many fields, and without whom life today would likely be worse.

1. R. A. Fisher. His work in population genetics laid the foundations for the vast productivity increases of twentieth-century agriculture. He was far from alone in this, but he stood apart from his contemporaries by inventing many of the statistical methods that would come to define scientific hypothesis testing. Without Fisher's innovations in statistics, large-scale medical research studies would be meaningless. All this after he established the basis for Mendelian inheritance of continuous characters.


Fisher strikes me as the Newton of the 20th Century: the scientist / mathematical innovator.

For the rest of Hawks' list, click here.


My published articles are archived at iSteve.com -- Steve Sailer

Will the NYT ever report anything bad about their blogger Steve Levitt?

Here's the abstract of a paper in press by economist Ted Joyce, followed by Joyce's cogent explanation of why it's important to keep harping on this subject.


A Simple Test of Abortion and Crime
Ted Joyce
Baruch College and Graduate Center
City University of New York
and
National Bureau of Economic Research

Forthcoming in Review of Economics and Statistics

A Simple Test of Abortion and Crime

Abstract

I replicate Donohue and Levitt’s results for violent and property crime arrest rates and then apply their data and specification to an analysis of age-specific homicide rates and murder arrest rates. The coefficients on the abortion rate have the wrong sign for two of the four measures of crime and none is statistically significant at conventional levels. In the second half of the paper, I present alternative tests of abortion and crime that attempt to mitigate problems of endogeneity and measurement error. I use the legalization of abortion following the 1973 decision in Roe v. Wade in order to exploit two sources of variation: between-state changes in abortion rates pre and post Roe, and cross-cohort differences in exposure to legalized abortion. I ind no meaningful association between abortion and age-specific crime rates among cohorts born in the years just before and after abortion became legal.

I. Introduction

The debate as to whether legalized abortion lowers crime leaped from academic journals to mainstream discourse with the huge success of Freakonomics.1 In the Chapter titled, “Where Have All the Criminals Gone?” Levitt and Dubner summarize academic work by Levitt and coauthor John Donohue, which shows that a one-standard deviation increase in the abortion rate lowers homicide rates by 31 percent and can explain upwards of 60 percent of the recent decline in murder.2 If one accepts these estimates, then legalized abortion has saved more than 51,000 lives between 1991 and 2001, at a total savings of $105 billion. But the policy implications go beyond crime. If abortion lowers homicide rates by 20 to 30 percent, then it is likely to have affected an entire spectrum of outcomes associated with well-being: infant health, child development, schooling, earnings and marital status. Similarly, the policy implications are broader than abortion. Other interventions that affect fertility control and that lead to fewer unwanted births—contraception or sexual abstinence—have huge potential payoffs. In short, a causal relationship between legalized abortion and crime has such significant ramifications for social policy and at the same time is so controversial, that further assessment of the identifying assumptions and their robustness to alternative strategies is warranted.


The New York Times more or less sets the agenda for the rest of the news media. If the NYT decides a story is fit to print, much of the the rest of the press will soon decide, what do you know!, that the topic deserves coverage. But if a tree falls in the forest and the NYT doesn't cover it ... This means the NYT has a particular responsibility to avoid giving in to conflicts of interest, which they have clearly succumbed to over the last two years in their refusal to report on any of the controversies swirling around their star columnist turned blogger Steven D. Levitt.


My published articles are archived at iSteve.com -- Steve Sailer

August 29, 2007

Graduate Record Exam scores by graduate field of study

A reader sends along this table from the Graduate Record Exam from ETS giving average scores by intended field of study in grad school. He includes an estimate of IQ from one of the popular conversion tables, although he didn't tell me which one.

One problem I saw was that the mean score for the Quantitative section is so much higher than for the Verbal section, and the standard deviation is also larger for Quant, that the combined scores were biased in favor of highly quantitative fields. So, I added three more columns on the right that show difference fro the mean in standard deviations and just take the average for verbal and quantitative compared to their separate means. That seems fair, since there's no evidence that verbal intelligence correlates lower with general intelligence, and it may well be the best surrogate for the g factor. So, that's how I sorted it, which moves philosophy up into second place behind physics.

That reminds me of how I wrote a review of a book by David Stove in 1999 making gentle fun of philosophy (well, maybe not that gentle: I referred to the "uselessness of philosophy"). I received a number of superbly articulate and intensely argued emails telling me I didn't know what I was talking about. You'll notice I've drawn in my horns on this topic ever since!

This table may not be fair to business students since perhaps the better ones tend to take the GMAT to apply to MBA schools.

Graduate Record Examination Scores






Mean

465

584






Standard Deviation

117

149







Verbal

Quant

Sum

IQ

Verbal SD

Quant SD

Avg. SD

Physics & astronomy

533

736

1269

133

0.58

1.02

0.80

Philosophy

590

638

1228

129

1.07

0.36

0.72

Mathematical Sciences

502

733

1235

130

0.32

1.00

0.66

Materials Engineering

494

727

1221

129

0.25

0.96

0.60

Economics

503

706

1209

128

0.32

0.82

0.57

Chemical Engineering

485

726

1211

128

0.17

0.95

0.56

Other Engineering

493

714

1207

128

0.24

0.87

0.56

Mechanical Engineering

469

724

1193

126

0.03

0.94

0.49

Other Humanities & Art

563

599

1162

124

0.84

0.10

0.47

Physical Sciences

486

697

1183

125

0.18

0.76

0.47

Engineering

468

719

1187

126

0.03

0.91

0.47

Electrical Eng

459

726

1185

126

(0.05)

0.95

0.45

Banking & finance

467

711

1178

125

0.02

0.85

0.43

Chemistry

486

680

1166

124

0.18

0.64

0.41

Computer & Infor Sci

466

701

1167

124

0.01

0.79

0.40

Civil Engineering

457

700

1157

124

(0.07)

0.78

0.36

Religion & Theory

541

589

1130

121

0.65

0.03

0.34

Industrial Engineering

440

707

1147

123

(0.21)

0.83

0.31

Earth, Atmos & Mar. Sci

495

636

1131

121

0.26

0.35

0.30

English language & lit

560

553

1113

120

0.81

(0.21)

0.30

Humanities & arts

545

566

1111

120

0.68

(0.12)

0.28

Arts-History, theory, crit

539

572

1111

120

0.63

(0.08)

0.28

Biological Sciences

491

631

1122

121

0.22

0.32

0.27

Political Science

524

588

1112

120

0.50

0.03

0.27

Foreign languages & lit

531

574

1105

119

0.56

(0.07)

0.25

Anthropology & Archeology

533

569

1102

119

0.58

(0.10)

0.24

History

542

557

1099

119

0.66

(0.18)

0.24

Library & Archival Sciences

536

542

1078

117

0.61

(0.28)

0.16

Architecture

475

610

1085

118

0.09

0.17

0.13

Natural Sciences -Other

474

598

1072

117

0.08

0.09

0.09

Secondary

485

578

1063

116

0.17

(0.04)

0.07

Social Sciences

487

565

1052

115

0.19

(0.13)

0.03

Agriculture

458

592

1050

115

(0.06)

0.05

0.00

Arts-Performance & studio

488

553

1041

114

0.20

(0.21)

-0.01

Life Sciences

462

581

1043

114

(0.03)

(0.02)

-0.02

Sociology

488

545

1033

114

0.20

(0.26)

-0.03

Other business

444

599

1043

114

(0.18)

0.10

-0.04

Business

442

592

1034

114

(0.20)

0.05

-0.07

Psychology

472

545

1017

113

0.06

(0.26)

-0.10

Higher

464

548

1012

112

(0.01)

(0.24)

-0.13

Communications

470

533

1003

111

0.04

(0.34)

-0.15

Curriculum & Instruction

459

546

1005

111

(0.05)

(0.26)

-0.15

Health & medical sciences

447

552

999

111

(0.15)

(0.21)

-0.18

Other social Science

465

527

992

110

0.00

(0.38)

-0.19

Business admin & mgmt.

438

561

999

111

(0.23)

(0.15)

-0.19

Education

449

534

983

110

(0.14)

(0.34)

-0.24

Accounting

408

585

993

110

(0.49)

0.01

-0.24

Evaluation & Research

450

530

980

109

(0.13)

(0.36)

-0.25

Public Administration

453

515

968

109

(0.10)

(0.46)

-0.28

Other Education

439

532

971

109

(0.22)

(0.35)

-0.29

Elementary

442

526

968

108

(0.20)

(0.39)

-0.29

Administration

426

522

948

107

(0.33)

(0.42)

-0.37

Home Economics

435

501

936

106

(0.26)

(0.56)

-0.41

Special

431

502

933

106

(0.29)

(0.55)

-0.42

Student Counseling

427

500

927

105

(0.32)

(0.56)

-0.44

Early Childhood

418

497

915

104

(0.40)

(0.58)

-0.49

Social Work

428

466

894

103

(0.32)

(0.79)

-0.55

***Permalink/Comments***

My published articles are archived at iSteve.com -- Steve Sailer