Chance News 59

From ChanceWiki
Jump to navigation Jump to search


I wish I could say, well, I told people correlation doesn't equal causation back in 1989, so I don't have to say it again.

Joe Palca, NPR science correspondent, lamenting the lay public's continual lack of understanding of how science progresses; this appeared in an article by Julia Galef entitled "Uncertainty in Science, It's a Feature, Not a Bug," The Humanist, January-February 2010.

Submitted by Paul Alper

Suddenly, we could analyze, and there was no logical end to analyzing the data ... The trick is to not go over the line. ... "Oh, those are the guys who tell us how so and so bats on alternate Tuesdays under a full moon." ... [Y]ou have to be careful not to believe that the statistics tell the whole story ... You can always find mitigating circumstances to any statistic or any fact. And you don't want to lose the beauty of the game. You want to appreciate a game for [the] game's sake.”

Steve Hirdt, Elias Sports Bureau VP, discussing obscure sports statistics/records provided by ESB to media sports commentators during broadcast games, in “Is There a Stat for That?”, The Wall Street Journal, December 23, 2009.
Submitted by Margaret Cibes

In a civil suit, gambler Terrance Watanabe, who lost nearly 127 million dollars in 2007, claims that employees “plied him with alcohol and prescription drugs to encourage him to stay and gamble.” He faces criminal charges related to his remaining debt of almost $15 million.

One reason Mr. Watanabe was seen as so valuable to Harrah's, say … two of his handlers, is that he gravitated toward games with low odds, including roulette and slots. "He was considered a 'house' player because slots and roulette are house games -- they have terrible odds for the player," says [one of the handlers]. "And the way he played blackjack, he made it a house game. He made such bad decisions on the blackjack table."

Alexandra Berzon, in “The Gambler Who Blew 127 Million Dollars”, The Wall Street Journal, December 5, 2009.
Submitted by Margaret Cibes


"I can show you a bar graph where free and reduced lunch has the worst test scores in the state of South Carolina... You show me the school that has the highest free and reduced lunch, and I'll show you the worst test scores, folks. It's there, period."

Lt. Gov. Andre Bauer of South Carolina, prospective candidate to succeed Gov. Mark Sanford, as quoted in

Bauer appears to be be confusing/reversing cause and effect. Or else we should investigate what is being put in to those school lunches to bring down test scores.

Submitted by Paul Alper

Calculating high school dropout rates

KC School District's dropout rate doesn't add up. Michael McShane, The Kansas City Star.

The Kansas City, Missouri school district had some amazing statistics to brag about.

The Kansas City School District recently announced a dropout rate of 5.9 percent. Compared with the dropout rate of 41.2 percent reported a year ago, it appeared as if the district was moving by leaps and bounds in the right direction to correct the problem.

These results, however, appear to be incorrect.

The Missouri Department of Education says when the Kansas City School District’s Class of 2009 started eighth grade in the fall of 2004 it had 2,629 members. When that class graduated this spring, 1,032 students earned diplomas. It doesn’t take a degree in mathematics to recognize that does not add up.

The calculation of a dropout rate is not too difficult.

It is a simple mathematical formula; take the total number of students who graduate and divide it by how many students started in eighth grade. If necessary, adjust that number for demographic movement trends and with a No. 2 pencil and a scientific calculator, anyone at home can estimate the graduation rate.

Here are the numbers you need for the calculation.

Let’s calculate it together. When those 2,629 eighth-graders were enrolled in the district, the total enrollment for the district was 26,968 students. When 1,032 members of that cohort earned diplomas there were 22,479 total students enrolled in the district.

If you don't account for migration, the graduation rate is 1032 / 2629 = 39%. Here's how to account for migration.

In that same period, the overall district enrollment declined by 16.65 percent, so it’s fair to reduce the number of eighth-graders to reflect that, which we can do by multiplying by 0.8335. After those calculations, the adjusted graduation rate of the district is really 47 percent.

Part, but not all of the discrepancy, can be accounted for by a change in time frame. The 5.9% represents an annual drop-out rate, not a rate across four years, a practice that Dr. McShane derides.

The district’s using that number as its dropout rate is the equivalent of your credit card company telling you the monthly rather than the yearly interest rate. It may make you feel better, but you’re still going to pay big.


1. How would you convert a yearly dropout rate to a four year dropout rate?

2. The adjustment for migration makes some assumptions. What are those assumptions? Are they reasonable?

3. Would it make sense to compute confidence limits for the dropout rate?

Submitted by Steve Simon

Graphing the politics of health care

The Senate’s health care calculations
New York Times, 18 November, 2009
Andrew Gelman, Nate Silver and Daniel Lee

Using some beautiful statistical graphics, the authors discuss the politics of the health care debate. One graphic explores a putative relationship between senators' positions on health care and public opinion in their home states. However, the relationship is shown to disappear when a third variable is accounted for, namely President Obama's 2008 margin of victory in each state. A second graphic uses data maps to illustrate public opinion on health care broken down by age, family income and state.

A November 19 post from Andrew Gelman's blog provides a fascinating description of how the graphics evolved, including the collaboration with graphics professionals at the New York Times as they worked to transform the originals into an Op/Ed Chart. We read there that:

Our graphs took months of effort, but the Times versions were not immediate either. We had to go back and forth several times to get the clarity we all wanted. I'd like to think, though, that our effort was not wasted: by being able to make a bunch of graphs that were informative for us, we were able to home in on the story.

For some of the history of those early months, see the October 12 post.

Additional reactions to these graphs can be found on the Junk Charts blog.

Submitted by Bill Peterson

Ill health news

10 trends in health care journalism going into 2010
HealthNewsReview Blog
Gary Schwitzer

Schwizter is an associate professor at University of Minnesota School of Journalism & Mass Communication, and the publisher of Health News Review, a web site which monitors the accuracy of news stories related to medicine. The project is funded by the Foundation for Informed Medical Decision Making.

From the blog post referenced above we learn that "An updated look at the first 900 stories reviewed on shows that

    71% fail to adequately discuss costs,
    71% fail to explain how big (or small) is the potential benefit,
    66% fail to explain how big (or small) is the potential harm,
    66% fail to evaluate the quality of the evidence,
    60% fail to compare new idea with existing options."

As for television in particular, "After 3.5 years and 228 network TV health segments reviewed, we can make the data-driven statement that many of the stories are bad and they're not getting much better."

With regard to the US Preventive Task Force's mammography recommendations, "There was some excellent journalism done on the issue last week, but it was overwhelmed by and drowned out by the drumbeat of dreck shoveled out by many news organizations - including in much (not all) of what was provided on network TV." Further, "The week was certainly a setback for the nation's understanding of science, of evaluation of evidence, of the potential harms of screening tests." Much of this thanks to "The public relations machinery of the American Cancer Society, the American College of Obstetrics and Gynecology - and other groups that opposed the USPSTF recommendations" which "helped the anti-USPSTF message rule the media all of last week."

Discussion Questions

1. As an exercise in futility, discuss with your sister, girl friend, wife, mother the USPSTF recommendations regarding breast cancer screening.

2. Do the same with your brother, boy friend, husband, father regarding the recommendations for prostate screening.

3. Why is there alleged a "public relations machinery" against the USPSTF?

Submitted by Paul Alper

Another medical news blog

Robin Motz alerted us to the blog Medicine: Facts and Fictions, which is subtitled "Corrections to and explanations of medical stories in the news."

In particular, Robin highlighted an interesting post on meta-analysis that describes potential pitfalls, including Simpson's paradox.

Dying from TV?

During the period January 11-13, 2010, the global public was bombarded with the results of an Australian study in the medical journal Circulation, which linked television viewing hours to risk of death, even for people who exercised regularly. A Wall Street Journal article included a chart of risk increases associated with various viewing times [1].

Here are some headlines from around the world:
“Watching TV Linked to Higher Risk of Death”, The Wall Street Journal
“Watching TV shortens life span, study finds”, LA Times
“Watch more TV, die younger, study finds”, ABC News
“Researchers: Too Much Television Points to an Early Death”, Voice of America
“TV can cut your life short, says study”, Toronto Star
“Watching TV may shorten life for couch potatoes”, Economic Times
“How watching too much television can kill you … literally”, Scottish Daily Record‎
“Beware! Watching TV cuts short your life”, Times of India
“Watching TV ‘increases risk of death’”, Irish Health
“Researchers point to early deaths from too much TV”, Zimbabwe Star‎
“Too much TV leads to an early grave”, The Australian

Two Wall Street Journal bloggers [2] commented:

So if we don't watch TV at all, the "risk of death" goes to zero? Excuse me? So death is not inevitable? What a thought!

The risk of death is already 100% so I'm not sure how they managed to say it can be increased.

A former colleague of the contributor had a great comment:

What about sick people? Don't they watch TV a lot?

For an analysis of the study by the British National Health Service, see “Turn on, tune in, drop … dead?, January 12, 2010.

Submitted by Margaret Cibes

Measuring happiness

“The Drag of Devising a State-by-State Mirth Meter”
by Carl Bialik (“The Numbers Guy”), The Wall Street Journal, December 24, 2009

This article discusses the difficulties involved in measuring how happy people are.

In 2008 Gallup and Healthways began a 25-year project with the goal of polling 1,000 Americans nearly every day about their “health and happiness.” Preliminary results were posted in Fall 2009 in the Journal of Research in Personality. (See “Gallup-Healthways Well-Being Index” for an article about the project.)

Starting in 2005, the CDC began an annual survey of more than 350,000 adults and reported in Science the results of asking: “In general how satisfied are you with your life?” The CDC study controlled for socio-economic differences among respondents. (See “Behavioral Risk Factor Surveillance System” for the CDC’s data and methodology.)

According to Bialik, a British economist commented that:

A state that has a lot of married, wealthy people is likely to rank high in happiness, but not because its residents have chosen the ideal place to live. In fact, wealthy, married couples tend to be happy anywhere. If you are single and not wealthy, moving to a more happily-ranked state isn't likely to lift your spirits much ….

Another economist is quoted:

Economists have typically played fast and loose with psychological terms.

Submitted by Margaret Cibes

Frugality of economists

“Secrets of the Economist’s Trade: First, Purchase a Piggy Bank”
by Justin Lahart, The Wall Street Journal, January 2, 2010

This article discusses economists’ tendencies to frugality in their personal lives.

According to Lahart, recent research by two University of Washington economists found that:

[E]conomics majors were less likely to donate money to charity than students who majored in other fields. After majors in other fields took an introductory economics course, their propensity to give also fell.

And a 1981 paper by two University of Wisconsin sociologists found that:

[E]conomics students showed a much higher propensity to free ride [take more than their fair share of something when circumstances permit] than other students.

Apparently economists tend not to be gamblers either.

One year, Yale University economist Robert Shiller, who'd never gambled in his life, found himself at a casino [in New Orleans]. He says that was because Wharton economist Jeremy Siegel realized that by using coupons offered to conventioneers, they could take opposing bets at the craps table with a 35 out of 36 chance of winning $12.50 each. Over two nights, Mr. Shiller netted $87.50. …. He hasn't gambled since.

See the paper “Why are economics students more selfish than the rest?”, University of Washington, November 2009.
Also see an abstract of the paper “Economists free ride, does anyone else?”, University of Wisconsin, 1981.

Submitted by Margaret Cibes

Terrorism risks

“Crunching the Risk Numbers”
by Nate Silver, The Wall Street Journal, January 8, 2010

In this article Silver discusses the risks of terrorist attacks and cites a few statistics to back up his claim that the risk of a future airline terrorist attack is minimal compared to other risks.

The chance of a Westerner being killed by a terrorist is exceedingly low: about a one in three million each year, or the same chance an American will be killed by a tornado [while the] Department of Homeland Security's budget is 50 times larger than that of the weather service.... This is not to suggest that no efforts should be made to stop "conventional" terror attacks. But surely we must understand that, at best, we will reduce the risk from an extremely small nonzero number to a slightly smaller nonzero number.

Silver also claims that the risk of an airline terrorist attack has been lower during the 2000 decade than it was during each of the previous five decades, relative to the number of commercial flights.

He considers the 9/11 event as a “horrible outlier,” which is unlikely to be repeated and which “would almost certainly be thwarted by brave passengers (and secure cockpit doors)” if it were attempted again. And he believes that “improved vigilance and intelligence,” along with a reduced threat from “once-threatening terrorist organizations” will mitigate future threats.

Silver does feel, however, that terrorism involving nuclear weapons is a legitimate concern and he cites some statistics from a Harvard scholar that “there is greater than a 50% likelihood of a nuclear terrorist attack in the next decade, which he says could kill upward of 500,000 people.”

Submitted by Margaret Cibes

Bogey-man study

“Misreading the Green: Study of Tiger's Toll Misses”
by Carl Bialik (The Numbers Guy), The Wall Street Journal, January 7, 2010

This article discusses the effect of Tiger Woods’ performance on the stock prices of his corporate sponsors and advertisers, as reported in a recent study by two UC Davis professors. (See Shareholder Value Destruction following the Tiger Woods Scandal, January 4, 2010.)

The study concluded that as much as $12 billion in capitalization had been lost by sponsors and advertisers as a result of the publicity about his car accident, infidelity, and decision to stop golfing indefinitely. A graph [3] is provided of the timeline of Tiger’s statements/actions with corresponding changes in the DJIA and the stock prices of Nike, PepsiCo, and Electronic Arts.

The paper’s authors acknowledge that there are problems with their findings, some of which have been identified by critics. Issues of concern include the timing of the recorded stock price drops relative to news reports, the presence of other factors that may have influenced the stock price drops, and the relatively small number of companies involved with Woods.

Bialik also notes:

[T]he initial finding wasn't statistically significant. …. They ran more sensitive statistical tests, some of which did clear the 5% significance threshold.

The authors plan to address their concerns before submitting the report to a peer-review process.

See Bialik's pre-article comments in “Tiger Woods and Market-Moving Events”, The Wall Street Journal, January 6, 2010.

See also a blogger's remark [4] (and his Tiger vs. DJIA charts) in response to those pre-article comments:

"Check out my own research into Tiger’s performance against the DOW. It is hsyterically [sic] accurate over the life of Tiger’s career [5]

Submitted by Margaret Cibes

U.S. presidents take on Super Bowls

“44 VS. XLIV”
by Nick Paumgarten, The New Yorker, January 18, 2010

The author discusses the work of a Philadelphia sportswriter in searching, tongue-in-cheek, for coincidences related to numbers. Most recently he identified Barack Obama as the 44th U.S. president, Hank Aaron as baseball’s wearer of shirt #44, and the 2010 Super Bowl as the 44th such match.

The sportswriter has set up an online website for hypothetical head-to-head contests between U.S. presidents and their correspondingly numbered Super Bowls, “America Bowl”.”

The criteria are vague. The Presidents are judged, for the most part, by their accomplishments, and the games by their competitiveness, with points on either side for iconicness.

In the past he developed a formula that gives the age, in human years, of any car; a boxing statistic that gives the “knockouty”-ness of a boxer in terms of the percentage of scheduled rounds he fights; and a “bracket” showing Obama Cabinet nominees in relation to America’s favorite vegetables.

Submitted by Margaret Cibes

Contributor’s note: Obama is president #44 if one counts Grover Cleveland twice (#22 and #24).

Value of Vikings

“What Price Vikings Fandom? Funny You Should Ask” by Conor Dougherty, The Wall Street Journal, January 16, 2010

This article discusses a study by two economists of the value of the Viking franchise to Minneapolis-St. Paul, amidst a 2009 controversy about spending public monies on a new stadium for the team.

The two used "contingent valuation methodology" …. The result: The Vikings' "welfare value" is $702,351,890— $530.65 for each of the roughly 1.32 million households in Minnesota. The study was conducted in 2002, and the figures are not adjusted for inflation (or for the recent acquisition of quarterback Brett Favre). …. In the 2002 off-season (to minimize in-season emotions), [the two researchers] mailed 1,400 surveys to households across Minnesota, capturing both fans and nonfans.

While the survey [6] contained 30 questions,

the so-called welfare value was generated from a single yes or no question: Would you be willing to pay $X out of your own household budget for the next year to make a new stadium possible?

For a copy of the study, see “Estimating Local Welfare Generated by a Professional Sports Team: An Application to the Minnesota Vikings under Threat of Relocation”, February 28, 2008. The paper was published in the Southern Economic Journal, July 2009.

Submitted by Margaret Cibes

Football action (or inaction) time

“11 Minutes of Action”
by David Biderman, The Wall Street Journal, January 15, 2010

This article reports the results of a WSJ study of how air time was spent in four recent football broadcasts of “four games on four networks on one weekend in late December.”

Of the 185 minutes of broadcast time, the “average amount of time the ball is in play on the field during an NFL game is about 11 minutes,” with about 60 minutes of commercials and 75 minutes of players waiting for action. The remaining time was spent on commentaries, interviews, background clips, and replays. The study estimates the rate of inaction to action as about 10 to 1.

Readers can click on sortable charts and tables [7] to see how the minutes are accounted for in the WSJ’s sampling of the four games.

The WSJ’s study is said to have been corroborated by at least two other researchers, one in 1912 and another in 2010.

Submitted by Margaret Cibes