CHANCE News 5.09

(17 July 1996 to 14 August 1996)


Prepared by William Peterson, with help from J. Laurie Snell, Fuxing Hou, Ma.Katrina Munoz Dy, and Joan Snell, as part of the CHANCE Course Project supported by the National Science Foundation.

Please send comments and suggestions for articles to

Back issues of Chance News and other materials for teaching a CHANCE course are available from the Chance web site:


Note from Laurie Snell: Bill Peterson kindly agreed to do chance news this time while I was on vacation. I think you will agree that he did a fine job!

"The lottery: A tax on people who flunked math."
-- Monique Lloyd

From our readers:

The lottery quote appearing on the header was sent to us by Meredith Warshaw, who commented:

"Thought your readers might appreciate [this] definition from someone on the TAGFAM (families of gifted/talented kids) email list. I imagine that could stir up a fair classroom discussion :-)"

The discussion might compare this definition with "Seen from a rut, the lottery is essential" in Chance News 5.08.

Tom Moore sent us the following article and discussion question.

Statisticians take calculated risk trying to shed nerd image.
The Chicago Tribune, 3 August 1996, p1.
Sabrina L. Miller

This article describes how the ASA put on a stat theme day for Chicago school kids, which apparently turned out to be a bust. A talk on "Sex, Drugs and Statistics" had no sex, discussed drugs through the job description of a pharmaceutical worker at Merck, and presented a statistics lecture via overhead projector. A talk on "Statistics in Sports" discussed confidence intervals in the context of an estimate that Chicago Bulls basketball star Scottie Pippen would shoot 80% for the season--a percentage that the audience obviously recognized as unrealistic. Bored students repeatedly interrupted the presentations with wisecracks and laughter.

Organizers of the event were aware that their first-time attempt at high school outreach would be risky. A teacher at the school noted that, while the idea of making statistics accessible was good, the delivery needed work: "You have to know how to get these kids interested. You can't just stand up there and lecture like a college teacher with an overhead projector for 40 minutes."

How do you think this activity could have been made more successful?

Web addresses from Chicago talks.

We've just returned from the Joint Statistical Meetings in Chicago, where we heard a number of excellent presentations on statistical education--none of which recommended lecturing for 40 minutes with an overhead!

When we noticed people scrambling to write down URLs, we promised to collect web sites that were referenced at the (non random) sample of talks we attended. Links to these will be included on the Chance Database. Please let us know of any that we missed!


Stochastic Visualization and the Internet. R.W. West

Interactive Instruction on the Web. B Narasimhan


Integrating Computer Oriented Assigments... .E. Mansfield
emansfie@alston.cba.ua.edu [e-mail for Kentucky Derby data]

Datasurfing on the World Wide Web. R. Lock

Fishing for Data Using the Net. S. Turner

The Statistical Instruction Internet Pallette. J. Behrens

A Prototype Multimedi Module. R. Heckerd, et al.

A New Approach to Resource Sampling. K Portier et al


Experiences with Alternative Assessment Techniques.
Beth Chance


The Internet and Reproducible Research. J. Buckheit


Workshop Statistics. A. Rossman

NOTE: Not all of the addresses worked as we wrote them down at the conference. When we got stuck, we found the AltaVista search engine very helpful. (Name AND Institution AND keyword from talk was often successful). The above have all now checked out.

Armed with the abstract from the conference, readers can probably find talks we've missed by the same strategy!

Here are three items from Marilyn vos Savant's column.

Ask Marilyn.
Parade Magazine, 21 July 1996, p6.
Marilyn vos Savant

A reader asks: "Say someone offers you the following bet: He will toss three coins all at once, if they all turn up heads, he'll give you $10. And if they all turn up tails, he'll give you $10. But if they land with either 1) two heads and a tail or 2) two tails and a head, you have to give him $5. Now--without stopping to think about it--should you take his bet?"

Marilyn answers, that because this someone probably did stop to think about it, it's a bad idea to take his bet. She adds that in this case he will make about $5 for every four tosses of the three coins, so in the long run you'll lose.

Can you verify Marilyn's calculation?

Ask Marilyn.
Parade Magazine, 4 August 1996, p7.
Marilyn vos Savant

A reader asks: "Can you come up with a ballpark figure for the cost of government regulation and taxes that are included in the purchase price of a product?"

Marilyn replies that "our research" indicates that about 9% of consumer prices is attributable to federal regulations, 18% to federal taxes, for a total of 27%. Noting that state and local regulation and taxes vary widely, she suggests they add about half again as much, bringing the total to about 40%. Finally, some experts add "indirect costs" which bring the figure to 50%.

Who do you suppose is the "we" responsible for what Marilyn calls "our research"? How might you check on these figures?

In this same week's column, a second reader asks: "Say that Tom studied a lot of mathematics in college and was campus chess champion, too. If that's the case, which of the following statements is more likely to be true? 1) Tom is now a mathematician. 2) Tom is now a mathematician and plays chess as a hobby."

Marilyn answers that, although it seems counterintuitive, the first statement is more likely because it "includes" the second: it is true of a mathematician whether he plays chess or not. The counterintuitive aspect is what Kahneman and Tversky call the "conjunction fallacy," and the example here is reminiscent of their famous "Linda problem." But to have the full punch of that problem, the present example needs to include a characteristic for which Tom's profile is not "representative"--e.g., 1) Tom is a waiter. 2) Tom is a waiter and plays chess as a hobby.

Polling: Box populi.
The Economist, 27 July 1996, p73.

This is a review of the book "The Voice of the People" by James Fishkin (Yale University Press). Fishkin, a professor of government at the University of Texas at Austin, argues that governments have lost touch with the voice of the people and that debates on complex issues are increasingly dominated by instant opinion polls, media sound bites and advertising campaigns by special interest lobbies.

To determine what the people would believe if they had access to deeper information on the issues, Fishkin proposes a technique called "deliberative polling," which he recently implemented to explore British views on the future of the monarchy. Early in July, he brought together a randomly selected group of 300 British voters. They were initially polled to get a baseline, after which they were presented with 3 days of presentations by politicians, constitutional experts and journalists, before being polled again.

[Partial results can be found in the AP wire report "The world in brief: Monarchy a good deal?", The Atlanta Journal and Constitution, 29 July 1996, p5. At the beginning of the discussions 26% believed that the monarch should no longer be head of the Church of England. In the final poll, that figure was up to 56%--a dramatic change!].

There is a charming reference here to a 1940s film called "Magic Town", which depicts a Midwestern town named Grandview whose citizens' opinions always statistically match those of the country. Jimmy Stewart plays a pollster who uses the town as a shortcut for measuring national public opinion. But when the citizens learn what is happening, they feel obligated to make the most informed choices possible. They arrange for their own surveys, providing library reference materials at every polling booth. The reviewer opines that the citizens of Grandview would have been enthusiastic about Fishkin's approach!


1. In the deliberative poll on the monarchy, how do you think the organizers selected the politicians, constitutional experts and journalists who made presentations?

2. Beyond the usual difficulties in polling, what problems do you see in justifying an inference of the form "Here is our estimate of what the population would think if it had time to fully educate itself on the issue"? Is there a parallel here with treatments in a medical experiment?

3. Do you see any potential for deliberative polling in reforming the political process in this country?

Here is a Web site for those interested in the most up-to-date information from traditional polls:


PoliticsNow advertises itself as the place "to find the latest political news from ABC News, The Washington Post, National Journal, Newsweek, and The Los Angeles Times on the race for the White House to the battle for control of Congress and beyond."

The website includes a feature called "Poll Track", which is regularly updated with results from national and state polls. Some brief selections follow:

STATE POLLS Conducted in Late July

 State     Clinton  Dole   Date           Source

Michigan 49 39 July 28-30 Mason-Dixon

Michigan 58 35 July 24-29 EPIC/MRA

Missouri 49 37 July 28-30 Mason-Dixon

New Jersey 47 35 July 28-30 Mason-Dixon

New York 57 35 July 27-29 Mason-Dixon

North Dakota 39 47 July 25-27 Mason-Dixon

South Dakota 41 45 July 25-27 Mason-Dixon

Virginia 44 46 July 26-28 Mason-Dixon


Source          Date       Sample   Clinton  Dole

CBS/NY Times Aug 3-5 N=NA 59 36

NBC/Wall St J Aug 2-6 N=NA 55 35

ABC Aug 1-5 N=1514 55 39

The full Web site also provides extensive charts and graphical displays of the results. This is a potential source of data for student projects.

Business Bulletin
The Wall Street Journal, 25 July 1996 p1.

Decisions go sour...

Paul Nutt, professor at Ohio State University's Fisher College of Business, has done research indicating that business managers fail about 50% of the time on decisions ranging from deciding what products to sell to office renovations.

DISCUSSION QUESTION: Does this mean that businesses should be tossing coins rather than paying MBAs to make their decisions?

Beating the competition...

A study by professors at the Wharton School at the University of Pennsylvania shows that marketing managers may sometimes put their competitive urge ahead of profits. When asked in a product pricing exercise to choose between a low-price, high-profit strategy and a high-price, low-profit strategy, most subjects chose the high-profit route. But when a second group was told that, while the lower price would let them double their profits, it would let their competitors do even better, 60% of the subjects sacrificed profit in order to beat the competition.

One explanation for the results: since market share correlates with profit, managers may assume that it causes profits. Says Wharton professor J. Scott Armstrong, "Psychologists call this social comparison theory. Parents call it sibling rivalry. Managers call it warfare."

What is the analogy with sibling rivalry?

Study blames cot deaths on smoking parents.
Reuters, 25 July 1996, p8.

Researchers reporting in the British Medical Journal estimate that crib deaths would be reduced by 61% if smoking were eliminated from babies' environments. The two year study included every case of crib death in three regions of England. Mothers of the 195 babies who died were questioned, and comparisons were made with mothers of 780 babies whose babies lived. It was found that 62% of the mothers of babies who died> smoked, compared with 25% of the mothers of babies who lived.

Researchers added that risks were greater if a mother smoked before birth of her baby as well as after, and that risk increased with the level of exposure to smoke after birth. Having a smoker as a father was also a problem.

All of the data appearing in the article are given above. How do you think the 61% figure (for the potential reduction in crib deaths) was arrived at?

Unconventional Wisdom: New facts and hot stats from the social sciences.
The Washington Post, 28 July 1996, C5.
Richard Morin

White bigotry, black self-hate.

In national opinion polls, nearly one in six whites will agree with the statement that "blacks have less inborn ability to learn than whites". This is disturbing, but perhaps not surprising. What is surprising is that one in eight blacks also say that blacks are less intelligent. These results come from General Survey Data collected since the early 1970s by the National Opinion Research Center at the University of Chicago. Some other surprises: Morin reports that one in six blacks say that white homeowners should be "legally allowed to discriminate against blacks" when selling their homes.

The director of the survey says such results are reminiscent of the famous survey 50 years ago when white and black children were shown black and white dolls and asked which they preferred. Many black children chose the white dolls as being "prettier." One explanation for such trends is that blacks may have on some level internalized the prejudices of a predominantly white society. The researchers also suggested that the recent results may be due in part to respondents misunderstanding the questions.

The actual question on housing (reported at the end of the article) asked whether respondents would favor a law allowing a "homeowner to decide for himself whom to sell his house to, even if he prefers not to sell to blacks." If Morin's phrasing had been used, do you think the responses would have been different?

Gold Medal Smiles

Spanish psychologists analyzed the facial expressions of 22 Olympic gold-medal winners from the Barcelona games. During the moment when they were handed their medals, they smiled about 70% of the time; they smiled less than 10% of the time immediately before the presentation and afterwards during the playing of their national anthems. This despite the fact that the athletes reported feeling intensely happy throughout the ceremony. These findings are consistent with other research indicating that smiling is a social function rather than a spontaneous reaction to feelings of happiness.


Writing in the Journal of Sport and Social Issues, Catrioni Higgs and Karen Weiller report that during the 1992 games, NBC devoted 56% of its coverage to men, 44% to women. (This is billed as an unofficial Olympic record in its approach to equal coverage.) In gymnastics, tennis, rowing and cycling, women were featured the majority of the time; men were more prominently featured in track and field. Here are some of the data (air times in hr:min).

Sport   Total Air Time % Male % Female

Basketball 18:23 74% 26%
Track/Field 9:10 63 37
Gymnastics 5:48 16 84
Swimming 3:35 48 52
Volleyball 3:05 75 25
Diving 2:11 44 56
Cycling 0:52 40 60
Kayaking 0:27 81 19
Tennis 0:18 33 67
Rowing 0:12 33 67


1. If you take away basketball (where, as the article mentions, intensive coverage was lavished on the men's "Dream Team"), how do the sexes compare overall for the remaining sports?

2. What reasons can you suggest for different sexes being favored in coverage of particular sports?

Time for a Reality Check on the Deficit.
The Wall Street Journal, 1 August 1996, p1.
Roger Lowenstein

Last year, President Clinton had a standoff with Republicans Congressional leaders over their respective plans to balance the budget. At one point, Congress shut down the government and then threatened to not raise the debt ceiling, which would have forced the country to default. Yet, despite the fact that no agreement was reached on how the budget should be balanced, the deficit has gone down dramatically. A year ago, the deficit for 1996 was projected at $160 billion, about the same as in 1995 but down from $290 billion in 1992. But most recent estimates put this year's deficit at only $117 billion. The article cites a stronger than expected economy and stock market for the drop.

We discussed the difficulty of projecting deficits in an earlier Chance News (5.02, Item 11). The present article is even more severe in its criticism. An unnamed economist, who publishes long-term, macro-economic forecasts for a living, is quoted as saying that such forecasts are a "joke." Lowenstein asserts that: "If General Motors doesn't know how many cars it will sell next year, neither does the Congressional Budget Office. Nor, then, does it know the sum total of cars, trucks, terrorist-prevention devices and other goods and services of GM and other companies combined. If one-year projections can miss by a third, seven-year estimates are virtually meaningless." He recommends that planners focus instead on obvious long-term problems such as Social Security and Medicare, rather than arguing over detailed projections about exactly when the budget will balance.

Is Lowenstein arguing that we should completely abandon long-range forecasts? Would you agree?

Making the grade? Disasters have US weighing airline safety rankings.
Chicago Tribune, 4 August 1996, p3.
Andy Pasztor and Bruce Ingersoll

An article reviewed in Chance News 5.07 looked at airline safety in the wake of the ValuJet disaster. Now, with the recent crash of TWA flight 800 off Long Island, even more attention is being paid to this issue. The Federal Aviation Administration (FAA) and Transportation Department are reportedly considering plans to publish periodic safety rankings of US airlines. Preliminary proposals range from a single strict numerical safety ranking, to a sorting of carriers into categories based on safety systems, security measures and maintenance records.

Not surprisingly, the airlines vigorously oppose any such system. Indeed, there is general agreement that the data now collected by the FAA can be confusing. Says David Hinson of the FAA: "If you fall because of turbulence during a flight and break your arm, that's considered and accident, the same as fatalities in a crash." Furthermore, because crashes really are infrequent, an airline that looks bad during one review period could look sterling during the next.


1. Would you prefer to fly on an airline that had one crash, but no other accidents during the last year or on one that had no crashes, but a dozen instances of passenger injuries from turbulence and other in-flight disturbances? Does a crash in the air "count" the same as one on take-off or landing?

2. A previous discussion question (Chance News 5.07, Item 21) noted that some observers feel air travel is so safe that crashes can be viewed as chance events. Would ranking airlines then ignore W. Edwards Deming's advice that we should not declare winners and losers on the basis of the variability inherent in a process? or does that logic not carry over from the manufacturing context to human safety issues?

3. Speaking about counterterrorism, aviation-security consultant Frank McGuire says here: "On a scale of 1 to 100, the FAA rates a 30. I think the public should forget about total security on an airplane." Do you think McGuire is sure the rating should not be 28 or 32? What do you suppose his rating system means?

Warfare 2020: Ones and zeroes, not bombs and bullets, may win tomorrow's battles.
US News & World Report, 5 August 1996, p34.
Richard J. Newman

This piece is part of a Special Report on how technology is shaping the future of war. One theme here is that electronic linking of military hardware will dramatically reduce the number of soldiers needed to wage a war. The opening two-page spread features a graphic showing historical data transfer rates (in words per minute) and the number of soldiers needed to cover 10 square kilometers.

War        Year   Rate (wpm)   Technology    Soldiers


Civil War 1865 30 telegraph 38,830

WWI 1915 30 telegraph 4040

WWII 1945 66 teletype 360

Gulf War 1991 192000 computer 23.4

Future 2010 1.5 trillion computer 2.4


1. Does it make sense that data transmission rates were the same in WWI as in the Civil War? How do you explain the 10-fold reduction in number of soldiers required per unit area?

2. Make some plots of the data for 1865-1991. How would you describe the trends?

3. We tried a log transform on the transfer rates and computed 10/(#soldiers) to find the area a single soldier could cover. Then a plot of area vs. log(rate) looks roughly linear. Would you be comfortable extrapolating this to log(1.5 trillion) to predict the number of soldiers required?

4. How do you think the data transfer rate for 2020 was estimated? The associated number of soldiers?

School choice data rescued from bad science.
The Wall Street Journal, 14 August 1996.
Letter to the Editor, by Jay P. Greene and Paul E. Peterson

Greene is an assistant professor of political science at the University of Houston, Peterson is the director of the program in education policy and government at Harvard. The two have reanalyzed data from Milwaukee's six-year-old school choice experiment, where low income families were given the opportunity to use publicly funded vouchers to pay for private schools of their choice. Green and Peterson conclude that the program has been successful in raising student performance. They dispute the findings of John Witte of the University of Wisconsin, whose claim that there are no educational benefits has been widely quoted by teachers unions.

Greene and Peterson explain that the design of the Milwaukee experiment provided an excellent opportunity for statistical analysis. The Wisconsin legislature had ruled that, if choice schools were over-enrolled (it turned out that they were), then applicants would be admitted at random. The result was to create randomized treatment and control groups. Comparisons between these two groups show reading scores of the voucher group were an average of 3 and 5 percentile points higher than the control group during the third and fourth years of the study. Math scores for the corresponding periods were 5 and 12 points higher. During the first two years, the choice group actually averaged a fraction of a point lower. The authors say these early differences were not significant, adding that it takes time for the educational benefits to accumulate.

Witte's results are criticized because he compared the choice students with a cross-section of public school students, who on the whole were much less disadvantaged.


1. Greene and Peterson call the third- and fourth-year results "substantively significant." Do you think this phrase refers to statistical significance or practical significance?

2. The letter compares Witte's methodology to the infamous 1936 presidential poll in the Literary Digest, which published a prediction based on 2.2 million responses to a 10 million letter mailing. Is this analogy apt?

3. Playing devil's advocate, the authors write: "But could it be that students who weren't doing well in choice schools left after two years, leaving behind third- and fourth-year students who would have scored better no matter where they went to school? To answer this question, we checked to see whether the third- and fourth-year students' scores had differed significantly from those of the group as a whole during the first two years. They had not, confirming that the substantial effects of choice schools were due to accumulated learning over three to four years." Do you agree with this argument? After all, there was no difference between the choice group and the control group during the first two years, either.

Omaha hands I-Cubs fifth loss in six games, 6-2.
The Des Moines Register, 16 August 1996, Sports p1.
Randy Peterson

The the manager of the Iowa Cubs (the AAA, American Association farm club of the Chicago Cubs) is upset about the team giving up too many runs to the opposition with 2 outs. The paper supports the complaint by noting that they had given up 224 runs (39.8% of the total runs they've given up) with two outs. As of Friday they had played 126 games.


1. Runs can be scored when there are 0, 1, or 2 outs. If we naively assume that over the long run 33.3% of the total can be expected in each situation, would the data reported above really seem out of line?

2. Does the 33.3% each assumption seem reasonable? How might you go about checking it?

Please send comments and suggestions for articles to


CHANCE News 5.09

(17 July 1996 to 14 August 1996)