Please send comments and suggestions for articles to:
jlsnell@dartmouth.edu
Back issues of Chance News and other materials for
teaching a CHANCE course are available from the Geometry
Center Mosaic (http://www.geom.umn.edu/) in their
Online Document Library.
========================================
It is time for us to reclaim our numbers,
our truth.
Cynthia Crossen
========================================
IN THIS NEWSLETTER
* FROM OUR READERS
* OTHER INTERNET SOURCES
* ABSTRACTS OF NEWS ARTICLES
* 1. Space station faces danger from flotsam.
* 2. Testing the blood 3 ways.
* 3. Embattled tobacco: One maker's struggle.
* 4. Were you mislead?
* 5. Psi in psychology.
* 6. Completing math literacy.
* 7. Early warnings, early worries.
* 8. Implant study too small for final word.
* 9. PREGNANCY and breast cancer pose lethal risk.
=======================================
FROM OUR READERS
Goran Djuknic sent us the Bureau of Consular Affairs
[Public Notice 1974] describing the rules for the visa
lottery. He noted that, at first, it is stated that
every applicant will have an equal chance of winning but
later every applicant "within a given region" has an
equal chance of winning. The following newspaper
article describes this visa lottery
<<<========<<
>>>>>==========>>
Immigrants' American dream: winning a visa.
The Washington Post, 1 July 1994, B3
Ann O'Hanlon
Over 9 million people have applied since June 1 for a
lottery which will give visas to 55,000 immigrants.
The purpose of the lottery is to balance a skewed
distribution of immigrant nationalities resulting from
giving preference to those with family ties and job
skills. Countries that sent more than 50,000 legal
immigrants to the United States in the past 5 years
are not eligible.
In this article, a State Department spokesman said, "it
is all random and nobody's chances are better than
anybody else's". As Djuknic points out, this is not
correct because the number of visa numbers allotted for
each region is chosen to give a greater share of numbers
to low admission regions. Obviously, one's chances in
this lottery depend upon both the allotment for a region
and the number of people who apply in this region.
The procedure for applying is very simple but swindlers
are making money by putting ads in papers read by
immigrants such as "Green Card!!! Try Your Luck!! $20 a
person and $35 per family.
<<<========<<
>>>>>==========>>
Jim Baumgartner suggested the following book:
Tainted Truth: the Manipulation of Fact in America
by Cynthia Crossen, Simon and Shuster 1994.
This book might well be required reading for students
in a CHANCE course.
Cynthia Crossen argues that a lot can go wrong with
statistics that cannot be blamed on the whims of chance.
Her many insightful observations include:
For very good reasons having little to do with
statistics, Coca Cola taste studies show that
people prefer Coke and Pepsi Cola studies show
that they prefer Pepsi.
Polls are a politician's weapon and are frequently
designed and interpreted accordingly.
A study paid for by the tobacco company is not
likely to conclude that second hand smoke is
dangerous and medical researchers funded by drug
companies may not be completely immune to bias.
Parameters in risk models can and often are
chosen judiciously to make the outcome agreeable
to the developer of the model.
Expert witnesses in the courts are paid a lot of
money to try to reach different conclusions from
the same statistical data.
Each chapter has numerous examples. Here are chapter
titles and examples from each chapter that I particular
enjoyed.
Chapter 1: The Study Game
Example: The author asked Gallup five questions
about credibility and information. Gallup
responded by making the questions into a survey
which they carried out for a modest fee ($4,500).
Chapter 2: The Truth about Food.
Example: the oat bran mania of the 80's and the
stampede against Alar started by the 60 Minutes
report.
. Chapter 3: Numerical Lies of Advertising:
Example: An account of the taste tests carried
out by Coca Cola and Pepsi companies.
Chapter 4: False Barometers of Opinion:
Example: Polls taken at the time of the Clarence
Thomas hearings indicated that people did not
believe Anita Hill. It is likely that these polls
influenced the outcome of the Thomas appointment.
However, similar polls a year later indicated that
Anita Hill was at least as believable, if not more
so, than Clarence Thomas.
Chapter 5: False Truth and the Future of the World
Example: The environmental battle between
disposable and cloth diaper industries. A 1988
study by the cloth diaper industry served as
ammunition for opponents of disposable until
this study was neutralized by a 1990 study
produced by Proctor & Gamble. This in turn
was followed in 1991 by another study sponsored
by the cloth diaper industry showing that
cloth diapers were environmentally superior.
Chapter 6: Drugs and Money
Example: Two conflicting studies on the
effectiveness of a drug based on the
same data were submitted simultaneously
to the New England Journal of Medicine.
The author who found the drug effective had
large grants from the pharmaceutical
companies and had his paper accepted.
The other with no grants had his paper
rejected by the NEJM but then accepted by
the Journal of the American Medical
Association but this did not lead to a
happy ending.
Chapter 7: Research in the courtroom
Example: The account of the Dalkon Shield
case is interesting but I would have
preferred DNA finger-printing.
Chapter 8: Solutions
Example: I liked the recommendation
that high schools and colleges teach
critical assessment of such news. In
other words, Take a chance on CHANCE.
<<<========<<
>>>>>==========>>
OTHER RELEVANT INTERNET SOURCES
Oncolink (The University of Pennsylvania Multimedia
Oncology Resource) is a prize winning Mosaic that
provides all kinds of news on cancer. The address
is http://cancer.med.upenn.edu/.
Oncolink has lots of interesting studies, EPA reports
survey articles, etc. many of which are related to
Chance News. The treatment of cancer and power lines is
the most complete discussion of this problem that I have
see anywhere.
I would encourage you to browse around yourself but here
are a few things I found in my browsing.
Cancer and Power Lines, emf etc.
Spiritual Well-Being: A Review of the
Research Literature
Dioxin Reassessed - Part 1,2
Decision Support for Patients with Breast Cancer
FDA Statement on Nicotine and Cigarettes
Cancer News, Warnings, etc.
<<========<<
>>>>>==========>>
The Electronic Newsstand
gopher://gopher.internet.com:2100/11/
This newsstand has a wide variety of journals some of
which have articles relevant to Chance News. Typically,
the journal includes the table of contents of the
current issue and one or two articles from this issue.
This is the case, for example, for the journals
"Discover", "The Economist" and the "Skeptical Inquirer"
of special interest to us. Occasionally you find more
complete set of articles in the current issue as, for
example, is the case with "The Scientist".
It is fun to browse, and in our own browsing we found
the following article for this chance news:
Trials and errors in breast-cancer research
The Journal of NIH Research, July 1994
Nancy Touchette
The author states, "It wasn't meant to be fought this
way. But leaders of the National Cancer Institute (NCI),
the chief strategists in the War on Cancer, are caught
in a crossfire." This crossfire started with the
revised mammography-screening guidelines, and was
followed by charges that the Breast Cancer Prevention
Trial may have seriously understated the potential risk
of tomoxifen, and the charges that the NCI tolerated
improper enrollment of patients in other breast-cancer
trials. In the process of describing this crossfire we
learn a lot the current state of our knowledge on some
of the most important on-going cancer studies.
=======================================
CURRENT NEWS
=======================================
Space station faces danger from flotsam
The New York Times, 27 June, 1994, pg. A1
William J. Broad
This article discusses NASA's estimate of the chance
that the international space station, due to be built
from 1997 to 2002, will be penetrated by "space junk"--
dead satellites, shattered rockets, and other man-made
debris. The chance of being hit is given at 1 in 5 over
the course of its construction and expected 10-year
life. The article explains that the "probability means
that, if there were five such stations, experts would
expect one to be hit." For comparison, the overall risk
of a catastrophic collision that causes the death of
astronauts or the destruction of the station is given at
"1 chance in 10".
The article states that there are "perhaps 150,000
objects that could penetrate the space station," of
which 7000, "ranging from the size of a school bus to
the size of a baseball," are tracked by the military.
The remaining objects are too small to track. The size
of the planned station itself is 361 by 290 feet--
larger than a football field.
DISCUSSION QUESTION:
The Russian space station Mir has been in orbit for
eight years, and apparently has suffered no penetrations
from space junk. How could you determine the chance of
this happening, (assuming the probabilities given in the
article are correct)?
<<<========<<
>>>>>==========>>
Testing the blood 3 ways.
The New York Times, 9 July 1994, Section 1 pg. 8
Gina Kolata
This article explains the blood test reported in the
hearing leading to the judge's decision that O. J.
Simpson must stand trial for murder. In this test, blood
found at the scene of the crime was compared with that
of the suspect O. J. Simpson and the two victims, his
former wife and Ronald Goldman.
Serology proceeds by looking at proteins in the blood
that occur in people in different forms. The most well
known protein determines the ABO grouping or blood type.
Simpson's blood type was A as was his former wife.
Ronald Goldman, the other victim, had type O.
The second protein they looked at, called ESD, has three
variants, and in this case all three had the same type.
The third protein, PGM, has 10 different forms. Here
Simpson had the same type as the bloodstain found at the
scene of the crime, and Mrs. Simpson and Mr. Goldman had
different types.
In an accompanying article, it is stated that estimates
of the frequencies for these types were obtained by
using blood tests run by the Los Angeles police
department. Naturally, these estimates were multiplied
together to obtain an estimate that 1 person in 200 had
the types of proteins that Simpson had. The defense
observed that between 40 and 80 thousand people in
Los Angeles should also have these three types.
One of the few attempts to estimate these from national
data was published in 1987, but this study did not have
enough data to include Asians or Hispanics so was hardly
applicable to Los Angeles. From the results of this
study it is estimated (again by multiplying the
individual estimates) that among the white population 1
in 150 would have the types of the proteins that Simpson
has and, in a black population, 1 in 500.
DISCUSSION QUESTIONS:
1. The prosecutor tries to narrow down the 40,000 to
80,000 people in Los Angeles that have the same blood
characteristics as Simpson by asking "But how many of
these would have a bloody glove in their house?" The
defense responds: No, you should ask "How many have
some any bloody or suspicious object such as a knife."
Which one is correct?
2. An Los Angeles Times article discussing the
possibility of a death sentence if Simpson is found
guilty remarked: "The death penalty is rarely imposed
in spousal murders. Indeed, of the 2,812 men on Death
Row nationwide, only 34, or 1.2%, killed their wives or
ex-wives". Are these statistics convincing?
3. How would you determine if multiplying the frequency
of the three protein types is reasonable?
<<<========<<
>>>>>==========>>
Embattled tobacco: one maker's struggle.
The New York Times, 16, 17, 18 June, 1984
Philip J. Hilts
Based on documents from the archives of the Brown &
Williamson Tobacco Corporation, this three-part series
describes how the major American tobacco companies
chose to publicly downplay or even deny the results of
studies, many conducted by their own research
departments, on the health risks and physiologically
addictive properties of smoking cigarettes.
For example, the second article states that after an
internal study from 1963 found evidence of significant
physiological effects from smoking, considerable effort
was put toward creating what the company called "a
device for the controlled administration of nicotine"---
a "safer" cigarette with equal (or even greater)
nicotine levels. A top researcher for Brown &
Williamson's sister company, British-American Tobacco
(Batco), described Batco in 1967 as being "in the
nicotine rather than the tobacco industry." But when
questioned in April of this year by the House
Subcommittee on Health and the Environment, each of the
seven top executives in the American tobacco industry
stated that nicotine is not addictive and that
cigarettes may not cause cancer.
Meanwhile, on 17 June, 1994 a spokesman for Brown &
Williamson is quoted as saying: "The tobacco industry
was and is just as interested in research on smoking and
health as those outside the industry. Our position
continues to be that there are health risks
statistically associated with smoking, but there is no
conclusive evidence of a causal link between tobacco use
and disease."
The third article includes a "Chronology of Concern"
that lists some of the results of major studies on
tobacco and disease since the 1950's, along with actions
taken by tobacco companies and government agencies.
DISCUSSION QUESTIONS:
1. Referring to the statement by the industry
spokesman, above, what do you think is meant by the
phrases "statistically associated", "causal link" , and
"conclusive evidence"?
2. In testimony before congress, Andrew W. Tisch,
chairman of the Lorillard Tobacco Co., stated that "We
have looked at the data and the data we have been able
to see has all been statistical data that has not
convinced me that smoking causes death." What do you
think of this remark?
<<<========<<
>>>>>========= =>>
Were you mislead?
The Boston Globe, 3 July 1994, p7.
Paid advertisement, Philip Morris USA
This advertisement reprints an article from Forbes
MediaCritic entitled "Passive Reporting on Passive
Smoke", by Jacob Sullum, the managing editor of Reason
magazine. (This is clearly related, though not
identical, to Sullum's recent National Review article;
see CHANCE News, June 10, 1994.) The central theme is
that the media have been totally uncritical in their
acceptance of the EPA's claims of evidence linking
environmental tobacco smoke (ETS) to cancer. For
example, the EPA examined 30 epidemiological studies
looking for a link. While most of the studies found a
positive association, this association was statistically
significant in only six. Why, Sullum asks, has the
issue of significance been ignored in press reports?
Sullum also notes that the risk ratios associated with
ETS are an order of magnitude lower than those for
smokers. He complains that the news has been reported in
a way that leads people to think that the evidence is
just as strong for passive smoking as for active.
Recent studies indicate that the average male smoker is
20 times more likely to develop lung cancer than a male
nonsmoker; the corresponding risk ratio for women is 10
to 1. By contrast the EPA estimated that a woman living
with a smoker is 1.19 times as likely to develop lung
cancer as a woman who lives with a non-smoker. Sullum
states that, with risk ratios this small, it is
difficult to rule out other confounding variables, such
as diet and pollution. James Enstrom, a UCLA professor
of epidemiology, is quoted as saying: "You're talking
about risk ratios so close to 1.0 that it's really
beyond the realm of epidemiology."
DISCUSSION QUESTIONS:
1. At one point, noting the media's failure to explain
that an association was not statistically significant,
Sullum explains that this means "the probability that
the result occurred purely by chance was greater than
5%." Is this an accurate description?
2. In a section entitled Weasel Words, Sullum complains
about phrases describing evidence that 'suggested there
may be an increased...risk' or 'shows an apparent
pattern' or 'appears to be associated.' He states that
"in rigorous science, close doesn't count". Comment.
3. Sullum writes that "no study has ever found
'statistical proof that secondhand smoke caused cancer
with certainty.' (In fact, it is impossible for an
epidemiological study to provide such proof)." Does
this mean that all epidemiological research is
worthless? What would constitute "statistical proof
with certainty?"
<<<========<<
>>>>>========= =>>
Psi in psychology.
Skeptical Inquirer, Summer 1994, pg. 351
Susan Blackmore
A recent paper in one of psychology's most prestigious
academic journals, Psychological Bulletin (Vol. 115,
1994, 4-18.) forms the subject of this article by Susan
Blackmore, a parapsychology researcher for the past 20
years. The paper, entitled, "Does Psi Exist? Replicable
Evidence for an Anomalous Process of Information
Transfer", concerns recent research on the existence of
ESP by parapsychologist Charles Honorton and
psychologist Daryl Bem. Blackmore remarks, "Bem's high
profile and the respect he is accorded by psychologists
will ensure that [their paper] is taken seriously."
Honorton and Bem's paper concerns experiments in sending
and receiving "telepathic" information using a
procedure, pioneered by Honorton, known as a "ganzfeld".
In the experiment, a person called the "receiver" is
deprived of any outside sensory stimuli, while another,
the "transmitter", concentrates on one of four randomly
selected images. In general the ganzfeld is designed to
limit the possibility of fraud or sloppy technique. The
paper published in the Psychological Bulletin claims to
have produced results with a 40 percent success rate,
while a 25 percent success rate would be expected by
chance alone.
For more information about the history of ESP research
and a thorough description of the ganzfeld procedure,
see the article
Inside Story: Talking Heads.
The Guardian Weekend Page, 25 March, 1994, pg. 34
Simon Beckett.
<<<========<<
>>>>>========= =>>
Completing math literacy; teachers attend 'U' seminar on
probability and statistics
Star Tribune, 1 July 1994, pg. 3B
Jim Dawson
This article from a Minneapolis newspaper describes the
recent workshop for high-school teachers about teaching
the CHANCE course held at the University of Minnesota.
The students were 30 junior and senior high school math
teachers from Minnesota.
In the article there are examples of topics covered in
classes, such as margin of error and the reliability of
political polls, and DNA fingerprinting in the courts.
A typical workshop activity is also described in some
detail: the design and execution of an experiment to
determine the best of 12 chocolate-chip cookies.
A syllabus for this course can be found on the
CHANCE Mosaic.
DISCUSSION QUESTIONS:
1. Here is an example of a discussion question about
polling and margin of error used at the workshop.
A poll in the Star Tribute says that "one can be 95%
confident that error due to sampling will be no more
than plus or minus 4.4 percentage points." A poll in the
New York Times taken in Iowa says that "In theory, in 19
out of 20 cases the results based on such samples will
differ by no more than 3 percentage points in either
direction by what would have been obtained by
interviewing all adult Iowans." These are both attempts
to explain the concept of margin of error. Do they
amount to the same thing? If not, which do you think is
the more accurate description? Where do the 4.4 and 3
percent come from?
2. Mr. Dawson wrote that the second of the two polls
explains margin of error this way: `In theory, in 19 out
of 20 cases the results based on such samples will
differ by no more than 3 percentage points in either
direction.' How does this differ from the explanation
given in question 1?
3. According to the article, one participant apparently
used a theory "from a discreet [sic] mathematics course
he had once taken" to help with the chocolate-chip
cookie experiment. What do you suppose is taught in a
"discreet" mathematics course?
<<========<<
>>>>>==========>>
Early warnings, early worries
The Economist, 18 June 1994, Pg. 91
This is a remarkably complete story about problems
related to medical screening.
The article begins with the medical problems caused by
false positive and false negative outcomes. Such
outcomes can lead to additional expensive tests,
unnecessary and sometimes harmful treatments, and
emotional problems. The issues are illustrated in terms
of a number of specific diseases and screening
processes.
The author discusses current problems associated with
genetic testing and future problems anticipated as these
tests become more common. Again specific diseases and
screening processes are discussed.
The author remarks that the ability to develop new
screening tests is running ahead of new cures and
clearly thinks that screening is in danger of getting
out of control.
<<========<<
>>>>>========= =>>
Implant study too small for final word.
The New York Times, 28 June, 1994, pg. A16
Letter to the Editor
Sidney M. Wolfe, M.D., and Joanne C. Mott
This Letter is in response to a 16 June news article
which reported on a study by the Mayo Clinic. The
study, conducted to determine if there is a link between
silicone breast implants and connective tissue diseases
such as rheumatoid arthritis, found no evidence of such
a link. The authors criticize the study for not
addressing problems which they claim are "unequivocally
caused by silicone implants", including delayed
diagnosis of breast cancer and implant rupture.
They also condemn the study for its small sample size,
stating that there is "...a low probability that such a
small sample will detect a statistically significant
increase in connective tissue diseases among women with
breast implants." For example, the authors claim that
the probability that the study would detect a doubling
of the risk of these diseases is 31 percent.
DISCUSSION QUESTION: How do you think the 31 percent
figure was determined?
<<<========<<
>>>>>==========>>
Pregnancy and breast cancer pose lethal risk
Los Angeles Times, 24 June, 1994, pg. A3
Thomas H. Maugh II
This article reports on a study published in The Lancet
by the M.D. Anderson Cancer Center in Houston which
found that women who are pregnant when diagnosed with
breast cancer are 3.26 times as likely to die from the
disease as women who are not and never have been
pregnant. If the woman is not pregnant, there is still
a higher risk of mortality if she has been pregnant
during the past five years, with the highest risks
associated with more recent pregnancies. The results are
based on studies of 407 women in their 20s.
There has been evidence that younger women with breast
cancer generally fare worse than older women, and,
according to Dr. Vincent F. Guinee of M.D. Anderson, the
results from this study show that "pregnancy itself is
responsible." However, the article states that the
increased mortality rate for pregnant women should apply
to women of all ages.
DISCUSSION QUESTIONS:
1. The article says that when the researchers "adjusted
for the stage of the disease at the time of detection--
the size of the tumor, the number of lymph nodes
involved and so on--the risk was reduced, but the women
were still 2.83 times as likely to die. What do you
think this means?
2. According to Eugenia Calle of the American Cancer
Society, women in their 20s represent less than 1% of
all breast cancer cases. On the other hand, the article
states that one in every 2,426 women under 30 develops
breast cancer, compared to one in 96 for women aged 45.
Over a woman's life span, the risk is one in nine. What
do these numbers mean?
<<<========<<
>>>>>==========>>
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
CHANCE News 3.09
(11 June to 8 July 1994)
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Please send suggestions to: jlsnell@dartmouth.edu