An Introduction to Statistics
Major Lessons we should Learn in this Course
Title: Sep 1 2:56 PM (1 of 9)
The study of statistics is supposed to help
students work with data in their varied
academic disciplines and in their unpredictable
later employment. Students learn to work with
data by working with data. (Moore, [Link])
Data are more than mere numbers they
are numbers with a context that should
play a role in making sense of the numbers
and in stating conclusions. (Moore, p. ix)
Statistics, more than [other branches of]
mathematics, depends on judment for
effective use. (Moore, p. x)
Data are numbers within context... The context
engages our background knowledge and allows
us to make judments. (Moore, p. xx)
Title: Sep 1 1:27 PM (2 of 9)
Data Beat Anecdotes
An anecdote is a striking story that sticks in our
minds exactly because it is striking. Anecdotes humanize an
issue, hut they can he misleading.
Does living near power lines cause leukemia in
children? The National Cancer institute spent 5 years and $5
million gathering data on the question. The researchers
compared 638 children who had leukemia and 620 who did
not. They went into the homes and actually measured the
magnetic fields in the children’s bedrooms, in other rooms,
and at the front door. They recorded facts about power
lines near the family home and also near the mother’s
residence when she was pregnant. Result: no connection
between leukemia and exposure to magnetic fields of the
kind produced by power lines. The editorial that
accompanied the study report in the New England Journal
of Medicine thundered, “It is time to stop wasting our
research resources” on the question.
Now compare the effectiveness of a television news
report of a 5year, $5 million investigation against a televised
interview with an articulate mother whose child has leukemia
and who happens to live near a power line. In the public
mind, the anecdote wins every time, A statistically literate
person, however, knows that data are more reliable than
anecdotes because they sysrematically describe an overall
picture rather than locus on a few incidents.
Title: Sep 1 1:34 PM (3 of 9)
Beware the Lurking Variable
The Kalamazoo (Michigan) Symphony once advertised a
“Mozart for Minors” program with this statement: “Question:
Which students scored 51 points higher in verbal skills and 39
points higher in math? Answer: Students who had experience in
music.”
Who would dispute that early experience with music is
good for you? The skeptical statistician, that’s who. Children
who take music lessons and attend concerts tend to have
prosperous and welleducated parents. These same children are
also likely to attend good schools, get good health care, and be
encouraged to study hard. No wonder they score well on tests.
The children’s family background is a lurking variable when
we talk about the relationship hetween music and test scores. It
is lurking behind the scenes, unmentioned in the symphony’s
publicity. Yet family background, more than anything else we
can measure, influences children’s academic performance.
Perhaps the Kalamazoo Youth Soccer League should
advertise that students who play soccer score higher on tests.
After all, children who play soccer, like those who have
experience in music, tend to have educated and prosperous
parents. The message is worth repeating: beware the lurking
variable, because almost all relationships between two variables
are influenced by other variables lurking in the background.
Title: Sep 1 1:34 PM (4 of 9)
Where the data come from is important
The advice columnist Ann Landers once asked her readers, “If
you had it to do over again, would you have children?” A few
weeks later, her column was headlined “70% OF PARENTS
SAY KIDS NOT WORTH IT.” Indeed, 70% of the nearly
10,000 parents who wrote in said they would not have children if
they could make the choice again. Do you believe that 70°o of
all parents regret having children?
You shouldn’t. The people who took the trouble to write
Ann Landers are nor representative of all parents. Their letters
showed that many of them were angry at their children. All we
know from these data is that there are some unhappy parents out
there. A statistically designed poll, unlike Ann Landers’s appeal,
targets specific people chosen in a way that gives all parents the
same chance to he asked. Such a poll showed that 91% of
parents would have children again, It matters a lot whether data
come from a haphazard poll or a statistically designed survey: if
you are careless about how you get your data, you may
announce 70% “No” when the truth is close to 90% "Yes.”
Title: Sep 1 1:34 PM (5 of 9)
Variation is Everywhere
The company’s sales reps file into their monthly meeting. The sales manager rises.
“Congratulations! Our sales were up 2% last month, so we’re all drinking champagne
this morning. You remember that when sales were down 1% last month I fired half of
our reps.” This picture is only slightly exaggerated. Many managers overreact to small
shortterm variations in key figures. Here is Arthur Nielsen, head of the country’s
largest market research firm, describing his experience:
"Too many business people assign equal validity to all numbers printed an
paper. They accept numbers as representing Truth and find it difficult to work with
the concept of probability. They do not see a number as a kind of shurthand fisr a
range that describes our actual knowledge of the underlying condition."
Title: Sep 1 1:34 PM (6 of 9)
Title: Sep 1 1:45 PM (7 of 9)
Conclusions are not Certain
Most women who reach middle age have regular mammograms
to detect breast cancer. Do mammograms really reduce the risk
of dying of breast cancer? To defeat the lurking variable,
doctors rely on experiments (called “clinical trials” in medicine)
that compare different ways of screening for breast cancer. The
conclusion from 13 such trials is that mammograms reduce the
risk of death in women aged 50 to 64 years by 26%.
On the average, then, women who have regular
mammograms are less likely to die of breast cancer. But
because variation is everywhere, the results are different for
different women. Some women who have yearly mammograms
die of breast cancer, and some who never have mammograms
live to 100 and die when they crash their motorcycles. Statistical
conclusions are “on the average” statements only. Well then,
can we be sure that mammograms reduce risk on the average?
No. We can he pretty confident, but we can’t he sure. Because
variation is everywhere, conclusions are uncertain.
Statistics gives us a language for talking about uncertainty that is
used and understood by statistically literate people everywhere.
Title: Sep 1 1:34 PM (8 of 9)
Chapter 1:
Picturing Distributions with
Graphs
Title: Sep 1 1:16 PM (9 of 9)
