One of my rare departures from crime and thriller novels.
(As usual, a note on full disclosure. I flunked my first
stats class. As a transfer student in the pre-computerized registration days, I
just needed hours to be eligible for varsity athletics. About the only course
that fit my schedule was a course called Social Statistics taught by a barely
fluent international faculty member. I had no idea what stats were, got lost by
midterms and stopped going. A bit older and wiser and now in grad school, stats
just clicked (kudos to a GREAT teacher of introductory stats) and I aced my way through multiple stats courses. Taught stats most
every semester when working in a university.)
Dr. Wheelan is a professor of public policy and economics at
Dartmouth and previously wrote Naked Economics. This book takes a peek behind
the curtain of how statistics can be used and misused. He points out that the
old line, “lies, damn lies, and statistics” should probably be “lies, damn
lies, and bad data” because it’s not the statistics that mislead people.
Statistics are just a tool. If something’s confusing, it’s more likely the
fault of the data, usually because of bad sampling.
For about 12 weeks this winter, my wife hosted a Bible study for
women at the house, which meant I was kicked out and spent the bulk of my
evenings at Barnes and Noble. This book was displayed on a table on the route to the
little boy’s room. The engaging part of this book is Dr. Wheelan’s presentation;
think one part John Corey (Nelson DeMille) or Elvis Cole (Robert Crais) and one part Darrell Huff (author of
the now infamous (to statisticians at least) How to Lie With Statistics, ©1954, of which I have an original).
The book breaks no statistical
ground. What it does do, exceptionally well, is scrape off the glamour (or
crust, you choose) about how stats are used to twist or disguise a topic to
meet some agenda. His examples are drawn liberally from economics (national productivity, wealth), education (rating schools and teachers), and sports
(who’s the best QB). He also throws gasoline on the fire that is the US News
and World Report rankings of colleges/universities and doctors/hospitals
showing how the results can be manipulated. Not to mention the habits of those most grievous of liars – make that manipulators of the truth – politicians.
Actually thought one of the best
chapters was on something most find simple and quasi-dull – descriptive stats
(mean, median, standard deviation, standard error) by using as his example the
health trajectory of Jay Gould, the late evolutionary biologist. Dr. Gould was
diagnosed in 1982 with mesothelioma where he learned the diagnosis had a median
eight-month survival; half live less than eight months, half live longer. Thus
ensued a discussion about data distributions, skewness, and responders vs.
non-responders to treatment.
Were I still teaching stats in college,
this book would form part of my trilogy of resources: the textbook, Huff’s How
to Lie, and Wheelan’s book. The examples are so ridiculously clear, presented
with a wink and nod that no one who makes the effort will regret. This wouldn't be a primary text for a class, but it would be the ideal source of real world examples of the wise use of statistics. Yeah, it's that good.
BTW. Dr. Gould eventually
succumb to an unrelated cancer . . . in 2002, 20 years after being told he
probably had eight months to live.
East Coast Don
No comments:
Post a Comment