The answer lies with the story of Herbert Needleman, a professor (first at the Harvard School of Public Health, and subsequently at the University of Pittsburgh) who did pioneering work on the effects of low-level lead exposure on children’s intelligence. Needleman’s conclusions were challenged, and he was forced to submit to a formal inquiry at the University of Pittsburgh—not because his data or his conclusions were faulty, but because of complexities and misunderstandings surrounding the regression methods he had used to analyze the data.
Glymour cautions against
drawing simplistic conclusions from complex real-life stories, but notes that Needleman’s
story highlights the problems inherent to applied statistics; and that science as practiced is frequently more sensible
than are the pieties of official scientific methodology. Finally—and here’s
where the title comes in—Glymour writes: “justice is not done when sensible
scientists are tried by untenable standards, any more than it was in the
seventeenth century.”
Each
of the chapters in this compact, acerbic book tells a similar story, set in
places as diverse as Glymour’s hardscrabble hometown of Butte, Montana; public
schools; NASA’s Mars mission; and (in the case of an imagined conversation on
climate change with Glymour’s father-in-law, an old-fashioned conservative) a
dinner table.
Drawn
from real life, these stories are too complex to be parables, but they possess
a parable’s acuity in conveying an underlying truth. What they collectively
demonstrate is that
well-intentioned science can be untrustworthy, and that this dubious science
sometimes supports practices central to our daily life, from forest fire
forecasting to medical research to public education.
Most of these stories involve mistakes or abuses of methodology, but sometimes they are simply the product of the very ordinary human frailty of wanting something to be true. Grimly funny, the stories together convey the dangers in mis(applying) science—and demonstrate how that practice can create Galileos in places as apparently unlikely as Pittsburgh.
