BE CURIOUS

Audio-book Review
 By Chet Yarbrough

(Blog: awalkingdelight)
 Website: chetyarbrough.blog

The Data Detective

By: Tim Harford

Narrated by: Tim Harford

Tim Harford gives listeners a practical application of “Thinking Fast and Thinking Slow” in the art of statistical analysis. Sounds boring, just as the title “The Data Detective” but in this day of media overload Harford castes a warning. Be skeptical of conclusions drawn by statistical data, whether accumulated by business interests, science nerds, or algorithms. Think slow because thinking fast obscures understanding of statistical analysis. Above all, be curious when reading a statistical analysis that either adds or subtracts from your understanding. With that admonition, Harford offers ten ways to question the veracity and truthfulness of statistical analysis.

Harford argues it is important to investigate a writer’s qualifications as an analyst, and the “how, why, and when” data is collected. As the famous economist Milton Friedman said, “Statistics do not speak for themselves.” Or, as Mark Twain made famous, “There are three kinds of lies: lies, damned lies, and statistics.” It appears Harford agrees with Friedman, but not Twain, because he believes understanding a statistical study can reveal possible or at least probable truth.

Harford gives an example of statistical reports that correctly correlated smoking with lung cancer. Cuyler Hammond’s and Daniel Horn’s 1952 statistical study led to the 1964 Surgeon General report that confirmed cancer’s correlation with smoking. The disheartening story Harford tells is the tobacco industry’s purposeful effort to deny correlation. The tobacco industry’s methods were to suggest other causes, like auto exhaust or other carcinogens, as likely causes of lung cancer. They created doubt, whether true or false, which poisons belief in statistical studies.

Harford explains personal investigation based on curiosity and detective work is necessary if one is looking for a probability of truth.

American free enterprise is created to produce product, service and jobs while making enough profit to stay in business. Sometimes those goals interfere with truth.

Harford explains there are many statistical studies purporting rises in crime, inequality, poverty, and medical health that need to be closely examined for validity. He argues every conclusion drawn from statistical surveys that contradict interest-group’ or individual’ belief should be closely examined. The methodology of a good statistical study must be understood within its era, its compiler’s biases, its stipulated human cohort, its conclusion, and its tested repeatability by others.

Harford challenges the supposition that violence has increased in America. This is undoubtedly music to the ears of elected officials who resist national gun control measures. Harford and the famed psychologist, Steven Pinker, suggest statistical analysis shows violence of earlier history is greater than in the 21st century. Harford acknowledges this is no comfort to the heart-rending reality of a child lost to suicide by gun or the horrendous school shootings of the last 3 years. As Horford explains statistics do not register human grief. Statistics are an impersonal unfeeling view of human life.

Harford does not read statistical surveys as truth but as a roadmap for discovery. He looks at a statistical survey like a detective searching for details. Who are the gatherers of the statistics? How were they collected? Why are they relevant? What period do statistics represent and do they relate the present to the past? Without answers, Harford argues statistical surveys are no better than propaganda.

Harford offers a graphic example of the context needed to clearly illustrate the value of statistical studies. The history of America’s invasion of Iraq and its human cost is dramatically and comprehensively revealed in one statistical picture.

Harford advances his view of the metaverse and its growing role in the world. He gives examples of Target’ and Costco’ algorithms that tells a father his daughter is pregnant, and infers a wife’s husband is cheating. A Target algorithm sends a note to a father about the pending birth of a baby based on his daughter’s purchases at the store. Costco sends a rebuy message for condoms to a wife when she calls and explains they never use condoms. Both stores apologize for sending their notes and say their stores made auto-response mistakes. Harford notes email apologies are a common response of stores that use similar algorithms.

The last chapters of Harford’s book reinforce the importance of statistical studies by recounting the history of Florence Nightingale’s heroic hospital service in Turkey during the Crimean war (1853-1856). Harford explains Nightingale’s interest in mathematics and association with luminaries like Charles Babbage (an English polymath that originated the concept of a digital programmable computer). Nightingale’s hospital service and interest in mathematics lead her to correlate patient’ diseases with causes. The hospital to which she was assigned by the U. K. was without proper food and water. The hospital was dirty, and disease ridden. She had two objectives. First to have food and water supplied, and second to clean the hospital. Her statistical analysis made her realize cleaning was as important as food and clean water in reducing contagion among her patients. Like the statistical analysis of smoking and cancer changed smokers, Nightingale changed nursing.

This is a disturbing book that shows the power of media and how it can mislead as well as inform the public. With poorly or intentionally misleading statistical studies, opposing interest groups harden their political beliefs.

Harford concludes with an appeal to discordant interest groups to be curious about why they disagree with each other.  Reputable statistical analysis can improve one’s belief in probable truth and decrease echo chamber‘ adherence of disparate interest groups.