Bests & Worsts Reviews from Amazon

On page 107 of this book, in a box labeled "Selection bias lives on in infamy," one reads that "What was arguably the all-time greatest example of selection bias resulted in the embarrassing 1948 Chicago Tribune headline `Dewey defeats Truman.' In reality, Harry Truman trounced his opponent." The author goes on to tell us that "...the telephone polls conducted tended to favor Dewey because in 1948, telephones were generally limited to wealthier households, and Dewey was mainly popular among elite voters." Ms. Seethaler has confused the 1948 election with that of 1936, when the Literary Digest straw poll predicted incorrectly that Landon would defeat Roosevelt. Selection bias might be a reasonable explanation for the Literary Digest debacle, but not for the 1948 polls, as shown by the 1949 Mosteller Commission report to the Social Science Research Council. Worse, the explanation of why the polls were wrong is pure fabrication. Among other things, there were no telephone polls in 1948, much less in 1936, when the Literary Digest relied primarily on mail ballots. Telephone polling did not become the norm until the 1970's. So much has been written about the 1936 and 1948 presidential election polls that it is hard to believe any competent researcher could get the story so wrong. Given the title and stated purpose of this book, the lapse is simply jaw-dropping. How can one trust anything else one reads there? Beyond that, the author's grasp of statistical concepts is tenuous at best, she frequently makes arguments that treat unsupported assertions as established facts, the writing is deadly and the book is organized like a collection of crib notes. I first found it in my local library, but decided to buy a copy as an illustration of the worst of science journalism. I can't think of any other use for it.