home.gif (1194 bytes)grades.gif (1215 bytes)assignments.gif (1284 bytes)feedback.gif (1254 bytes)discboard.gif (1264 bytes)

syllabus.gif (1124 bytes)terminology.gif (1142 bytes)lectures.gif (1112 bytes)resources.gif (1130 bytes)jmp.gif (1086 bytes)

 

title.gif (3960 bytes)

 

 

1936 Literary Digest Poll

A convenience sample gone bad

The 1936 Presidential election was between Alf Landon (Republican) and Franklin D. Roosevelt. The Literary Digest sent out 10 million (!) ballots to

subscribers
people in the phone book
auto registry records
voter registration records

The return rate was 23% (which is incredibly high for a volunteer survey, often return rates are below 5%) meaning 2.3 million ballets were returned. For reference, the US Census counted 123 million Americans in 1930 and 132 million in 1940. Most modern Gallup polls are based on surveys of 1,000 Americans.

The poll predicted Roosevelt would get 43% of the vote. But Roosevelt won by a landslide getting 62% of the vote.

Why was Literary Digest so wrong? Their sample was a sample of convenience and the sample was not representative of the target population (American adults). In short, their sample over-represented wealthier Americans (people with phones and cars) who, even then, tended to be Republicans.

George Gallup accurately predicted the 1936 election. He also accurately predicted the outcome of the Literary Digest poll before the results were tabulated! He used quota sampling to select his sample which made his sampled population much closer to his target population. This was the beginning of George Gallup's successful career. Today the term "Gallup poll" is familiar to most Americans.

 

 

E-mail Mr. Callahan at stat110@edcallahan.com with questions or comments about this web site or about the class itself.

This page was last modified on October 17, 1999.