Passage
Read the full passage. Bracketed letters like [A] mark the four positions for the insert-text question.
The Poll That Asked Too Many People
¶1 A poll's credibility, most people assume, comes down to how many people it asks. A survey of ten million respondents should be more trustworthy than one of only a few thousand, simply because it counted more opinions. Twice within twelve years, that assumption produced one of the most confident wrong predictions in American political history. The lesson pollsters eventually learned was not that a large sample is useless. A sample's size guarantees almost nothing on its own. What matters is whether the people counted actually represent the people being predicted about. Getting that right proved far harder than simply counting responses.
¶2 In 1936, the magazine Literary Digest ran the country's best-known election poll. It had correctly predicted the winner of every U.S. presidential race since 1920. The method was simple: mail millions of postcard ballots to names drawn from telephone directories and automobile registration lists. That year it mailed ten million ballots and received nearly 2.4 million back. It confidently predicted that the Republican challenger would defeat President Franklin Roosevelt in a landslide. Roosevelt instead won one of the largest landslides in American history. The magazine's sample was enormous. In the middle of the Great Depression, however, owning a telephone or a car was still a mark of relative wealth. The households that owned them leaned far more Republican than the country as a whole. A rival forecaster, George Gallup, had used a sample of only about fifty thousand people. It was deliberately chosen to match the nation in age, income, and region, and Gallup predicted Roosevelt's win correctly. Size had lost to structure.
¶3 Gallup's technique, known as quota sampling, fixed target categories in advance: so many farmers, so many factory workers, so many residents of each region. Interviewers then kept interviewing within each category until the finished sample looked like a small mirror of the national population. It solved the flaw that had wrecked the Literary Digest poll. A sample no longer needed to be large to be accurate. It only needed to be representative of the traits that actually predicted how someone would vote. Pollsters finally had a specific, repeatable answer to a question that had previously been guesswork: exactly whom should a survey ask?
¶4 Quota sampling spread quickly. By the 1940s, several organizations were using it, including Gallup's own and two competitors. They forecast elections, tracked consumer habits, and gauged public attitudes toward the war. Interviewers carried lists of quotas into neighborhoods and stopped once each category was filled. They were confident that the method that had beaten Literary Digest would keep working. For more than a decade, it did. Quota sampling became the accepted foundation of American survey research. Newspapers printed its election forecasts as though the outcome were already decided.
¶5 That confidence broke in 1948. [A] Gallup and two rival pollsters again used quota sampling to forecast the presidential race. All three predicted that Thomas Dewey would defeat Harry Truman by a comfortable margin. [B] Truman won, and one newspaper had already printed a morning edition with the headline "Dewey Defeats Truman. Photographers later caught Truman holding up a copy of it and laughing. [C] The investigation that followed found that quota sampling's flaw was not the categories themselves but the interviewers who filled them. [D] Given freedom to choose exactly whom to approach within each quota, interviewers tended to pick people who were easier to reach. Those people leaned more Republican than the category as a whole. Pollsters replaced quota sampling with probability sampling, a method in which every person in the population has a known, fixed chance of being selected and interviewers have no say in who actually gets asked. That shift, from interviewer choice to random chance, has remained the foundation of credible survey research ever since.
Questions
11 questions — every TOEFL Reading question type, in test order.
Answer
Answer
Answer
Answer
Answer
Highlighted: "Pollsters replaced quota sampling with probability sampling, a method in which every person in the population has a known, fixed chance of being selected and interviewers have no say in who actually gets asked."
Answer
Answer
Answer
Answer
That error, once again, had nothing to do with how many people were counted.
Where would the sentence best fit?
Answer
Twentieth-century American pollsters twice made confident, widely publicized election predictions that turned out to be badly wrong, and each failure taught them something new about what actually makes a poll's sample trustworthy.