Which of the following is NOT true regarding polling?
A. A poll that showed Candidate X with 50 percent of the vote and Candidate Y with 45 percent of the vote, with a margin of error at 3 percent, means that we can be 97 percent sure that Candidate X will win.
B. The larger the sample, the smaller the margin of error, if the sample is randomly drawn.
C. A sample of about 2,000 respondents can be very representative of the entire U.S. population if it is randomly drawn.
D. People will often give answers to questions that they know little about.
E. How the question is worded can lead a person to give significantly different answers.
The correct answer to this is Option A. Option A shows an incorrect understanding of what margin of error means. Margin of error does not refer to confidence levels. Let us look at what margin of error really means.
Let us say that we take a poll using a random sample of likely Republican primary voters in Iowa. We find that 22% of them say that they will vote for Donald Trump. But we did not ask every likely Republican primary voter in Iowa so we do not know for sure exactly what percent support Trump. Therefore, there is going to be a margin of error that will depend mostly on how big our sample is and how randomly it was selected. Let us say that our margin of error is plus or minus 4%. That means that we are reasonably sure that the actual support for Trump is between 18% and 26%. So, for the example given in the question, it means that Candidate X could have anywhere from 53% of the vote to 47% and Candidate Y could be anywhere from 42% to 48%. This is the correct meaning of margin of error, which is why Option A is the correct answer.