A researcher wants to conduct a political poll of American voters. She really wants to know how people are going to vote in the upcoming election. She's also a bit delusional and thinks she can do whatever she wants in the amount of time she wants. So, she sets her sights on collecting... a sample of all Americans who say they are going to vote. a sample of all registered voters. a sample of all Americans who voted in the last election and have registered to vote in the upcoming election. the population of all registered voters.

She wants to know how people will vote in the upcoming election, not just the opinions of anyone who talks about voting or of everyone who is registered (since many registered voters don’t actually show up at the polls). The best way to focus on those who actually vote is to target people who have a history of voting and are registered to vote again. This is why a sample of all Americans who voted in the last election and have registered to vote in the upcoming election is the best choice. Option 1 (all Americans who say they are going to vote) could include people who are overly optimistic or who might not actually vote on election day. Option 2 (all registered voters) includes many people who might not vote, and option 4 (the population of all registered voters) is essentially trying to get a census – something that isn’t practical for a poll. Thus, the answer is: A sample of all Americans who voted in the last election and have registered to vote in the upcoming election.