Interpret what a poll's margin of error actually tells you about sampling variability.
22 original sample statistics and margin of error questions in the SAT Ranker bank, with instant scoring and an Elo-based rank once you sign up. Below are a few free samples with full explanations.
A poll of 1000 randomly selected voters found that 38% support a proposed policy, with a reported margin of error of 3%. Which choice best interprets the margin of error?
A margin of error describes the expected range of variation between a sample statistic and the true population value due to random sampling, not a guarantee, a proof of methodology, or a nonresponse rate.
A poll of 200 randomly selected voters found that 58% support a proposed policy, with a reported margin of error of 5%. Which choice best interprets the margin of error?
A margin of error describes the expected range of variation between a sample statistic and the true population value due to random sampling, not a guarantee, a proof of methodology, or a nonresponse rate.
A poll of 800 randomly selected voters found that 48% support a proposed policy, with a reported margin of error of 5%. Which choice best interprets the margin of error?
A margin of error describes the expected range of variation between a sample statistic and the true population value due to random sampling, not a guarantee, a proof of methodology, or a nonresponse rate.
A poll of 200 randomly selected voters found that 39% support a proposed policy, with a reported margin of error of 2%. Which choice best interprets the margin of error?
A margin of error describes the expected range of variation between a sample statistic and the true population value due to random sampling, not a guarantee, a proof of methodology, or a nonresponse rate.
A poll of randomly selected voters estimates that 46% support a measure, with a margin of error of 3 percentage points at the 95% confidence level. Which statement is the best interpretation?
The margin of error gives an interval around the estimate, here 46% ± 3%, which runs from 43% to 49% and plausibly contains the true population value. The 46% is exactly what the sample reported, so the interval describes the population rather than the sample.