The table shows hypothetical results from a random-sample poll of likely voters in a two-candidate race. Each margin of error is at the 95 percent confidence level.
| Group | Sample size | Candidate X | Candidate Y | Margin of error |
|---|---|---|---|---|
| All likely voters | 1,200 | 47% | 46% | ±2.9 points |
| Ages 18-29 | 200 | 55% | 38% | ±7.1 points |
| Ages 30-49 | 400 | 47% | 46% | ±5.0 points |
| Ages 50-64 | 350 | 44% | 50% | ±5.3 points |
| Ages 65 and older | 250 | 46% | 49% | ±6.3 points |
Based on the data in the table, which of the following statements is most accurate?
- ACandidate X leads among all likely voters by more than the poll's margin of error
- BVoters aged 50-64 prefer Candidate Y by a gap too large to be explained by sampling error
- COnly among voters aged 18-29 is either candidate's lead clearly beyond sampling errorCorrect
- DThe estimate for voters aged 65 and older is more precise than the estimate for voters aged 30-49
Why C is right
A margin of error gives the range within which the true population value probably falls because of sampling error. Among voters aged 18-29, Candidate X's plausible range (55 ± 7.1, or 47.9-62.1%) does not overlap Candidate Y's (38 ± 7.1, or 30.9-45.1%), so the 17-point lead is larger than sampling error can explain. In every other row the two candidates' ranges overlap. Overall, the 1-point gap is well inside ±2.9. For ages 50-64, 44 ± 5.3 reaches 49.3 and 50 ± 5.3 drops to 44.7. For ages 65 and older, the 3-point gap is inside ±6.3. Ages 30-49 differ by only 1 point.
Why the other 3 choices miss
- A
- Misreads the overall row. X leads by just 1 point (47% to 46%), well within the ±2.9-point margin of error, so the race is a statistical tie.
- B
- Compares the 6-point gap with the ±5.3 margin as though the margin applied to the gap. Each candidate's share carries its own ±5.3 range, and those ranges overlap (X up to 49.3%, Y down to 44.7%), so sampling error could explain the gap.
- D
- Reverses how precision works. The 65-and-older group has a smaller sample (250 vs. 400) and so a larger margin of error (±6.3 vs. ±5.0), which makes its estimate less precise, not more.
When pollsters report that a survey has a '95% confidence level,' they mean that
- A95% of the eligible population was contacted and offered the opportunity to participate in the survey
- B95% of respondents gave accurate and honest answers, and 5% are estimated to have provided false responses
- Cthe pollster is 95% certain that no sampling bias occurred in the respondent selection process
- Dif the same survey were conducted many times using the same sampling method, 95% of the resulting confidence intervals would contain the true population valueCorrect
Why D is right
A 95% confidence level is a frequentist statistical concept: if the survey procedure were repeated many times on independent random samples from the same population, approximately 95% of the calculated confidence intervals would contain the true population parameter. It does not mean the pollster is 95% certain their specific interval is correct — that probability is either 0 or 1 (the true value either is or is not in the interval). This distinction is frequently misunderstood by journalists and the public, who often interpret it as a probability about the specific poll's interval.
Why the other 3 choices miss
- A
- Incorrect because no poll contacts 95% of the eligible population; polls contact small samples (typically 600-2,000 respondents) and the confidence level refers to the sampling procedure's long-run accuracy, not the fraction of the population surveyed.
- B
- Incorrect because the 95% confidence level is a statement about the sampling procedure's long-run properties, not about respondent honesty rates; dishonesty or social desirability bias is a separate measurement concern addressed differently.
- C
- Incorrect because a 95% confidence level says nothing about sampling bias; confidence intervals assume random sampling and are invalid (regardless of stated confidence level) if systematic bias exists in sample selection.