Which approach best reduces uncertainty in polling?

Which approach best reduces uncertainty in polling?

• A. Ignoring nonresponse and using a small, non-random sample.
• B. Increasing sample size, ensuring random sampling, and adjusting for nonresponse.
• C. Asking only friends and family of respondents.
• D. Using a single open-ended question only.

Answer: (B)

Uncertainty in polling comes from not sampling the entire population and from who actually responds. The best way to cut that uncertainty is to use a larger, randomly selected sample and to adjust for nonresponse. A bigger sample lowers the margin of error because the estimate’s precision improves as sample size grows. Random sampling ensures every individual has a known chance of being included, making the results more representative of the population. Adjusting for nonresponse, typically through weighting, helps fix biases that occur when certain groups are less likely to respond, so the sample better reflects population characteristics. By contrast, ignoring nonresponse and using a small, non-random sample introduces bias and higher uncertainty, asking only friends and family of respondents skews results with a biased sample, and relying on a single open-ended question reduces measurement reliability and precision.

Uncertainty in polling comes from not sampling the entire population and from who actually responds. The best way to cut that uncertainty is to use a larger, randomly selected sample and to adjust for nonresponse. A bigger sample lowers the margin of error because the estimate’s precision improves as sample size grows. Random sampling ensures every individual has a known chance of being included, making the results more representative of the population. Adjusting for nonresponse, typically through weighting, helps fix biases that occur when certain groups are less likely to respond, so the sample better reflects population characteristics. By contrast, ignoring nonresponse and using a small, non-random sample introduces bias and higher uncertainty, asking only friends and family of respondents skews results with a biased sample, and relying on a single open-ended question reduces measurement reliability and precision.