MEASURING PUBLIC OPINION

MEASURING PUBLIC OPINION

  • Measuring public opinion is a huge and lucrative business. But it is important to know that polling is essentially a statistical representation that projects likely outcomes or positions. The quality of those outcomes is determined by the quality of the modeling the polling is using.

  • Opinion polling communicates to both politicans and the electorate.

  • Every part of the sampling process atters and can indicate to you the reliability of the data being presented.

  • Polling is conducted using a sample of respondents, but how to achieve a sample matters in the applicability of your returns.

  • Representative v.random sampling

  • Selection bias, occurs when a sample is used that is not reflective of the population as a whole.

  • Sample sizes impact tdata reliability, a sample size too small may not reflect public opinion.

  • Sampling errors attempt to convey the margins of accuracy for data returns.

  • How the questions are written and othered impact respondents answers, therefore impacting the true applicability of the data return.

  • Questions can be poorly worded, or only offer two options for responses - when a respondent would choose a third option if available.

  • The internet and phones better enable the reach of polling and sampling.

  • However, you can sign up to be sampled, meaning that polls have a higer representation of those who are more involved in government.

  • The returns also must be taken at face value, assuming the persons sampled are being truthful.

  • Much like questions can be engineered to illicit certain responses, respondents can engineer responses tailored to what they believe is the more popular answer.

  • Social desirability bias

  • Polling itself can also have a sizable impact on election outcomes, as respondents can be interpreted by people as identifying a winning candidate - which they in turn support because they were identified as winning or leading.

  • Polling has also evolved beyond internet and phone response, with companies investing significantly in data mining and big data collection to better predict human action and more specifically tailor messaging to influence behavior.

  • These data collection strategies are used to enhance the accuracy of polls and by politicians and supporters to achieve your vote.

  • Predictive polls are often wrong

  • Due to the myriad of factors we outlined about both the construct of the poll and the reliability of respondents, polls should be considered predictive rather than fact