Public Opinion Research and Polling Resource Centre

  • Home
  • Resources
  • Public Opinion Research and Polling Resource Centre

What is Public Opinion Polling?

A public opinion poll is considered scientific because the results are drawn from research that is conducted on a representative sample in order to make generalizations about a population being studied. The statistical theory of generalization only applies to sample surveys that are based on probability sampling. Probability samples are statistical theory of generalization only applies to sample surveys that are based on probability sampling. Probability samples are based on the principle of randomness. Random means that each unit in the population has an equal or at least measurable chance of being included in the sample. The principle of randomness provides a sound theoretical basis for projecting the results to the general population, which is called statistical inference. 

On the other hand, you cannot draw generalizations about the population from which non-probability samples are drawn. However, the predictive ability of polls from non-probability samples has been just as accurate in recent years, opening up new debates as technology, mixed-methods and new sampling strategies offer arguably improved responses and better quality data. Non-probability sampling does not involve random selection. With non-probability samples, we may or may not represent the population well, and it will often be hard for us to know how well we have done so. All online samples are non-probability samples. However, the results from a sample drawn from good access panels (as judged by polling results) could resemble those obtained from a probability sample, i.e., it might be a representative sample while not having been randomly selected. 

The results of a public opinion poll often help measure, understand and report on the attitudes, perceptions, and behaviours of the population being studied. Public opinion polls are just a measurement tool. However, as accuracy remains one of the ten core principles of marketing research, when reporting on public opinion polls, it is important to report the margin of sampling error and confidence level on probability samples. When reporting on non-probability samples, it is important to report the researcher’s use of other indicators to indicate the level of and possible sources of error in the sample. 


How do election polls affect voters?- CTV

How to report on polls responsibly - Poynter institute

Elections 2020 - Field guide to Polling- Pew Research Centre

Surveys and Polling- American Statistical Association and SciLine

Guides on polling for media and news outlets- American Association of Public Opinion Research (AAPOR)

Guidance for media and journalists looking to report on data from public opinion polls- Market Research Society (MRS), UK

What is public opinion polling and why is it important?- Gallup

The polling report- Polling Report.com

American Historical Association - Public Opinion Polls

A podcast on polling - CBC


There are many reasons polling paints an incomplete picture often. According to market research, polls can be inaccurate if there are flawed “weights” in the sample, shoddily constructed questionnaire design, or minimal contact rates. 

·       Sampling issues in polling - Pew Research Centre

·       Election polling data - Stanford University

Organizations like the Pew research centre and the American Association of Public Opinion Research reveal that polls in the US mainly concentrated on 3 groups of people, the Adults, the Registered voters(RV), and the likely voters(LV).

How to define likely voters - Gallup

Can likely voter models be improved? - Pew

Likely and unlikely voters - AAPOR 2004


One of the most remarkable applications of survey research is in election polling. Election polls attract great attention for their perceived ability to provide a strong indication of the outcome of elections based on scientific polling. However, as polls should only be seen as a measurement tool, the most important function of election polls is to serve a better understanding of the public’s perceptions of a campaign and election for journalists as well as citizens. Some resources below expand on the role of polling in elections. 

·       Roper Centre for Public Opinion Research - Congressional Elections

·       Survey research - election polling

·       AAPOR Education Election Polling Resources


A sampling method is biased if every member of the population does not have equal likelihood of being in the sample. How can one choose an appropriate sample? 

·       The New York Times polling standards 

·       Gallup World poll methodology


The margin of error is a statistic, expressing the amount of random sampling error in the results of a survey. Usually, a poll includes a sample of respondents rather than the total population, which leads to the random sampling error. So, the margin of sampling error describes how closely we can expect a survey's result to match the outcome relative to the true population's value.

·       Understanding the margin of error in polls - Pew Research Centre

·       Margin of error for journalists of survey polls - Journalists' resource


Exit polls and their uses

The name “Exit poll” refers to the poll of voters taken just after they exit from the poling stations. The results from exit polls can be very useful to the media and can have several advantages. Exit polls can act as a valuable safeguard against counting fraud. Since the voters cast their actual votes anonymously, we can get additional details from exit polls, like how different groups of people voted, and why they voted.

·       CNN Explainer on exit polls 

·       AAPOR explainer on exit polls

Nonresponse Effects 

For many years, a survey's response rate was viewed as an important indicator of survey quality. Many observers believed that the higher the response rates in the sample, the more accurate the survey result. Non-response is widely recognized as a potential source of error that can reduce the accuracy of all types of polls. Non-response becomes a critical issue when non-response falls to below 70 percent. National Elections Studies and the General Social Survey reveal that in-person interviewing has non-response rates of 25 to 35 percent, whereas in telephone surveys receive a nonresponse rate of at least 10 percentage points higher.

·       AAPOR on response rates overview 

·       Possible consequences of non-response for pre-election surveys - Pew Research Centre 

·       The causes and consequences of response rates in surveys 


·       https://www.adl.org/education/resources/tools-and-strategies/election-polls

·       https://www.washingtonpost.com/news/the-fix/wp/2016/01/07/why_the likely voter is the holy grail of- olling/

·       https://thehill.com/opinion/campaign/516655-mellman-the-likely-voter-sham

·  https://www.mrs.org.uk/pdf/CIPR%20MRS%20RSS%20Guidelines%20for%20using%20statistics%20in%20communications%20CIPR.pdf%20

·       https://courses.lumenlearning.com/math4libarts/chapter/sampling-methods/

·       https://www.scribbr.com/methodology/sampling-methods/

·       https://journalistsresource.org/tip-sheets/reporting/public-opinion-polls-tips-journalists/

·       https://news.harvard.edu/gazette/story/2020/11/harvard-experts-weigh-the-good-and-bad-of-political-predictions/

·       https://www.surveypractice.org/article/11736-sample-size-and-uncertainty-when-predicting-with-polls-the-shortcomings-of-confidence-intervals

·       https://www.pewresearch.org/methods/u-s-survey-research/collecting-survey-data/

·       https://www.bostonglobe.com/2020/11/13/opinion/problem-with-polling/

·       https://www.bbc.com/news/election-2019-50716626

·       https://www.pewresearch.org/fact-tank/2016/11/02/just-how-does-the-general-election-exit-poll-work-anyway/

·       https://pprg.stanford.edu/wp-content/uploads/2007-TSMII-chapter-proof.pdf

Please refrain from sharing sales pitches posting on unrelated matters and keep comments cordial and to the point. Those participating must abide by the Terms and Conditions.  

*Before creating new topics, you must sign up for MRIA communications here 

This forum is empty.

© Marketing Research and Intelligence Association

Powered by Wild Apricot. Try our all-in-one platform for easy membership management