In April 2016, I started capturing information about public opinion polls that were released to the public. Many but not all of these polling numbers enter the public domain through the media but many are not newsworthy enough to get any media attention. Of course, many earn media attention because of an explicit media-polling firm sponsorship relationship, which may or may not involve a financial remuneration (this is of course, not disclosed).
Some highlights
- Between April and the end of 2016 there were 706 unique releases. This represents 78 per month or about 2.6 per day.
- Focusing only on national surveys, reveals 443 releases between April and the end of 2016. Almost 50 per month.
- Online (panel, including representative panels) is the most used data collection method for individual releases representing 57% of all releases with IVR (24%) and telephone (16%) next most used.
While the public was not equally likely to read or see information about all of these releases, because of geography and because many never were covered in the media but these polls are the concrete record of the business of reading our minds!
Check out the sway below to see some of the numbers.
What is included or not included in the dataset?
To make this manageable the data includes only surveys based on a consumer sample. Any survey of businesses or employees are excluded as well as most multi-country surveys. Consumer surveys with special populations are included as long as they were not based, to the extent this is possible to identify, on membership lists. For example, a survey of parents would qualify but a survey of members of a travel site were not included. The reason is to focus this on surveys that are implicitly about generalizations to the population not to the universe of a particular membership.