Web polls are inadequate. Anyone can take part and skew the sample.
A survey conducted by the Angus Reid Institute through the Angus Reid Forum should not be confused with the millions of “quick polls” (or “instant web polls”) that appear in websites all over the world. These “quick polls” serve to take the pulse of visitors to a specific website, but are neither scientific nor descriptive of any particular population. While the “quick polls” do elicit the opinion of web users, their results—usually accompanied by a disclaimer— should never hold the clout of a representative survey. The polls conducted by the Angus Reid Institute through the Angus Reid Forum are always accompanied by a section that describes the methodology in detail; number of interviews, dates, and margin of error. From time to time, “quick polls” are used as a device to recruit new panel members by the Angus Reid Forum. However, their results are never reported to the public.
An online panel cannot represent the entire population.
One of the misguided observations about online polling alleges that certain segments of the population are underrepresented, making a credible sample unattainable. The Angus Reid Forum has enough panelists from every basic demographic group in Canada to guarantee that samples are representative of the entire adult population. Still, gathering interviews is only one step in the process. Every survey requires a procedure known as weighting. This exercise, an essential component of public opinion research, entails interpreting the responses in accordance with up-to-date social and demographic data. This ensures that the make-up of a constituency (a city, a province or state, or the whole country) is carefully represented within the sample, in areas such as gender, age, and income.
Online surveys use “convenience sampling”
Quality market research panels do not use the techniques used by convenience sampling methods. Survey respondents are not recruited on the basis of proximity to the researchers or the product or some other definition of convenience nor, critically, by survey-specific advertising that would allow panel members to self-select to participate in any given survey. That is not how quality market research panels work.
An online panel does not allow for random selection.
Other critics claim that online polling is biased, because people choose to become panelists in order to respond to surveys themselves, instead of being picked at random through a list of telephone numbers. This argument erroneously assumes that every person will be asked to take part in every online poll at every time. Invitations for particular surveys are sent to a randomly selected portion of the Angus Reid Forum, to ensure a representative sample.