The annual conference of the American Association of Public Opinion Researchers is running this week in Denver, and Edison Research is proud that our team is giving four different presentations on best practices in survey research.
One of these presentations is from Edison VP Randy Brown, who is documenting the issues that come from implementing surveys online that attempt to measure internet behaviors. .
As you can see in Randy’s report, streaming behaviors will be overstated in a survey implemented online because it is difficult to reach lighter internet users from internet sample frames, even if one is using high-quality internet samples. Beyond that, about 10% of Americans still have no online access and are entirely invisible to online research.
The report documents the steps we take to make sure that our research represents the total population – using our Share of Ear® studies as his example.
One prime example of why these steps need to be taken can be seen with regard to listening to the content produced by America’s broadcast radio stations. Share of Ear® determines whether listening to radio content is coming via the over-the-air signal (whether analog or HD) or from the station’s streams.
Our estimate is that 8% of the combined listening to broadcast radio content is from the streams and 92% is from over-the-air.
Nielsen has produced similar estimates. These numbers would be vastly higher (and incorrect) if we did not take the steps we do to correctly represent all listeners and all listening.
This is not, of course, to say that radio streams are unimportant – in fact they are crucial. It is to say, however, that one can be misled by estimates that are not designed to fully represent a population, as Randy discusses in his talk.
It is worth noting that we see quite a difference based on the type of content. For news, sports and personalities, streams comprise 12% of the total listening, whereas for music the streams are 6%. Perhaps this speaks to streaming being more vital for radio’s more unique aspects.