Telephone surveys – inching to the to the deadpool?

The image “” cannot be displayed, because it contains errors.The National Center for Health Statistics put out some numbers on surveying — The general conclusion is that using “landline” only methods for data collection will skew the analysis. Here is a direct link to the report.

The key findings are :

  • 1 in 6 homes (~15%) are “Cell Phone Only” — that is they simply DO NOT have a landline.
  • 1 in 8 homes (~13%) receive “almost all” calls on their cell phone even though they have a landline

If you just look at those two numbers, ~30% of households essentially rely on cell phones as their primary voice communication device. And this is the last 6 months of 2007. I can only imagine, the first six months of 2008 this has only gone up, especially with the launch but both Sprint and At&T for “unlimited” minutes. I can also say from personal anecdotal evidence, I finally convinced my wife what we really don’t need a land line — so my own residence does not have a landline anymore. We were getting too many telemarketers anyways.

The other paradigm shift that is happening is that businesses (brick n mortar) are accepting emails as the preferred communication model — Eastside Bavarian where I change oil emails me and I can email them to request an appointment — they even send me a survey after every service! I hardly ever talk to my mortgage broker or banker anymore — we simply rely on email to communicate with each other.

The third component that is facilitating this shift is the “Smart Phone” market is expanding rapidly – Blackberry, iPHones etc.  — All these phones are becoming communication assistants rather than being a phone. I use my iPhone for email, text and phone (in that order.)

Now, what does all this mean for research? — The big issue is that of sample bias. Traditionally online research was considered as an after-thought to phone based research. Now with potentially a 30% sample bias error, phone surveys will become increasingly difficult to conduct and more importantly pass the sample bias validation test. That is the general conclusion of the report also.

Here is another interesting blog post from one of my competitors – Check it out. It addresses the same issue.