A few weeks back I took an all-day seminar with Don Dillman, “How Visual Design and Layout Influence Responses to Questionairres.” It was a great course and I definitely recommend taking the opportunity to do anything similar with Dillman or Odum if the opportunity presents itself.
In addition to some great walk-throughs on the power of design to elicit greater rates of survey response, and the importance of harmonizing design elements across multiple modes in survey designs (i.e., web, mail, phone), Dillman also made a pretty shocking (even to him!) point about what his latest research showed: namely, that mail surveys are (still?again?) the best method:
Postal delivery sequence file (DSF) provides all residential addresses and may now be our best household address frame.
When you give this a minute to think, it’s not all that outlandish. Despite huge increases in Internet connectivity – even among older and rural populations – it remains far from universal, and any given channel online (e-mail, SNS) is only going to present a relatively small and self-selecting share of the population. Further, there’s no centralized database of “online users”, and those with the biggest files (Facebook, MySpace, Google, Yahoo! and Microsoft) sure aren’t giving you access to them, Mr./Ms. Academic Researcher. Landline use continues to decline and cell-phone-only-households, with the protection of the Federal Do Not Call Registry, continue to move into a patchwork non-contactable space.
But nearly everyone still has a street address, and even if it’s not always correlated reliably with a person-name, it’s the best way to reach the biggest and most generalizable share of the population. So in a world where more and more of our interactions and identity are moving online and mobile, to spaces where we increasingly control access, how can researchers hope to build generalizable samples of the population?
Let’s step back for a minute and talk about the U.S. in five years. Just as most of the population now has a cell phone, most of the population will have a smart phone/iPhone-like device that will handle voice communications, e-mail, SNS, microblogging, etc. [A point for future discussion is just what this will do to the differential effects of media channel as observed in the media effects literature] It will be the pivot point for all of our communications and personal identity information – we’ll increasingly be using it as an identity storage and verification device for airport check-in, payment and receipt of payment, and a half-dozen other things that now seem outlandish and will soon seem mundane. It’ll be how we carry who we are, and how we tell others about that, for any manner of transactions and interactions.
But that identity will also be floating, a bit. There’ll be several big databases – the mobile companies, Apple, Google, Facebook, etc. – but, again, they won’t be distributing Yellow Pages. Your identity will be relational and transactional and contingent but always subject to change and shift depending on your satisfaction with service provided. Which is good, but presents an increasing challenge to anyone who wants to find some kind of “everyone” (e.g., Census takers, public opinion researchers, etc.). What’s needed is a tether, that also is contingent and user-controlled, but is based on a stable hook.
The DSF can provide that hook. Most people will continue to have a street address – indeed, even the homeless can provide some manner of address that would interface with the DSF – even as the majority of their communications are mediated through shifting electronic interface. A user-controlled and -verified system of tying your various communication methods – contingently – to a physical address could allow users the ability to better control access to all manner of modes of communication and contact. Physical address and solicitation could become tokens that would then be entered into whatever other interface you wished (Amazon for deliveries, Gallup for polls, IRS.gov for taxes), allowing the third-party only what permissions you desired but also providing the verification layer that you are indeed [a person]. Of course this would raise all sorts of new issues about interface and self-report data, but given Dillman’s very promising results – >50% response rates to online surveys via mail solicitation (and >70% via mail) – this is certainly worth thinking about more extensively.