Editor's note: Tim Macer is a U.K.-based independent specialist and adviser in the use of technology for survey research.
While persuading respondents to complete a survey even once is the problem most research and fieldwork companies face on a daily basis, a new interviewing solution from Cincinnati-based OpinionOne is so pleasing to use that the problem is more likely to be preventing your sample from lining up to take the survey again.
CAVITM (Computer Assisted Visual Interviewing) gives self-completion interviewing a completely different face, with an appealing visual environment that has the futuristic feel of computers in science fiction movies, especially when CAVI is running in touch-screen mode. But this is neither fiction nor virtual reality - CAVI is a serious and versatile data collection system for kiosk interviewing, CAPI and Web which is attempting to re-invent what is meant by computer-assisted self-completion interviewing.
CAVI breaks the link with paper, perpetuated by so many text-based online interviewing systems, by creating an entirely visual means to gather data. In doing so, it provides a new opportunity to tackle a whole forest of today's thorny issues, such as respondent satisfaction, speed, consistency, modal bias, and difficult sampling issues like the exclusion of minorities whose reading skills are limited. It does this by tackling something almost all the other systems conveniently ignore: how to make the appearance of the interview, the size of the text, the size and shape of the buttons and the exact rendering of the colors and pictures consistent on every computer on which it appears.
This is easier to control when using your own equipment, but a technical minefield on the Internet, where your painstakingly perfected Web interview can be rendered a complete mess because the respondent's computer is configured differently. Trying to fix it is like wallpapering the ceiling with paper that doesn't want to stick.
CAVI has an built-in browser sniffer that diagnoses in a couple of seconds whether a respondent's PC is capable of meeting the study's requirements; these requirements can be varied according to the demands of the project. It also takes over the whole screen, to ensure that no other visual clutter can introduce bias, and renders the interview perfectly and consistently.
The system is designed to work in three principal contexts: touch-screen kiosk interviewing; CAPI, ideally where the bulk of the interview is self-administered; and on the Web. Whether you need a keyboard or a mouse depends entirely on the kind of interview you create - the right interview will even work on handheld computers or PDAs. Most questions are presented as a grid of buttons, which may either contain text - usually one individual answer, or a graphic. Products, logos, pictures or concepts can all be selected simply by clicking or touching the relevant button. A range of feedback encourages the respondent and increases accuracy, from a satisfying Star Wars-style R2D2 chirp whenever a selection is made through to optional full voiceover on both questions and answers. This means that respondents with poor literacy skills or even young children could complete the interview with ease. The system also provides complete multi-language capabilities to the UNICODE standard, including non-Roman writing systems.
 |
 |
In addition to the text or picture buttons, there is a good range of other question types, from free texts to complex constrained sum questions. But the real strength of the product is in these button questions. There is even one type of question which simulates product facings on a supermarket shelf simply from appropriately scaled product shots.
There is a strict rationale to the CAVI screen, one which aims to ensure that the respondent's focus is always directed first to the question, then to the task of making an appropriate selection for the answer, then to how to navigate to the next question. It employs several neat tricks, like immediate elimination of the option just chosen on multiple-choice screens, and a wide range of randomization techniques. It can also display time scope questions, where images are shown for a measured time only. It integrates with audio and video in a variety of ways and offers a choice of conjoint methods too.
My initial reaction, on only seeing the interviewing interface was "Well, anything is possible if you put enough effort into the programming," assuming that CAVI was an entirely custom-built solution for each survey. I was very surprised and reassured to learn that CAVI comes with its own rapid development survey authoring environment called Oasis, which rivals any of the better known survey authoring tools in terms of functionality. Oasis comes in both Web-based and locally installable versions. It neatly combines graphical tools and visual objects with code snippets you write in a cut-down version of Java: a mix of point-and-click with some syntax attached.
The hallmark CAVI screens are built up simply as a series of layers, with templates readily available to simplify the task and provide a dependable starting point. While there are constraints, there is sufficient latitude in the design tool to place objects on screen pretty much however you want. An excellent preview window gives you immediate feedback on every change you are making. Some users direct this preview window to a second monitor for a really clear view.
To design a survey from scratch in Oasis definitely takes a little longer than designing most CATI surveys, as there is more to consider. An experienced programmer can still put together moderately complex surveys in around two days. However, the process definitely calls for different skills and resources: the programmer needs to have an eye for design - and a beefy PC. Java eats up memory, so 512MB is an absolute minimum (though actual interviews require 64MB or less).
Perhaps the most intriguing use of the technology is in kiosk mode in locations with good public access. To kick-start this, OpinionOne has partnered with mall interviewing facility owners to create a network of CAVI stations in over 60 malls across the U.S. - picked to provide balanced, representative coverage of the U.S. population - on which you can deploy your survey. Of course, you can also install your own station, or lease them from OpinionOne on short- or long-term contracts.
Burke Inc. uses CAVI in its information, communications and entertainments (ICE) research division for video testing of TV ads and characters in shows, sophisticated conjoint-based trade-off exercises and price sensitivity testing. "The overall breadth of this technology expands what you can do market research-wise," says Cary Nadel, vice president of Burke and head of ICE research. "It can allow companies to do more multi-modal research. By using this, the one thing you have eliminated is any bias from the visual appearance."
Burke uses it on its own network of kiosk interview stations, which will handle around 300 interviews a day. "If we need greater throughput we can pick up those additional interviews in the malls, using OpinionOne's network," says Nadel. "The alternative is to broadcast to consumers' homes, with a follow-up interview. With this, we can do all the work in the same time frame for 20 to 25 percent less cost."
Burke has integrated the system with its own Digital Dashboard reporting system, a process Nadel found was surprisingly straightforward. "Their data streams feed right into our reporting engine. If the fieldwork ends at 6:15, at 6:16 you can see if the pie slice has grown wider or the bar chart has grown taller."
Nadel has found the method does effectively boost response. "We've seen it does make the interview more enjoyable and less tiresome," he says. "Fatigue can be a huge factor, regardless of the methodology. If you have a more appealing visual environment, it speeds up the individual process and, watching people, I can see they pay attention more closely and read things a bit more carefully. My gut feel is this enhances the quality of the information we get."
The Bank of New York has installed a CAVI kiosk at two of its branches to gather customer feedback. For Rick Miller, the bank's research manager, a major attraction was the system's ability to do more than collect survey data: it also works as a customer information resource and provides specific marketing information and messages to customers visiting the branch. "With this kind of investment you like to do as much as you can," Miller says. "If it is positioned right, customers will punch the buttons out of curiosity. Another attraction was that it was different - very space-age - and not what you would expect find in the conservative world of banking."
Miller reports that the data from CAVI has been consistent with data collected in the past using more traditional methods. "It was a new way of doing research, but the best part is it did not differ from what we had done in the past," he says.
OpinionOne rightly deserves praise for developing something that is innovative yet practical, that enhances the interview experience, and therefore the status of research in the eye of the public. The respondent cooperation issue cannot be solved by technology alone, but it is bound to help if you have a kit that looks like it belongs to the future and not like some relic from the past.