Editor's note: Michael Mitrano is a principal at Transition Strategies Corporation, a management consulting and mergers and acquisitions advisory firm serving the research industry.

Back in the old days, when researchers specified questionnaire length in card and people wanted column binary files, survey research data processing equaled tabulation. This work was first outsourced. As technology grew less costly, companies began to bring it in-house to be done by a group of people often called spec writers. These people were typically neither researchers nor programmers by education or experience. Instead, they were specialists trained in the often-arcane software packages and analytic conventions of market research. They had almost no interaction with people involved data collection, taking their instructions instead from project managers and analysts.

Along came CATI

When computer-assisted interviewing first came on the scene, some companies brought in or developed technical staff to focus on it fully. Many others gave it to their DP department (or someone in the department) and said "This is the new thing - we need to make it work." Sometimes that was successful (after a while) but in other cases it ended in failure. The software sat there - untried or felt to be unworkable. Why? Every company situation is different, but I have seen some common themes.

While both CATI (or CAPI or Web data collection programming) and tabulation involve using computers to handle survey research data, they are very different in many respects. CATI involves capturing the sometimes-messy interaction of interviewer and respondent in a structured way, while tabulation starts with an already-structured file. Tab spec writers work largely by themselves with data files and tab plans, while CATI programmers interact constantly with phone center managers, sampling people and project managers. Tab work is done after the fact, off-line, and can be done over and over until it is right. Computer-assisted data collection of any type occurs in real time, and a programming error creates bad or missing data that often cannot be recovered. CATI programming involves working late and being called at home at all hours. Very few tab spec writers get calls at midnight.

Because of these differences in work environment and required skills, many companies set up separate staffs to handle the questionnaire programming and the back-end tabulation. That was particularly common when companies used one software system for interviewing and a different one (by a different developer) for tabs. This looked like efficient division of labor. I thought so, too, at the time, but later I realized that it was bad for our business and our clients.

The case for integration

These days, I think that almost all companies that do both computer-assisted interviewing and tabulation in-house should integrate those activities, so that one software system is used and, whenever possible, one individual handles both phases of each project.

Three developments in our industry make this increasingly important:


  • Schedules are tighter. Years ago, projects could move step-by-step through the research machine, with each step being largely completed before the next starts. That's out of the question today - work must move ahead on many aspects of a study at once. When one person is handling both interviewing and tabs, he or she can begin the tab work even before the questionnaire is final because he or she knows which parts of the questionnaire are likely to change and which aren't. Those changes that most be made will be made with tabulation in mind.
  • Questionnaires are more complicated. The power of computer-assisted methods combined with the complex modeling and methods (e.g., online conjoint) that some researchers now use have led to very complex questionnaires and large data files. As questionnaires are changed during the field period and from wave to wave, they become even more complicated. Communicating that complexity from a questionnaire programmer to a tab spec writer creates many opportunities for error. If one person handles both parts, that opportunity for miscommunication is removed.
  • Budget pressures are greater. Reducing the number of people involved in a project reduces its cost. By having fewer people to brief, get changes to, monitor, and hand-hold, project management costs are reduced.

Implications of integration

These benefits don't come without a few costs. First, the people doing the work need to have both the attention to detail of a good tab spec writer and the people skills of a CATI programmer. They need a more complete understanding of the whole research process. This may mean people with more education and higher pay.

Having one person handle most studies "from soup to nuts" places a premium on having one software system to handle both data collection and tabulation. This greatly reduces the amount of programming conventions that the person has to learn and (at least in theory) facilitates the movement of data and data definitions from interviewing to the tables. This trend works to the benefit of companies that have a full suite of products (SPSS, CfMC, and Pulse Train, to name a few) and against the firms that offer only data collection or tabulation software. Many of the data collection only products are new, Internet-based tools developed by people who are not familiar with the production DP needs of market research. Many of the tab-only products have been around since the minicomputer days and some are extremely outdated.

If you try to integrate separate computer-assisted interviewing and tab functions, it will not be easy. You will need a lot of cross-training and an extended transition period. Some staff members will strongly resist the change, for fear that they won't be able to master the "other" part of the work. If you persevere, in the end, most will master it - and they will gain the increased satisfaction that comes from seeing a job through from beginning to end and being a closer part of the project team. At the same time, not everyone has to be the best at everything. Some studies (e.g., mail surveys) won't have that real-time interviewing component and other ones (e.g., surveys for some government or academic users) don't require the research company to deliver tables.

As client needs change and the power of our technology grows, the trend will move away from a departmental, sequential focus and toward more simultaneous, coordinated work by a smaller group of well-trained professionals.