How PC Survey Software Changed Market Research
In the 1960s and 70s what survey software there was ran on mainframes. This software was cost prohibitive for all but the very largest research firms, so firms used time-sharing, basically renting the use of a program on mainframe. The research firm where I worked in the early 80s paid over $1000 per study for time-sharing. Of course we had to build that price into what we charged for research projects.
Around 1980 minicomputer survey software made its appearance. But that was still cost prohibitive for most research firms, costing about $50,000 for both the hardware and software. And that was just for data processing software, not telephone interviewing.
The first PC survey software appeared in 1983. The Survey System was one of those programs. The first version of The Survey System was user-hostile by current standards, but still easier to use than a mainframe of that era. And it created attractive tables, and not much else available for the PC did so at that time. Most of the other survey software from that era have long since faded away. The Survey System has evolved over the years, becoming easier to use and offering vastly more capabilities, such as telephone, web and in-person interviewing.
The PC survey programs for the first time made it practical for most research firms to own their own data processing abilities. This ability led to better data analysis, because researchers didn’t have to consider the cost of each individual table, as they did with time-sharing. Another factor leading to better research was that the researchers themselves use their PCs to try different forms of analysis and make changes as they went, rather than needing to spell out specifications for data processing personnel. Using a PC was very much more efficient.
Of course there can be too much of a good thing, and now it has perhaps become too easy for people to do their own surveys. With some services offering rudimentary online survey abilities at no cost, maybe people are doing surveys themselves, rather than hiring experienced researchers. This has led to lots of bad research, because creating good surveys is a skill. Just because one can type a question and some answer choices doesn’t mean that they are worded in ways that will lead to useful research conclusions. Surveys with badly worded questions, or even with good questions presented in the wrong order can lead to misleading results. We offer some survey design principles for those interested in learning more on how to make a good survey.