| Commercial organisations involved in market research often have entrenched methods of data analysis that have been around for a considerable length of time. It may be more efficient for them in the long term to upgrade to a more advanced method of analysing data, but the investment in time and training may give them reservations. Furthermore, members of staff who are accustomed to their role may resent the change. Here Erica Lucas presents an interview with a person who works in a market research company that has recently started to use SPSS. We will refer to the interviewee as MR. | |
| EL: | Could you describe the way that you worked in the past? |
| MR: | People in my role (execs) design schedules to collect data, which is entered into a database program. The data processing department then produce a printout of all variables cross-tabulated by any break variables of interest. The execs extract the data from this (very large) printout, and the more senior members of the team then look at what they have extracted, try to identify a "story", and perhaps ask them to create some more charts, and discard some of the ones they made previously. |
| EL: | What sort of training was provided when SPSS was introduced? |
| MR: | At a basic level: Everybody in the organisation is given general training about the capacity of the software, and how it can be used to improve their analysis and client deliverables (charts, graphs, etc.). They are given the option of having data in SPSS format for free, in addition to their normal paper/electronic data tables. Some people remember to take it. Most rarely use it (though a handful do). At a more advanced level, a series of follow-up courses are offered, relating to syntax, tables and multiple response variables. |
| EL: | Were people scared or worried about the change in working practice? |
| MR: | People have nothing in particular against using SPSS; they are not scared of it, and have no objection to its introduction. Indeed, it would potentially make the hands-on executive’s job more interesting, allowing them to the means to "think through" an analysis instead of churning out charts and graphs and hoping that something might emerge. Yes, SPSS would allow a more inductive approach to analysis, rather than the somewhat run-of-the-mill deductive method, which is currently the standard. |
| EL: | What about those people in the data processing department? Are they worried that they may be "surplus to requirements"? |
| MR: | One might wonder if they aren’t a bit worried that we will steal their thunder, make their role obsolete, but they have no reason to worry, and they know it. Those people can be confident that as long as they clean the data, name the variables, and know their stats, we are unlikely to supersede them in our capabilities of manipulating data. And as long as we have to concentrate on making things look good for clients in a format that both we and they are used to, we will never have time to learn to do all of those things, much as we are all aware that it would make our lives more interesting and do wonders for our personal development. And as long as we have to cut costs, new and radical working methods will not be at the cutting edge, even if they are capable of being there. |
| EL: | In what ways does SPSS compare badly with the previous practice? |
| MR: | There are one or two practical difficulties, regarding what SPSS can do, and what is expected. Multiple response variables, for example, are something that SPSS does not handle particularly well, but that are used in market research a great deal for looking at various levels of brand awareness and consumer behaviour. In one study, there are likely to be 9 or 10 multiple response sets, each of which imply a huge number of variables in SPSS if the more accessible analysis method is to be used. The tables of figures that SPSS produces contain many superfluous figures. They are interesting to the researcher, but not to the client. Moreover, when the researcher is told he or she can produce deliverables at the touch of a button, they are likely to be disappointed when they cannot. The new Version 8 does not compare to the charting capabilities of PowerPoint. Graphs cannot be manipulated once created – they are not "linked" to the data as such, because the data, in its processed form can only exist in a cosmetic format. Even the colour scheme is fixed once the graph is created. One final point is that it takes time to familiarise oneself with a data set that one has not created, and with variables that one has not named. |
| EL: | Are people entrenched in the old way of doing things? |
| MR: | Inertia does play a part in it. This is not a conscious inertia, but one which results from many other constraints, most notably time, resulting from that, habit. Time pressure and habit make inertia a necessity for the time being. The data processing people are quite aware of the time pressure the execs are under. People are used to paper tables these contain pre-specified cross breaks for all responses. Any further cross breaks can be run later. Any complex modelling is sent to the department in the organisation that deals with complex modelling. This is all pre-arranged with the client at an agreed cost. Using SPSS would require learning how to use it, and over and above any amount of training, the only real way to learn how to use it is in a hands-on situation. However, nobody has time to learn to use SPSS in a hands-on situation, even if they would like to. Furthermore, anybody who has learnt to analyse deductively does not have time to learn to analyse inductively. But what data processing can be most confident about, is that they know all about statistics, whereas most execs have a basic knowledge of what is necessary for speaking to their clients, and how to do significance tests. |