The items of information gathered as the basis of marketing research are usually referred to as data and we speak of primary data and secondary data. The former is information collected by means of a research programme carried out for a specific purpose. The latter is information that already exists, because it was collected as part of a previous research operation or for some different purpose.
Secondary data can be found inside a company, in sales records in particular. When used for marketing research, such data probably need reorganizing. For example, sales of a particular product will often be listed customer by customer, whereas the research might call for a geographical breakdown. Alternatively, many external sources of secondary data are available, in government departments, trade associations, professional bodies, the press, specialist research agencies and many other sources. The increasing availability, power and flexibility of computers makes it increasingly easy for this information to be made immediately available to decision makers. An increasing range of databases can now be accessed on-line.
Some research agencies operate syndicated research programmes in special fields. These are research programmes set up on a co-operative basis and paid for by contributions from each of the companies taking part. Usually it is possible to ‘buy into’ such a programme and thus gain access to data already collected. Alternatively, agencies sometimes mount a programme of research and offer the results for sale to anyone interested. Trade associations often make certain information freely available to their members but sell it to ‘outsiders’.Increasingly all this kind of information is built into a total ‘Marketing Information System’ which is constantly up-dated.Identifying relevant sources of secondary information, extracting the relevant data and analysing it is usually referred to as desk research.
The Collection of Primary Data
If the information required for a particular marketing research project does not already exist as secondary data, we have to determine the best way of collecting it. There are three fundamental approaches to the collection of primary data – observation, experiment, and survey. It is the third approach that most people normally associate with market research. The first two, however, also have an important role to play in certain circumstances.
It is sometimes better to watch what people do rather than to ask them what they do. This has the advantage that it eliminates any problem of interviewer-bias and avoids the difficulty that people do not always remember their actions – especially trivial ones – very clearly. For example, a hidden camera may be the best way of establishing how customers move through a shop, and a tape recorder the best method of establishing the sales approach used by salesmen. Similarly a physical count is normally used to establish the volume of traffic on key roads and the volume of different brands sold by important retail outlets (increasingly via electronic point of sale – EPOS – installations).
Simulation of a real situation may often be a better way of assessing likely future behaviour than asking people hypothetical questions. It is notoriously difficult to get reliable answers about possible future behaviour; but, if for example, we want to know which of two possible packages housewives would prefer, we can put them side by side in a real or dummy shop, give a group of housewives a shopping list and money to spend and see which pack they choose. Similarly, a way of assessing children’s preference for one toy as against another is to give a group of children a ‘selection of toys to play with and see what happens (the way in which they play can also yield valuable insights). Test marketing is of course an example of experiments as a means of obtaining marketing research data.
Basic Types of Survey
If it is necessary to obtain primary data by survey, three methods are available. They are (a) personal interviews, (b) telephone interviews, and (c) postal questionnaires. In general, the cost decreases as we go down this list, but so does the reliability and the extent of the information that can be obtained.Personal interviewing is the most versatile and can fairly readily be carried out on the basis of a properly selected sample. A large number of detailed questions can be asked, and the answers can be supplemented by the interviewer’s personal observations if required. But the cost per interviewer is high, and the degree of planning and supervision required adds further to the cost.
Telephone interviewing enables many people to be reached quickly over a wide geographical area. For this reason it is widely used in industrial marketing research. Its drawbacks are that, generally speaking, only short interviews of an impersonal nature can be carried out. Answers can be keyed straight in for computer analysis, giving further cost reductions.
Postal questionnaires are relatively very cheap. However, the response rate (number of people who return properly completed questionnaires) is usually very low, which introduces its own form of bias.
Other Survey Methods
For particular purposes, variations of the following surveying techniques have proved valuable.
When continuing research is required, the panel method is often used. This differs from the ad hoc enquiry in that the same group (or panel) of informants is used to provide a series of answers over a period of time. This arrangement is particularly valuable when the need is to establish trends. Disadvantages are that it is difficult to maintain over a long period a panel that is truly representative; panel members may gradually become self-conscious and the information they provide no longer a spontaneous expression of their personal views. Panels are used extensively in listener/viewer research and the retail shop audit panel is a well established source of information.
A small (typically around eight) and carefully selected group of people are brought together to discuss a particular topic. The interviewer does not normally pose specific questions but intervenes only to ensure that the discussion stays on subject and that all important aspects are discussed. Because interpretation of results can be difficult, interviewers (or more correctly discussion leaders) are frequently qualified psychologists.
Discussion groups cannot be regarded as properly representative and statistical analysis is normally impossible. However, they have the advantage that (a) they are relatively inexpensive, and (b) the dynamic group situation may bring out information that would not have been foreseen by someone constructing a questionnaire. The technique is particularly valuable in obtaining information rapidly and inexpensively, e.g. as a guide to copywriters and product development groups and as an aid in constructing questionnaires for pilot surveys.
This had a strong vogue some years ago but is now much less popular. It uses methods adapted from clinical psychology in an attempt to establish motives for behaviour and opinions. The methods used include word association, ink-blot tests, and sentence completion tests.
While in theory such methods can give deep insight into human attitudes, in practice much doubt has been cast on the validity of the results. To carry out such tests thoroughly is very expensive of highly trained manpower, since interviews must be individually conducted and each can last several hours.