Judgement sampling

Judgement sampling is a move further away from the unbiased attempt to provide a representative sample of the population to be surveyed. In theory, it is only the random sampling methods that do this. Quota sampling is an attempt to replicate what random sampling achieves at a lower cost, and it has some success in doing this. An element of judgement comes into the method in deciding which quota controls should be used in selection of the sample. Nevertheless, the basic aim of the method is to attain the representativeness of random samples.

Judgement sampling, like quota sampling, is a form of purposive sampling but differs from both quota sampling and random sampling in that the attempt to achieve statistical representativeness of the population as a whole is largely abandoned. The aim is still to produce data representative of the population to be sampled, but judgement is used in the sample selection procedure to make the data more useful to the decision maker.

Judgement sampling is commonly applied in industrial and trade research when a few large manufacturers or retailers may dominate the market. In this case the sample might include all the major manufacturers or retailers in a trade and then a sample of other organizations. This will ensure the inclusion of those major organizations whose activities are of such significance in the marketplace that any survey not including all of them could not hope to give a valid picture of what is happening in that market.

A manufacturer of china included all the major department store buyers in a sample together with a random sample of specialist china outlets. The purpose here, as is usually the case in trade and industrial research, was to give additional weight to the views of more important members of the trade in terms of their size or share of turnover.

A similar approach may be taken by an industrial supplier surveying its own customers. Once again, the supplier may wish to seek the views of all the important customers and sample the rest.

Judgement sampling may be the only practical approach for sampling populations if no sampling frame can be constructed for random sampling and insufficient data is available about the population for quota sampling. This is often the case in the preliminary stages of a survey and judgement sampling is the main technique used in small-scale exploratory research surveys. A programme of 'key interviews' might be carried out to generate some ideas about what the views of members of the population might be with respect to the topic under discussion. This is the sampling basis for the interviews described in Section 6.2.1.

The manufacturer of a new type of hard-wearing floor covering conducted a depth interview survey among architects working for local authorities. After interviews with only 10 architects about the new material, it seemed that the concept of the material was generally acceptable. The case quoted is an actual example in which the company went ahead and launched the product on the basis that most of the architects interviewed had expressed an interest in it. Unfortunately, that interest was not directly translatable into product sales in the marketplace. The company is now in the process of improving its sampling procedures for a second attempt to survey the market, since its own belief in the product remains unshaken. This time it will not be using a qualitative approach to make quantitative decisions. This highlights the danger of judgement sampling as a form of research. Such research is so small scale and unrepresentative that false readings can be obtained._

Judgement sampling is used as the basis for surveys using the method of group discussions. As discussed in Chapter 6, group discussions are a very widely used form of research. They give the decision maker some insight into and understanding of the attitudes, opinions and feelings of members of the market about the organization's own, or competing, products. In a survey of housewives about consumer goods it is not unusual for a qualitative survey using group discussion to include only six to eight groups. Eight groups containing eight people is only 64 individuals, and so inevitably the findings cannot be 'representative' in the same sense implied by the use of random or quota sampling methods. However, the attempt is made to generate as wide a range of views as possible by using judgement to decide the type of individuals who will be used for each group.

In order to give a breadth of opinion a cake manufacturer commissioning a survey including eight groups spread them as follows. Four were held in northern towns and four in southern towns. Each group of four interviews was split into two of middle-class respondents and two of working-class respondents. Each pair of groups included one with respondents aged under 30 and one with respondents over 30. The attempt here was to collect the widest possible range of views from a national, class and age spread of respondents._

Typically, judgement sampling is used in the conduct of smaller surveys. When these are particularly small and qualitative in nature, it is most important not to confuse the attempt to represent a wide range of views with the concept of 'representativeness' implied by random and quota sampling methods. If judgement sampling is used in small-scale surveys the results should never be used in a quantitative way. If half the members of a set of group discussions agree on a particular point, it should not be assumed that 50 per cent of the target population will also agree with that point.

Judgement is used in sampling as a way of overcoming practical difficulties or limitations in using other sampling methods. It is also used for 'toe in the water' exercises and, just like using a toe to measure the temperature of water, it is equally unreliable if used in place of a thermometer. It produces a 'feel' rather than a measurement. A toe can give some idea of general hotness or coldness but if placed in the edge of the sea or a cold eddy of a river will not give a reliable measure for the whole body of water. In the same way, the results of small-scale judgement sample surveys cannot be used as representative of the population as a whole, although they do give some guidance about what the population might be thinking, feeling or doing.

Team LiB

Continue reading here: Postal or selfcompletion questionnaires

Was this article helpful?