Surveys. The experimental style

Surveys

The aim of a survey is to obtain information which can be analysed and patterns extracted and comparisons made. The census is one example of a survey in which the same questions are asked of the selected population (the population being the group or category of individuals selected). The census aims to cover 100 percent of the population, but most surveys have less ambitious aims. In most cases, a survey will aim to obtain information from a representative selection of the population and from that sample will then be able to present the findings as being representative of the population as a whole. Inevitably, there are problems in the survey method. Great care has to be taken to ensure that the sample population is truly representative. At a very simple level, that means ensuring that if the total population has 1000 men and 50 women, then the same proportion of men to women has to be selected. But that example grossly oversimplifies the method of drawing a representative sample, and if you decide to carry out a survey, you will need to consider what characteristics of the total population need to be represented in your sample to enable you to say with fair confidence that your sample is reasonably representative.

In surveys, all respondents will be asked the same questions in, as far as possible, the same circumstances. Question wording is not as easy as it seems, and careful piloting is necessary to ensure that all questions mean the same to all respondents. Information can be gathered by means of self-completion questionnaires (as in the case of the census) or by means of questionnaires, schedules or checklists administered by an interviewer. Whichever method of information gathering is selected, the aim is to obtain answers to the same questions from a large number of individuals to enable the researcher not only to describe but also to compare, to relate one characteristic to another and to demonstrate that certain features exist in certain categories. Surveys can provide answers to the questions What? Where? When? and How?, but it is not so easy to find out Why? Causal relationships can rarely if ever be proved by survey method. The main emphasis is on fact-finding, and if a survey is well structured and piloted, it can be a relatively cheap and quick way of obtaining information.

The experimental style

It is relatively easy to plan experiments which deal with measurable phenomena. For example, experiments have been set up to measure the effects of using fluoridated toothpaste on dental caries by establishing a control group (who did not use the tooth-paste) and an experimental group (who did). In such experiments, the two groups, matched for age, sex/gender, ratio of boys to girls, social class and so on were given a pre-test dental examination and instructions about which toothpaste to use. After a year, both groups were given the post-test dental examination and conclusions were then drawn about the effectiveness or otherwise of the fluoridated toothpaste. The principle of such experiments is that if two identical groups are selected, one of which (the experimental group) is given special treatment and the other (the control group) is not, then any differences between the two groups at the end of the experimental period may be attributed to the difference in treatment. A causal relationship has been established. It may be fairly straightforward to test the extent of dental caries (though even in this experiment the extent of the caries could be caused by many factors not controlled by the experiment), but it is quite another matter to test changes in behaviour. As Wilson (1979) points out, social causes do not work singly. Any examination of low school attainment or high IQ is the product of multiple causes:

To isolate each cause requires a new experimental group each time and the length and difficulty of the experiment increases rapidly. It is possible to run an experiment in which several treatments are put into practice simultaneously but many groups must be made available rather than just two.. The causes of social phenomena are usually multiple ones and an experiment to study them requires large numbers of people often for lengthy periods. This requirement limits the useful-ness of the experimental method. (Wilson 1979: 22)

So, the experimental style does allow conclusions to be drawn about cause and effect, if the experimental design is sound, but in education and the social sciences generally, large groups are needed if the many variations and ambiguities involved in human behaviour are to be controlled. Such large-scale experiments are expensive to set up and take more time than most students working on 100-hour projects can give. Some tests which require only a few hours (e. g. to test short-term memory or perception) can be very effective, but in claiming a causal relation-ship, great care needs to be taken to ensure that all possible causes have been considered.

⇐ Предыдущая 1 2 3 456 7 Следующая ⇒