Analysis Of Quantitative Data:
If a survey has been well prepared and piloted, and successfully carried out, then the data analysis should be relatively straightforward. It is essential that the responses are condensed to numerical data by coding, and the figures can then be collated - either manually or, for larger surveys, by computer - and statistical analyses carried out to identify dependent and independent variables, and relationships, and to test their significance. There are three main types of numerical data, and the distinctions between them are important when they are to be analysed statistically.
¨ Nominal data. This term can be explained by considering, for example, the response to a question asking respondents to identify their gender. The possible responses could be coded "1 = female" and "2 = male". The numbers themselves are arbitrary and do not signify anything - they could well have been picked at random. Their use is purely to provide consistency in the treatment of responses to each particular question. This is known as nominal data, and is no more than labelling different categories, with no obvious ordering. Statistical measures such as medians and modes and frequency measures are used to summarise key features of nominal data.
¨ Ordinal data. If a particular question has a range of answers, such as an attitude statement on a Likert scale, the response selected is known as ordinal data. This type of data signifies classifications which have a relationship with other, similar classifications in terms of their order.
¨ Interval data. This is where the interval between any adjacent points is the same as the interval between any other adjacent points, such as the difference between salaries of £10,000 and £20,000 and between £20,000 and £30,000. Statistical measures such as the mean, standard deviation and variance are used to summarise key features of interval data.
It is not our objective here to go into the detail of the different types of statistical analyses, which may be applied. But, make sure you are clear about the most appropriate methods when designing the questionnaire.