Hutcheson, G. D. (2011). Measurement Scales. In L. Moutinho and G. D. Hutcheson, The SAGE Dictionary of Quantitative Management Research. Pages 184-187.
There are many different types of information and ways in which this information may be categorised and represented as data. Although a number of different schemes for representing data have been proposed thatutilise a variety of categories and sub-divisions (see, for example, Agresti and Finlay, 1997; Barford, 1985; Harris, 1999; Lindsey, 1995; Sarle, 1995; Verzani, 2005), in this chapter I will distinguish between just a few distinct scales of measurement which will allow a wide range of graphical methods and statistical analyses to be applied. It is worth noting at this point that the process ofrepresenting information using particular scales of measurement is not always obvious, as some information may be legitimately classified in a variety of ways depending on the properties of the information, the coding scheme used to represent this, the number of observations recorded, the type of analysis to be used and the specific research questions being asked. The classification of data intodifferent scales of measurement is not, therefore, an exact science. I will, however, concentrate on practical considerations by showing how a wide range of information may be profitably classified for analytical purposes.
Measurement scales need to be accurately identified in order to...
appropriately code the data select appropriate analytical methods for the coded data(i.e., techniques designed for use on data recorded on a continuous scale are not used with data recorded on an ordered categorical scale) draw appropriate conclusions from the analyses (this requires one to distinguish between the scales of the attribute and the data).
Data can be broadly divided into two main categories; numeric data and categorical data, with each of these categories furtherdivided into two; numeric data into continuous and count, and categorical data into ordered and unordered (this is by no means a full list of data types, but is one that differentiates data on the basis of a number of distinct statistical techniques that may be applied).
Some information can be represented directly using numbers and the resulting data can be described as numeric.I will deal here with two main categories of numeric data – continuous data and count data. Continuous data A data point (a single observation) on a continuous scale can, in theory at least, assume any value between the highest and lowest points on the scale. The only restriction on the number of values, is the accuracy of the measuring instrument. For example, the weight of a person can bemeasured fairly crudely in pounds using a set of bathroom scales, or measured much more accurately in grammes using a professional set of medical scales. A person can, within certain limits at least, be any weight. Other types of information that may be regarded as continuous are people's ages, salaries, IQ scores, percentage marks in examinations or time spent engaged in certain activities. Twodifferent types of continuous data are commonly distinguished, ratio and interval scales, although for analytical purposes such a distinction is not usually important, as few statistics are applicable to only one of these scales. The difference between ratio and interval data is simply that ratio data has a “real” zero point whereas interval data does not. One may, therefore, legitimately talk aboutratios for ratio data but not for interval data (for example, 10 grammes is twice as heavy as 5 grammes, but 10 degrees Celsius is not twice as hot as 5 degrees Celsius).
Count data A data point on a count scale can assume any positive number (integer). Similar to continuous data, the value of the data points are meaningful as a count of 5 is half that of a count of 10, and if you add 2 onto a...
Ler documento completo
Por favor, assinar para o acesso.