4. Data, units and variables of analysis

Summary

Data is the basic input for data analysis. A data point is the smallest unit of information, the raw material from which knowledge is produced and evidence-based action can be taken with a certain measurable degree of uncertainty.

It is useful to think of a data point as whatever can be placed inside a cell (of a spreadsheet): a number, a letter, a word, a true or false. Whether something counts as a data point depends on the analysis itself, because that data point arises from the intersection of 2 very important concepts: the unit of analysis and the variable.

The unit of analysis is the object or subject on which the analysis is carried out. The variable is the characteristic of interest of that object or subject, on which a data point will be observed or measured. The former is usually identified with each of the rows (of a spreadsheet), the latter with the columns (see table), although this need not always be the case. The opening sequence of any data analysis is therefore to identify the unit of analysis and the variable or variables of study, and to obtain one or more data points from those variables.

variable 1 variable 2 variable 3
unit 1 data point data point data point
unit 2 data point data point data point
unit 3 data point data point data point

Variables can be of 2 types: quantitative or qualitative. The former can in turn be continuous or discrete, and the latter ordinal or nominal. The importance of recognizing which type of variable is at hand lies in anticipating the techniques and limitations that apply in each case. The most flexible and powerful variables are continuous quantitative ones, because they can be discretized, ordered or turned into qualitative ones, whereas the reverse is not possible.

In addition, classifying variables can involve not only their type but also whether or not they result from a random process, that is, whether chance plays a part at some point in obtaining them. This brings a complete change in the way the variable is analyzed, as well as in the conclusions that can be drawn.

To sum up, it is very important for a data analyst to take part in, or be fully aware of, the thorough definition of the unit of analysis, the identification of the variables and the collection and organization of the data, since these 3 points are the foundation on which any analysis is entirely built.

Further reading

  • Los datos. Aguirre, Niño and Simonetti, Estadística aplicada en las ciencias sociales y humanas (in Spanish), pp. 13–15.
  • Las variables. Aguirre, Niño and Simonetti, Estadística aplicada en las ciencias sociales y humanas (in Spanish), pp. 15–17.
  • La primera organización de los datos: la matriz de datos. Aguirre, Niño and Simonetti, Estadística aplicada en las ciencias sociales y humanas (in Spanish), pp. 25–27.
  • Tidy Data. Hadley Wickham, Journal of Statistical Software, sections 1, 2 and 3.
  • Tipos de variables. Lind, Marchal and Wathen, Estadística aplicada a los negocios y la economía (in Spanish), 15th ed., pp. 8–9.
  • Variables aleatorias. Lind, Marchal and Wathen, Estadística aplicada a los negocios y la economía (in Spanish), 15th ed., pp. 189–190.
  • Types of Data. STAT 414: Introduction to Probability Theory, Penn State, lesson 1.4.
  • El concepto de variable aleatoria. Canavos, Probabilidad y estadística. Aplicaciones y métodos (in Spanish), pp. 52–53.