See Manuals methodology

How to operationalize and measure variables?

Having already specified the concepts you are interested and produced a list of dimensions for each concept, the next stage is to operationalize and measure your variables. The dimensions of a concept form the basis for a list of indicators of the concept: for instance, the indicators of prejudice might include ‘negative attitudes or behaviour towards women’; ‘negative attitudes or behaviour towards ethnic minorities’ etc. Note that while the concept itself and the dimensions of the concept are abstract, the indicators refer to behaviour that can be observed.

You must also decide what level of measurement to use for each indicator. It is important that the variable is exhaustive, meaning that every observation can be placed into one of its categories.

As you may end up with several variables related to different dimensions of a single concept, it may be necessary to combine these variables in some way so as to arrive at one overall measure. An ‘index’ is a measure which combines several different pieces of information. For example, an IQ score is an index based on a person’s response to a large number of questions. Similarly, an index of ‘prejudice’ can be made based on a person’s responses to several questions related to this concept. An index is usually an ordinal measure. It is also possible to arrive at a composite measure that is nominal; this is known as a ‘typology’. For instance, you may combine several indicators of political ideology to develop a four-fold typology of political parties.

Given the complexities involved, it is necessary to assess how good the final measurements are. The two most important criteria are reliability and validity. Validity refers to bias in the measurement: is it really measuring what it is supposed to measure? Reliability refers to the certainty that the same measurement would be made if the research was repeated. Further information on measurement reliability can be found here and here.


Basic readings

  • Babbie, Earl (2004). The Practice of Social Research (12th edition). Belmont: Wadsworth/Thomson. Chapters 5 & 6.
  • De Vaus, David (2001). Research Design in Social Research. London: Sage. Chapter 2.
  • Shadish, William R., Thomas D. Cook and Donald T. Cambell (2002). Experimental and Quasi-Experimental Designs for Generalized Causal Inference. Boston: Houghton Mifflin. Chapters 2 and 3.

Additional readings

  • Collier, Laporte and Seawright (2008) Typologies: Forming Concepts and Creating Categorical Variables, in: Box-Steffensmeier, Brady and Collier, Oxford Handbook of Political Methodology, Oxford University Press.