Evaluation Cookbook

A typical question using a Likert Scale might pose a statement and ask the respondent whether they Strongly Agree - Agree - Undecided - Disagree or Strongly Disagree.

The responses elicited may be coded e.g. 1-2-3-4-5, but this remains just a coding. It makes no sense to add a response of agree (coded as 2) to a response of undecided (coded as 3) to get a ‘mean’ response of 2.5 (what would it mean?). So how can you analyse data from a Likert scale?

The data collected are ordinal: they have an inherent order or sequence, but one cannot assume that the respondent means that the difference between agreeing and strongly agreeing is the same as between agreeing and being undecided.

Descriptive Techniques
 Summarise using a median or a mode (not a mean); the mode is probably the most suitable for easy interpretation.
 Express variability in terms of the range or inter quartile range (not the standard deviation).
 Display the distribution of observations in a dotplot or a barchart (it can’t be a histogram, because the data is not continuous).

Inferential Techniques
Investigate differences between (the medians of) comparable groups using non-parametric methods e.g.:
 for two unrelated samples – Mann Whitney test;
 for paired samples – Wilcoxon signed rank test;
 for three or more samples – Kruskal Wallis test.
Investigate any association between two different sets of observations using a chi-squared test of association.

Nora Mogey
Coordinator, LTDI,
Heriot-Watt University.