Named after Italian economist Vilfredo Pareto, the Pareto principle states that 20% of the causes (the “vital few”) are reasonsible for 80% of the consequences. Translated to the context of quality control, this means that most problems (e.g., delays, products that are out of spec, etc.) can be attributed to a small number of underlying causes. Identifying those causes (and fixing them first) can therefore be an important step in improving a process efficiently. To help differentiate the “vital few” from the “trivial many”, it is customary to present the data as a Pareto chart.
Example: Not Landing a Quotation
For our Pareto chart example we reproduce Figure 6.4 from the book “Six Sigma in Practice” (also available in Dutch: “Six Sigma in de Praktijk“) authored by Rijk Schildmeijer and Paul Suijkerbuijk. The data concern the likely causes of not landing a quotation, as assessed by an improvement team. Imported into JASP, the data are as follows:
In order to produce a Pareto chart in JASP, we navigate to Descriptives (on the ribbon), select Descriptive Statistics and then Basic Plots. This is where we find the Pareto chart (or Pareto plot, as it is called in JASP). The Pareto plot GUI looks as follows:
Note that all three boxes have been ticked; the “count variable” Occurrences lists the relevant frequencies. The resulting plot looks as follows:
The Pareto chart is really quite straightforward to interpret. The chart consists of two main elements. (1) a bar chart gives the frequency of occurrence for each cause (first y-axis). Note that the causes are ordered in terms of decending frequency, with the most frequent category positioned on the left; (2) a line gives the cumulative percentage (second y-axis). In the GUI, we selected the option “Pareto rule 80%” and hence the graph includes the dashed yellow line that separates the “vital few” causes (that contribute 80% of the problems) from the rest. JASP allows the axis scale and axis titles to be edited.
A Pareto chart can be useful because it may attend the analyst to what problems are really important to solve. However, its diagnostic value should not be overestimated:
- The causes are usually ordered in terms of frequency, not in terms of importance. Some defects may be more serious than other defects, and to take this into account demands that the defects are weighted in some way.
- Some causes are easy to fix, and others are difficult or even impossible to fix; concentrating on those is foolish. For instance, air traffic delays are often caused by bad weather. Although this cause is definitely part of the “vital few”, there is nothing that an airline company can do about it. On the other hand, a small cause (e.g., suboptimal boarding procedures) may contribute only 0.5% to delays; but when this cause is relatively easy to address it may still be worth fixing it.
- At any point in the improvement process, a decision needs to be made how best to spend limited resources. Ideally this requires a decision analysis that maximizes expected utility. A Pareto chart is but a single step in a more comprehensive assessment.
References
Schildmeijer, R., & Suijkerbuijk, P. (2023). Six Sigma in Practice. The Lean Six Sigma Company.
Schildmeijer, R., & Suijkerbuijk, P. (2023). Six Sigma in de Praktijk. The Lean Six Sigma Company.
Authors
Eric-Jan Wagenmakers (CEO) and the JASP Services BV Team.
