Pareto Analysis for Surveys
So how can Pareto analysis be used on surveys? Any time you collect discrete attribute data (also known as categorical data) you can apply Pareto analysis to your survey data. For instance, if you are gathering information from customers regarding types of problems they've had with a product or types of improvements they'd most like to see, you have attribute data that you can subject to Pareto analysis.
You might also be collecting other types of data that are not themselves categorical, such as satisfaction ratings, that can then be broken out in your data set into categories. For instance, if you are gathering data on customer satisfaction with technical support calls, you can break the data out by support representative, by product type, or by customer type.
Regardless of the specific data you gather, the process for conducting Pareto analysis for surveys is the same.
1. Aggregrate your data so that you have a count of the number of occurrences in each category.
For example, tabulate the number of poor satisfaction ratings for each product type or support representative, or tabulate the number of customers who experienced each type of product problem.
2. Create a Pareto chart from the tabulated data.
You can do this using a program like Minitab or with an add-in for Excel like QI Macros. Depending on the program you use you may even be able to skip step 1 and have the software create your Pareto chart from raw data.
3. Examine your chart to determine whether the Pareto principle applies.
A Pareto chart displays data in order from the most frequent category to the least frequent and includes the cumulative percentage. So, for instance, if the data in the first two of nine categories account for 83% of the total data set, the Pareto principle applies. If it takes more than half the categories to reach about 80% of the data, the principle does not apply.
4. Determine the appropriate next step based on the result of your analysis.
If in fact the Pareto principle applies, you would be wise to focus on those first few categories that make up the majority of the results either in selecting projects to charter or in determining how to focus your existing DMAIC project. If the principle clearly does not apply, consider breaking the data out differently. For instance, instead of looking at product complaints by product type, analyze them by customer type or type of problem.
This image shows an example of a Pareto chart (click on it to view it a larger version):