Data Collection Methods: Techniques, Tools, and Challenges

Last Updated on 16 September 2023

A key task of the measure phase of DMAIC is to collect data, and so your data collection method can have a huge effect on the project. If you collect data efficiently and accurately, it can help you have the maximum possible effect with your process improvement, and therefore have the biggest impact on your organization.

 Types of Data Collection Methods

In Six Sigma methodology, data collection plays a crucial role and different types of methods can be used to gather reliable and accurate data.

• Surveys: One of the most common data collection methods, surveys consist of a set of questions used to gain insights from stakeholders. The information is usually simple to analyze and is cost-effective to collect.
• Interviews: This qualitative method offers a deeper understanding of the issue at hand. Interviews provide details about a person’s behaviors, experiences, and opinions which may not be identified through other approaches.
• Observational Studies: A method wherein data is collected by directly observing processes. This allows for a real-time understanding of the workflow and identifies issues or bottlenecks in the process.
• Document Review: Reviewing existing documents or data related to the process can provide valuable historical or contextual information.
• Focus Groups: Similar to interviews but conducted with a group, focus groups can yield a wealth of ideas or solutions to an issue discussed collectively.
• Experimental Studies: These involve manipulating variables in a controlled environment to understand cause-effect relations.

Successful Six Sigma projects bank on a robust data collection strategy. Using a combination of these methods can help maintain a balanced approach to data collection, ensuring excellent project outcomes.

Data coding

It can take a long time to record numbers e.g. -995,-997,-998,-990

You can make it easier to record if you ‘code’ them by moving them to a simpler number range, e.g. in this case adding 1000 to each number. This then becomes 5,3,2,10 which is much easier to record, and should be quicker with fewer mistakes. Your standard deviation will remain the same, and the mean will just move by your adjustment, so to get the ‘real’ mean you just undo the adjustment. If your coding includes multiplying you just need to do this to the standard deviation as well.