Provides a graphical means for viewing or comparing (for example, before/after) values of one variable versus categorical values of a second variable.
|When to Use||Purpose|
|Start of project||Bar charts are a useful tool for viewing and communicating the effects of different levels of a discrete variable on a process output of interest. For example, you could view average yield of a chemical process versus different urns to see which urns have the lowest average yield. This technique can help pinpoint areas of improvement and communicate the need to process stakeholders.|
|Mid-project||The first rule of data analysis is to graph the data before running any statistical tests. Use bar charts to assist in evaluating inputs and setting aside the unimportant inputs. For example, you can compare the relative frequencies of a problem occurring as a result of different levels of a discrete process input to determine whether changes to the input have any effect on the process output. Further, you could compare the frequencies of the problem at various levels of the first input variable under two settings of a second input variable to determine whether the two inputs act together to affect the output (this action is called an interaction). If no effects are seen, the inputs are often eliminated from consideration because they are likely to have little importance to the project.|
|End of project||Compare defect rates, COPQs, or other similar metrics for different levels of a process input, usually to verify these are constant. For example, if the project improvement implemented new SOPs for processing medical claims, you could compare defect rates for processors in different locations to verify they applied the SOPs consistently at all locations.|
Summarized values of one variable versus categorical values of a second variable. Note, you may also have an optional second categorical variable with two levels.