Understand Variation in Data

It's hard to overstate the importance of data in the improvement process. Without data, there is no objective way to measure the success of your efforts. Data can confirm that you are on the right track, or reveal that you need to try something else. Data is the lifeblood of improvement efforts.
 
But learning from data is not always as straightforward as it might seem. It's important to understand how to take meaning from data once it's been compiled. This involves understanding variation in data.
 
There is always variation in data, whether the data measures something as simple as the daily temperature or as complex as the success of a surgical procedure. Variation, according to Walter Shewhart, known variously as the Father of Statistical Quality Control and the Grandfather of Total Quality Management, can be viewed in two ways: either as an indication that something has changed (a trend), or as random variation that does not mean a change has occurred. Understanding the nature of the variation is paramount in decision-making about improvement efforts.
 
Because variation is normal and constant, data must be plotted over time to be useful, say the authors of The Improvement Guide: A Practical Approach to Enhancing Organizational Performance. It is only by plotting data over enough time — both before and after a planned changed is implemented — that you can judge whether the variation is random or forms a pattern that indicates that a meaningful change has occurred.
 
While a data line that trends sharply upward or downward is easy to read as an indication of success, it is often the case that the data are more subtle and require more careful interpretation. With variation appearing on either side of the change, the trick is to compare the variation by looking at averages before and after. If the data after the change continue to fall within the same range as the data before the change, then more fundamental changes are needed to bring about true improvement.
 
The chart below is an example of how data plotted over time can reveal when change occurred despite ongoing variation. While the number of patient beds being used in this hospital varied throughout the five-year period, it is clear from looking at the chart that the average number of beds in use dropped at about the time a group of local physicians opened an emergency clinic. This helps managers understand not only what happened, but why.
 
Graph_Theimprovementguide_Fiveyearsofmonthlydataonpatientbeds.gif

Source: Langley G, Nolan K, Nolan T, Norman C, Provost L. The Improvement Guide.
Average Content Rating
(0 user)
Please login to rate or comment on this content.
User Comments