Practicing the Three Phases of Improvement! – The Assessment, Phase 1/Step 2: Analyzing Data
How can we practice improvement?
As explained in our first post “Practicing the Three Phases of Improvement!“, a Process Improvement Exercise has three important and necessary phases, 1) The Assessment Phase, 2) Problem Solving Phase, and 3) Implementation Phase. Furthermore, as explained in our last post “Practicing the Three Phases of Improvement! – The Assessment, Phase 1/Step 1: Collecting Data,” the Assessment Phase has two steps, Collecting Data and Analyzing Data. We devote this article to the “Analyzing Data” part of the Assessment Phase. The goal of this step is to analyze the appropriate and useful information gathered during the earlier collection part of the Assessment Phase.
Collecting Data – As we learned in the earlier post, collecting the right data is important. We are looking for the appropriate data related to the processes and problems we are tackling. Our collecting efforts should match our data collection plan and answer the Who, What, When, Where, How, and Why questions. During the collection and analysis step of a continuous improvement effort, we will become aware of potential problems to solve. However, during the Assessment Phase our goal is not to solve the problems but rather to plant Big Red Symbolic Flags on the problems and Big Green Symbolic Flags on examples of excellence.
Understand the Meaning of the Data – “Nothing is so simple that it cannot be misunderstood” (Freeman Teague, Jr.). Although collecting and analyzing data may overlap, the analyzing step of process improvement is all about understanding the meaning of the data. Be careful, often we can contaminate good information by adding details fabricated by our own mind’s eye (our imaginations). Just understanding the meaning of data can be difficult because of our mental models, personal biases, wishful thinking, and bad habits. To optimize and improve, collect the required truthful and appropriate information then analysis it in unbiased ways.
Types of Information – Information may be either qualitative or quantitative. Quantitative data is countable, relates to quantities, and contains quantifiable numbers. Qualitative data is more text or narrative based and may be more conversational in nature, feeling based, and harder to quantify. However, we can quantify qualitative data (known as quantifying intangibles) and with today’s understanding of data science we can paint elaborate pictures using large batches of data.
When practicing process improvement, collect and analyze both qualitative and quantitative information related to processes and systems to improve. Remember the goal is to find problems (planting Red Flags) and examples of excellent (planting Green Flags). Nevertheless, in each case, the analysis requires a logical approach.
Analyzing Data – Data Analysis…”is a process of inspecting, cleaning,” transforming, … (using statistics,)…and modeling data with the goal of highlighting useful information, suggesting conclusions, and supporting decision-making.”[i] Analyzing both quantitative and qualitative data requires two hats, one is more of a statistical approach while the other is more of a content approach. Therefore, “with all data, analysis and interpretation are required to bring order and understanding.”[ii] (Taylor-Powell and Renner)
Aim each improvement project at the target of discovering what good looks like. Every analysis should have a “why?” Analysis is agile or dynamic not static. Although we may focus our analysis on the questions we want answered, we should expect the questions to change as you learn more.
Analyzing Data Sub-step 1 – Inspecting and Understanding the Data – Weather qualitative or quantitative, all data is not created equal and quality data is important when suggesting logical improvements. However, even high quality data should focus on the goals of the improvement effort (the why). Therefore, it is important to understand the goals and the quality of the information you are collecting as well as the environment in which you are working. We will be better able to judge the value of data as we learn more about:
- the way the information was collected,
- the population and sample size of the data,
- the questions asked,
- the accuracy and truthfulness of the data,
- cycles and timing of the data,
- the start/stop times of the data collections,
- the way the processes flow,
- the way the culture operates,
- the life cycle of the organization,
- who the stakeholders are, and
- other important background issues.
Analyzing Data Sub-step 2 – Categorize Information – The best way to understand a chaotic situation is to organize information into categories. Throughout the process, identify themes and patterns, and then organize the data into coherent categories. Sometimes this is referred to as coding or indexing the data. Often sub-categorizations are helpful. Many times the organization of data starts during the collection part of the analysis phase and continues throughout the problem-solving phase. Commonly categories and patterns emerge as you read the information you have collected.
Analyzing Data Sub-step 3 – Cleaning Data – Ensure the data is correct and unbiased. Throughout the collection and analysis process, remove any misleading, duplicated, or poor quality data. Remove data polluted with inappropriate or error prone information. Remove or repair misfiling, spelling, or calculation errors.
Look for inconsistent data. Entering the same good data in different ways often wrecks our ability to sort data or analyze information. For example, one may express an idea in different ways using different groupings, definitions, formulas, or units of measures. This could happen by accident or by manipulation. People may also enter the same expression of an idea in several ways. For example, one might enter (1) New York City, (2) New York, New York, (3) NY, NY, or (4) New York, NY. Therefore, inspecting, cleaning, and repairing data becomes an important step in analyzing. Remove or repair errant information.
Analyzing Data Sub-step 4 – Modeling and Visualizing Data – Modeling data in our case is simply analyzing data using simple statistical tools, numerical scales (i.e., Likert 1 to 5 ratings), pictures, graphics, and possibly simulation. Nonnumeric or qualitative data is often harder to statistically measure while quantitative data is easily analyzed using statistical methods. In fact, every time we collect numbers, we have statistics. Nevertheless, it is what we do with it that matters.
Although, understanding statistics can be complicated, just knowing the basics takes you a long way in understanding the meanings of observations and metrics. For example, most of the time, just knowing the mean, median, range, minimum value, maximum value, standard deviation, and curves will tell us most of the story. In addition, the more you apply statistical tools the more useful it becomes. In school, I must have taken ten or so classes related to statistics; however, it wasn’t until I stated using and teaching stats that it all made sense.
Statistics – In the words of the Scottish poet, Andre Lang, “An unsophisticated forecaster uses statistics as a drunken man uses lamp-posts – for support rather than for illumination.” The goal of analyzing information is to illuminate the current situation. Remember, the goal of statistics in the analyzing step of any improvement project is merely to find red flags (places to improve) and green flags (examples to share). The statistical information does not solve the problem but once you plant the flags, you know better where the problem solving should begin.
Visualization – With either qualitative or quantitative data, pictures are worth a thousand words (or numbers). Once we collect, categorize, clean, and model the data – visualization is a way of thinking about and communicating data using graphical means.
Every mathematical formula creates a picture and every well-defined and well-constructed picture adds understanding. Most pictures are simple, like pie and bar charts or the statistical curve above. Likewise, even the newer more sophisticated Data Science and Business Intelligence (BI) tools create usable data pictures that simplify data as those below. Be creative and draw pictures to help you analyze data, see patterns, and communicate your points. See the Business Intelligence Picture to the right (http://images.fastcompany.com/upload/Wheel.jpeg) as one example using data as a visualization tool and to the left (http://files.visualization.geblogs.com/visualization/files/2011/02/GE_CDSS_State.png) as another example.
Identify Patterns and Interrelationships Between Systems and Processes: As we collect, categorize, and clean data we look for patterns and interrelationships between and within the groupings. Modeling the data will often illuminate hidden issues. Statistics and visualization techniques may also bring to light patterns, interrelationships between groupings of data, and system loops. Systems loops within an organization describe the interlinked processes that are required for organizational success. A department may have the perfect process in place but it may never be effective until other departments feed useful materials and/or actuate information.
Analyzing Data Sub-step 5 – Interpretation – Interpretation comes from building on our knowledge to find the meaning in the data. As we gain experience with data and interpretation, we become better at seeing trends and opportunities. However, in the words of Albert Einstein, “The only source of knowledge is experience.” Therefore, the best place to find the real-world meaning in the data that we have painstakingly collected is from the experience learned by those doing the work. From another old saying, ”Learn from the mistakes of others. You don’t have time to make them all yourself.” Therefore, check everything, asks others for meaning, and learn of opportunities for improvement from not only your own personal experiences and observations but also from others (front-line workers, doctors/nurses, suppliers, vendors, patients/customers and others doing real work).Patterns may also show the relative importance of data by the number of times a theme appears. Likewise, Interdependence and independence may help us understand cause and effect relationships. For example, productivity data may relate to the distance that nursing staff travels. Just drawing the trips on a faculty layout (as in the Spaghetti Diagram to the right) may help to show problems to solve. The more spaghetti like the tangled paths taken are, the more opportunity we have for improvement.
How can we practice improvement?
To improve, good usable data has to be collected and then analyzed. How can you participate? Keep a look out for ways to improve and be open-minded. As you have ideas, do not jump to conclusions; instead, think in terms of information to collect that will help validate your concerns. Then analyze the information using logical analysis tools. Share your observations with those in your organization and those on the your Process Improvement teams. Participate in data collection and the follow up analysis, also, as we will learn later, follow up with good problem solving steps. Finally volunteer to help implement well-thought-out solutions.
Westbrook Stevens, LLC has teamed with the RGP Nashville Office, Lipscomb University’s College of Computing and Technology, and VMEdu’s 6Sigmastudy to offer ongoing Lean Six Sigma Training in 2016. Come join us and earn your Lean Six Sigma and Six Sigma Belts.
We meet half days on Fridays from 2:00 PM CST to 6:00 PM CST on the Lipscomb University campus in the College of Computing and Technology. During these interactive workshops, you can earn a bunch of PMI PDUs as well as Lean Six Sigma and Six Sigma belts.
To learn more click on this link! http://www.e-wbs.com/healthcare-lean-six-sigma-training/.
[ii] Ellen Taylor-Powell and Marcus Renner, Analyzing Qualitative Data, 2003, as published by the University of Wisconsin- http://learningstore.uwex.edu/assets/pdfs/g3658-12.pdf (Downloaded 2012-08-15)