Methods & Tools
QA Resources
Quality Assurance Tools and Methods
Descriptive statistics enable us to understand data through summary values and graphical presentations. Summary values not only include the average, but also the spread, median, mode, range, and standard deviation. It is important to look at summary statistics along with the data set to understand the entire picture, as the same summary statistics may describe very different data sets. Descriptive statistics can be illustrated in an understandable fashion by presenting them graphically using statistical and data presentation tools.
When creating graphic displays, keep in mind the following questions (IHI 1995):
- What am I trying to communicate?
- Who is my audience?
- What might prevent them from understanding this display?
- Does the display tell the entire story?
Several types of statistical/data presentation tools exist, including: (a) charts displaying frequencies (bar, pie, and Pareto charts, (b) charts displaying trends (run and control charts), (c) charts displaying distributions (histograms), and (d) charts displaying associations (scatter diagrams).
Different types of data require different kinds of statistical tools. There are two types of data. Attribute data are countable data or data that can be put into categories: e.g., the number of people willing to pay, the number of complaints, percentage who want blue/percentage who want red/percentage who want yellow. Variable data are measurement data, based on some continuous scale: e.g., length, time, cost.
Return to Resources
Back to top
To Show |
Use |
Data Needed |
Frequency of occurrence:
Simple percentages or comparisons of magnitude
|
Bar chart
Pie chart
Pareto chart
|
Tallies by category (data can be attribute data or variable data divided into categories)
|
Trends over time |
Line graph
Run chart
Control chart
|
Measurements taken in chronological order (attribute or variable data can be used)
|
Distribution: Variation not related to time (distributions)
|
Histograms
|
Forty or more measurements (not necessarily in chronological order, variable data)
|
Association: Looking for a correlation between two things |
Scatter diagram |
Forty or more paired measurements (measures of both things of interest, variable data) |
Return to Resources
Back to top
Bar and pie charts use pictures to compare the sizes, amounts, quantities, or proportions of various items or groupings of items.
Bar and pie charts can be used in defining or choosing problems to work on, analyzing problems, verifying causes, or judging solutions. They make it easier to understand data because they present the data as a picture, highlighting the results. This is particularly helpful in presenting results to team members, managers, and other interested parties. Bar and pie charts present results that compare different groups. They can also be used with variable data that have been grouped. Bar charts work best when showing comparisons among categories, while pie charts are used for showing relative proportions of various items in making up the whole (how the "pie" is divided up).
Teams may choose from three types of bar charts, depending on the type of data they have and what they want to stress:
Simple bar charts sort data into simple categories.
Grouped bar charts divide data into groups within each category and show comparisons between individual groups as well as between categories. (It gives more useful information than a simple total of all the components.)
Stacked bar charts, which, like grouped bar charts, use grouped data within categories. (They make clear both the sum of the parts and each group’s contribution to that total.)
Return to Resources
Back to top
Step 1. Choose the type of bar chart that stresses the results to be focused on. Grouped and stacked bar charts will require at least two classification variables. For a stacked bar chart, tally the data within each category into combined totals before drawing the chart.
Step 2. Draw the vertical axis to represent the values of the variable of comparison (e.g., number, cost, time). Establish the range for the data by subtracting the smallest value from the largest. Determine the scale for the vertical axis at approximately 1.5 times the range and label the axis with the scale and unit of measure.
Step 3. Determine the number of bars needed. The number of bars will equal the number of categories for simple or stacked bar charts. For a grouped bar chart, the number of bars will equal the number of categories multiplied by the number of groups. This number is important for determining the length of the horizontal axis.
Step 4. Draw bars of equal width for each item and label the categories and the groups. Provide a title for the graph that indicates the sample and the time period covered by the data; label each bar.
Return to Resources
Back to top
Step 1. Taking the data to be charted, calculate the percentage contribution for each category. First, total all the values. Next, divide the value of each category by the total. Then, multiply the product by 100 to create a percentage for each value.
Step 2. Draw a circle. Using the percentages, determine what portion of the circle will be represented by each category. This can be done by eye or by calculating the number of degrees and using a compass. By eye, divide the circle into four quadrants, each representing 25 percent.
Step 3. Draw in the segments by estimating how much larger or smaller each category is. Calculating the number of degrees can be done by multiplying the percent by 3.6 (a circle has 360 degrees) and then using a compass to draw the portions.
Step 4. Provide a title for the pie chart that indicates the sample and the time period covered by the data. Label each segment with its percentage or proportion (e.g., 25 percent or one quarter) and with what each segment represents (e.g., people who returned for a follow-up visit; people who did not return).
Caution
Be careful not to use too many notations on the charts. Keep them as simple as possible and include only the information necessary to interpret the chart.
Do not draw conclusions not justified by the data. For example, determining whether a trend exists may require more statistical tests and probably cannot be determined by the chart alone. Differences among groups also may require more statistical testing to determine if they are significant.
Whenever possible, use bar or pie charts to support data interpretation. Do not assume that results or points are so clear and obvious that a chart is not needed for clarity.
A chart must not lie or mislead! To ensure that this does not happen, follow these guidelines:
- Scales must be in regular intervals
- Charts that are to be compared must have the same scale and symbols
- Charts should be easy to read
Back to Resources
/10-4-2001/spacer.gif){kind=spacer}
