Control Charts allow auditors to distinguish normal variation from signs of actual performance problems. By plotting data points over time and comparing them to calculated upper and lower control limits, auditors can quickly determine whether a process is operating within acceptable boundaries. This structured method helps organizations identify emerging issues, assess their significance, and decide when corrective action is warranted.
This lesson is a preview from Graduate School USA's Analysis Techniques for Auditors Course.
There are several approaches to creating a control chart. The following four-step process is suggested by Peter Mears.
Step 1: Determine What to Measure
The first step in constructing a Control Chart is identical to the first step in creating a run chart:
- Identify one key performance aspect you want to track over time, or against some base other than time
- Choose the time interval of data collection (minute, hour, day, etc.)
- Decide on the period of time over which you will collect data
Possible performance aspects include the following:
- Volume (how much over a specified period)
- Cycle time (how long something takes)
- Errors and defects (how many are incorrect)
- Waste (how much is reworked or rejected)
Example: Forest Service Vehicle Fuel Budget
The USDA’s Forest Service has a fleet of over 3,000 vehicles which are assigned to its employees for use in performing their assigned tasks. The Forest Service replaces its fleet of vehicles every four years. The Forest Service budgets for fuel for its vehicles based on the miles per gallon average given by the manufacturer, coupled with expected total miles. To track the actual experience with its vehicles, the Forest Service calculates the miles per gallon on a weekly basis for a predetermined group of vehicles for 20 weeks per year. The sample consists of 10 vehicles from each of the 5 regional offices.
What it measures:
- Miles per gallon
- Weekly for 20 weeks
Step 2: Collect the Data
Collect data using a data collection instrument. The instrument should provide for recording data (1) on the total quantity of whatever is being measured (e.g., transactions), (2) for each selected time interval. The instrument should also include the date of collection. If the concern is with errors, the instrument should provide for recording both the total quantity and the error quantity. For example, if the objective is to determine error rates you would gather the following information:
- Date (within time interval)
- Number inspected
- Number defective
- Percent defective
Example: Forest Service Vehicle Fuel Budget
The person responsible for each vehicle included in the sample records odometer readings and gallons of gas used during each week. This information is reported to the central office. Using this information, the central office calculates the average miles per gallon for the fleet. The calculation for the 10 vehicles for 10 of the 20 weeks is shown in the table below. We use the 10-week experience to make the problem manageable.
To create a Control Chart, we want to know the actual average miles per gallon by week for the five regions as a whole. To do this, and create plot points for each week, we add the data for each week for all five regions (from the table on the prior page) and divide the sum by five.
Step 3: Plot the Data
After you have taken at least 20 samples and calculated the value for each, create the control chart. You create the plotting scale on the vertical (Y) axis of the graph. The scale should reflect whatever is appropriate for your particular measurement. Then create a plotting scale on the horizontal (X) axis with a point for each sample data.
You then plot the individual values on the graph (see chart on the following page). The next task is to compute the average or expected level or use a predetermined average level. When computing the average, add all the individual samples and divide the result by the number of samples taken.
Example: Forest Service Vehicle Fuel Budget
For the Forest Service example, we would create a control chart with “average miles per gallon” as the scale on the vertical (Y) axis and “weeks” as the scale on the horizontal (X) axis. We would then plot our data on the chart. As our predetermined average, we would use the manufacturer’s miles per gallon, which is given as 19.8.
Step 4: Calculate the Control Limits
Control limits will tell you if your process is in control (variations within tolerance limits require no further investigation). Think of control limits as individual boundary lines. As long as the data points are within the boundary lines, everything is “OK.” However, when data points are outside the boundary, alarms should go off, and you will need to investigate why the boundary has been crossed.
Control limits will normally be determined as an amount of variation from the average. The allowable variation is calculated as an acceptable variation from normal. It should be based on some sound criteria such as industry standard or some other acceptable determinant. For our Forest Service example, a 15% variance from the manufacturer’s miles per gallon standard is considered acceptable. Therefore, any plot point that falls outside of this variance is considered out of control or outside the acceptable boundaries and should be a candidate for investigation.
Example: Forest Service Vehicle Fuel Budget
The Forest Service expects actual average miles per gallon for its fleet to be within 15 percent of the manufacturer’s published miles per gallon, which is 19.8. With this, we compute the upper control Limit as 22.8 (19.8 x 115%), and the lower limit as 16.8 (19.8 x 85%). These are plotted to the control chart as a dashed line. Any plot point that falls outside the control limits is considered out of control and a candidate for investigation as to the cause.
Follow-Up: Decide on Next Steps
Any follow-up tasks depend upon whether or not any points are outside of the control limits.
If all points are within the control limits:
- There is no need for investigation
- Sample periodically to see whether performance stays within the control limits
- If some plot points are approaching the control limits, consider process improvements to forestall going out-of-control
If one or more points are out-of-control:
- Investigate and take steps to eliminate the cause(s)
- Review to ensure changes have had a positive effect (causes for out-of-control position have been eliminated)
- After corrective action, take a new sample to determine if all points are within tolerance
Example: Forest Service Vehicle Fuel Budget
For the Forest Service example, seven plot points are outside the tolerance limits (see page 38). Therefore, use of the vehicles should be investigated to determine what is causing the wide variances in performance and the variances between expected and actual. A first step would be to plot a Control Chart using the same parameters for each region as well as for the average. This will give an indication of whether the problem is system-wide or restricted to one or two regions.
