Methods for handling low frequency managerial data

摘要

The typical manager is faced with data ranging from sales and cost figures to absenteeism and accident rates. There are significant benefits to be gained by simply plotting the data and understanding the principles of statistical thinking. Some helpful rules will be covered that provide necessary guidance for managers. Handled correctly, the majority of necessary decisions can be made from simple graphical depiction of business process performance. However, there are times when a more rigorous method for making databased decisions is called for, requiring the introduction of formal tools. Statistical treatment of managerial data has traditionally handled the data in the same fashion as manufacturing data. Unfortunately, most managerial data is infrequently occurring. Monthly and quarterly data points are the norm, with weekly and daily data being luxuries. Furthermore, the premise of much of the control chart treatment of managerial data has presumed that managers are prone to overreact to the noise in their data, and that traditional control charts are useful in subduing this problem. This paper will take the position that low-frequency managerial data, combined with the needs of managers in general, requires a change in thinking about the balance of risks associated with control charting this type of data. Modified control chart rules, as well as a Change-Point Analysis approach will be discussed as useful tools for assisting managers in their decision-making.