3.5 Problems with Common Estimation Techniques

Considering the widespread poor results from software estimation, it shouldn't be a surprise that the techniques used to produce the estimates are not effective. These techniques should be carefully examined and thrown out!

Albert Lederer and Jayesh Prasad found that the most commonly used estimation technique was comparing a new project with a similar past project, based solely on personal memory. This technique was not found to be correlated with accurate estimates. The common techniques of "intuition" and "guessing" were found to be correlated with cost and schedule overruns (Lederer and Prasad 1992). Numerous other researchers have found that guessing, intuition, unstructured expert judgment, use of informal analogies, and similar techniques are the dominant strategies used for about 60 to 85% of all estimates (Hihn and Habib-Agahi 1991, Heemstra and Kusters 1991, Paynter 1996, Jørgensen 2002, Kitchenham et al. 2002).

Chapter 5, "Estimate Influences," presents a more detailed examination of sources of estimation error, and the rest of this book provides alternatives to these common techniques.

Additional Resources

Goldratt, Eliyahu M. Critical Chain. Great Barrington, MA: The North River Press, 1997. Goldratt describes an approach to dealing with Student Syndrome as well as an approach to buffer management that addresses Parkinson's Law.

Putnam, Lawrence H. and Ware Myers. Five Core Metrics. New York, NY: Dorset House, 2003. Chapter 4 contains an extended discussion of the importance of predictability compared to other project objectives.