Six Days Until the Release of Solving for Project Risk Management!

The release of my book Solving for Project Risk Management: Understanding the Critical Role of Uncertainty in Project Management is now less than a week away! Leading up to the release, I have been writing individual blog posts about each chapter. On Monday, I wrote about Chapter 1 and on Tuesday about Chapter 2. Those two chapters deal with the enduring problems of cost growth and schedule delays that affect projects of all types. This long-standing problem indicates that projects are inherently risky and that the project management profession has not dealt with this problem effectively. This motivates the need for better risk management. A key part of credible and sound risk management is the quantitative measurement of risk, which is the subject of Chapter 3, titled “Beyond the Matrix: The Cost and Schedule Risk Imperative.” The key point of this chapter is to motivate the use of quantitative methods in risk analysis. Simple point estimates are not enough, nor are qualitative methods such as the use of risk matrices or heat maps of risk.

In this post, I am going to discuss why single estimates of cost and schedule are not enough. Even when these estimates represent an average value, such as a mean or median, there is more to understanding the potential outcomes for resources than just an average. The sport of baseball is a good example that more than just averages
should be considered. Between 1900 and 2019, among players with at least 250 at bats in a season, the mean batting average has fluctuated around a value equal to .275. A .275 batting average means that a hitter gets a hit in 27.5% of his at-bats (this does not include reaching base by being walked or hit by a pitch). There are long trends upward and downward that can last a few decades, but batting averages have fluctuated up and down over the years around this value.

However, the variation around this mean value has not stayed constant. As measured by the standard deviation,
the variation in batting average among batters has been declining steadily since the nineteenth century, when play began. The last person to hit .400 or better in Major League Baseball was Ted Williams, who hit .406 in 1941. The four players to come closest since have been Tony Gwynn, who hit .394 in 1994 for the San Diego Padres in a strike-shortened season; George Brett, who hit .390 for the Kansas City Royals in 1980, another strike-shortened season; Rod Carew who hit .388 in 1977 for the Minnesota Twins; and Larry Walker, who hit .379 for the Colorado Rockies in 1999. From 2000 to 2020, no one has had an average higher than .372. D.J. LeMahieu had the highest batting average of any hitter between 2011 and 2020, leading the American League with a .364 average for the New York Yankees in a COVID-19 shortened season of 60 games.

Long ago, exceptional batters hit .400 or better. Due to less spread in the range of batting averages, the very best batters no longer hit .400 after World War II. Measured in terms of the number of standard deviations above the mean, George Brett’s 1980 batting average of .390 can be considered more exceptional than any batter since Ted Williams’s 1941 season. The declining trend in standard deviation is due to several reasons. Baseball has become accessible to a wider group of people. The sport has matured. Training methods have optimized over time, and with increases in salaries there is a financial incentive for players to play well. In 2019, the average major leaguer made more than $4 million a year. As pitchers and batters improve, it gets harder to exceed the average. Just knowing that the average batter hits .275 only tells part of the story.

Likewise, single point estimates of cost and schedule are not enough to adequately describe the potential range of outcomes. We know from historical cost growth and schedule delays that all projects bear significant resource risk, which requires us to consider variation around point estimates as well. These are best considered as shapes, which are described by probability distributions. For more about this topic, check out my book, now available for pre-order from Amazon and Barnes and Noble.