Avoiding Data Pitfalls

Chapter FourPitfall 3: Mathematical Miscues

“Calculation never made a hero.”

—John Henry Newman

How We Calculate Data

There are heroes and there are goats. As the epigraph states, calculation may have “never made a hero” – a status more commonly attributed to women and men who pull off daring acts of bravado in spite of great odds – but failure to calculate properly has definitely made more than one goat.

Infamous examples abound, such as the disintegration of the Mars Climate Orbiter¹ on September 23, 1999, due to atmospheric stresses resulting from a problematic trajectory that brought it too close to the red planet. The root cause of the faulty trajectory? A piece of software provided by Lockheed Martin calculated impulse from thruster firing in non-SI pound-force seconds while a second piece of software provided by NASA read that result, expecting it to be in Newton seconds per the specification. One pound-force is equivalent to 4.45 Newtons, so the calculation was off by quite a bit.

These cases remind us that to err truly is human. And also that it can be really easy to get the math wrong. It's a pitfall we fall into so often.

We make calculations any time we apply mathematical processes to our data. Some basic examples include:

• Summing quantities to various levels of aggregation, such as buckets of time – the amount of some quantity per week, or month, or year
• Dividing quantities in our data with other quantities in our data to produce rates or ratios
• Working with ...