---

What Is 'Variance'?

• Variance is the degree of fluctuation in a measure or set of data.
• It’s possible to quantify the size of the fluctuations present in a data set.
• Measures of Variance are helpful in identifying patterns when you can't examine all of the variables.

Josh Kaufman Explains 'Variance'

Measures that fluctuate at random are often difficult to analyze. Unit sales figures are a good example: If you sell 3,017 units on Monday, 2,967 on Tuesday, and 3,142 units on Wednesday, how do you estimate your “normal” daily sales? Even more challenging: if your sales increase or decrease at random each day, how can you tell if a new Marketing or Sales strategy is working?

Variance is the degree of fluctuation in a measure or set of data. Measures of Typicality, like means or medians, are a natural place to start, but they don’t capture the range of fluctuations you’re likely to experience. Without establishing the degree of normal fluctuation, you’re not equipped to make good Decisions.

In Understanding Variation: The Key to Managing Chaos, Dr. Donald J. Wheeler describes an approach called “statistical process control” that makes analysis of Variance possible. Starting with a mean or median, it’s possible to quantify the size of the fluctuations present in the data set. This gives you a typical value, an upper normal boundary, and a lower normal boundary: a range of expected normal values.

Once you’ve established an upper and lower bound, you’re in a much better position to make Decisions. Data points that break the upper bound or lower bound are statistically significant, as are a series of consecutive data points that are close to the boundary. Returning to the daily sales figure example: if your normal upper bound is 3,500 units per day, it’s worth looking into a 4,000-unit sales day to see what led to that result. Likewise, if you have five consecutive days that exceed 3,500 units, that indicates you should keep investing in your new marketing plan: it’s working.

Measures of Variance are helpful in identifying opportunities or problems when you don’t have direct access to or visibility of all of the inputs and outputs. Defect rates are a good example: if you sell a Product, you won’t be able to test every unit, and Sampling may not catch problems that occur in use. If you track the Variance of your customer-support requests, return requests, and warranty requests, you can take immediate action if you see an increase that exceeds normal variation.

---

"Keep in mind how fast things pass by and are gone—those that are now and those to come. Existence flows past us like a river: the 'what' is in constant flux, the 'why' has a thousand variations. Nothing is stable, not even what’s right here."

Marcus Aurelius, philosopher and emperor of Rome, second century BCE

---

---

Share this concept:

https://personalmba.com/variance/

---

---

---