tl;dr Many data statements are factually true but practically meaningless. Two-point comparisons and color-coded dashboards hide whether anything has really changed. To know if performance is improving–or just fluctuating–you need to plot the dots and use Process Behavior Charts to…
The average NFL score is around 44-46 total points per game — but that number hasn’t always been stable. Has scoring truly surged, or are fans reacting to short-term noise? TL;DR: NFL scoring trends reveal a real system shift–not just…
tl;dr: Headlines about baseball trends–like Bryce Harper’s performance, game length, or declining attendance–can be misleading. Using Process Behavior Charts, we can distinguish between meaningful signals and routine noise. This post shows how to apply that thinking to baseball metrics, with…
TL;DR: Home runs in MLB haven’t just “gone up”–the data shows a real system shift, not random year-to-year variation. Process Behavior Charts help separate signal from noise, revealing when it’s worth investigating what truly changed instead of overreacting to single-season…
tl;dr: Two data points aren’t a trend. NCAA Final Four TV ratings swing up and down every year, but Process Behavior Charts show this variation is predictable noise–not something to explain or overreact to. Metrics only become meaningful when viewed…
tl;dr: Year-to-year changes in Oscars TV ratings are usually just noise, not meaningful signals. Blaming or praising the host (or lack of one) based on small fluctuations is a mistake. Looking at longer-term trends with tools like Process Behavior Charts…
Here is a new article that I posted on LinkedIn... don't worry, it's not political, but it does look at recent voter turnout rates through a statistical lens. And I think it's a lens that's helpful for looking at metrics in our own organizations:
"How to Tell the Difference Between ‘Signal’ and ‘Noise’ in Voter Turnout Numbers or Your Organization’s Metrics"
I'll be presenting a free webinar on Wednesday, October 31st at 1 pm ET... as you might notice, that's Halloween day!
Here's the description from the KaiNexus website:
Metrics & Statistics Don't Have to be Scary
Why do we have to be skeptical or careful when we see a chart with just four data points? Does such limited data help prove that our consulting work or our Lean Sigma project has made a difference?
I started my career in manufacturing, so that's just one reason I'm interested in the topics of offshoring (sending manufacturing work overseas) and what's now called "reshoring," or bringing jobs and factories back to the U.S.
Somebody at A.T. Kearney sent me a link to their recent report on reshoring, with the headline:
Reshoring in Reverse Again
US manufacturers are not exactly coming back in droves. In fact, the 2018 Reshoring Index shows that imports from traditional offshoring countries are at a record high.
When you learn to look at data and workplace metrics through the "Process Behavior Chart" methodology, you learn to be skeptical of text descriptions like "a record high." Does "a record high" mean that the data point is statistically meaningful? Not always.
Are minor league games taking longer? Well, yes and no, depending on the league.
My next book, Measures of Success, is about the use of this Process Behavior Chart methodology in the workplace, as applied to our performance measures.
This post explores some data, how asking for more data can be more helpful, and how to use charts to evaluate a metric over time.
As I blogged about yesterday, things went well at the Oscars... or, at least, no errors were made in the announcements. But that thing that didn't go well was the TV ratings.
Two Data Points Are Not a Trend
The headlines I saw had a lot of two-data-point comparisons. Headlines sometimes gave the percentage decrease in viewers or how many million fewer viewers there were. Many talked about "record low" but if you're tracking a metric "record low" or "all-time high" doesn't mean there's a "special cause." That "record low" could still be noise in the system.