Improvement Insights Blog
If It Don’t Fit You Must Forget
Trendlines are often fake news. How can you separate fact from fiction? It’s easy.
“I’m Jay Arthur, author of “Lean Six Sigma Demystified” and QI Macros [software].
“One of the things that really irritates me about Excel is it’ll add a trend line to any data, but it does not add a “goodness-of-fit” metric automatically. It’s called “R squared,” and R squared should be like .8 or 80% fit in general.
“I’ve seen lots of charts and lots of posters at lots of improvement conferences where the goodness of fit metric is less than 50%. What?! So it’s not really an improvement. Just because the line goes down doesn’t mean it’s… [an improvement]. That’s all fake news. Until you have a goodness-of-fit metric on there, it’s just all fake news. Don’t believe any of it.
“For some reason I was thinking about this: “If it don’t fit, you must forget.” If it don’t fit, you must forget.
“So I don’t believe this stuff that’s out there; don’t believe it if it doesn’t tell you what the goodness-of-fit metric is. If you get over .50 you might be getting somewhere, right? There might be a process change which causes the 50%, but the only way to know that is to draw a control chart of that; most people are using line, bar and pie charts: the Three Stooges of charts.
“Let’s go up a notch. Let’s go to grad school. Let’s start using advanced tools.
“That’s my Improvement Insight for this week. Let’s go out and improve something.”
