‘What’s the Trend?’

Wrong question!

In my last column, I showed how hidden special causes can sometimes create the appearance of common cause, but the purpose of common-cause strategies is to deal with this and smoke them out. When there's an underlying structure to how these data were collected, or when one can somehow code each individual data point with a trace to a process input, stratifying the data accordingly can many times expose these special causes.

Even the simple coding of individual points on a graph can be every bit as effective as the more formal tool of a stratified histogram. I’m going to take the issue of understanding variation in count data further in the next couple of columns. I’ll begin here by looking at two scenarios.

…

Want to continue?

By logging in you agree to receive communication from Quality Digest. Privacy Policy.

Create a FREE account

Forgot My Password

Comments

Ban The Phrase "Trend Line" From Business!!!

Excellent article, Davis, and good examples, as usual. The hairs on the back of my neck always stand up when someone in a meeting presents a chart of data with a "trend line". Unfortunately, the software companies have made it far too easy to add a so-called "trend line" line that has no meaning to a mass of data. People love to add a trend line, no matter how slight, to support whatever premise they are trying to prove. "Oh, look! We're trending in the right direction." I usually ask for the raw data and 99% of the time, find that a scatter plot with 95% confidence limits for the least squares fit line clearly shows that the "trend" is not significant. Pass the barf bag, please. I have learned to hold my tongue in meetings (If you know me, you know that ain't easy for me!) and instead have private discussions with the perpetrator of "bad statistics", even though the damage is already done. The downside is that this usually means more work for me! OK..."Teach a man to fish and he will"...continue to ask you to analyze his/her data for him/her. Run charts as very easy to plot and analyze. We MUST find a way to use them and, in most situations, be able to ban the phrase "trend line" from presentations. Time-ordered data is precious and should be properly analyzed.

Cause & Effect Diagram

What's wrong with using a C&E diagram to deal with common causes?

I find them very useful.

Rich D

"Vague"

If you want to do a HUGE Ishikawa diagram brainstorming "What causes medication errors" or "What causes accidents," be my guest.

I'd rather do a high-level stratification FIRST to find out the 20% of the process causing 80% of the problem first, THEN doing an Ishikawa diagram.

Three sources could get exposed: (1) a certain department ALREADY DOING GOOD WORK could have a problem with one particular medication/accident type (due to a unique input), (2) certain departments might have an OVERALL problem with their "safety" or "medication prescribing" process, and (3) there might be certain accidents or error types that are being made by EVERYONE -- which would be a system problem. These will be far more FOCUSED issues.

Plus...to dicover this, you will have to collect a lot LESS data than what would result from a huge Ishikawa no doubt implementing "vague" ideas suggested by "good people"...and getting "vague" results...and making a lot of people mad in the process.

If this makes you feel any better, Mea culpa! I learned the above from Joiner's brilliant "Fourth Generation Management" book...and I'm making a lot fewer people mad these days by not wasting their times collecting data that doesn't ultimately help them or anyone else.