W. Edwards Deming created the high quality car industry in Japan when they implemented his statistical QC procedures.
Yep and he received Japan’s highest award by their Emperor.
Kudos to you for learning who Deming was. Deming is not only a hero in Japan, he’s also a giant of history when it comes to the US Census.
He was a statistical genius coming up with methods of counting and sampling that were extremely efficient and accurate.
Why count 100% when counting a fraction using efficient unbiased sampling gets the same result within say 0.5%?
Put into another context, why should we buy the latest and greatest LED monitors with 240 FPS when our human eye can’t tell the difference between 240 and 120 FPS? The limitations are based on sampling rates. This is what underlies Deming’s genius.
But there are other mathematical geniuses. A full MRI has in the past taken an hour in the tank. Statistical methodology using only 10% of the full scan data acquired under a smart sampling algorithm requires less than 5 minutes in the tank and cranks out images that expert radiologists can’t tell the difference from the full data scans.
Thousands of examples exist that render pointless a wasteful brute force approach.
Here’s a simple example to drive the point home:
Joe and Don are tasked with counting the number of bushels of corn grown in a hundred acres of land. They divide the land into two equal parcels of 50 acres each.
*** Database of related information ***
“The two parcels are known to be equivalent in terms of soil capacity but within each 50-acre parcel, there are variations in soil fertility gradients. A soil gradient map is available. But on average when comparing John’s parcel with Jim’s parcel, the conditions and capacity are equivalent.”
Slow Joe adopts a brute force method where he counts the ears of corn on every corn stalk. It takes him a month to get it done.
Smart Don adopts a sample plan that accounts for soil fertility striations on a gradient map. It takes him 3 days to get the job done.
Slow Joe and Smart Don compare results. Slow Joe comes up with an average 130 bushels per acre. Smart Don comes up with an average 128.5 bushels per acre.
Slow Joe says Smart Don is using the wrong approach and that Smart Don’s results were just ‘lucky’. The next year they switch parcels and the results are Slow Joe calculates 129 bushels per acre and Smart Don comes in with 131 bushels per acre. Slow Joe is stubborn and won’t change. Besides he’s not into all that fancy math and all that.
Smart Don has utilized a database of related information to get the job done in about one-tenth the time.
It’s all about sampling, efficiency, and testing.