Stats Made Easy

Practical Tools for Effective Experimentation

Sunday, October 28, 2007

TV show misguided in mocking triangle graphs


One of my favorite TV shows, NBC’s Emmy-winning 30 Rock, stepped over a line last night with this sarcastic dialog:

Liz (comedienne Tina Fey) says to her boss Jack: You are a suit who feeds off creativity and hard work of other people and turns it into…pie charts and triangle graphs!

Jack (actor Alec Baldwin): What is a triangle graph?

Liz: I don’t know. It sounded real!

I remember being astounded at my first exposure to a triangle graph in a materials science class taught by University of Minnesota professor Chris Macosko . It was a ternary phase diagram showing the recipe for 18-8 stainless steel – commonly used for knives, spoons and forks – kitchen flatware. Virginia Tech’s Laboratory for Scientific Visual Analysis provides pretty pictures and explanation on how to read triangle graphs. They are a wonderful tool for pinpointing optimal formulations of foods, paints, chemicals, cosmetics, plastics and all sorts of products. For example, see this primer on mixture design that details a fun experiment I did with Pat on making a bouncier play putty .

However, I can appreciate how it must seem totally weird to the vast majority of people that a graph could be triangular. This episode of “30 Rock” had Jack getting after Liz to get on the fast track for attaining his elite financial level by making some investments. That got me thinking of how our company’s 401k advisor always promotes the changing mix of where to put your money, depending on how many years remain to retirement. I used our Design-Expert software to produce this triangle graph showing how the balance between stocks (A-X1), bonds (B-X2) and money market funds (C-X3)might change with age. It shows a flag set at age 30, which being near the top of the triangle is heavily weighted (77 percent) toward stocks. A second flag shows an investment portfolio for a 58-year old, mostly in money market funds.

Note how the triangle graph forces the mix to one-hundred percent. That is what makes them so neat for a chemical engineer like me – the kettle is always kept full, no more and no less. This financial example is just off the top of my head, but maybe a smart New York City business guy like Jack in “30 Rock” could make something out a triangle graph like this. Food for thought…

PS. I visited General Electric’s corporate R&D center in the mid ‘90s to help their Six Sigma Tiger Team leader for DOE get approval for installing Design-Expert. I was amazed to see big displays of NBC’s lineup of new. Ironically, one of them was “3rd Rock from the Sun,” which I realize now must have been a play off 30 Rockefeller Plaza – NBC’s headquarters.

Monday, October 22, 2007

The lottery of life versus death

I am enjoying a long weekend vacation in the Hill Country of Austin, Texas this weekend with my wife and youngest daughter. We took a side-trip east of the metro area to the town of La Grange where we came across a State historic site called Monument Hill. It commemorates the 1842 Mier Expedition that led to Sana Anna’s infamous decree that these invading Texans be decimated via lottery. The Mexican army put this command into effect via a lottery – 17 black beans were put into a pot with 159 white ones. Can you imagine having your life at forfeit for the ten percent chance of pulling a black bean? As one soldier after the other endured his lottery, did the remaining ones recalculate their odds? I went through something like this, but without the horror of anticipation, by surviving a heart attack and then learning that one in five do not. According to Texas Handbook Online, William A. A. (Bigfoot) Wallace did something much more useful than calculating his chances: He observed that the black beans were the larger and fingered the tokens successfully to draw a white bean.

Life really seems at times to be largely a matter of chance. Monument Hill overlooks the lovely Colorado River of Texas. The three of us followed a wooded trail along the bank that a park ranger recommended. Suddenly my two ladies leaped ahead of me. They had nearly stepped on a coral snake that I must have startled from my lead position. I mentioned this to the ranger upon my return and he just shrugged his shoulders and said we ought to expect snakes when walking the woods in Texas. However, although the ranger allowed that the coral is perhaps the most venomous snake in North America, its mouth is very small, so it cannot bite very effectively, and furthermore -- this reptile is very shy. Nevertheless, we three naïve Minnesotans learned a lesson: Never mess with Texas – especially wearing sandals. Waiting in the Minneapolis airport, I was chuckling at the Austinites pacing around in their boots. Now I understand the reason for this choice of shoe.

PS. In my 8/10/06 blog titled “Stats that will be the death of you” I reported the odds of an American dying of snake bite: 1 in 2 million.

Sunday, October 14, 2007

Treat or Trick? Drinkers punked by brewers?


My report last month on Mixture Design Brews Up New Beer Cocktail—Black & Blue Moon generated many favorable comments from formulators wanting to apply the methodology for their product development. For what it’s worth from a very limited tasting panel, a blend of Sam Adams Black Lager and Blue Moon white (wheat) beer came out significantly better than either brew alone.

Budweiser, the least expensive beer, did not fare as well in this blending experiment. Thus when I brought up my study at a gathering of industrial statisticians last week, the one from Anheuser-Busch (Bud's brewer) expressed little enthusiasm. She suggested I try their Jack’s Pumpkin Spice Ale. I found this seasonal ale in the selection of beers at a bonfire last night and gave it a try. From the reviews I’ve read, it looks distinctive when poured into a glass, but straight out of the bottle it tasted no different than any other beer. Several other party-goers tried it, but none could discern anything special.

In their landmark article “Influence of Beer Brand Identification on Taste Perception (Journal of Marketing Research, Vol. 1, No. 3, Aug., 1964, pp. 36-39), Ralph I. Allison and Kenneth P. Uhl demonstrated that “beer drinkers cannot distinguish among major brands of unlabeled beer.” The experimental subjects rated their favorite brands significantly higher than other beers. But when they tasted and rated the same beers without labels ("nude"), the differences were not significant. As noted in my article on mixture design (link above), my taste test on beers a few decades later came up with the same result –perception dominated reality. Based on his comment about the recent merger of Miller and Coors, former Beer Writer of the Year Don Russell, better known as Joe Sixpack, evidently would agree with this premise that branding influences perceived taste:

“In beer-enthusiast circles, there’s a lot laughing going on, because, if I’m not mistaken, the name of the company is going to be MillerCoors. Beer enthusiasts refer to these types of beers as BudMillerCoors — just one word. They’re two thirds of the way there now. People who really love beer really do not distinguish them at all.”

So the challenge continues for beer lovers to remain skeptical about advertising claims by the big beer brewers. Put them to the test by asking a buddy to serve you nude, that is, without the label. Only then can you judge your taste preference objectively. Otherwise you don't know Jack. (I must try 'his' Pumpkin Spice Ale again in a fairer test!)