“A bigger effect produced in a study with a big margin of error is more impressive than a smaller effect that was measured more precisely.”

– Stephen Ziliak, economics professor

However, this legal analysis of the ruling cautions that statistical significance remains relevant for assessing materiality of an adverse event.

Given all this, we can be certain of only one thing – more lawsuits.

What I always look for along with such predictions is the margin of error (MOE).  The Humphrey Institute pollsters provide these essential statistical details: “751 likely voters living in Minnesota were interviewed by telephone. The margin of error ranges between +/-3.6 percentage points based on the conventional calculation and +/-5.5 percentage points, which is a more cautious estimate that takes into account design effects, in accordance with professional best practices.”**  Note that the more conservative MOE (5.5%) still left Dayton with a significant lead, but just barely at 12 points (vs 5.5%x2 = 11% overlap of MOEs).

Survey USA, on the other hand, states their MOE as +/- 4%.  They provide a very helpful statistical breakdown by CPO versus landline, gender, age, race, etc. at this web posting.  They even include a ‘cross-tab’ on Tea Party Movement – a wild card in this year’s election.

By tomorrow we will see which polls get things right.  Also watching results with keen interest will be the consultants who advise politicians on how to bias voters their way.  Sunday’s New York Times offered a somewhat cynical report on how these wonks “Nudge the Vote”.  For example, political consultant Hal Malchow developed a mailer that listed each recipient’s voting history (whether they bothered to do so, or not), along with their neighborhood (as a whole, I presume).  Evidently this created a potent peer pressure that proved to be 10 times more effective in turning non-voters into voters!  However, these non-intuitive approaches stem from randomized experiments, which require a control group who get no contacts (Could I volunteer to be in this group?).  This creates a conundrum for political activists – they must forego trying to influence these potential voters as the price paid for unbiased results!

“It’s the pollsters that decide. Well, a poll can be skewered [sic #]. I can go out and get you a poll on anything you want and probably get the results that I want just in how I conduct it.”

– Jesse Ventura, professional wrestler (“The Body”) and former governor of Minnesota

# Evidently a Freudian slip – him being skewered on occasion by biased polls.

* “Poll parsing” column by David Brauer, page 15B.

** From this posting by Minnesota Public Radio

What caught my eye in the critique by Steven Strogatz – an applied mathematics professor at Cornell, was the deception caused by “disestimation” (as Proofiness author Seife terms it) of the results from Minnesota’s razor-thin 2008 Senate race, which Al Franken won by a razor-thin 0.0077 percent margin (225 votes out of 1.2 million counted) over Norm Coleman.  Disestimation is the act of taking a number too literally, understating or ignoring the uncertainties that surround it; in other words, giving too much weight to a measurement, relative to its inherent error.

“A nice anecdote I like to talk about is a guide at the American Museum of Natural History, who’s pointing at the Tyrannosaurus rex.  Someone asks, how old is it, and he says it’s 65 million and 38 years old.  Sixty-five million and 38 years old, how do you know that?   The guide says, well, when I started at this museum 38 years ago, a scientist told me it was 65 million years old. Therefore, now it’s 65 million and 38.  That’s an act of disestimation.  The 65 million was a very rough number, and he turned it into a precise number by thinking that the 38 has relevance when in fact the error involved in measuring the dinosaur was plus or minus 100,000 years.  The 38 years is nothing.”

-          Charles Seife (Source: This transcript of an interview by NPR.)

We Minnesotans would have saved a great deal of money if our election officials had simply tossed a coin to determine the outcome of the Franken-Coleman contest.  Unfortunately, disestimation is embedded in our election laws, which are bound and determined to make every single vote count, even though many thousands in a State-wide race prove very difficult to decipher.

1. A team of U.S.-British researchers announced earlier this month that they deciphered previously-illegible scrawling by African explorer David Livingstone, which he made 140 years ago under desperate circumstances using the juice of local berries.  See the image enhancement in this article by New Scientist Tech.  Given the depressing content of Livingstone’s laments, it may be just as well he used ephemeral ink.
2. The Dead Sea Scrolls, now on exhibit at the Minnesota Science Museum (see this picture, for example), were written with extremely durable black ink (well over 2000 years old!) comprised of lamp black (soot), gum Arabic and flaxseed oil.  According to this Numerica entry on the chemistry of ink a red version was made by substituting cinnabar (mercury sulfide ? – HgS).  That must have been used by the editor overseeing publication of the Scrolls. ; )
3. Printer.com suggests that we all save ink by favoring certain fonts over others.  For example Century Gothic* uses 30 percent less ink than Arial.  As a general rule the serif fonts do better than the sans serif ones.  An article by Dinesh Ramde of Associated Press on 4/7/10 reported that a school of 6,500 students, such as the University of Wisconsin-Green Bay, can save up to $10,000 per year by switching to an ink-stingy font.  To really make a statement about their support for Earth, UW-GB ought to go with the “holey” ecofont.  However, rather than going to something so ugly, perhaps the best thing for all concerned about going green would be to be prohibited from printing anything and just hand-write what’s absolutely essential to put on paper (or papyrus).

Unfortunately the people funding consultants and researchers do not want to hear any equivocation in stated results.  However, it’s vital that experts convey the possible variability in their findings if we are to gain a true picture of what may, indeed, transpire.

“Error is to be expected and not something to be scorned or obscured.”

– Trevor Butterworth

“Most of the month-to-month changes are not only nonsignificant in a statistical way, but they are often straddling zero, so you can’t even infer the direction of the change has been accurately represented.”

-          Patrick O’Keefe, economic researcher

The uncertainty stems for the use of sampling as a cost-saving measure for government agencies and ultimately us taxpayers.  For example, field representatives covering 19,000 geographical units throughout the U.S. only sample 1 out of 50 houses to see whether they’ve been sold.

The trouble with all this uncertainty in statistics is that it ruins all the drama of simply reporting the point estimate. ; )

*(See “It Is 90% Certain That Unemployment Rose. Or Fell.” and a related blog on “What We Don’t Know About the Economy” )

Specific statistics like this, when used indiscriminately by strongly-biased people, give statistics as a whole a bad name.  However, those who are not duly diligent in vetting inflammatory stats are just as much to blame as the originators misleading them.

“It is proven that the celebration of birthdays is healthy. Statistics show that those people who celebrate the most birthdays become the oldest.”  — Widely quoted as stemming from a PhD thesis by S. den Hartog (perhaps too good to be true!)

Today I learned of another concept attributed to the great Italian economist*: Pareto inefficiency.  The Concise Encyclopedia of Economics explains that a “Pareto-optimal allocation of resources is achieved when it is not possible to make anyone better off without making someone else worse off.”  I found this detailing by The New School which is too much for me to completely digest, but my attention was caught by this heads up:

“An economy can be Pareto-optimal, yet still ‘perfectly disgusting’ by any ethical standards.”

 – Harvard Economics Professor Amartya Sen (1970)

So, while I am enticed by the idea that we can make most everyone (80 percent?) better off without making the others (20 percent?) worse off, I remain skeptical.  However, having seen what a focused quality improvement team can do with the aid of Pareto charts at a micro level, I remain hopeful that some big strides can be made at the macro level for health care nationwide.

*Vilfredo Pareto

So it seems that the number nerds will rule after all — just like they always dreamed when being belittled by the bullies who thought math and stats were simply a waste of time. Statisticians rule!

As I’d expected, her regression analysis (using the SPSS software) showed a positive correlation of “frequent church goers” voting for Bush (0.166 R²) and negative for “population with college degree or higher” (0.293). However, the highest correlation was seen with per-capita income, which surprised me by being negative – the more the voter earned, the more likely they were to NOT vote for Bush. I always thought that the Republicans were the party of the rich. But from this data one must conclude that they mainly appeal to poor, less-educated church-goers! (Please do not take the previous two ‘tongue-in-cheek’ statements seriously, I am only making a humorous point about how misleading statistics can be!)

I don’t give too much credence to any of this – mainly due to my great skepticism of using statistics to dissect historical data and generate inferences on cause and effect relationships. However, it makes me curious as to the driving forces of today’s party politics in the USA. That’s about all I figure that regression of happenstance data really offers – some food for thought that may lead to more rigorous investigation.