this 
ANALYST     Professional     Personal

*

First, why this page, why focus on The Bell Curve? First came curiosity and then came serendipity:

The residuals, however, showed that HM made a Type I error, a false positive--asserting a relationship between their model--scores on the AFQT, SES, and AGE--and POVERTY when, in actuality, there wasn't: HM's model predicted none of its cases of interest correctly and so no interpretation of their first analysis is warranted. This selfsame model being used throughout the book suggests analytical termination dust; it would be an interesting project to find out conclusively. I value the many really, really interesting e-conversations this page has generated over the years.

Professional

Call my philosophy of measurement "interpretive." Rene Magritte's

Magritte's  _The Fair Captive_ *The Fair Captive shows sky and ocean in the background and an empty picture frame resting on an easel in the foreground [ignore the flaming tuba, OK? life's messy and analogies, worse].

*The fair captive, the ocean, is perfectly captured within the picture frame, no distortion, unfiltered, the thing-in-itself--unchanged by the artist's mood at the time, the way s/he lays brush to canvas, choice of medium, the things s/he doesn't see that are there and the things s/he does that aren't.

The painter must form the object by brush, the photographer by angle, and quantitative researchers by using "scores" of one sort or another. Every painting results from an act of interpretation, as does every data analysis. And then there are the interpretations of our viewers and our readers (not to mention ourselves). We rejoice in the multiple interpretations of the painting ... we too often assume their absence in the analysis: data analysis, like art, is an act of interpretation, one whose goal is to make that act as rigorous as possible. The center of knowledge--as pointed out by Kant back in the 18th century--is the knower; the knower is human; knowledge is fallible. It's an a priori: expect it, try to mitigate it, check your work, look at residuals. Knowledge is a slippery slope.

Personal

I like data (mountains, too): In 1995, I put up my first web page--oohing and aahing about real-time data starting to appear online, a data renaissance: CERN, more and more .gov files consisting of only telephone and fax numbers--wrong kind of data--replaced by seismic activity in gopher space, CEISEN, the census and more--medline where it belonged in the public domain--who would have thought it'd turn out like this. I currently maintain several pages about doing dumb things with numbers, among them one on advertising "research," another on my former state's high school exit "exam" (Alaska's HSGQE), and another on the U.S. Department of Education's idiotic "response" to the state lunacy. Besides, my kid was once my ISP (not recommended during the teen years).

*Formative years*

The first one was 48 years ago. Soon after acquiring the requisite skills, I decided all I wanted to read about were those who are called the US's Native Americans (who can explain the tastes of children? perhaps because of a family rumor that a tiny piece of my mitochondria was Choctaw). Quite by accident--I didn't know that "Indian" had more than one meaning--I checked out the Bhagavad Gita, Krishna's advice to Arjuna on the battlefield (English translation). By the time I finished the book, I understood every individual word--and, even after rereading, knew that I knew not what the book was about: I knew what "right" and "livelihood" meant but not "right livelihood" (and "acting without attachment"??). A second formative event that year was my US K-12 curriculum including a part of US history titled, then, "Manifest Destiny," US history from the Louisiana Purchase to from-sea-to-shining-sea (without any mention that almost the entire cost of the white westward migration was paid for by stealing Native American lands (even 10 year olds can make rudimentary ethical distinctions)). There are few who want to conform more to what the group thinks, to belong, than 10 year olds--I was no exception--but although my teachers meant well, I knew the history books wrong.

*Present*

Each day I cross this lovely bridge over even lovelier Whatcom Creek on my way to the gym. And, although I've listened almost exclusively to internet radio for the last 5 years, I'm delighted to have discovered 89.3 on my local FM dial (not to mention a number of Canadian stations). I've wondered why Bellingham bills itself as the "city of subdued excitement." I'd speculated that it was due to the absence of a truly hot hot sauce on local grocery shelves; someone probably wiser said because "We are all drugged from the mercury that is coming from the water" (in Lake Whatcom, the source of the city's drinking water).

*Darwin Award Nominee*

... My car had dead-ended in muck 61 degrees north and 151 west, couldn't get unstuck, left a note, grabbed my then digitally utilitarian ibook, and started walking: 10 minutes later I heard branches breaking in the underbrush behind the closest trees, a sound that caught my attention and exactly paralleled me, turn after turn in the road for the next 20 minutes ... until a passing biker--blessed be harley--gave me a lift, one hand on the back bar and the other on the ibook, back to civilization's nearest outpost, the Bishop Creek Bar: Its patrons pulled me out, and I bought several rounds, more a matter of gratitude--which I was exuding mightily--than of etiquette. I'll never know--I later learned that that area was prime brownie habitat--and already knew that moose, the only other thing big enough to make that sound, browse as they go--what was on that bear's mind (they're individuals, too). I can live without knowing. I am fine, albeit with grayer hair--but the wussy ibook later died, a dozen windows open, all empty, all white, like the scream in that Edward Munch painting, except rectangles instead of ovals.

*My bones*

p(TII_Error)=tiny # My bone density, measured by a dexascan, compared to U.S. women overall, is over one standard deviation above the mean and, compared to my age-matched peers, over two (the small x's above the x axis under the curve to your left represent my dexascan scores). Those deviations above the mean mean that my bone density "score" is greater than

  • 85% of my countrywomen and

  • 98% of my peers, women my age.

We know the distribution of dexascan "scores" is normal because of the law of large numbers. That means that the probability of a Type II error, a false negative--my classified as not having osteoporosis when, in actuality, I do--can be calculated: the probability of my having osteoporosis is a very, very small number (possible but unlikely). Empirical data, however, give a different impression: I have "risk factors" like being hypothyroid--and I'm less tall than I was: I've lost 2 inches compared to the national average loss of 1 inch. Should I, individually, take a fosamax-like pharmaceutical? I'm basing my decision not to on Fisher's calculations, i.e., osteoporosis is *not* one of my concerns. Although not in agreement with frequentist reasoning, in this case, Fisher's probabilities can be usefully applied to subjective decision making.

*Dumbest Deed Done Using UNIX*

After a few hours working on a cloned file, I typed !c, remembering the previous command as cat, having forgetten cp: the shell remembered.

*Closest I've come to reducing the history of science to a limerick*

Ayer proclaimed scientific truths educible,
And Skinner proclaimed them reducible.
But Kuhn had the gall to suggest most, if not all,
Are quite simply irreproducible.

*Favorite joke heard in 3 segments over 10 years whose conclusion I don't know*

A barkeep asks 17th-century philosopher Rene Descartes, famous for his cogito ergo sum, if he wants a beer: "I think not," says Descartes and disappears.

Further down the bar, two atoms look up--and in walks Rutherford, looking pissed and carrying a particle gun. One atom falls to the ground, saying "It looks bad ... I've lost an electron." Are you sure asks the other: "Yes, I'm positive."

The barkeep, astounded, turns and asks the closest patron if he saw what happened: "I'm uncertain" says Heisenberg.

*Household Hints*

Before throwing your used sonicare toothbrush head away, use pliers to rip off the magnet: it's strong enough to hold up my former car's bumper on the side a caribou ran into with, of course, the help of the faux silver tape. And remember, if you have those round glass imac speakers, wide-mouthed canning jars make lovely speaker stands.


I hope your data are doing well today (I bet you didn't realize that constants are stored in the many new data warehouses).

Root Variables Subjects Replication & Fit Analyst Documentation