Beware of statistics

From Spurious Correlations, which has other examples, including a correlation between U.S. spending on science, space and technology and suicides by hanging, strangulation and suffocation.

Per capita consumption of mozzarella cheese (US)
correlates with
Civil engineering doctorates awarded (US)


Upload this chart to imgur

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Per capita consumption of mozzarella cheese (US) Pounds (USDA) 9.3 9.7 9.7 9.7 9.9 10.2 10.5 11 10.6 10.6
Civil engineering doctorates awarded (US) Degrees awarded (National Science Foundation) 480 501 540 552 547 622 655 701 712 708
Correlation: 0.958648

A new path to college math

A Los Angeles-area community college is trying the Carnegie Foundation’s alternative path to math success, an algebra-and-statistics mix called Statway. Success rates for remedial students — normally very low — are rising.

A Denver university is taking back remedial education from community colleges in hopes of boosting success rates.

No silver bullet for remedial woes

Reformers are transforming — sometimes eliminating — remedial education at community colleges, but fixing remedial ed will be “vastly more complex” than they think, argues Hunter R. Boylan, who runs the National Center for Developmental Education.

Virginia’s community college system raised success rates for unprepared students by lowering math demands for non-STEM majors. Carnegie’s Pathways reforms focus on statistics and quantitative reasoning rather than advanced algebra.

Success paths for all

How can high schools ensure graduates are college- and career-ready, asks an Education Next forum.

Students need multiple pathways, writes Robert Schwartz, a Harvard professor emeritus who coleads the Pathways to Prosperity Network. “We have allowed a very important idea—that all students need a solid foundation of core academic knowledge and skills—to morph into a not-so-good idea: that all students need to be prepared to attend a four-year college,” he writes.

If we follow a cohort of 8th graders, roughly 2 in 10 will drop out before high school graduation, and another 3 will graduate high school but choose not to enroll in postsecondary education. Of those who do go on and enroll in four-year institutions, nearly 4 in 10 will drop out before attaining a degree. Of those who enroll in community colleges, roughly 7 in 10 will drop out. The bottom line: by age 25, only 33 percent of the cohort will have attained a four-year degree, and another 10 percent will have earned a two-year degree.

Many good jobs require some education beyond high school but not a four-year degree, Schwartz writes. He likes the northern European model: “All students pursue a common curriculum up through grade 9 or 10, and then choose between an academics-only pathway leading to university and a more applied-learning pathway leading to a vocational qualification.”

Instead of letting students choose their path, we “force march all students” through a math sequence leading to calculus, a goal few will achieve and even fewer will need, he writes.

Yet most community college students and many university students aren’t prepared for college algebra. “In my view, the vast majority of students in two- and four-year institutions would be much better served by getting a solid grounding in data, statistics, and probability in high school,” he writes.

Common foundational skills are essential, writes Cynthia G. Brown, senior fellow at the Center for American Progress. All students should take a college-prep curriculum, but high school students could choose “curricular options that fit their interests, skills, and plans for the future.”

Algebra or statistics?

Most new students place into remedial math at California community colleges. Eighty percent will never pass a college-level math course. Some colleges have boosted success rates by teaching statistics and quantitative reasoning, rather than algebra, to non-STEM students.

Florida colleges will let students opt for college-level courses, even if they’ve done poorly on a placement exam. Instead of letting students ignore the placement results, let them try the test again, a graduate student suggests.

NCEE: Only 5% need calculus

Only 5 percent of students will use calculus in college or the workplace, concludes a new report on college and career readiness by the National Center on Education and the Economy. Most community college students could succeed in college courses if they’ve mastered “middle school mathematics, especially arithmetic, ratio, proportion, expressions and simple equations.” Many have not.

The report calls for providing an alternative track — less algebra, more statistics — for high school students who aren’t aiming at university STEM degrees.

In a few years, high school diplomas in North Carolina will show whether a graduate is prepared for a four-year university, a community college and/or a career.

Will this boy graduate high school?

subwayThe other day, on the subway in NYC, I saw this ad. It turns out there has been some commotion over it. (Approved and defended by Mayor Bloomberg, it is part of New York City’s recent campaign to raise awareness about teen pregnancy.)  I would like to add my own two or three objections to the mix.

First, this is an example of the “precision fallacy” in statistics. (That’s the best term I could find; there may be better.) Specifically, the ad confuses the individual’s probabilities with those of the group. It may be that “kids of teen moms are twice as likely not to graduate than [sic] kids whose moms were over age 22,” but this probability doesn’t hold for individuals.

Second, adults put words in this child’s mouth (and banal words at that). A baby or toddler would not say anything remotely close to this, unless someone had prepped him to do so.

That brings up a larger problem: from a young age, children are trained to describe themselves in statistical terms, at school and elsewhere. They learn to say, “My growth in such-and-such a skill is 30 percent,” or “I was one of the sixty percent who had the right answer.” In measure, in the right context, this may be fine–but when it’s the dominant lingo and mode of thought, it crowds out substance and meaning. (I wrote a satirical piece about this tendency.)

Beyond that, I did not bear this child as a teen, nor did 99.999999 percent of NYC subway riders, in all likelihood. (For all we know, this kid’s mom might have a chauffeur.) The “you” is not a real you, nor the “I” a real I. Yet here’s a tear-streaked face bringing sadness to a passenger’s day–and to what end?

What good does it do even for the target audience, teens who might get pregnant or father a child? If I were a teen looking at the picture, I’d want to wipe the little boy’s cheeks. I’d want to take out a book and read to him. Yet I wouldn’t be able to do so. I might dream of being a parent one day–and, if I were foolhardy enough, I’d want that day to come soon.

Worst of all, this ad gives the impression that the boy’s existence is a mistake and his fate sealed (or at least tipped in a direction). This is wrong. Once a child comes into the world, he or she is no mistake. Nor do we know what that child’s life will be.

Of course teen pregnancy is no light matter, no matter how it’s handled. I imagine many involved with the ad had good intentions. Still, it  fails to inform, enlighten, or persuade. And what a sad-looking kid.

No room in class? Earn credits online

California college students could bypass wait lists and earn credits online under a bill introduced by a Democratic legislative leader. State colleges and universities would be required to accept credits from faculty-approved online courses for about 50 high-demand, lower-level classes with long wait lists.

Nearly two-thirds of community college students place into remedial math. Half of students in Statway — Carnegie’s intensive, yearlong developmental math pilot  –passed a college statistics course in the second semester. By contrast, only 5.9 percent of non-Statway remedial students at the same community colleges earned college math credit in their first year; that rose to 15.1 percent in two years.

Study: Hybrid class works for college students

College statistics students in a hybrid class — online instruction plus a one-hour face-to-face session — performed slightly better than the control group and spent 1.7 fewer hours per week on the course, write William G. Bowen, Matthew M. Chingos, Kelly A. Lack and Thomas I. Nygren in Education Next.

“The effect of the hybrid-format course did not vary when controlling for race/ethnicity, gender, parental education, primary language spoken, score at the standardized pretest, hours worked for pay, or college GPA,” the authors report.

Half the students who participated come from families with incomes less than $50,000 and half are first-generation college students. Less than half are white, and the group is about evenly divided between students with college GPAs above and below 3.0.

Carnegie Mellon designed the hybrid course, which was taught at public universities in New York and Maryland.

Students in the control group received three to four hours of face-to-face instruction each week.

Moving to the hybrid model could cut the costs by 19 percent to 57 percent, depending on whether professors do all the teaching or assign sections to teaching assistants, the authors estimate.

Teaching the quantified student

 “I am a bad teacher” wrote Sujata G. Bhatt in Valerie Strauss’ Washington Post blog in the school test-taking season of 2011.  Education reformers want to use data to drive instruction, reform and accountability, wrote Bhatt. “At what cost? Does this data really represent learning and knowledge?”

Since then, she’s embraced data, Bhatt writes in The Quantified Student.

She teaches in a high-poverty Los Angeles school. Many of her students aren’t fluent in English. In the fall of 2010, her fourth graders were particularly unprepared.

Since California’s standardized test for fourth graders measured skills almost all my students needed, I analyzed its requirements, broke them down into core concepts, and then worked and reworked these concepts with the students until they felt a sense of mastery over them. My daily job consisted of finding different, creative ways of approaching, teaching, and reteaching the same core skills so that most all students could incorporate them into their cognitive toolkits.

It worked. The students succeeded wildly. They returned to me for fifth grade with heightened confidence. They saw something new in themselves: the reward of effort and the joy of success.

They also came back with questions about “how many more points it would take to get to the next level, how many more problems they’d need to get right to get those points.”  They saw the test as a game they wanted to win.

Teaching the same cohort in fifth grade, she looked for ways for her students to explore their interest in data. 

We used math websites like TenMarks that enable students to learn about their own learning even as they practice new skills. We analyzed information graphics and dove into ways of presenting numerical information. We explored how numbers shape our understanding of ourselves and the world. And much of their enthusiasm and curiosity for these tasks came out of their interest in numbers from standardized testing.

She now believes standardized testing can help teachers understand how well they’ve taught and help students become “agents in their own learning.”

Testing — and evaluation systems built on test scores — need to get a lot better, Bhatt writes. But it makes more sense “to work to create better data than to fight data.”

Data analysis is an increasingly significant and empowering way of making sense of the world. All sorts of professions use data to interpret their work and decide upon courses of action. Why shouldn’t we in education?

In the high tech world there’s a growing movement called “The Quantified Self.” With quantified self models, adults use data to change habits and behaviors–to lose weight, exercise more, to calm themselves.

“Why not help our students become makers and masters of their own data, and help them use it to propel their own learning forward?” Bhatt asks.

The Measured Man is a fascinating — and somewhat alarming — Atlantic profile of Larry Smarr, an astrophysicist, computer scientist and highly quantified human.