[NOTE: The instructions on reenacting the gummy bear experiment are not yet available. We will post as soon as they are available, and apologize for any inconvenience to you.]
If you visit Robert L. Rooke Professor of Statistics George Cobb's introductory statistics class, make sure to duck - otherwise you might be hit by a flying gummy bear. Be careful where you sit too, since class members could be measuring puddles of pink slime. At first glance, launching candy via catapult and experimenting with a glue and Borax mixture seems more appropriate for a preschool classroom than a college one. But for Cobb, these activities are tools to teach students with diverse academic backgrounds the basics about statistical methods ranging from experimental design to regression analysis.
Offered this semester for the first time, Statistics 140 makes use of an innovative hands-on curriculum, which Cobb and two colleagues designed and outline in their textbook Statistics in Action (to be published in summer 2000). Having students collect data and observe real-life applications for statistical methods seems only natural for Cobb, who views statistics as "a way of thinking about the world" that focuses on context and meaning. In this regard, statistics is radically different from mathematics, he says. "Mathematicians are the developers with bulldozers, needing to clear context out of the way in order to build their deductive structures. Statisticians are the archeologists with whisk brooms, sifting through the details in order to reconstruct the hidden story of how the data came to be."
Cobb contends that when students have a hand in creating data, they are more interested in analyzing them and learn from being part of the production process. In the gummy bear exercise, for example, students used a catapult made of Popsicle sticks, rubber bands, and blocks to launch the bears into the air, exploring such principles as sources of variability, the value of protocol to keep things constant, and the use of randomization to protect against bias. When it came to the data collection, students contended with incidents such as catapult misfires and the most effective way to record them.
Class members apply what they learn from their experiments to data taken from current news and the physical, biological, and social sciences. For example, they measured the percentage of change in input in relation to the percentage of change in output, considering the question of how much slime it takes to produce a puddle of a given diameter. It turns out that the relationship depends on the consistency of the slime, and provides a literal example of economists' notion of elasticity. The production metaphor extends from slime to more serious relationships: between body mass and brain mass for animal species, and between GNP per capita and literacy rates for countries.
The class is also working with computer-simulated data, which enable students to answer the versions of the question that statistical logic rests on, "What will happen if I repeat this a large number of times?" Simulation brings these ideas within the reach of those who do not have the background in mathematics needed for a more theoretical approach.
Student reaction to the class has been overwhelmingly positive. "Professor Cobb's experiments give us real experience--it's much easier to understand the concepts when you are figuring out the ratio of red M and Ms in a bag of plain vs. peanut than when a professor is dryly writing on a chalkboard," says politics major Katherine Gordon '99. "I love this class--I'm almost wishing, as a second-semester senior, that I'd majored in statistics." English major Sarah McKinney '00, who was initially apprehensive about Statistics 140 because math is not her strength, is doing fine. "The experiments not only make the class more entertaining, but they are a useful way to illustrate basic statistics," she says.
This new MHC course is part of a larger reform effort that is changing the way statistics education is taught in schools and colleges across the country. As one of the leaders of that effort, Cobb has planned and administered two projects funded by the National Science Foundation--a 1991 conference for statisticians with major projects in statistics education, and a three-year series of week-long workshops for mathematicians who teach statistics.
Mount Holyoke has contributed to the national effort. Its Faculty Grants Program has provided travel support that has enabled Cobb to serve for seven years as chair of the Joint Committee on Undergraduate Statistics of the Mathematical Association of America and the American Statistical Association. Cobb currently serves on the National Research Council's Committee on Applied and Theoretical Statistics, making Mount Holyoke the first liberal arts college in the country represented on that committee. This summer he will deliver the keynote address when 120 high school and college teachers meet to grade the advanced placement exam in statistics.
"A lot of the reform effort has focused on technology," says Cobb, who agrees that computers have had a big impact, "but in large part the computer is important because it frees us to think less about details and more about meaning." In his published articles and conference presentations, Cobb puts his emphasis less on technology, more on curriculum - how the big ideas fit together--and on "what the experience of learning statistics feels like."
Acquiring quantitative literacy, "the ability to handle numbers and their context effectively," is more important during the information age than ever before, Cobb says. If his students have fun in the process, and statistics' undeserved reputation as a colorless pursuit is invalidated, that's fine with him too.
Copyright © 1999 Mount Holyoke College. This page created and maintained by the Office of Communications. Last modified on April 9, 1999.