Arnold Kling  

Should Empirics Determine the Curriculum?

Single Payer in the U.S.?... Electoral Deconcentration...

Edward Glaeser writes,

thorough general education requirement on the scientific approach to society would require two courses. First, students should take a course that teaches the crafting of rigorous hypotheses. ...

Second, students should take a class on evidence and statistical inference. This could either be pure statistics or empirical tools taught through the lens of a particular topic. Decent citizenship of the world is incompatible with statistical ignorance.

(emphasis added; thanks to Tyler Cowen's pointer)

I am a big fan, as is Steven Pinker, of education in statistics. Thanks to computers, which store lots of data to feed statistical models and which use statistical methods to solve many important classes of problems, statistics is growing in importance. If you think that calculus is anywhere near as important to know as statistics, I believe you are at least a generation out of date.

Speaking of statistics, if I were looking at Harvard's curriculum decision problem, I would focus on data. Suppose that you had a ranking for the students in the most recent graduating classes that told you how well they turned out in terms of critical thinking skills. What is the correlation between the courses that a student took and his or her ranking? If there is no correlation, then maybe curriculum does not matter. If there is a strong correlation, then maybe the courses that are correlated with high thinking skills belong in the core curriculum.

If Glaeser and Harvard are serious about the value of empiricism in the curriculum, then they should base the curriculum on some empirical data. I know that correlation is not causation, but then, neither is guesswork. Eh, Professor Mankiw?

Comments and Sharing

CATEGORIES: Economic Education

COMMENTS (4 to date)
RogerM writes:

I agree with Arnold on the importance of statistics, but I'm curious as to how he would rate data mining techniques, such as neural networks, trees and other machine learning methods, compared to statistics.

Raphael Apollo writes:

You need basic calculus (integration, sequences, etc.) to do anything above rudimentary statistics. So if you want to do anything with statistics yourself (rather than just understand when people use the words 'significant at the x% level) you will need basic caculus.

Statistics may also be harder to teach than calculus. To teach stats well, you need a very clear understanding of the link between the mathematical formalism and the real-world situation you are modelling. Not very many people can do this well (fewer than can teach calculus as a pure math class). Kahneman and Tversky showed that even mathematical psychologists fail on fairly simple problems of applying statistics (see "Judgment under Uncertainty"), so I think it would be difficult to find enough people to teach stats well to make a class worthwhile.

Do you know of any data on the relative efficacy of different college courses? My experience is that academics (nearly as much as students) think shoddily about this issue. I commonly here people say "if I hadn't been forced to take this class then I never would have read X, which I loved reading". But if you hadn't been forced to take that class you would've have read something else which might have been even better than X!

RogerM writes:

For stats teachers, I highly recommend the Electronic Textbook at this web site:

Don Robertson writes:

Yes. Statistical analysis applied to everything is very important. I took statistics in 1970. It was one of my favorite courses. For the uninitiated, statistical analysis can be an eye-opener concerning the analysis of the efficacy of every paradigm.

It was statistics that provided me the breakthrough that permitted the formulation of Robertson's Law:

No matter the problem, no matter the solution, if it's an exclusively empirical solution applied to the real world, the resulting biproducts of the processes of empirical reason will give rise to problems tenfold that of the original problem.

It is also statistics that will show those who have no health insurance actually live longer than those who do. (There's definitely something nefarious implied there.)

It is also statistics that has demonstrated cell phone users do indeed incur a higher risk of brain cancer. (Don't worry. I'll have long since passed away before I have to look at all those bloated heads walking up and down the street asking, Is that my phone or yours?)

It is statistics that will eventually demonstrate that while the lifespan of the average American has increased over the years, it has reached its peak, and will begin to decline, mostly because statistical analysis has already shown greater indebtedness, longer working hours and a greatly decreasing standard of living due to increasing competition for the illusive ideals Americans seek, freedom, wealth, and respect.

What goes around, America, is generally what comes around.

As a postscript I'd like to add:

The Boston area is the intellectual colostomy bag of an intellectually ill nation.

I know. I went to school in Boston and lived there for twenty years.

You can feel sorry for the horrific depravation of all those Harvard kids. Harvard professors are so high on ego, they walk on clouds to keep from stepping in the ubiquitous dog poop in Cambridge.

Don Robertson, The American Philosopher
Limestone, Maine

An Illustrated Philosophy Primer for Young Readers
Precious Life - Empirical Knowledge
The Grand Unifying Theory & The Theory of Time
Art Auctions:

Comments for this entry have been closed
Return to top