Nurturing Whiz Kids

...and this times the Planck constant...

In this day and age, brains not brawn will determine the ability of hitherto highly affluent Western countries to secure their continued prosperity. It is no surprise, then, that they are fretting about their domestic level of what connoisseurs call ‘educational productivity’.

One focus of attention has been the so-called high-achieving students. After all, they often go on to create considerable wealth—and not only for themselves. Microsoft’s Bill Gates, Google’s Sergey Brin and Facebook’s Mark Zuckerberg were all high-achieving high-school students. And yet, in the US even the best-performing US states (and even high-performing demographic groups) have a lower percentage of high achievers in maths than most other OECD countries.

How much of this potential is there? And where is it being nurtured? CESifo Research Network member Glenn Ellison and his colleague Ashley Swanson have just released a CESifo Working Paper that sheds some light on these questions by delving into the heterogeneity in high math achievement across schools in the US. Their findings will surely provide insights into improvements that educational reforms could bring.

They start by defining what a “high-achieving” maths students is: one who would score 800 on the maths portion of the SAT test, the standardised test for college admissions in the US (800 is the highest possible score on the maths SAT and is achieved by around 1% of SAT-takers).

The primary data for their study is the Mathematical Association of America’s AMC 12 contest, which consists of a 25-question multiple-choice test on precalculus topics given annually to over 100,00 US students at about 3,000 high schools, and is explicitly designed to distinguish among students at very high achievement levels.

After controlling for demographic differences, they find that there is a great deal of variation in the number of high-achieving maths students produced by schools with similar demographics. Less surprisingly, they also find that demographics are very strong predictors of high maths achievement: areas where there are many highly educated parents and schools with many Asian-American students are much more likely to produce high-achieving students.

One fairly startling fact that they unearthed is that while income is positively correlated with high achievement in the full national sample, it is negatively correlated in the sample of relatively high-quality schools which offer the AMC tests; in other words, there are fewer high-achieving maths students in more-affluent areas (and fewer with very high SAT scores as well).

Another striking finding is that while there are many schools (38%) that produce far fewer high-achievers than the average, there is a ‘thick tail’ of schools, around 2%, that produce high-scoring students at more than five times the average rate. Sadly, their estimates also show that many schools are extremely unlikely to produce high-achieving girls. Still, the 99th percentile high school in their sample produces high-scoring girls at more than ten times the rate of an average school with comparable demographics. That might provide pointers for reforms elsewhere.

The authors check whether self-selection of high-ability students into certain public schools might explain the ‘thick upper tail’, and find that such is not the case. The reason for this, they suggest, may be that almost all schools see it as their responsibility to provide English and Maths courses to do well on the SAT, whereas there is much less uniformity in whether schools encourage gifted students to develop the more advanced problem-solving skills necessary to do well on the AMC.

They believe that the high-achieving maths students today in the US’s high schools may be just a fraction of the number of students who have the potential to reach such levels. The number of high-achieving students, they conclude in a hopeful note, would increase substantially if low-achieving schools could be brought up to average. The way to do it? Just look in detail at what those schools in the ‘upper thick tail’ are doing right, and emulate it.

Glenn Ellison and Ashley Swanson, Heterogeneity in High Math Achievement Across Schools: Evidence from the American Mathematics Competition, CESifo Working Paper No. 3903

• Other CESifo Working papers by Glenn Ellison
• Other CESifo Working Papers dealing with education
• Other CESifo Working Papers dealing with schools