In my latest Globe and Mail piece, I summarized a study by Sa Bui, Steven G. Craig, and Scott Imberman on the effectiveness of gifted education. The authors look at students in a large urban American school district who were evaluated for gifted programming in grade five. They ask: Who does better on the grade 6 and 7 standardized tests, the students who just made it into the gifted program, or the ones who fell just below the gifted threshold?
The authors have an impressive amount of data: standardized grade 5 test scores for 5,500 students either side of the gifted cut-off point before the gifted programming begins, and the same students’ grade 6 test scores, one year later. They have similar information for 2,600 grade 7 students.
To test the effectiveness of gifted education, they measure how far each student was away from the gifted cut-off. The authors then estimate grade 6 and grade 7 standardized test scores as a function of distance from the gifted-eligibility threshold and some other controls. A "regression discontinuity" analysis is used to figure out if those who make it into the gifted program experience a jump in educational outcomes.
It's easier to explain with a picture than with words:
For reading and language - the green line and the red line - there is no jump in the test results at the gifted threshold. There's a kink, but not a shift. From this, the authors conclude that “students exposed to gifted-talent curriculum for the entirety of 6th grade plus half of 7th grade exhibit no significant improvement in achievement.” This is despite the fact that the students in the gifted-talented program have more educational resources coming their way – they are in classes with higher performing peers, are more likely to be placed in advanced classes, and more likely to attend a gifted-talented magnet school.
The lack of improvement in reading and language can be explained in a number of ways. The less able "gifted" students might feel discouraged by being in the bottom of the class and thus put less effort into school. The standardized test scores shown in the figure above might be measuring innate ability rather than what is taught in school. Reading and language scores may be more influenced by home environment than what is taught in the classroom.
But what is really striking is the suggestion that math results actually *fall* for those identified as gifted.
There are three possible explanations for the decline in math results. The first is explained here: about 5% of the time, random variation will generate a result that is statistically significant at the 95% confidence level.
The second possible explanation is that gifted programs teach math badly. This is not implausible.
Jump Math is a new approach to math education developed by John Mighton. Mounting evidence finds that using the Jump Math approach leads to large improvements in students' achievement levels and mathematics capabilities. The key to the Jump Math approach is starting with simple problems and building intuition and confidence through repetition; breaking everything up into simple micro-steps. To the extent that gifted math education is even further from the Jump Math approach than regular math education - less repetition, fewer simple, intuition- and confidence-building problems - it may be leading to even worse outcomes.
A third explanation of that downward jump in the blue line - the fall in math results - is that the students at the bottom of the gifted group had lower levels of innate mathematical ability than students at the top of the non-gifted group. Bui, Craig and Imberman find that, once they add in controls for lagged achievement levels, the fall in mathematics achievement levels is no longer statistically significant, which is compatible with the differential-ability story.
It's easy to think of reasons why students who are talented at math might be less likely to be identified as gifted than students who are talented in reading or language. Math scores may receive less weight than reading/language scores when students are identified as gifted. There may be students - and this is a huge issue in Canadian school districts with large immigrant populations - who are gifted in math, but speak English as a second language, so cannot be identified as gifted.
(As I understand it, in my school district there is actually a rule that prevents students from being evaluated for gifted status until they have been in English-language education for three years. This means, for example, that the best math student in the local high school might not be able to get a space in the gifted class. Unfortunately I cannot find this written down anywhere, so I may be mistaken.)
I could go on, here is my take-away: There are great wads of resources thrown at gifted education, and little evidence of positive results for border-line gifted students. [Update: "great wads" might be an exaggeration.] My own interpretation: gifted programs aren't producing results for two reasons. The first is that border-line gifted students get discouraged. The second is that students who are talented in specific subject areas, e.g. math, are not being identified as gifted.
There's a simple solution - more flexible gifted programming. Allow students to be identified as math-gifted or language-gifted or arts-gifted or music-gifted, and develop specific programming for those needs. The best mathematician in the school might be lousy at languages, and vice versa. More flexible gifted programming would allow, for example, new Canadians to access gifted math programs while, at the same time, receiving remedial English instruction.
And forget about classes like gifted physics 11 and gifted calculus 12. Few students take these courses because it's generally harder to get high grades in a gifted class, and good grade 12 grades pay off in terms of university admission and scholarship offers. Plus there is so little difference between the ability distribution of students in Gifted Physics 12 and regular Physics 12 that it is hard to imagine that there are significant advantages to differentiating between the two groups. It's a waste of resources.
But change will not be easy. As my mentor Julian Le Grand used to say: Hell hath no fury like the middle class in defence of its privileges.