What Works In Education: Scientific Evidence Gets Ignored

nbauman writes "According to Gina Kolata in the New York Times, The Institute of Education Sciences in the Department of Education Sciences, U.S. Department of Education, has supported 175 randomized controlled studies, like the studies used in medicine, to find out what works and doesn't work, which are reported in the What Works Clearinghouse. Surprisingly, the choice of instructional materials — textbooks, curriculum guides, homework, quizzes — can affect achievement as much as teachers; poor materials have as much effect as a bad teacher, and good materials can offset a bad teacher's deficiencies. One popular math textbook was superior to 3 competitors. A popular computer-assisted math program had no benefit. Most educators, including principals and superintendents, don't know the data exists. 42% of school districts had never heard of the clearinghouse. Up to 90% of programs that seemed promising in small studies had no effect or made achievement scores worse. For example a program to increase 7th-grade math teachers' understanding of math increased their understanding but had no effect on student achievement. Upward Bound had no effect."

Comments

No shocker there

[hedwards]

I've yet to see a competently written math book. Most of them are written by and for people with PhDs in mathematics. They'll show one example, fail miserably to explain what they did in any clear way, then later they will refer back to it as what they did in example 3. And the student is expected to be able to figure out what they did. Sure, given sufficient time, a student could reverse engineer the problem, but it's also trendy for teachers to hand out way too many problems as homework, without permitting the students time to understand.

I remember when I was in middle school and high school, the schools were using "integrated math." Which is to say we didn't have algebra, geometry or trig, we had all of them at once and we would start over again the next year. The problem is that just as we were beginning to grasp one of them, we'd move onto the next subject, and the next year, we'd have to start over as we hadn't mastered the material the last time we saw it.

Re:No shocker there

[Anonymous Coward]

I've yet to see a competently written math book. Most of them are written by and for people with PhDs in mathematics.

Sounds about right - as a programmer, I've always been appalled by how math is taught. If we taught programming the way we taught math, every program would be unmaintainable. Think about it:

- One letter identifiers for everything. Algebra teaches you to always use x, y, z for variable names. Calculus teaches you to do it for function names. If you run out of those, use greek letters, or just start making up symbols.
- Everything is named after who discovered it, not what it does. Pythagoras's theorem, Newton's method, L'hopital's theorem, Cartesian co-ordinates, Euler's number...
- Formulas are always crammed into a single line, without being spaced out. And without any in-line comments. You're forced to try and understand the entire formula in one shot, rather than piece by piece.

MatLab is a perfect example of why mathematicians shouldn't program. You can look at the source to certain functions (like calculating Euler's number) to see this in action.

Re:No shocker there

[MobyDisk]

Obligatory XKCD:
http://xkcd.com/1201/
For those not sure why this is appropriate: the hypothetical teacher explains the rule, but not how to apply the rule or what use it is. We have all had professors like this. It is why teaching is a whole separate skill from the trade itself.

Richard Feynman on textbooks

[MobyDisk]

Richard Feynman's story on textbooks was eye-opening: http://www.textbookleague.org/103feyn.htm
(Thanks BobTree)

Variable students

[gurps_npc]

Part of the problem is that what works for one student doesn't always work for another.

You can't expect a child with dyslexia to learn from the same program that works for an excellent reader. Less serious learning issues have similar effect.

One thing I never understood is why we don't have a public boarding school option for those kids whose parents clearly are the problem.

If your parents are homeless, drug addicts, or convicted felons, you have about a 50% drop out rate. If we just offered them public boarding schools, we could save those kids - at far less cost over the long term than what those drop outs will end up costing the government.

Boarding schools can go for as low as $25k / year, vs regular schools at half that while a year in prison costs over $100k If just save just one out of 8 of those kids from a life of prison, we come out ahead.

Re:So much does not work

[JWW]

But there is a wonderfully effective way to use computers in education. You look at student's marks. You then look at the pattern of the marks as the student's pass through various teachers.

What?!!! That would allow you to actually truly measure teacher performance and effectiveness. It would make bad teachers absolutely impossible to miss.

WE CAN'T HAVE THAT NOW CAN WE???

Re:D.A.R.E has no benefit

[alexander_686]

Maybe.

There was a study about 10 years ago that showed zero correlation between teacher pay, teacher effectiveness, and academic results between states.

Probably what is more important is teacher requirement, training, and management. If you increased teacher pay today without changing the above you would increasing pay to those who are already teachers or are likely to become teachers.

I have issues with how teacher’s pay is structured. The initial pay is low and most of the benefits are at the backend so it encourages marginal teachers to become entrenched and discourages middle aged people from making a career switch into the profession. (I think there is a rich vein of potential people who hold masters in math, science, or engineering who would make great teachers but don’t want to deal with the initial low pay and would not qualify for some of the bigger retirement packages.)