Sign In

Communications of the ACM

[email protected]

Computers in Education Should Target Test Scores


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
Geeky Ventures Founder Greg Linden
Does it help to bring computers into the classroom?
 
So far, there seems to be little evidence that it does.  The New York Times, for example, recently reported that "schools are spending billions on technology, even as they cut budgets and lay off teachers, with little proof that this approach is improving basic learning."
 
In a second article, the New York Times expanded on that, pointing to a 2009 federal study that showed that almost all major math tutorial programs "did not have statistically significant effects on test scores" in independent testing.

Little proof of improvements, that is the problem.  There is a lack of convincing data showing that integrating computers into schools improves the schools.  Often, the computers yield little in the way of measurable results, and proponents are forced to suggest that there are unmeasurable benefits, such as breadth of skills, being achieved.
 
The question is why.  As computer scientists, why do we think that unmeasurable gains are worthwhile?  Why not build software that shows measurable benefits?
 
Computers excel at boring, repetitive tasks like testing and grading.  Even the easiest step of simply using computers to make it easier for teachers to administer practice tests likely would increase test scores, as repeated practice testing has been shown to significantly increase test scores.  
 
Beyond the trivial, computers can provide personalized coaching on practice tests.  When students err on a problem, computers can explain the way to reach the solution.  When students make similar mistakes on a group of problems, computers can recognize a misunderstanding of an underlying concept, explain it, and offer additional practice.  Computers offer an opportunity for fine-grained, individualized, personalized coaching.
 
Teachers benefit from being freed from some of the tasks they find unpleasant.  Grading is not fun.  Making lots of new practice tests is not fun.  Explaining to each student why each individual answer is wrong is time consuming and not fun.  When computers take over these tasks, it frees up teacher's time for collaborative problem solving and other much more creative instruction. 
 
Computers are good at these unpleasant tasks like practice tests and coaching based on practice tests. We know practice tests boost test scores.  We know teachers don't enjoy doing practice tests.  So why not have computers do it?
 
Computers in education have failed because they cannot prove their value.  Rather than attack the metric, we should embrace it.  Computers in the classroom should yield measurable improvements to test scores.

Comments


Anonymous

I think the Khan Academy method is probably an effective use of computers in education. Not just the Khan lectures, but testing and reviewing all points in an area until the student can answer ALL points correctly, THEN proceeding to the next section, and not before.


Anonymous

Oh, yes, God bless the holy fricken' test score! The reason districts are spending money on computers is because they are required to do so because of the stinking tests!! All of them are computer based and without moderate to high end machines they cannot take the tests! I teach and I HATE those tests. Not because of the test itself, but because of the ridiculous level of importance placed on them. Therein lies the real sin, the true detriment to public education - High stakes testing! - Dante T.


Greg Linden

Hi, Dante. That's a good point, that there likely are problems with the current tests as a metric. But the point of this article is not to defend standardized tests. The point is that, whatever metric we have, computers in the classroom better be able to show measurable improvements in it.


Displaying all 3 comments