New Way of Testing Pupils

13. February, 2011

Remember the horror of tests at school? The attempts to prepare, the insanely short hours of the test itself, the dreading wait for the results.

In Switzerland, pupils in the 8th grade can take a “Stellwerk Test” which works a little bit different. It’s a web based test. Instead of presenting all pupils the same questions, everyone gets different ones. Depending on the answers (correct or wrong), the program will select a more simple or more difficult question. After a while, this selection process will level out. At that point, it’s possible to calculate the level of understanding that each individual pupil has on the topic.

Unlike traditional tests, it doesn’t matter (much) if you can’t answer a question. Also, the leveling out is individual. Some pupils can finish the test in 30 minutes, others need 2 hours. That doesn’t mean the “slow” pupils are dumber; their understanding is just more “uneven”. That means the test is more fair than the traditional tests. Teachers also don’t have to come up with genuine questions every year (or make sure the questions can be kept confidential if they always use the same ones). Stealing the tests in advance doesn’t get you an advantage.

Makes me wonder when this kind of test can be used for more than math and physics. Not every school can afford a super computer like the one necessary to run Watson. Not yet.

Hidden JUnit features: @Rules

8. October, 2010

@Rules seem a better solution than @RunWith to do some special work before/after a test. The release notes mention a couple of ideas:

  • Notification on tests
  • Setting up or tearing down resources, especially when they are used in multiple test classes
  • Special checks performed after every test, possibly causing a test to fail.
  • Making information about the test available inside the test

Related articles:

%d bloggers like this: