Well, why are you asking that question?

Yes, basic mental arithmetic, like spelling, is an important skill that you can really only learn by practice. The Moodle quiz can be set up to let students do as much practice as they want. The question is, how you give them enough of an incentive so that they actually do it, and do it without just typing all the answers into Google.

Well, to rule out Google, you could use quite a stringent time limit. After some practice, doing it in your head is much faster than typing it into Google.

Another option is to use Moodle 90% of the time, but on occasion to the test in class, on paper and pencil. Than student will quickly work out that the only way to do well in the in-class tests is to have done enough practice on the other tests.

Also, try to make the Moodle tests like a fun game, not like a boring serious test. After all, some people pay good money for 'brain training' 'games' for games consoles that are just mental arithmetic. That just shows that exercising your brain can and should be fun. So perhaps don't use Moodle quiz at all. Find some free Flash game that tests basic arithmetic (and which does not have undesirable adverts) and use that instead.

On a related note, I read an interesting paper recently. The authors had constructed a computer program to teach long-hand addition and subtraction like 6523 - 1280. What made this clever was that the system tried to analyse the mistakes the student made, and so diagnose what systematic mistakes they were making in their calculations. For example, perhaps all their mistakes could be explained because they did something weird every time they tried to carry one. The paper found that the majority of student errors were systematic, rather than being careless mistakes. The system could often predict what wrong answer the student would give to a problem in advance.

On the one hand, this struck me as terribly clever. On the other hand it completely misses the point about what maths teaching should be. If students are taught long-hand subtraction as an algorithm to follow, with no justification, then of course they can make all sorts of strange errors if they get one step wrong. On the other hand, if you teach them to really understand what they are doing - what subtraction means - then they should not be able to have weird bugs in their subtraction algorithms.

As it happens, this paper was from 1978, so thanks to calculators and Google, that sort of maths teaching is now an anachronism, but the computing was, if anything, ahead of its time. The paper is Brown and Burton "Diagnostic Models for Procedural Bugs in Basic Mathematical Skills" Cognitive Science Vol 2, 1978. Sadly only the abstract is available online. I had to go to the OU library and take a copy on dead trees.