John, I dunno what's in the flickering mind, but these are extremely worthwhile questions. If, say, the project works with 3rd graders who are in a school with even a nominally effective program (sans OLPC), you'd expect to see pretty huge gains in writing and spelling (basically from not-writing to writing), and depending on the teacher, gains in creativity simply based on mastery of and comfort with school procedures and that sort of thing.

(If a 14-year-old had a 50% improvement in writing, however that might be measured, it would be a more substantial achievement.

One of the reasons that donor specify _independent_ evaluations is that the evaluator then has an interest in actually measuring performance rather than describing it, and--if there's enough $ available--setting up an evaluation that generates experimental results of the kind that Wayan and you are describing. Until we see some independent and authoritative evaluations, it's difficult to imagine advising a Ministry to do anything other than keep monitoring the situation and wait until there are some results that count.

AND, pursuant to another post, this is important precisely because $ is limited; if you're going to sell a cadillac item (which all ICT is in schools) to a poor-country government, it's shameful to do so without experimental evidence and full cost accounting.

@John Smith:
"Are there other factors involved? For example are people just "working harder / longer"? "

Actually, I think this is the most likely mechanism.

All the improved skills are limited by sheer practise.

To get good reading skills a child has to read a lot. For writing skills she has to write. Mathematics are only learned by doing it. As far as I understood the situation in Peru, there was a lack of reading and math materials. And without interesting texts, children stop reading. Without new math questions and exercises, there is only root learning (just memorize all the questions and answers).

The XO solves the question of these source materials. Correcting on a word processor is MUCH easier than on paper. So we can expect more time spend actually formulating and writing than rewriting and correcting.

So the laptops are doing exactly what is needed: Getting children to spend time reading and writing and forcing them to actually solve math questions instead of regurgitating the learned answers.

No "digital magic" needed. Anything that gets the children to spend more time practising these skills will do the trick.

Whether the learned skills will be worth more than the cost of the program is another question. One which will be extremely difficult to answer. What is the value of a litterate (text and math) work-force?

Winter

Although, student:teacher ratio, per pupil spending etc are important factors, they can not solve all the problems by themselves. Look at rural versus city schools for student:teacher ratio, and performance in relation to school-district incomes that is far from linear. Education performance does NOT depend on any SINGLE factor. If it was the “solution” should have been found years now, but is clearly not.

The big “secrete” is that ANY single action is destined to have ON AVERAGE pretty comparable outcomes! Simply because some students learn better by more teacher attention/guidance (smaller classes), others understand better by reading alone (libraries/books), others by interactively implement something (experiments, computers, discussion groups), others if they are “forced” to practice (authoritarian teaching) others if they are “moved” to practice (engaging teaching) etc. The idea to play one against the other is simply wrong. Any change that is done at the EXPENSE of the others is likely to fail. Any change that is done ON THE TOP will have some effects, but never the “advertised”. The perfect and ubiquitous teaching approach does not exist simply because its subjects are VERY diverse. For these reasons the idea that “something else would have worked better, IF implemented” is totally irrelevant.

What is important is to find out initially on average and ultimately at the individual level, in which setting something may have a better cost/benefit ratio than the next thing. For individual assessments we clearly have some time to go but for the average we (can) have implementations that we can then assess with the local “standardized” tests and have a cost/benefit score. However, even for the best of them we should not extrapolate from one setting to another. What “works better on average” in Peru will not have the same result in the US and what works in rural areas of the US will not have the same result in inner city districts. What is an education improvement in Nigeria may be irrelevant for UK and vice versa. The “education goals” in two societies can be dramatically different or even “opposing” and a positive outcome in one may not be considered as such in another. So until we find out which pre-existing parameters where affected by a specific implementation, “’transplantation” is destine to fail. Unless we know exactly WHY something happened, we can neither reproduce it nor avoid it.

So let’s find out how the SPECIFIC implementation works and try to figure out the parameters responsible for success or failure. Based on that we can then identify similar settings and act accordingly. Shooting down an approach a priori, is as mistaken as assuming that will solve all the problems magically.