As shown by Descartes and the great national debate on the School organized some ten years ago by Claude Thélot, who had collected no less than 50,000 contributions, " common sense is, with opinions on the school, the best thing in the world shared . "
Behalf of their respective visions of the school, many ministers of Education have themselves ceased, for 30 years, to shake Education.
For one, the key is in learning to read. For another, it is the individual support. For a third, this is the motivation of teachers, the number of students per class, the pace of school or even the amount of fat available on the mammoth.
Anything goes, so. But what is the common point between these different "visions"? Is that true or false, they are unfounded. They are not based on scientific facts but on a priori prejudices, as the term used by Descartes.
" Considering how many there may be various opinions regarding the same matter, which are supported by learned people, but there can never be more than one that is true, I réputais almost everything wrong for n 'that was likely . "
" Therefore , Descartes said, speaking to our ministers with almost 400 years early, I can not approve any such messy and restless moods, which, being called neither by birth nor by fortune in the use public affairs, do not leave it to always some new idea reformation . "
The experimental method invented by Descartes in the Discourse on Method is not a scientific theory, but a way to decide between the theories (by experience) and to advance in scientific discovery (by dividing a problem appearance in several complex problems simpler).
From Descartes, scientific progress is continuous regardless of the pace of discovery. Even the recognition that an idea is false is often a useful step. So if I take the case of the digital equipment of schools, there is no serious study on the benefits that students may withdraw from this equipment.Is, therefore, this equipment is unnecessary, in which case the budget can be made available for other investments more attractive, whether it is useful and it is then important to say why it is useful to identify its context of use, usage optimal material where it should be used, etc..
Until now, the method described by Descartes remained virtually unenforceable in teaching for two main reasons:
The complexity of validation of the theory : the opposite of the hard sciences where experiments were often able to quickly determine the validity of a theory, validate a theory on the teaching evaluations it has until now heavy, expensive, time and complex. Consequently, these assessments can not be made in small numbers and could not really influence policy statement, the duration of the evaluation is generally much greater than the longevity of the Minister.
The vagueness of the criteria : where, in the exact sciences, the criteria are measurable and usually accessible to experiment, to observe the data are complex to define in the case of education.How to judge with certainty the level of a student? the quality of a teacher? In a method? These terms have they same meaning? And if we can provide an initial response - a response or imperfect - to the previous questions, how to quantitatively observe that "the child is good about yourself at school," allowing it "to express his creativity ", as advocated by some currents?
But it turns out that two recent scientific developments will allow to apply the experimental method in school.
Although intimately related to the ongoing digital revolution, they have never, to my knowledge, been linked. The progress they make it possible to consider are immense. School pedagogy, almost static since the days of Aristotle, who simultaneously invented the lecture, small classes and resource materials, will be able to follow a path of continuous improvement, continuous and observable, comparable to that science has followed Since writing the Discourse on Method.
The first is the random method of assessment developed by a French researcher, Esther Duflo.
The second is the advent of "big data" ie the possibility of using large datasets of information collected on students. These data are now exclusively used for publicity purposes by companies such as Google or Facebook. But they can also be used decisively to improve teaching.
The evaluation method random
The random method is based on assessments made on small groups (a few hundred students) whose characteristics are identical to begin with. One of these small groups adopt a "new process" (eg it will use a method of reading a promising new). Then compared, on specified criteria, the performance of these groups (reading speed, comprehension, etc..).
In recent years, Esther Duflo, random method was used to evaluate the effects of policies against poverty with remarkable results. The techniques it has developed can be applied to school.
With means very low, the random method gave, in India, more information on the uses of digital technology in all developed countries.
" When comparing schools with computer with others, we find that students do better in the first. But is it really due to computers? The difference in results can not be explained does not rather by the fact that it is usually in urban schools, already relatively advantaged, which are equipped with computers?
An experiment to isolate the real impact of computers. In an Indian city, all schools were equipped, but generally, they stayed in their box, because teachers were not trained or do not have the appropriate software. We selected a group of schools randomly and have made available a software trainer and mathematics, so that all children benefited from a few hours a week of computing. A year later, we compared the performance. Schools that had been collected using computers better results in mathematics.
Many experiments can be conducted in the field of education, to test the impact of various factors high. One study for example showed that hiring a person in charge of tutoring has an impact equivalent to the purchase of computers, while being much cheaper . "
Current assessment processes are cumbersome, costly, time consuming and can not influence education policy. By replacing, or supplementing, by a large number of micro-randomized evaluations inexpensive, quick to make, well targeted and methodically determined, we can have a remarkably effective tool, to obtain the first results in few months, to influence policies and spending more public money.
With appropriate assessment tools, school can become a process of "optimized under constraints."As the process of evolution is constantly improving the performance of living beings, all initiatives can be evaluated, and selected the best and widespread.
The key to success of this policy guide is the speed of execution (of the order of weeks) and the amount of assessments in place - to borrow a term used in computing, in agility.
This process could be established by the Assessment Division of Education, under the control potential of a large body of Digital, consisting primarily of engineers on the model of a Mining or Bridges and would aim to complete the digital reform in the French administration and school.
The use of "Big Data"
While for now the main digital applications used in schools refers to those uses, the assessment in schools is, in my opinion, the most important and least understood of the ongoing digital revolution.
In the last decade, computers, interactive whiteboards, digital workspaces or other digital binders have returned to school, without really knowing, moreover, in the absence of any method, what is their impact on the level of students. Moreover, these technologies are designed to improve learning almost unique and not to evaluate it.
But from the moment you gather enough data (known here in petabytes, that is to say, one million gigabytes), statistics helps to identify the "laws" general causes. In a polemical article, but founder, entitled "The End of theories: the deluge of data makes the scientific method obsolete", Chris Anderson has shown how these data can be used not only to conquer advertising markets, as did Google but perhaps also to get rid of that was previously the aim of all science, namely the model (usually called "scientific theory").
Google does not attempt to model the behavior of users or to "understand" the content of Web pages, but uses statistical tools to draw correlations (patterns) from which advertising banners are proposed. No model, the correlation is sufficient.
What relationship with the school?
Current assessments established by the National Education are resented because exceptional, complex and cumbersome and generate very little data, in quantity, compared to data capture mechanisms set up by Google or Facebook, which are them voluntarily - and often wildly - used by students from their computer, digital tablet or mobile phone.
It is time to use these data in a general interest objective and integrate these technologies massively in education, as a minimum to evaluate students.
In the context of randomized evaluations that I have already mentioned, these devices make digital mobile student assessment and school policies much more quickly, easily and economically to the point that it becomes possible today, by multiplying the micro-assessments, to really run the whole school policy.
Then , the use of these materials must be systematized in order to obtain as much data to evaluate students and to collect these data evaluation. To the extent reasonable, most evaluations, assignments, questionnaires should be requested for students in digital form, in a format allowing statistical tools of "big data" being used.
Tools for querying students digitally (eg in the form of multiple choice questions) already exist but are hardly used in higher education, and piecemeal. From the moment the volume of data is sufficient, where they are consistently implemented throughout schools, these tools will not only determine the actual level of students with a depth much greater than the current testing techniques, but also determine amounts of educational laws, or rather, correlations, and new non-contestable.
It should be noted, because the objection will obviously be made, that "Big data" is not necessarily "big brother". All this data is personal but can - and probably most should - remain anonymous. The question of the identity of the student who answered matter to the statistical tool, what is important is to know all its data on its entire school career.
It will become possible to identify the "best teachers" (those who make the most progress the average level of their students and not those, probably starting with the best students, the best performers bachelor's degree) and s' learn from their methods to train other teachers. The phrase "good teacher" is perhaps not a big meaning in absolute terms, but it is possible also to determine, through correlation, what is the teacher who best fits a student determined, what is the students who learn better by reading than listening, what is the one who is able to enjoy the more personalized support ...
In short, the "big data" will give us very valuable information to improve teachers, students, teaching methods, referral mechanisms, etc.. It is impossible at present to predict how big will this progress.
It is just possible to predict the country to gain the most advantage of the digital revolution will be the one best able to use these new tools for data capture and evaluation for, says Descartes:
" Those who walk very slowly can proceed much further, they always follow the right path, as do those who run away from it "
This is a new age of the school opens.
