Every month we read news about the shortcomings and failures of US education. If you believe the press, the elementary school in America is not able to teach students even basic knowledge, the knowledge given by the secondary school is clearly not enough to enter college, but the students who managed to hold out until graduation are absolutely helpless behind its walls. But some very interesting statistics have recently been published, showing that in at least one particular aspect, this opinion is very far from the truth. Despite the well-known problems of the American secondary education system, graduates of American colleges specializing in computer science turned out to be excellently developed and very competitive specialists in comparison with their foreign competitors.
The study, conducted by an international team of researchers, compared US college graduates with high school graduates from the three largest countries to which the US outsources software development: China, India, and Russia. These three countries are famous for their first-class programmers and winners of international competitions, their reputation is impeccable, and the successful actions of Russian and Chinese hackers are constantly reflected in the news. In addition, there are large domestic software markets in China and India, served by a large number of local specialists. All of these factors make programmers from these three countries a very relevant benchmark against which to compare American graduates. At the same time, many students from these countries come to study in the United States.
The study does not claim to be comprehensive and, in particular, does not compare the results of Americans with the results of graduates from other developed liberal democratic countries like the United States. So it cannot be said that the results obtained in it can be generalized in favor of the unequivocal success and total dominance of the American education system throughout the world. But the countries considered in the study were analyzed very deeply and carefully. In these three countries, the researchers selected a total of 85 different educational institutions by random sampling among the "elite" and "simple" universities involved in computer science. With each of these universities, the researchers agreed to conduct a voluntary two-hour exam among final year students specializing in programming. The exam was prepared by ETS specialists, , consisted of 66 multiple-choice questions, and was conducted in the local language. The questions included discrete data structures, algorithms and estimates of their complexity, problems of information storage and transmission, general programming and program design problems. Tasks were not tied to any particular programming language and were written in abstract pseudo-code (similar to how Donald Knuth does in his work "The Art of Programming"). A total of 6847 Americans, 678 Chinese, 364 Indians and 551 Russians took part in the study.
According to the results of the exam, the results of Americans turned out to be much better than the results of graduates from other countries. Despite the fact that when compared with overseas peers, American students enter college with markedly poorer performance in mathematics and physics, by the time they graduate from college, they consistently score significantly better test scores. Of course, we are talking about purely statistical differences - the results of students depend not only on the college, but also on individual abilities, so the results of different graduates of even the same college can be fundamentally different and an outstanding graduate of a βbadβ college can be much better than a bad graduate of an βeliteβ college. Β» university. However, on average Americans scored 0.76 standard deviations better than Russians, Indians or Chinese. This gap turns out to be even greater if we separate the graduates of βeliteβ and βordinaryβ universities and compare them not in one pile, but separately - elite Russian universities with elite US colleges, ordinary Russian universities with ordinary American colleges. Graduates of βeliteβ educational institutions, as expected, showed much better results on average than graduates of βordinaryβ schools, and against the background of a smaller spread of marks among different students, the differences between students from different countries became even more pronounced. Actual results top universities in Russia, China and India turned out to be about the same as the results Conventional American colleges. Elite American schools turned out to be on average as much better than elite Russian schools as Russian elite universities are on average better than conventional βfence-buildingβ colleges. It is also interesting that the study did not reveal statistically significant differences between the results of university graduates from Russia, India and China.
Figure 1. Average test results, normalized to standard deviation, for students from different countries and different groups of universities
The researchers tried to take into account and exclude possible systematic causes of such differences. For example, one of the tested hypotheses was that the best results of American universities are simply due to the fact that the best foreign students come to study in the United States, while only worse guys remain in their home country. However, the exclusion of those for whom English is not their native language from the number of "American" students did not change the results in any way.
Another interesting point was the analysis of gender differences. In all countries, boys performed noticeably better than girls on average, but the gap found was significantly smaller than the gap between foreign graduates and Americans. As a result, American girls, thanks to better education, turned out to be on average noticeably more capable than foreign boys. This seems to indicate that the observed differences in the results of boys and girls arise mainly from cultural and educational differences in approaches to teaching boys and girls and not from natural abilities, since a girl with a good education easily outperforms a guy who was taught not so well. Because of this, the fact that female programmers in the US go on, on average, are paid significantly less money than male programmers does not appear to be related to their actual abilities.
Despite all efforts to analyze the data, the results obtained in the study, of course, cannot be considered an indisputable truth. Although the researchers made every effort to translate all the tests perfectly, the company that created them still initially focused on testing American students. It cannot be ruled out that the excellent results of the Americans may be due to the fact that for them such questions were simply better known and more familiar than for their foreign peers. However, the fact that students in China, India and Russia with completely different educational systems and tests showed approximately the same results indirectly indicates that this is probably not a very plausible hypothesis.
Summing up all that has been said, I would like to note that in the USA today 65 students graduate from computer science every year. This number has increased significantly in recent years, but remains far from China (185 software graduates annually) and India (215 graduates). But although the United States will not be able to stop βimportingβ foreign programmers in the foreseeable future, this study shows that American graduates are much better prepared than their foreign competitors.
From the translator: I was touched by this study and I decided to transfer it to Habr, since my personal 15-year experience in IT, unfortunately, indirectly confirms it. Different graduates, of course, have different levels of training, and at least a dozen truly world-class talents graduate annually in Russia; however average alumni results, mass the level of training of programmers in our country, alas, is pretty lame. And if we move away from comparing the winners of international competitions with an Ohio State College graduate to comparing more or less comparable people, then the difference, unfortunately, is impressive. Let's say I studied at Moscow State University and I read the research of MIT students - and this, alas, is a completely different level. Education in Russia - even programming education that does not require capital expenditures - follows the general level of development of the country, and with the general low level of salaries in the industry, in my opinion, it only gets worse over the years. Is it possible to somehow reverse this trend, or is it definitely time to send children to study in the States? I invite you to discuss this in the comments.
The original study can be read here:
Source: habr.com