What's wrong with China's higher education?
Application Report – CMSC838S
Chang Hu
March 23, 2006
1. Motivation
China used to be one of the most advanced countries in science and technology. These days its youth still enjoy a high reputation in various high school science contests. On the other hand, however, it's surprising that no Chinese scientist has ever been awarded the Nobel Prize in the country. What happens between high school and research labs in China?
2. Dataset and Method
A set of 70 countries (including China) are chosen as candidates. The countries are compared in terms of the following criteria:
Physics, one of the most historical fields in science, has been picked as the representative of science as a whole. To measure high school students' performance, compiled overall score on IPhO (International Physics Olympiad, [1]) has been selected. These scores are compared against the number of Nobel Prize laureates [2] in the same country. Higher education status in each country is measured by the number of Top-500 universities in it [3]. Basic statistics of countries, like GDP per capita and area are estimated values between 2001 and 2005 [4].
3.1. Future scientists? Not really
Two juxtaposed treemaps are shown below (figure 1). China is one of the best countries in terms of IPhO performance. It seems to imply that those bright students (indicated by bright color) in IPhO will translate into more Nobel Prize laureates, but actually it's not the case. In Nobel Prizes China has not been so successful, indicated by its darker color.
Fig 1. A comparison of IPhO and Nobel Prize performance in Treemap
size - actual area of the country
color (upper subfigure) - IPhO score (brighter - higher score)
color (lower subfigure) -number of Nobel Prize laureates (brighter - more laureates)
The following treemap (figure 2) shows a better performance in IPhO (larger area) doesn't seem to prepare young future scientists for better scientific discoveries towards Nobel Prize (brighter color). In fact, the United States stands out with very good Nobel Prize performance (very bright color), although it has similar performance in IPhO as China does.
Fig 2. Nobel Prize performance and IPhO performance
size - IPhO score (larger - higher score)
color - number of Nobel Prize laureates (brighter - more laureates)
3.2. What's behind Nobel?
IPhO seems to be not so relevant to Nobel Prize, so what is actually behind Nobel Prizes? An overview of all the data in HCE may shed the light. By ranking the data in terms of linear correlation coefficient, it could be discovered that the strongest correlation lies between the number of top-500 universities and the number of Nobel Prize laureates (see figure 3).
Fig. 3 Linear correlation between top-500 univ. and laureates
(upper subfigure) linear correlation ranking of data
(lower subfigure) linear correlation between top-500 univ. and laureates
3.3. What's wrong with China's higher education?
From above, it seems to be safe to draw the conclusion that China has not been doing well in the Nobel Prize because of its poor higher education. On contrary, however, China is one of the countries doing well in higher education relatively to its wealth. In the following treemap (see figure 4), it is shown that China has more top-500 universities than some wealthier countries - in the treemap China is the only middle-sized node with very dark color.
Fig. 4 China's high education being not so bad
size - number of top-500 univ. (larger - more univ.)
color - GDP per capita (brighter - higher GDP)
So, what's wrong about China's higher education?
4. concluding comments
It would be great if I can get the emigrant population of the country, or number of Chinese students over top 500 universities. For example, a further look into Chinese PhD students in the U.S. [5] could be like the following (Figure 5):
Fig. 5 Percentage of PhDs staying in the US
Pink squares ¨C percentage of Chinese students
Blue disks ¨C percentage of students from other countries
There is growth in percentage of staying Chinese students, with a jump around the year 1990. This fact may also help to explain why United States is indeed an outlier in terms of top universities and Nobel Prize laureates.The Treemap is a good tool to visualize 3-dimensional data with one nominal dimension and two numerical dimensions (or any of them categorical). With its rank-by-feature functionality, HCE is more useful to provide overviews of relationship in data.
I also found out what the user could see (and thus interpret) is very much related to how the data are presented. Take the "GDP vs. Top-500" problem for example (see figure 5), if the number of top-500 universities is color-encoded, the United States would stand out as the country with most top-500 universities. On the other hand, if the GDP per capita is color-encoded, China will stand out as discussed in section 3.3.
Fig 6. Different interpretation resulting from different encodings
size (upper subfigure) - GDP per capita
color (upper subfigure) - number of top-500 univ.
size (lower subfigure) - number of top-500 univ.
color (lower subfigure) - GDP per capita
It seems to me that by this example, users are more sensitive to color contrasts.
Honestly speaking, one should be careful to draw a conclusion that there is something wrong with a country's higher education, just by looking at the data provided here. However there is indeed a big debate about the difference between China's higher education and its more successful counterparts - as an international student I can tell there is such a difference.
5. reference