China may be set to overtake the US in scientific output as soon as two years from now, according to an alarmist BBC news article:
The country that invented the compass, gunpowder, paper and printing is set for a globally important comeback.
An analysis of published research – one of the key measures of scientific effort – reveals an “especially striking” rise by Chinese science.
The study, Knowledge, Networks and Nations, charts the challenge to the traditional dominance of the United States, Europe and Japan.
The figures are based on the papers published in recognised international journals listed by the Scopus service of the publishers Elsevier.
In 1996, the first year of the analysis, the US published 292,513 papers – more than 10 times China’s 25,474.
By 2008, the US total had increased very slightly to 316,317 while China’s had surged more than seven-fold to 184,080.
Previous estimates for the rate of expansion of Chinese science had suggested that China might overtake the US sometime after 2020.
At first glance, looking at the trendlines – and keeping in mind that trends do not necessarily go on forever – the thing I initially found discouraging is less China’s rise in academic output, than America’s decline in this rather terrifying graph:
The first place to look is the x-axis, where statisticians with pet theories all too often jigger the graphs by playing with scales. But the Beeb (and the Royal Society) get a break here – the scale is internally consistent (although cutting it off above 35% does tend to sharpen the trendlines).
So next we look at what we’re actually measuring: The rate of change in scientific articles’ citation rather than absolute output, far less quality.
We are not told the absolute number of citations there have been in the academic and scientific literature between the US vs China. This is crucial because we are measuring rates of change, and the identical number of events on a small data sample represents a much higher rate of change than it would on a large one. But if the rate of citations tracks the rate of output – and that’s is a mighty big if, because quality continues to lag output in the PRC – we can estimate that (interpolating the chart above from 2008 to 2013) if China experiences an average 14% rate of growth on 108,040 articles published in 2008, they will have grown to 209,851 in 2013. The US on the other hand, starting at the higher figure of 316,317 and growing at “only” 17% per year will publish 370,091 if the trendlines remain constant.
By 2013, this would yield 76% more US articles published in than Chinese, representing a four percent rise over the 72% US advantage in 2008. See also: sensitivity analysis.
There may be an important point on scientific competitiveness to make with these findings. Unfortunately, by comparing apples to lawnmowers – and throwing in a “scare” chart – the BBC doesn’t make it.
Cui bono from this method of data representation?
Professor Sir Chris Llewellyn Smith, chair of the report, said he was “not surprised” by this increase because of China’s massive boost to investment in R&D.
Chinese spending has grown by 20% per year since 1999, now reaching over $100bn, and as many as 1.5 million science and engineering students graduated from Chinese universities in 2006.
“I think this is positive, of great benefit, though some might see it as a threat and it does serve as a wake-up call for us not to become complacent.”
Always follow the money.
Most R&D in America is privately funded, with the government kicking in 29% in 2007 (pdf). The same is true in the UK: 2007 showed a government expenditure of 31%. This is just a two percentage point difference, but using the BBC’s methodology it represents a 6.5% difference.
Round it up to ten.