Categories
Analysis Contents Highlights Ranking Monitor

Analysis QS Subject 2020

This ranking measures, first the result of their academic reputation survey, the second, the results of their employer reputation survey. The other two are the number of citations per paper (citation rate) and a variant of the Hirsch index over five years, which purportedly measures both productivity and the fourth?, highly cited knowledge production, being a measure of h papers with h citations. This means that highly productive research groups are represented better than in mean citation rate.

What does this ranking measure?

Download the complete table of results

The QS Subject ranking measures four indicators in variable weightings

The first is the result of their academic reputation survey, the second, the results of their employer reputation survey. The other two are the number of citations per paper (citation rate) and a variant of the Hirsch index over five years, which purportedly measures both productivity and the fourth?, highly cited knowledge production, being a measure of h papers with h citations. This means that highly productive research groups are represented better than in mean citation rate.

The ranking uses its own proprietary schema to categorise research into five main areas of knowledge, with 48 subcategories. This schema is now an option in Scival, meaning that it is possible to some extent to replicate the datasets used to produce this ranking, which we did
here in the spreadsheet attached.

How did the state universities perform?

Table 1 – QS subject 2020 – USP

UniversidadeÁrea de
conhecimento
PosiçãoPontuaçãoReputação por
empregadores
Citação
H-Index
Citações
por artigo
Reputação
Acadêmica
ArtigosAutoresFWCIH-5Citações por
publicação
Colaboração
Internacional
Colaboração
Nacional
Colaboração
Industrial
Clusters
de Assuntos
Méd.
Artigos/Cluster
USPArtes e Humanidades6378.483.143.672.583.6167719060.68161.920.4220.312013.98
USPEngenharia e Tecnologia8677.880.677.174.777.218323158810.97485.736.936.32.650636.21
USPCiências da Vida e Medicina797780.573.370.582.455414471211.241298.435.542.51.979869.44
USPCiências Naturais7678.580.483.175.876.621780160031.11948.548.732.42.754040.33
USPCiências Sociais e Administração6377.780.567.877.578.19355101970.74434.527.4381.126735.04

* The indicators Position, Score, Employer’s Reputation, Citation , H-Index, Citations per Article and Academic Reputation were extracted from the QS by Subject 2020. The other columns were obtained in the Scival database.

USP’s arts and humanities (which humanities? are extremely well respected across the world, both by employers and academics.
Many of the categories are not measured bibliometrically, and the apparently lower bibliometric performance is due to differences
between areas of knowledge. Architecture, history and philosophy are particularly well respected. Architecture, notably, has a much
higher rate of international collaboration, much more intensive clustering and a higher citation rate than the area average. For
Engineering and technology, USP specialises in smaller domains like mining and civil engineering, characterised by less
concentration of activities, as well as computer science, where highly concentrated research in a few areas leads to both larger
clusters and a higher h-score than other areas. Life sciences are characterised by heavy publication in concentrated themes. It is
the strongest area of the university in bibliometric terms, but also a much more competitive field than the others because of its high
global priority. USP’s main strengths are agricultural and veterinary sciences, and medicine and dentistry. In Natural sciences, USP
ranks highly in environmental sciences, physics and geography. It should be noted that while the first two follow the typology of a
natural science, geography obeys a publication logic much closer to a humanity such as history. Finally, in the social sciences, USP
ranks particularly well in sociology, sport science and statistics.

Table 2 – QS subject 2020 – Unicamp

UniversidadeÁrea de
Conhecimento
PosiçãoPontuaçãoReputação por
empregadores
Citação
H-Index
Citações
por artigo
Reputação
Acadêmica
ArtigosAutoresFWCIH-5Citações por
publicação
Colaboração
Internacional
Colaboração
Nacional
Colaboração
Industrial
Clusters
de Assuntos
Méd.
Artigos / Cluster
UnicampArtes e Humanidades13373.173.650.668.677.516775840.5771.118.822.40.11056.25
UnicampCiências Sociais e Administração16870.169.95477.871.9316032300.86304.82640.21.224912.69
UnicampCiências da Vida e Medicina1996869.652.669.676.2153501231801.08647.530.846.91.577119.91
UnicampEngenharia e Tecnologia13174.171.869.878.675.79408690201.04526.833.8384.147919.64
UnicampCiências Naturais13973.970.277.975.872.9962264081.25739.242.635.74.450619.02

* The indicators Position, Score, Employer’s Reputation, Citation , H-Index, Citations per Article and Academic Reputation were extracted from the QS by Subject 2020. The other columns were obtained in the Scival database.

All areas of Unicamp’s research rank within the top 200, with a special strength in engineering and technology. The
lowest ranking area, life sciences, has a higher average weighted citation rate and h-index higher than its top ranked
area. This reflects the care required in interpreting these rankings, where both variations in habits and in the
competitiveness of the ranking are not immediately clear. Unicamp has a remarkably even spread of articles per cluster,
and all are much lower than USP, meaning that while Unicamp has a much higher citation rate and h-index than USP, yet
it ranks nearly 50 places lower, in part because of its lower visibility. Unicamp’s areas of outstanding performance are
architecture and history in humanities, computer science, electrical, chemical and mechanical engineering, agricultural
science and biological science in life sciences, along with dentistry. In natural science, chemistry, physics and geography
are all strengths, while in social sciences, sociology and education are both strong. As observed in other rankings, the
state universities have broadly complementary strengths.

Table 3 – QS subject 2020 – UNESP

UniversidadeÁrea de
Conhecimento
PosiçãoPontuaçãoReputação por
empregadores
Citação
H-Index
Citações
por artigo
Reputação
Acadêmica
ArtigosAutoresFWCIH-5Citaçoes por
publicação
Colab.
Intl.
Colab.
Nacional
Colaboração
Industrial
Clusters
de Assuntos
Méd.
Artigos / Cluster
UnespArtes e Humanidades34163.26934.65068.216775490.4860.913.626.80.29605.05
UnespEngenharia e Tecnologia29265.863.86475.964.1659765060.99466.732.744.51.143815.06
UnespCiências da Vida e Medicina32361.26444.461.17120911186660.84545.426.950.7174528.07
UnespCiências Naturais23768.263.778.474.662.2868268321.26721041.438.51.147418.32
UnespCiências Sociais e Administração36962.564.248.877.761.2302437280.74314.925.345.70.123213.03

* The indicators Position, Score, Employer’s Reputation, Citation , H-Index, Citations per Article and Academic Reputation were extracted from the QS by Subject 2020. The other columns were obtained in the Scival database.

Despite relatively similar bibliometric profiles, Unesp struggles to gain as much visibility as the others, and as a result struggle to position as strongly. Its strongest area is the natural sciences, specifically in environmental science, chemistry, physics and geography. Unesp’s strongest reputation is for life sciences, although they are also well recognised for environmental sciences and geography. One particular place to pay attention to is that agricultural science has 42 papers on average per cluster, compared to an average of 28. This extra intensiveness translates into a much higher position than other life sciences. Environmental sciences, similarly, is the strongest area for Unesp in natural sciences, with a correspondingly large average cluster size.

What were the main determinants of success?

Analysing the bibliometric data, often statistically similar areas of knowledge have produced huge variations in position. Number of papers in an area of knowledge is important, as is the field weighted citation index and level of international coauthorship. However, for this
analysis we produced a novel indicator that appears to explain a large amount of the performance in this ranking. Scival uses a clustering methodology to identify topics based on bibliographic coupling, and groups those topics into larger topic clusters. When we take the
mean topic cluster size as an indicator, we notice two distinct tendencies; small areas of knowledge and small numbers of published papers rank more highly in the QS than those trending towards the mean number for the area of knowledge. This rule is completely reversed for sets of papers of over 1000 – for these, the larger the mean cluster size, the better the university ranks. This suggests that for small departments, there is a gain to operating in multiple fields of study in order to increase their profile and therefore their reputation score. When there is already an established profile from a certain number of papers, universities benefit from increased specialisation and concentration, improving bibliometric indicators and gaining repute. USP is much more effective in doing this than either Unicamp or Unesp, explaining why they consistently rank much more highly.

How can the universities improve in future cycles?

The universities can improve by encouraging local companies with whom they share close relationships to respond to the employer survey. Special attention should be paid to those with a publication, licensing or graduate employment relationship with the university. This would ensure that people who intimately know the university have a chance to respond. At the moment, just 1.7% of respondents are Brazilian, compared with 2.6% Japanese employers, 2.6% Russian employers, 3.6% French employers and 12.2% US employers.

Increasing the number of respondents would increase the number of votes for different Brazilian universities, rather than being concentrated on one or two institutions. For emerging areas, or areas of low publication, publishing in a variety of areas seems to increase the university’s visibility and therefore its reputation score. This would equate to a situation of low coordination and convergence of research, wherein researchers work more or less autonomously of one another. This helps universities to become visible on the ranking, but generally does not equate to higher position. There is a clear tipping point of
around 1000 papers in a set in which mean cluster size becomes a key factor in ranking position. When this critical mass is reached, a university performs better in all indicators when it is able to converge its research efforts into specialising in some areas that receive global recognition. The fact that in many cases USP has double the average cluster size means that it is spread less thinly across disciplines than Unicamp or Unesp, despite in many cases the bibliometric data suggesting similar performance.