É«ºÐÖ±²¥

Browse the full list of the top 50 institutions in the Latin America rankings

Last year, 76 institutions from Latin America submitted information as part of the data collection process for the Times Higher Education World University Rankings, but only 28 met our criteria for inclusion in the analysis. Most were excluded because they did not meet our minimum threshold on the number of academic papers that must be published.

We do not think the restrictions on entry to the World University Rankings give a good enough description of the vibrancy of the Latin American higher education landscape, so this year we have created our first dedicated ranking for the region, giving more universities an opportunity to shine.

Why do we have a publications threshold for the global rankings? We apply a threshold of 1,000 academic papers published over five years because three of our key metrics are based on bibliometric data: citations (demonstrating research impact), papers per member of staff (demonstrating research productivity) and the number of papers with an international co-author (which helps us determine a university¡¯s international outlook). In each case, we want the metric to act as an indicator for the research performance. To do that, we want to have statistical validity.

Think of the papers as a sample reflecting the research performance of the university. A sample of 1,000, if we were running a marketing survey (we aren¡¯t and this isn¡¯t a perfect comparison, but it gives an insight into the reasons why we do this), would give us a 95 per cent confidence that the result was within 3 per cent of the value we calculated.

For the world rankings, this approach makes sense. But when exploring an individual region, especially one where there is less of an established research culture, and where local research may not always be published in journals indexed by the major global journal databases, we have some flexibility.

The first change we have made to determine this new top 50 list for Latin America is to reduce the threshold on papers to 500 over the past five years. If we were back at the marketing example, this would push the error rate to about 4.3 per cent. (Of course, many of the universities we are assessing here are considerably above the threshold, so their conceptual error rate would be even lower.)

Among the things that we¡¯ve been looking into since the production of the World University Rankings 2015-16 last year is how different parts of the world have very different approaches to higher education. One of the ways that we have explored this is by clustering universities based on their reputation and their citation score by subject. This combination of subjective (reputation) and objective (citations ¨C I¡¯m pretty certain that there is a large subjective element to citations as well, but that is because I try not to view data categories in black and white) gives us some interesting insights.

One is that, when it comes to excellence, the best universities in Latin America might have more in common with Asian universities than with the ¡°usual suspects¡± in the West. In fact, the top Latin American universities in the World University Rankings 2015?16, the University of S?o Paulo and the National Autonomous University of Mexico, appear in our ¡°regional stars¡± cluster ¨C alongside Peking University, Kyoto University, the University of Tokyo and Seoul National University.

Given these facts, for 2016 we¡¯ve decided that this pilot Latin America University Ranking will use the same data points as the main World University Rankings but that we will adjust the proportions of each metric in the final result.

First, given the reduction in the threshold on academic papers, we will reduce citations from 30 per cent to 20 per cent. We will partly balance this by increasing the simpler measure of ¡°papers per member of staff¡± from 6 per cent to 10 per cent.

We will also adjust some of the teaching environment metrics to reflect the importance placed on teaching within the Latin American higher education systems, bringing the overall percentage of teaching up from 30 to 36 per cent.

But which data? We are in a fortunate position ¨C as we¡¯ve just finished collecting data from the 2016 Academic Reputation Survey (which closed in March 2016) and from universities across the world for the World University Rankings 2016-17, we are able to use those data for the first time.

So not only does this new Latin America ranking give us the most accurate view of these great universities to date, it is the first Times Higher Education analysis to use data that will also be used to create the World University Rankings 2016-17 this September, giving a tantalising glimpse into the forthcoming global rankings.

Duncan Ross
Director of data and analytics,
Times Higher Education