We can expect some big-name institutions to take a hit in the new World University Rankings.

Why? Because the rankings we will publish this autumn will be based less on subjective opinion and more on objective evidence.

As I have mentioned before, our old rankings relied too heavily on a survey that asked scholars to rate institutions based on their reputations. Although this survey had a limited response rate and was based on style, not substance, the results were worth 40 per cent of institutions' overall scores. Another 10 per cent was based on a survey of graduate employers. This meant that half of the old scores were based on reputation alone.

For 2010 and beyond, our new methodology will reduce the weightings given to subjective measures. Although we have invested heavily in more rigorous reputational surveys that look for the first time at both teaching and research - surveys that have received more responses than at any time under the old system - we will still reduce their influence in the final reckoning.

É«ºÐÖ±²¥

Under the initial proposals for our methodology, currently being refined in line with responses from the global academy, reputational measures are worth no more than 20 per cent of overall scores.

I have also set a cap to ensure that subjective elements are never again anywhere near the 50 per cent used in our previous methodology. This means that big names with big reputations that lack world-class research output and influence to match will suffer in comparison with previous exercises. Conversely, unsung heroes have a better chance of recognition.

É«ºÐÖ±²¥

We are clear that because the methodology for 2010 is so different, no direct comparisons should be made between 2009 and 2010. But we are also clear that some halos may start to slip from this autumn.

Phil Baty is editor, Times Higher Education World University Rankings. phil.baty@tsleducation.com

For the latest World University Rankings news, debate and social networking, see www.timeshighereducation.co.uk/worlduniversityrankings2010.