Based on data from Altmetric.com. Altmetric is supported by Macmillan Science and Education, which owns Nature Publishing Group.

Amid a wave of recent retractions, researchers are taking to social media to discuss a perennial favourite: a three-year-old paper looking at the relationship between a journal's impact factor and its retraction frequency (F. C. Fang et al. Infect. Immun. 79, 3855–3859; 2011). The 2011 report proposed a “retraction index”, a measure of the likelihood that a paper in a given journal will eventually be pulled from the literature. The authors looked at articles published from 2001 to 2010 in 17 journals and plotted the journals' retraction indexes against their impact factor. The result was clear: the higher the impact factor, the higher the retraction index. “You know 'high impact' journals? All that means is that work is more likely to be retracted,” tweeted Jon Tennant, who studies palaeontology at Imperial College London, earlier this month. David Basanta, a cancer researcher at the Moffitt Cancer Center in Tampa, Florida, responded on Twitter: “A case could be made that more people try to replicate the results.” See go.nature.com/f7b6ud for more.