Scientists who have attended meetings are more likely to cite work discussed in talks they could see in person, compared with results described in sessions that they could not attend1. That citation bump from in-person attendance accrues even for talks that conference attendees hadn’t planned on listening to.

Attending a talk is “really, really effective” for increasing the chance that researchers will cite the work, says study co-author Misha Teplitskiy, an information scientist at the University of Michigan in Ann Arbor. The study was posted on the arXiv preprint server on 5 May. It has not yet been peer reviewed.

Teplitskiy has long been interested in where scientists get their ideas, but struggled to measure this. Then he was introduced to David Karger, a computer scientist at the Massachusetts Institute of Technology in Cambridge, who had developed an app called Confer to help computer scientists plan their personal schedules at conferences. Scientists using the app can ‘like’ a talk by clicking on it.

Showing up

Teplitskiy, Karger and their colleagues used data from the app to deduce which talks 2,404 Confer users attended at 25 computer science conferences held between 2013 and 2020. The team assumed that a conference participant was more likely to attend a ‘liked’ talk if it took place in a session that did not conflict with anything else on the attendee’s schedule.

The authors then assessed the works cited by those conference attendees within two years of the meeting. They found that for attendees who had multiple schedule conflicts with a liked talk, the probability of later citing that talk was 2.6%. But for attendees who did not have a conflict, that figure rose to 3.8%.

After taking other effects into account, the authors found that meeting attendees cited liked papers 52% more often when they could see them in person than when they couldn’t. That’s “pretty sizable”, Teplitskiy says.

The analysis also found a similar benefit for citations of non-liked papers, an effect that the authors call serendipitous diffusion of information. Serendipitous diffusion accounted for nearly 22% of the overall dissemination of information brought about by presentations at the conferences.

Serendipitous spread

That was a surprise, Teplitskiy says. The team had assumed at the start of the project that serendipitously seeing a talk at a conference would have a small effect on later citation, if the effect could be measured at all. That the serendipitous spread of ideas can be measured and quantified is important, he says, because studying how ideas flow within communities is challenging. The study findings suggest that scheduling conflicts could be a useful tool to analyse this information flow.

The work adds to evidence that attending conferences in person is important for scientists, says Ina Ganguli, an economist at the University of Massachusetts Amherst. “We know that face-to-face interactions in settings like conferences matter,” she says. “It would be interesting to quantify how much knowledge diffusion and exchange is due to being at a conference in person or whether you can get similar effects in virtual formats.”

The work is important, because “the authors found a laboratory to study the effect of something unrelated to quality on a paper’s citations”, says economist Edward Van Wesep at the University of Colorado Boulder. “We all understand that presenting our work helps it get cited, but it’s hard to prove that this is true,” he adds.

Scientists pour time and money into organizing and attending conferences, Ganguli says, “so having evidence of the benefits like those shown in this paper, is important”.