More from Columbia University:

Donald Green, a professor of political science, has asked Science to retract an article he and a UCLA graduate student, Michael LaCour, published last year.  The paper demonstrated that an extended and open-ended conversation with a gay canvasser could change an opponent of same sex marriage into a supporter–and the change was, if not permanent, at least long term. The publication had generated a huge amount of attention (see, e.g., This American Life) , at least partly because it seemed to offer a way out of what often seem to be intractable political conflicts: human conversation.

Impressed, two graduate students at Berkeley, David Broockman and Joshua Kalla, decided to try to extend the study, and found that many of the details of the published research didn’t exactly work.  People responded to an online survey, for example, at far lower levels than the publication reported.  They approached Green, who urged them to investigate further, recruiting Yale professor Peter Aronow as their collaborator.  They’ve posted a timeline detailing the discrepancies they found.  Green also asked his coauthor for documentation.  When LaCour responded that he’d accidentally deleted some files and maybe didn’t present everything accurately, Green asked him to prepare a letter to Science asking for the piece to be retracted.  When the letter didn’t appear, Green wrote his own.  LaCour has promised a comprehensive response to charges that he fabricated survey results, surveys, and grants to support his idea.

At this point, there’s no evidence to support the notion that an extended conversation can change someone’s mind on an important issue, but the question remains.

Obviously, American opinion has changed on same sex marriage, and a change of this size means that a lot of individuals also changed their minds.

This is important, and not just on this issue.  We know from both substantive research and personal experience that individual opinions tend to be way too stable, immune to both new information and rational argument.  But change happens, somehow.

On same sex marriage, we have some suggestions of what matters from individual stories, particularly from people who have to explain themselves.

Hillary Clinton, between 2008 and 2015, evolved on the issue, starting with support for civil rights, then civil unions (an orphaned notion), then state-by-state reforms, and finally a Supreme Court decision establishing marriage rights.  She hasn’t described a moment of epiphany, and it’s also true that her evolution has basically tracked that of American opinion; certainly, opposition to same sex marriage would be a terrible problem for her in Democratic primaries.  On the other hand, we respond to the values of our communities, as well as proximate incentives like trying to get a job.

Senator Rob Portman also evolved in roughly the same period of time, announcing his support for homogamy some two years after his son came out to him.  Although Portman was in the mainstream of American opinion, his position was not the mainstream of his party, but ongoing conversations with his son, and presumably serious reflection, led him to reconsider a position that prevented someone he loved from enjoying basic rights available to most Americans.

This makes sense too.  Exposure and emotional connections matter.

But both Secretary Clinton and Senator Portman changed their positions over much longer periods of time than an extended porch conversation, and their new positions were buttressed by personal relationships and popular culture.  We still want to know how typical their processes were, and how such changes translate to positions on other issues like vaccines, climate change, or the designated hitter rule.

Systematic research should help answer these questions.

LaCour let us down.

But science really didn’t.  There are all kinds of norms and institutional safeguards to protect against bad science winning acceptance; none are foolproof, but together they produce more reliable information over time than anything else we’ve got.  LaCour tricked his collaborator.  The peer reviewers at Science accepted the paper, taking the scholars’ word that they had conducted the surveys they said they collected.  Because the findings were so interesting and unusual–20 minutes can change your mind on a hot button issue–the publication was high profile and provoked subsequent research.  We work with a norm that a scientist (social or otherwise) will provide data to others who want to replicate, challenge, or build upon his findings.  When those data were unavailable, the findings are disregarded–and, at least in this case, retracted.  Science marches on.

It’s frustrating that we can’t get clear answers to lots of questions that we care about: how do people change their minds?  is it healthy to cook with cocoanut oil?  It’s exasperating that we can find serious peer-reviewed articles that offer contradictory answers.

The promise is that over time a serious scholarly professional consensus emerges.  The LaCour/Green affair supports that promise.