Sunday, May 24, 2015

Coming out, electoral persuasion and academic lying

The tale of a political science experiment published last December in Science and then retracted last week fascinates me.

If you missed it, Michael J. LaCour and Donald P. Green originally reported that that

a short conversation with a gay canvasser appeared to significantly nudge California voters in a pro-gay-marriage direction, but that the effects were contagious within those voters' households and lasted at least nine months — the final point at which the researchers checked in with the study participants via online surveys.

A professor at Columbia, Green pioneered ground breaking experiments looking at what really happens in electoral mobilizations. Most of what he's written has seemed consistent with my experience of field organizing tactics. I've written about this at some length here and here.

But now it turns out that the UCLA graduate student who was the primary researcher on this particular study can't produce the data that backs up its conclusion. According to a "bewildered" Green, the reported canvassing by gay and straight canvassers was done -- twice. But LaCour has not shown any data from the supposed follow-up questioning evaluating the effect of these contacts. Most likely that step never happened. Subsequent researchers aiming to build on this experiment couldn't get any answers. The thing is FUBAR.

One reality I've always pointed out about Green's (and his various excellent collaborators') research is that it has been better at telling us what election workers can do to increase turnout (who votes) than at how to accomplish successful persuasion (getting people to vote your way.) The former can be evaluated by checking the lists of who voted after the election -- that information is in the public domain. But you can't know in most cases how they cast their ballots -- that is secret. This study seemed to have made a breakthrough in persuasion, except that the measuring never happened.

I can easily understand why the faked "results" seemed plausible. For literally decades, it has been axiomatic among gay activists that confronting our heterosexual neighbors with the fact of our orientation, in a friendly and respectful way, is the best course to move them toward acceptance.
  • As early as 1978, I remember being gripped by Amber Hollibaugh's tales of carrying her message to rural and suburban areas against Prop. 6 which sought to fire gay teachers. Against the advice of cautious political consultants, the group she was affiliated with insisted on putting their queer identities front and center. And they believed they'd made a dent in unconsidered homophobia. She later told that story in My Dangerous Desires. Somehow we won that round.
  • Anyone who has been trained in organizing in the last fifteen years has encountered (and usually practiced) the personal story telling methodology of Marshall Ganz, the legendary organizer who honed his craft with Cesar Chavez and the United Farmworkers before going on to the Kennedy School of Government and the first Obama campaign.
  • As far as I could discern from afar, after losing a referendum vote in 2009, marriage equality campaigners in Maine emphasized this personal story methodology in contacts during their successful initiative effort in 2012.
  • Above all, hasn't brave, repeated, ongoing "coming out" by all sorts of us gay and lesbian individuals been the essential motor of our progress toward a more normal life in U.S. society over the last 40 years? Sure it has, and that made the study seem plausible.
Except it wasn't really. The supposed effects were simply too large to credit easily, so others attempted to repeat them -- and discovered they still can't quantify persuasion.
***
I asked Erudite Partner how this faked study could have been published? How could LeCour have thought no one would try to replicate it? The EP is my source on all things academic. She says:

Studies are hardly ever replicated. He had no reason to think anyone would try. Budding researchers all want to publish their own new and exciting work, not check someone else's.

But this study seemed to have valuable real world implications, so someone was bound to replicate it -- or at least to build on it. And when they did, the false underpinnings showed up.

Green claims, highly plausibly, to have been deceived into endorsing it because all the apparatus of the research -- the initial canvassing and the re-contacting -- seemed sound. But the second step didn't apparently happen. Green too wanted his name on the breakthrough. His weight in the field gave the story additional heft.

Duke University sociologist Kieran Healy has pointed out that this electoral organizing study is particularly pernicious pseudo-science.

That paper got a lot of public attention, and people both heard about it and tried to learn from it. ...perhaps LeCour was ... like an undergraduate plagiarist who suddenly finds, to his horror, that his copied paper has won a university prize. It doesn’t much matter to me. But social science can have real consequences for people’s lives...

Apparently the YES organizers in Ireland adopted the "insights" of this study; fortunately, Ireland apparently didn't need the extremely strong effects from the emergence of gay neighbors that the study promised. "Coming out" -- introducing ourselves to the community -- clearly does work and did work in Ireland. How deeply and over what time frame has not been measured as conclusively as we were told.

No comments: