AG's Monkey Cage post on LaCour and Green's "Contact Changes Minds" study is great. He's essentially saying, "yes, great study and all but it is not really doing what you say it is doing re: changing minds." This line is pure gold:
"The experimenters were pushing at an open door. Or, to switch analogies slightly, they were jumping on a moving train."
BOOM. Read it all here:
Plus, it has ZERO comments so far and it was posted at 10:00am. Nobody has the grownup pants to debate Gelman?
Plus, it has ZERO comments so far and it was posted at 10:00am. Nobody has the grownup pants to debate Gelman?
Comments have been way down ever since the most to the post. I commented regularly at the old site. I'm too lazy to jump through the hoops to comment at the new one.
Not all of us feel like this dude:
http://xkcd.com/386/
Read between the lines. Gelman isn't going to heavily critique a new hire senior at Columbia. That he will post even a positively worded critique like this, is a HUGE signal. If this piece had authors at Stanford or Princeton instead, a complete take down would have been posted.
I don't understand how this is even remotely a take-down of the paper. There are strong effects at 9 months, several months after the court decisions shown in the graph. If any two events were going to sway large numbers of voters I would think it would be the court decisions and the media coverage that surrounded it, followed by a gay canvasser coming to your door.
So if Gelman's point is: these people would have changed their minds anyway, this just impacted the timing, my question is "when are all those people in the control going to change their minds if not after a major court case"?
If Gelman were right, and ML/DG's canvassers were picking low-hanging fruit that would have had their minds changed by anything, we would expect to see big effects early on (we do) and then the court decision would prompt changes in the control groups but not much in the treatment group--since the low-hanging fruit had already been picked by their canvassers. But we don't. In fact, the jump at the court decision looks to be biggest among the treatment group.
I understand he's trying to situate it with the prevailing public opinion literature, but I don't really get it.
Except Gelman is not wrong. The argument is misleading. They are saying they changed people's views on same-sex marriage when in reality people were already trending in that direction. The authors fail to address that little detail in their discussion.
Does not know how an experiment works. Does not understand control groups.
It's a low hanging issue, is Gelman's point. It's the only issue were this effect would be easily understood in terms of magnitude. Attitude change on abortion or environmentism, would have been more surprising.
Gelman's point is that there are a ton of contextual factors, encouraging change, especially in Los Angeles (television, Will and Grace reruns, news, political endorsement, supportive political elites). Think of these as 'nudges' to use Gladwell's non-science phrase. The canvassers may have broken down some factors which prevented them from working prior to the canvass, and allowing the 'nudges' to have a disproportionately large affect. In reality, both are encouraging the change, leading to the magnitude. And the canvassers may have only made social desirability more salient....
It would also be nice if they (DG and ML) had a theory as to why they got those results, then maybe--just maybe--we could test it.
If there is anyone willing to criticize a senior colleague publicly it ia A.G. Dude has no filter, as readers of the Monkey Cage know.
Read between the lines. Gelman isn't going to heavily critique a new hire senior at Columbia. That he will post even a positively worded critique like this, is a HUGE signal. If this piece had authors at Stanford or Princeton instead, a complete take down would have been posted.