I blogged the other day about this interesting study on whether a politician correcting himself or herself can influence attitudes.  See my original post for study details, but here's a bit of the results I just reread and found fascinating.

The dependent variable is belief in whether weapons of mass destruction (WMD) were in Iraq when the U.S. invaded.  The authors, using multiple regression, found no main effect for a correction embedded in a news report about WMDs.  But when they looked at (geek stats warning!) interaction terms they found some interesting stuff. 

Stay with me. It's worth it. 

The authors found a three-way interaction for correction, ideology, and how important the Iraq war was perceived.   In other words:

... the correction was effective in reducing misperceptions among conservatives who did not select Iraq as the most important issue, but its effects were null for the most strongly committed conservatives...

In other words, a correction in a news story about WMDs seemed to work for conservatives who didn't care much about Iraq -- their misperception went down -- but for those who really cared about the war, the correction led them to a greater misperception about the existence of such weapons.  I'm pushing it a bit because although the authors above called it null effect, but they also note just afterward that the result is awfully close to statistical significance.  So let's run with that.

There's a lot more to get into, but I've mined the study about as much as I plan to.  Got other stuff to write about.  For details about the study, see my previous post.