A cor­rec­tion only serves its pur­pose (to cor­rect our falsely-held beliefs) if we are pre­dis­posed to believe the cor­rec­tion itself. If we dis­agree with the cor­rec­tion, how­ever, it instead acts to actu­ally rein­force our incor­rect beliefs (the “back­fire effect”).

That’s the con­clu­sion drawn from research con­ducted by Bren­dan Nyhan, look­ing at how we avoid cog­ni­tive dis­so­nance in the face of cor­rec­tive infor­ma­tion (pdf).

Brendan’s research on cog­ni­tive dis­so­nance and cor­rec­tions has been nicely sum­marised by Ryan Sager in a cou­ple of posts: one that looks briefly at the effect of cor­rec­tions on mis­in­for­ma­tion, and another look­ing in great detail at the roots of the anti-vaccine move­ment.

We find that responses to cor­rec­tions in mock news arti­cles dif­fer sig­nif­i­cantly accord­ing to sub­jects’ ide­o­log­i­cal views.  As a result, the cor­rec­tions fail to reduce mis­per­cep­tions for the most com­mit­ted par­tic­i­pants. Even worse, they actu­ally strengthen mis­per­cep­tions among ide­o­log­i­cal sub­groups in several cases. […]

Test sub­jects read mock news arti­cles fea­tur­ing mis­lead­ing state­ments about well-known but ide­o­log­i­cally con­tentious sub­jects such as the pres­ence of weapons of mass destruc­tion in Iraq prior to the U.S. inva­sion. Half of their sub­jects read arti­cles includ­ing only the mis­lead­ing state­ments; half read arti­cles that also included a correction.

By com­par­ing the two groups of respon­dents, [it was] deter­mined that the ide­ol­ogy of the sub­jects tended to pre­dict reac­tions. Efforts to cor­rect mis­per­cep­tions were more likely to suc­ceed among those ide­o­log­i­cally sym­pa­thetic to the cor­rec­tion, such as lib­er­als to the notion that WMD were never found in Iraq after Sad­dam Hus­sein was deposed. But the cor­rec­tions tended to “boomerang” among those ide­o­log­i­cally pre­dis­posed to believe the erro­neous infor­ma­tion. Thus, con­ser­v­a­tive sub­jects who had read the cor­rec­tion were even more.

Every arti­cle Sager points to in these posts is worth read­ing, espe­cially Is Health Care Turn­around a Bad Bet?, How Facts Back­fire and Per­sis­tence of Myths Could Alter Pub­lic Pol­icy Approach.