The graph illustrates how Dutch voters interviewed in 1998 with varying levels of political sophistication used their attitudes toward the Christian Democratic Appeal (CDA) to evaluate the performance of Wim Kok’s 1994-98 coalition government, which included three other parties: PvdA (Labour), VVD (the Liberals, in the European sense of the word), and Democrats 66 (who maybe Pieter can explain, since I haven’t figured them out yet), but not the CDA.

The least sophisticated members of the electorate show a positive relationship, which indicates that they evaluated the performance of the coalition based in part on their attitudes toward a political party that hadn’t been in office in four years. In American terms, it would be essentially the same thing as a voter thinking George W. Bush was doing a good job because he likes Democrats (which, before you laugh, I’m sure I could find evidence of among a fair portion of the U.S. electorate). You can also see an exaggerated negative relationship among the most sophisticated voters, suggesting that there’s some sort of cognitive balancing going on (“I dislike the CDA so I must like the coalition’s performance”) independent of the feelings toward the other three parties, which are also controlled for in the model (and held constant in the graphs).

The point isn’t so much that they’re wrong, but rather that they’re relying on an outdated view of how Dutch politics works to make voting decisions. In terms of my dissertation topic, their heuristic (cognitive shortcut) for evaluating coalition performance is flawed—and this leads them to make incorrect decisions compared to their “fully-informed preferences” (how they’d behave if they knew everything that a highly sophisticated voter did about Dutch politics).

Honorable mention to Dad, who guessed it had something to do with magnets. I think he’s been watching too much Stargate SG-1…