A Quantum Diaries Survivor

Italians who followed the television programs after the end of votations in Italy last Monday were treated with a wildly uncertain string of projections on the results, made by a polls institute, Nexus.
Basically, Nexus initially presented predictions that were similar to predictions seen weeks before: Prodi would win with a 5 points margin. Then, after a few hours, the fork between the center-left coalition and the center-right parties started to narrow down considerably, until the prediction became a narrow win by Berlusconi’s coalition!

I think I understand the mechanics behind the several projections that were made.
Initially (say, at time -1, before urns are closed) Nexus had in their hands the pre-vote polls. Then, they started to get a few exit polls to add: a weighted average of the two provides a good way to use the information. But then, when the first real results come out, Nexus could determine the bias in their exit polls, and use it to make more accurate predictions.

The above mechanism explains why the first projections were what people were expecting, and the slow but steady divergence with the original pre-electoral polls. What happened, in fact, was that Berlusconi’s campaign was a winner: he convinced at least a million italians who would normally desert the vote to cast their vote in his favor for this time once, with his promises to cut several taxes that were given for granted by everybody (and whose removal would put in peril the services that the counties provide to citizens, like daycare - I wrote about that here already).