December 6, 2007

Curiouser and curiouser...

Quico says: The struggle to get to the real results from Sunday's constitutional reform referendum took a turn for the weird last night, after reader amieres struck again. This time he managed to harvest the 2006 presidential election data from the CNE's "divulgación" website. He warns, however, that not every table at last year's election corresponds to a table in Sunday's referendum.

In total, we now have 3 databases: the 2006 results, the 2007 first bulletin results, and a handy combination of the two. You can go ahead and download them (all three are zipped excel files):
amieres set out to analyze this data immediately. His preliminary result? sure you're sitting down for this.

The 13.6% of voting tables that were not included in Sunday's CNE Results Announcement (primer boletín) are from places that, in 2006, voted more heavily for Chávez than the national trend.

CNE's final result will probably show a smaller margin of victory for the No side than the 1.4% announced on Sunday night.

On the basis of the 2006 results for the missing tables, amieres projects the final outcome will be achingly close.

A second reader, Edgar Brown, came to a similar conclusion using a different method. He notes that in the tables not included in Sunday night's primer boletín, Chávez won the 2006 presidential election by 67.8% to 32.2%. But in the tables that were included in CNE's primer boletín on Sunday night, Chávez won last year by 62.5% to 37.5%. He also concludes that this suggests a narrower victory for the No side than the primer boletín showed.

We beseech all statistically literate readers to have a crack at falsifying this work.

To reiterate, only an acta por acta count can settle these matters definitively. Apparently UNT is working feverishly on that gold standard tallysheet by tallysheet verification as we speak. Lets hope they publish their results soon.

All we can do with the information we have is speculate on an informed basis. At this point, the stories going around about a much more comfortable win for the No side than was reported on election night are difficult to reconcile with the data.