Almost there

04.24.2004

Just finished running most of my statistical models. It took me until 5:30am, but it's almost finished. All except models involving electoral volatility, since I still haven't built the variable. I'll do that in Lima. Meanwhile, stats-guru Nenad is double checking some of my results. Thanks a million.

So far. However. Seems my basic hypothesis was correct. When controlling for other variables — effective threshold (calculates the lowest vote % needed to win a seat) and an MMP dummy variable (0 for elections before 1997, 1 for elections after 1997) — the "media luna" dummy variable is the best predictor for party system variables.

The "media luna" dummy reduces effective number of parties by almost 1.5 parties, reduces blank & null vote by 4.3% & increases votes for the three "systemic" parties (ADN, MIR, MNR) by 17.5%. In other models, the "media luna" dummy increases votes for parties that go on to become part of the government coalition by 16.5% & increases votes for previous government parties by 24.8%. Talk about regional differences!

Ended up falling asleep listening to Sweeder.

Posted by Miguel at 01:44 PM

Comments

And just like that - I've learned of a good band.

I hadn't heard of Sweeder before. Good stuff.

Posted by: tom at April 24, 2004 03:34 PM