Sunday, July 16, 2006Universal National Election Studies
At the Colombia Statistical Modeling blog, Gelman has posted a short summary and link to a statistical analysis of the 2000 election votes in Mexico. The paper has some interesting empirical findings but is a little short on theoretical or contextual analysis. [You know, the "why" question....] There's no clear theoretical or even Mexico-specific explanation for the observed pattern. This seems like a prime case of having the data and methods to run some cool models without doing the necessary research on Mexican politics to make good sense of the results. I'm sure journal reviewers (and Mexican specialists) who are impressed by the methods will help brainstorm explanations to explain the results...essentially doing the authors' work for them.
Given the availability of Mexican poll data, there's almost a cottage industry in testing hypotheses from American election studies on the Mexican data. Though there are exceptions, many studies are misspecified because in their rush to apply the their pet method to the new data, they forget to bother learning about Mexican politics. I'm not trying to make a case for Mexican exceptionalism; there probably are voting models that are nearly universal. But on the other hand, researchers need to understand that certain indicators that reliably measure something in the U.S. may not measure the same concept elsewhere. Just because decades of election studies have established the validity and reliability in the U.S. context doesn't mean that those same measures are sure to be valid and reliable elsewhere. We need much more work testing basic measurement in the Mexican context, IMHO.
I don't want to accuse Gelman of not doing his homework, but if he has, it's not evident in the draft of that paper.
posted by Michelle @ 4:48 PM,
- At 7/16/2006 5:00 PM, JB said...
Michelle, tenéis una complicada y delicada situación en México. Desde España los seguimos con atención peron como siempre llegan pocas noticias y faltaría ver la fiabilidad de las mismas!!
Thanks for your blog.
- At 7/16/2006 6:06 PM, said...
Michelle, my hat is off for all the statistical modelling, graphs, pie charts, etc. (pretty boring work, in my book) But, being a wee-bit old fashioned, I have to insist: vote by vote! This is the only way to solve this crisis that is escalating in Mexico.
If the minimum of correct scrutiny is not accomplished, this will only allow for the situation to deteriorate to the point that Mexico will become ungovernable. The corrupt PRI and PAN business elites don't know the fire they are playing with if they try to shove Calderon down the throats of the long-oppressed and demeaned masses of people who feel that they are robbing democracy in the country.
(Have you ever seen a Zapatist woman with a machine gun! Believe me, she'd make Bush shit his pants, the little chicken hawk.)
You can send the maquilas to China for lower wages, you can make fun of the 'wet-backs' and call the Mexicans fleeing to the US to help their families survive 'criminals'---but you don't mess with the people's land, and their ability to change their collective political destiny by the vote. If you do, you reap the whirwind.
Given the miserable job that elites in Mexico have done to secure a decent life for the majority of its citizens--given that they cannot secure the life, liberty, opprotunity and prosperity of the population with their failed neoliberal economic model, I can assure you that the Mexican people know that they can do infinately better.
Vote by vote! Not hard to conceptualize that a fair recount is doable--despite what the corrupt Calderon, Bush, and their capitalist cronies say. Vote by vote! Not one step back so that we can protect and safeguard the viablity and stability of the republic.