Thursday, October 26
Thursday, October 26, 2006 | 
Posted by
thistle
I normally frown on the use of multiple exclamation points, but I'm making an exception today in the title of this post because this open letter from the President of the American Statistical Association is very cool. It's cool because:
1. An academic discipline is reaching out trying to make the world a better place on an important issue with timely, practical, and reasonably cheap solutions.
2. Voting is a crucial right in a democracy, but a lot of us feel helpless about how to fix the problems with voting, especially now that there are stupid voting machines that can be hacked. The key breakthrough idea that the ASA is offering is that you can use random audits to provide data to estimate the size of the problem, instead of relying on anecdotal evidence or trying to find universal fixes. The idea of sampling is definitely under-used in public policy, which is almost always stuck in an entire-population-or-nothing mentality.*
3. The letter discusses the need to look at multiple sources of error at each step along the chain. The reality is that officials need to be focusing on minimizing error instead of devising systems that are completely error-proof, because it isn't possible to be error-proof. The point I'm inelegantly trying to make is like the debate about the census v. sampling for getting an accurate picture of the American population.
4. The president of the American Statistical Association is a woman. (I'm assuming that because the President is named Sallie).
______
* example = No Child Left Behind which mandates testing for every kid in the nation, when sampling approaches would allow testing with higher validity (like not multiple choice) and be a lot cheaper and not waste the time of millions of schoolkids.
1. An academic discipline is reaching out trying to make the world a better place on an important issue with timely, practical, and reasonably cheap solutions.
2. Voting is a crucial right in a democracy, but a lot of us feel helpless about how to fix the problems with voting, especially now that there are stupid voting machines that can be hacked. The key breakthrough idea that the ASA is offering is that you can use random audits to provide data to estimate the size of the problem, instead of relying on anecdotal evidence or trying to find universal fixes. The idea of sampling is definitely under-used in public policy, which is almost always stuck in an entire-population-or-nothing mentality.*
3. The letter discusses the need to look at multiple sources of error at each step along the chain. The reality is that officials need to be focusing on minimizing error instead of devising systems that are completely error-proof, because it isn't possible to be error-proof. The point I'm inelegantly trying to make is like the debate about the census v. sampling for getting an accurate picture of the American population.
4. The president of the American Statistical Association is a woman. (I'm assuming that because the President is named Sallie).
______
* example = No Child Left Behind which mandates testing for every kid in the nation, when sampling approaches would allow testing with higher validity (like not multiple choice) and be a lot cheaper and not waste the time of millions of schoolkids.
0 comments:
Post a Comment