Ooodles of old stuff including How-To's, Writers Resources, Directories of all Types, Technology Reviews, Health Information, Marketing Statisitcs, Affluent Markets, Hospital and Medical Market Data and more

Monday, September 04, 2006

Statistics in Life 6

Entry Six
Todays entry is about the United States Census. As everyone has heard, the government is now suing itself in the effort to decide if statistical sampling is to be allowed in the next census (as compared to an individual person by person count). Now this is an interesting question… and even more exciting is waiting to see how long this will be fought out and how much money will be spent to settle it (my bet is that it will cost as much as doing the population count in the first place).
Now, I am far from being an authority on matters like these, and I realize there are many variables in the census…but there are many variables in sending a man to the moon or charting the human genetic code. If statistical sampling methods are "good enough" to address these issues, it would seem a valuable tool to sample the population. Of course, this is totally aside from the constitutional legality of the issue, but taking a "numerical" perspective…I don’t see why this isn't used. After thinking about it I have come up with a few of my own ideas…you let me know if they are right..
1. There is no guarantee the samples are absolutely random samples. Homeless, institutionalized, and others may be left out. Then again it seems they are left out on the regular sample.
2. Some degree of "matching' may be needed to adjust for the under-representation of minority groups in some areas. Because of the relatively fewer numbers there is risk of further under-representing these groups.
3. People move around a lot. There may be risk of getting some skewed results because of mobility and that could hurt funding.


Post a Comment

Links to this post:

Create a Link

<< Home