[quote author="jumpcut" date=1253659696][quote author="EvaLSeraphim" date=1253657821][quote author="jumpcut" date=1253655854]These statistics are suspect. The Korean population went from 9% in 2007 to only 4% in 2008? Over half of all Koreans moved away from Irvine in one year? That's quite a mass exodus. I guess the new H-Mart and Zion markets drove all the Koreans away.
The government needs to start outsourcing these censuses, so someone with a brain can review them before publishing and see if they're way off.</blockquote>
Actually, if you took a closer look at what was profiled / linked, it appears to be the boundary of the Irvine Unified School District. Given that the City of Irvine is covered by two different school districts, this may explain the data "problem," which is not a problem at all.</blockquote>
If that's true, then the problem is even worse. They're comparing the same data set between 2007 and 2008. So that would mean the IUSD lost 50% of it's Korean population between 2007 and 2008. School funding is based, in part, on this census data. If I was a Korean parent of IUSD students, I'd be pretty upset.</blockquote>
If you didn't understand what you were reading, then yes. I was kinda hoping that my comment would cause you to not just look at the top line data but to really take a look at what you were looking at, and understand the how and why. Alas...
Let's go the <a href="http://factfinder.census.gov/servlet/MYPTable?_bm=y&-context=myp&-qr_name=ACS_2008_1YR_G00_CP5_1&-ds_name=ACS_2008_1YR_G00_&-tree_id=308&-redoLog=true&-_caller=geoselect&-geo_id=16000US0636770&-format=&-_lang=en">Irvine City numbers.</a>
First, note that the population of Irvine as a whole decreased. The numbers for the Korean population are about the same between the Irvine City and Irvine Unified data sets. Hmm... Does that mean half the Koreans up and left town? Not necessarily.
Before we even get to the tables, we see the following "For more information on confidentiality protection, sampling error, nonsampling error, and definitions, see Survey Methodology." That seems kind of important. I think they are trying to say that there are some qualifications to the numbers.
<blockquote>The data in this table are based on the 2008 American Community Survey 1-Year Estimates and/or the 2008 Puerto Rico Community Survey 1-Year Estimates. To maintain confidentiality, the Census Bureau applies statistical procedures that introduce some uncertainty into data for geographic areas with small population groups. The data in this table contain sampling error and nonsampling error. Data users who create their own estimates using data from American FactFinder tables should cite the Census Bureau as the source of the original data only.
For descriptions of response rates, coverage rates, sample size and allocation rates see quality measures. Quality Measures data are available in the B98 series of Detailed Tables in American FactFinder.
For the full documentation on the ACS sample design, estimation methodology, and accuracy of the ACS data, see the Accuracy of the Data (2008) [PDF].
For ACS definitions of subject characteristics see the subject definitions.</blockquote>
Ok, I think that is important. Let's check the areas they suggest we read. <a href="http://www.census.gov/acs/www/Downloads/ACS/accuracy2008.pdf">Here is the document discussing accuracy of the data.</a> But that's for stats people. Let's go simple: what is the margin of error? Lo and behold, <a href="http://factfinder.census.gov/home/en/epss/glossary_m.html">on these informal annual surveys, it is 10%.</a> (At least, that is how I read that sentence. Perhaps a stats person can help out.) In other words, any YoY change that is 10% or less, is statistically insignificant, i.e., nothing to be exercised about.
Or as the <a href="http://www.latimes.com/business/la-fi-census22-2009sep22,0,7464111.story">LA Times says</a> about the margin of error:
<blockquote>The data come from the Census Bureau's annual survey of about 3 million Americans, not the entire population. The survey's margin of sampling error is high enough to make it possible that the number of foreign-born people in the country actually remained unchanged from 2007 to 2008 rather than declined.</blockquote>
What I will say, Rep. Bachmann notwithstanding, is that everyone should encourage everyone else to participate in the 2010 Census. I would also encourage you to volunteer, particularly if you belong to a population that is typically resistant to government reporting or otherwise values privacy. The 10 year census is the most comprehensive and should be the most accurate. If you don't participate, you can't b!tch. ;-P
