Quote:
Originally Posted by Island_OnThe_Land
Lifeshadower: Very interesting post. I would like some additional explanation for your initial calculation. I get mechanically what you did: the sum of the squares of each group minus 1.
What you seem to have measured is "racial/ethnic balance" which I suppose can also be called "diversity."
As the raw figures show, LA County has the lowest percentage of WHITENON HISPANICS yet it is ranked 3rd in your "diversity" ranking system. If one were measuring what metro has the lowest % of nonhispainic whites than LA County would be tops. Many demogrpahers have used that as the measrue of diversity since nonhispanic whites were the largest group in the USA for so long. But I suppose in a changing world your method has merit too......

Sure thing
:
As I have repeated throughout the thread, the formula I used to calculate the diversity index comes from
probability theory and
sample size methodology. The US Census uses the exact same formula (check out the link on the first page) to calculate the same exact thing, except I threw out variables that would be too statistically insignificant for meaningful data collection (basically, any groups that have less than 0.5% of the total population).
Basically, what's being calculated here (and I repeat this again and again) is
what are the chances that any two people within a metropolitan area will be of different, US Census defined racial groups? If you look at all the indicies, they are all percentages. For example:
1) In the SF CSA, there is a 68.7% chance that any two people will be from two different racial groups
2) In the Houston CSA, there is a 67.6% chance that any two people will be from two different racial groups
3) In the Los Angeles CSA, there is a 66.4% chance that any two people will be from two different racial groups.
Simply having less of one group (like nonHispanic Whites) won't really affect the data IF another group simply takes up a disproportionate percentage. After all,
every group has unitary value (each group is WEIGHED the same) with no preference being given to the nonWhite groups. That's why simply stating that 'there are less White people in LA' doesn't necessarily make it the more diverse. After all, the Rio Grande Valley in Texas 80%90% Hispanic and no one ever accused it of being very diverse. If you want to weigh the groups different (like assigning Whites a value of 1; Blacks, Latinos, and Asians a value of 2) all you need to do is multiply the percentage of the affected group by the value. However, that would make talking about 'diversity' meaningless because there is no equity.
I think the issue troubling most people in this thread is the term 'diversity' which is an amorphous concept with no real meaning. I didn't calculate 'balance' or anything of that sort, but rather probabilities. This is why adding more than 10 groups would make any probability equation meaningless because it would make them ALL statistically insignificant. In other words, there will be an 99% chance of running into another group almost everywhere in the country. It is what it is.
Again, I will reiterate: The US is a diverse country as a whole. If there is already a 50% (or .5) percent chance of running into another race, then that's pretty good. Arguing over a few percentage points doesn't matter in the bigger scheme of things. It is what it is.
I didn't rank the data out of malice. The numbers spoke for themselves, and it organized ITSELF along those lines. If you have trouble with the rankings, 1) Do the math yourself and check again 2) Write to the US Census for better data collection 3)Come up with your own methodology that can be measured with data
I'm sick of CityData threads that become a series of anecdotal statements which can be faked, or easily challenged (
on topics that could be measured in real life (just check out the first 10 pages of this thread). You can't measure things like 'cultural influence' or 'most powerful cities', but you can measure things like 'economic power' 'GDP' 'racial diversity'. It will be exciting to see what the new US Census looks like this year, to see if anything really changed.