What do you think of Sonia Sotomayor's nomination to the Supreme Court??

[billyjoe]

Karel, "I am interested to hear how small government would combat racism."

I guess the minorities will have to fend for themselves and hope for good treatment by the majority. Just like 150 years ago. And we know how that worked.

-------------billy

[Karel]

Aha! Now we are talking. Unfortunately you give a very incomplete picture of the research. The point you cite is as I expected. But what does this mean for the New Haven case? The conclusion from the article is that the disparate result from the test for lieutenant is significant, that of the test for captain is not. (BTW, for the statistically completely illiterate, this does not mean that the lieutenants test was biased, but only that it tests as "probably biased", according to the methodology used in the article.)

Another point raised in the article is that the Uniform Guidelines actually allow statistical considerations like these in the assessment of bias in tests and mention the unreliability of the 4/5ths rule that was used as unreliable in cases were so few people were tested as here. The city of New Haven might have interpretated the guidelines too narrowly, or even wrongly.

New Haven also has made other mistakes, such as combining the results of the tests for the general assessment of bias, which is a bad thing to do.

The article indeed has a lot of over-my-head-ness, but in my opinion it has become a distinct possibility that the arguments and decisions of the city might have been an overreaction. It remains a complex case, but perhaps there now is a satisfactory way to reach a conclusion, preferably with other statistical experts chiming in. It really is a pity that the statistical arguments surface only now, when the case is under review by the Supreme Court. This should already have been available at the District level, and one question is why statistical significance testing was not proposed by either party.

I am interested to hear how small government would combat racism.

Regards,

Karel

[LemonButt]

Nitpicking aside, testing is a complex art, with a lot of statistics thrown in to spice things up. I doubt if anyone here is qualified to judge a test. Gut feeling doesn't count.

I actually have a BS in Applied Mathematics and I found this 23 page statistical analysis of the case: http://papers.ssrn.com/sol3/papers.c...act_id=1381293

From the abstract:
"A statistical study of the various criteria considered by the city and lower courts in their review of the data demonstrates that nearly 70% of the time a fair non-discriminatory test for either position will fail the government’s “80% rule” and at least 60% of the time both fair tests would fail this “four-fifths rule”."

I also used a Hypergeometric Distribution (http://en.wikipedia.org/wiki/Hyperge...c_distribution) and assuming each candidate has an equal chance of passing the test with 18 promotions available, the candidates that pass will only be proportional to the test taking population 6.199% of the time with non-whites underrepresented 32.4% of the time. That means 93.811% of the time, there will be a disparity to either whites, hispanics, or blacks. The odds of 17 whites, 1 hispanic, and 0 blacks passing is 0.0132% and the odds of 17 whites and 1 hispanic OR black passing is 0.0287%. Again, this is assuming each candidate has an equal chance of passing the test, which is obviously not a realistic assumption to start with.

Also, a test will always have skewed results. When a statistical model is created, it is using a continuous probability density function in order to approximate a discrete case, such as test scores. It would most likely be the normal distribution in this case (standard bell curve). The Central Limit Theorem (http://en.wikipedia.org/wiki/Central_limit_theorem) states that the larger a population, the less skewed the data becomes. Basically if you have two people taking the test, one black and one white, and the white person passes, it doesn't mean all blacks taking the test can't pass it. When statistical sample populations are analyzed, an unbiased estimator is used, which means instead of calculating average by (sum of scores)/(test takers) it is actually (sum of scores)/(test takers - 1) which produces more conservative numbers that are actually undercalculations instead of overcalculations. If we assume the New Haven Fire Department has 100 black firefighters, which it probably has many more, and only 27 blacks took the test, that means in order to create a statistical model to prove the test is racially biased within 5% standard error (5% or more is considered significant in statistics) and 95% confidence, you would have to test 80 (79.38 actually) black firefighters, which is nearly three times the number who actually took the test. If the number of black firefighters is higher than 100 (which is most likely is), then that number actually goes up.

In the end, the whole thing is a joke and government is screwing us again. They've created a system of "equality" which fails statistically more than 60% of the time. Just one more reason why I'm an advocate for small government.

[Karel]

I would be interested in seeing a test with a racial bias...seriously. How can a test be racially biased exactly? Are candidates left in the sun and the first person to sunburn gets bad marks? Are they asking questions about watermelon, fried chicken, and refried beans or what? To say a test is racially biased is ludacris in my mind unless someone can come up with some examples to prove me wrong.

The word is "ludicrous". Nitpicking aside, testing is a complex art, with a lot of statistics thrown in to spice things up. I doubt if anyone here is qualified to judge a test. Gut feeling doesn't count.

Here is a great article I found on the Sotomayor nomination compared to minorities in the past: http://www.americanissuesproject.org...relations.aspx

By the way, Wikipedia says the population of New Haven is 37% black, yet only 23% of candidates taking the test were black. It seems more likely that the fire department is racially biased in their recruiting than a written test or even a physical test being racially biased.

I don't think the article is all that helpful, but the point of the fire department itself being biased against minorities may be a good one, and it might be part of the problem. It doesn't help in finding a solution, however. The fact that the firefighters chose to challenge the city's decision as "racism" might point in this direction. Peanut's comments are also plausible.

In all seriousness, can anyone come up with questions on a test that would promote racial bias?

See above

Perhaps a better guideline would have been: whenever a test has a skewed result, it must be withdrawn or scrutinized. But then imagine the fun in the courts when it comes to judging a test for minority bias, when testing requires such specialized knowledge, with experts, and counter-experts, and general over-everyone's-head-ness.

Regards,

Karel

[peanuts]

The minorities that would have been able to pass an aptitude or ability test most likely would not be applying to be a fireman. Unless they really wanted to be firemen, they would most likely be getting jobs that pay higher and have better benefits. With these qualified potential firemen not applying, then you are left with the bottom of the barrel minorities that wouldn't be hired even if all the applicants were minorities.

And lemonbutt makes a great point regarding a test that IS racially bias. I'd like to see one as well.

[LemonButt]

I would be interested in seeing a test with a racial bias...seriously. How can a test be racially biased exactly? Are candidates left in the sun and the first person to sunburn gets bad marks? Are they asking questions about watermelon, fried chicken, and refried beans or what? To say a test is racially biased is ludacris in my mind unless someone can come up with some examples to prove me wrong.

Here is a great article I found on the Sotomayor nomination compared to minorities in the past: http://www.americanissuesproject.org...relations.aspx

By the way, Wikipedia says the population of New Haven is 37% black, yet only 23% of candidates taking the test were black. It seems more likely that the fire department is racially biased in their recruiting than a written test or even a physical test being racially biased. In all seriousness, can anyone come up with questions on a test that would promote racial bias?

[Karel]

i know that i want to know that i have the most professional and most qualified in my house at that moment. i really dont understand the premise of creating a better test. does she mean to say one that is more conducive to latinos or blacks to the point where it finally becomes more proportional.

Yes, but not to the point of compromising the test for ability.

is it racist or wrong because any group of people arent smart enough to answer the questions on the test and can it ever be the fault of the people taking the test not being qualified enough to pass the test with a high enough mark to win a spot legitimally.

No, that isn't racist, but that isn't the point. The point is that a test that supposedly tests for ability, but where one racial group underperforms, might be a test that not truly or only tests for ability. That is the problem the city of New haven is faced with. Federal guidelines even presume a violation of Title VII of the Civil Rights Act (about minority discrimination by employers) when a test gives such skewed results. Given the law and the federal guidelines, the city cannot accept the results of the test without being found guilty of violation of the anti-racism law. This must have the reason that the district judge gave summary judgment: it is impossible to see how the city could have acted otherwise within the law.

Of course it is not clear that the test was racist. The city had hired a consultant with the express order to eliminate potential race bias in the test. But because of the guidelines, the result condemns the test as racist. Conclusion: the guidelines are too strict, as even a non-biased test might from time to time result in racially skewed results (but should do that in both directions, which however will not be apparent in one-off tests like this one). But this is not something the courts can do very much about. Especially not as looser guidelines might mean too loose rules.

It really is a devil of a problem and I see no satisfactory solution.

if any group wants to increase their presence and move up the ranks they should have to be as qualified as everyone else who gets higher marks. my mind isnt necessarily made up but i know how i feel and who i sympathize with.

Regards,

Karel

[skiracer]

Like all things everything else political...people already have their minds made up on both sides of the aisle.

i know that i want to know that i have the most professional and most qualified in my house at that moment. i really dont understand the premise of creating a better test. does she mean to say one that is more conducive to latinos or blacks to the point where it finally becomes more proportional. is it racist or wrong because any group of people arent smart enough to answer the questions on the test and can it ever be the fault of the people taking the test not being qualified enough to pass the test with a high enough mark to win a spot legitimally. if any group wants to increase their presence and move up the ranks they should have to be as qualified as everyone else who gets higher marks. my mind isnt necessarily made up but i know how i feel and who i sympathize with.

[yoyomama]

Like all things everything else political...people already have their minds made up on both sides of the aisle.

[Karel]

For people who like facts, here's a description of Ricci v DeStefano, with extensive references.

The case, as I understand it from that article, is as follows. After a qualifying test for higher positions in the New Haven fire department, it appeared that no black candidates (27 of 116) qualified. The city decided that the test might have been flawed, in the sense of racially biased, and decided to throw out the results, "citing a desire to avoid violating Title VII of the Civil Rights Act." According to Wikipedia, "Federal guidelines presume discrimination when a test has such a disparate impact on minorities."

The case appeared before the district court, which ruled in summary judgment that the city had the right to do so. The Appeals court, in which Sotomayor served, but not as president, confirmed this ruling.

The Supreme Court now has the case under review.

My conclusion is that this not a case of racism. On the contrary, it is a case against the right (or rather: duty) of the city of New Haven not to act in a racist way. If the test was indeed racially flawed (which by the way is not clear from publicly available sources, as the test is not published), the firefighters who passed can not be considered "the best".

The role of the Appeals court is limited. It can not judge in the case, only whether the lower court has judged well. In the case for the Appeals court, Sotomayor stressed the question whether the city had the right to review the test for flaws and if possibe create a better test (note 6 in the article). Apparently this was not contested by the firefighters’ lawyer, and in my opinion that makes their case very weak.

Regards,

Karel

[IIC]

Sonia...She's just so GhettoFabulous http://www.urbandictionary.com/defin...hetto+fabulous

[skiracer]

You're probably right. The sad thing is he would be working in a field against his will and given shelter and three square meals a day as opposed to the current system where I work hard and get taxed against my will and the proceeds are given to someone else to provide them shelter and food who doesn't do anything productive to earn it. It's pretty sad when slavery is a more efficient system than welfare when it comes to punishment and incentives, isn't it?

interesting statement. i need more time to digest that idea. there is alot going on there with the entitlement programs and whether or not it is a positive thing for everyone regardless of white or black. but an interesting statement.

[yoyomama]

You're probably right. The sad thing is he would be working in a field against his will and given shelter and three square meals a day as opposed to the current system where I work hard and get taxed against my will and the proceeds are given to someone else to provide them shelter and food who doesn't do anything productive to earn it. It's pretty sad when slavery is a more efficient system than welfare when it comes to punishment and incentives, isn't it?

Woah...I think I'm going to pass on this one.

[LemonButt]

If the founding fathers were alive today they'd be having Obama picking cotten for them....and no I'm not black.

You're probably right. The sad thing is he would be working in a field against his will and given shelter and three square meals a day as opposed to the current system where I work hard and get taxed against my will and the proceeds are given to someone else to provide them shelter and food who doesn't do anything productive to earn it. It's pretty sad when slavery is a more efficient system than welfare when it comes to punishment and incentives, isn't it?

[yoyomama]

If the founding fathers were alive today, I'd bet a kidney that there would have been more tea parties and organized resistance, not only during Obama's administration, but Clinton's as well as Bush I and II.

If the founding fathers were alive today they'd be having Obama picking cotton for them....and no I'm not black.