September 27, 2012

FDNY passes La Griffe 101

Back in 2007, the Bush Administration sued the Fire Department of New York, which lost 343 men on 9/11, for disparate impact discriminating against blacks and Hispanics by using a firefighter hiring test that asked fairly hard questions about firefighting. Now, the FDNY has finally learned the lesson that La Griffe du Lion pointed out years ago: If you want to please the feds, you can minimize disparate impact (percentage point differential) of the hiring test by making it extremely easy, then hiring randomly. 

From the NY Daily News:
Scores for blacks and Hispanics soar on newly revamped FDNY test  
More than 42 percent of minority-group members earned grade of 97 or higher and likely to be hired by FDNY over next four years 
Blacks and Hispanics scored significantly better on the newly revamped FDNY exam, a shift that could lead to firefighter jobs for a many as 42% of the minority-group test-takers, city officials revealed Tuesday. 
The would-be minority-group firefighters far outscored those who sat for the previous three exams — the results of which were deemed discriminatory and tossed out by a judge. 
Bravest hopefuls who nail a grade of 97 or higher are considered likely to be hired by the FDNY over the next four years. 
Nearly half — 42.3% — of minority-group members who took the test cleared that hurdle, officials said. 
That’s much higher than top-scoring minorities who took the 1999 exam (14%), the 2002 exam (16%) and the 2007 exam (33%). 
The results of all three earlier tests were thrown out by Brooklyn Federal Judge Nicholas Garaufis in response to a 2007 lawsuit from the Justice Department and the Vulcan Society of black firefighters. ... 
The exam’s pass rate was high. Of the 42,231 people who took the test, 40,426 passed with a score of 70 or above. 
FDNY Battalion Chief Paul Mannix, an outspoken critic of the judge’s decision who believes the original lawsuit was bogus, said he still has questions about how the experts scored the exam. 
“On the face of it, I have no confidence in the test and the list that will come of it,” Mannix said.

Now, you might think this could make a good GOP campaign issue, but, of course, it was the Bush Administration that filed the lawsuit against the FDNY in the first place.

39 comments:

Anonymous said...

I believe we call that "race norming."

Anonymous said...

When you going to talk about Bibi's bomb?

NOTA said...

I noticed the article didn't mention what fraction of white applicants passed. If the graded it fairly, something close to 100% would be my guess.

Anonymous said...

Judge Garaufis must have passed a bar exam at some point in his life. I'd guess that this bar exam was at least as tough as the firefighter tests he's thrown out, its impact at least as disparate.

Two differences:

1) Neither Garaufis nor any other authority figure I'm aware of is waging a war on bar exams. He's a hypocrite.

2) Competence tests are far more effective at raising the quality of firefighting than at raising the quality of lawyering or judging. Case in point: Garaufis doesn't appear to be incompetent. After all, he IS getting what he wants in this case. What's defective is his motivation. He's like a firefighter who knows his craft well, but uses this knowledge to start fires. This sort of problem - defective motivation - is far more common among lawyers than among firefighters. Consequently, competence tests are more effective at weeding out bad firefighters than at weeding out bad lawyers. Society needs firefighter tests more.

Severn said...

The results of all three earlier tests were thrown out by Brooklyn Federal Judge Nicholas Garaufis in response to a 2007 lawsuit from the Justice Department and the Vulcan Society of black firefighters. ...



Would it be impertinent to inquire as to what sort of testing process led to Nicholas Garaufis becoming Judge Nicholas Garaufis, and whether that testing process had a disparate impact against blacks?

That's the "elite" for you ... the process which led to their own position is sacrosanct.

Anonymous said...

I suspect that the Bush administration was trying to punish NY.

Anonymous said...

Steve, you made a mistake in your summary of "La Griffe 101." If you want to minimize the percentage point difference between racial groups, then yes, you should either make the test very easy or very hard. For example, a 95% black and 99% white pass rate on an easy test or a 1% black 5% white pass rate on a hard test clearly minimizes the pass rate differential. However, if you want to minimize the disparate impact of a hiring test, then you can only make it very easy. Making it very hard (and then hiring randomly) will only exacerbate disparate impact.

For example, assuming intelligence is normally distributed, a white mean of 100, a black mean of 85, and a SD of 15 for both groups, a test with a white pass rate of 50% should have a black pass rate of ~16%. Making the test easier so that we have a white pass rate of 97.7%, we get a black pass rate of 84%. But making the test harder to reach a white pass rate of 2.2%, we get a black pass rate of 0.13%. As you can see, reducing the magnitude (white pass rate - black pass rate) can be achieved either by making the test really easy or really hard. However, as the test becomes harder, the ratio of whites passing to blacks passing always becomes larger, and as the test becomes easier, the ratio of whites passing to blacks passing always becomes smaller. In a disparate impact case like this one, they consider relative hiring rates and so a harder test would only exacerbate the "problem." In the hard test, a 2.2% white pass rate and 0.13% black pass rate gives a ratio of 17:1, while the ratio is only slightly above 1:1 when we make the test so easy that 97.7% of whites pass.

Steve Sailer said...

"Steve, you made a mistake in your summary of "La Griffe 101.""

Right. Thanks. I'll change it to take out the "or incredibly hard" part.

Anonymous said...

A 95.7% passage rate. That's some useful test. I wonder the judge would think about a bar exam with a 95.7% pass rate.

Anonymous said...

Steve, I don't mean to nitpick, but the text in parentheses should be something like (white-black pass rate ratio) rather than (percentage point differential). You CAN make the percentage point differential tiny with either a very easy or very hard test. I.e. a 99% white pass rate and 95% black rate on an easy test, vs a 5% white pass rate and 1% black rate on a hard test. In both cases we have a 4 percentage point differential. However in a disparate impact case like this one they look at how much the test affects an applicant's likelihood of getting a job, and here it is relevant to consider pass rate ratios. So, on a test with a 99% white pass rate and a 95% black pass rate, the pass rate ratio is ~1.05 and the test only slightly decreases the chances of blacks relative to whites. But on a test with a white pass rate of 5% and a black pass rate of 1% we see a white is 5x more likely to pass, and so the hard test would make blacks ~5x less common relative to whites, even though the pass rate differential in percentage points is small.

ATBOTL said...

The GOP and conservatives have dropped affirmative action as an issue. During the 70's, 80's and into the mid 90's, this was big conservative issue that candidates often used to lure white voters. What happened? We know how Gingrich scuttled efforts to roll back AA in Congress, by why did the talk radio types drop it too? Why didn't other politicians bring it up?

Anonymous said...

I wonder the judge would think about a bar exam with a 95.7% pass rate.

Or a law school which passed 95.7% of students.

Matthew said...

"The GOP and conservatives have dropped affirmative action as an issue."

Because some damn fool - perhaps it was Karl Rove - convinced them it wasn't electorally viable to press the matter. And yet the GOP still gets almost non of the black vote, and maybe 1/3rd of the Hispanic vote.

They don't have to press the issue during elections - just pass laws and issue executive orders to eliminate AA after winning.

Anonymous said...

Sample question.

If there's a fire...

A. pour gasoline on it

B. sing and dance

C. toast marshmellows

D. sit your ass on it

E. pour water on it

Anonymous said...

Like I said...

it's over, it's over, it's game over.

All we can wait for now is the Great Depression from liberal overspending and lunacy.

Yan Shen said...

I think the world would be a much better place if we all embraced cognitive elitism.

Anonymous said...

IT's a little more complicated. They didn't just use the "make the test pass/fail and make passing easy" strategy.

There are apparently two important thresholds--a 70 score is passing, but to actually be hired you need a 97 or higher. (Otherwise you sit forever on a list waiting to be called, and your number doesn't get called before the next round of tests).

We are told that "Nearly half — 42.3% — of minority-group members who took the test cleared that [97]hurdle, officials said."

Digging further into the article, "There’s now a total pool of 9,417 candidates who reached the 97 mark and are likely to be considered for hiring within four years.....The exam’s pass rate was high. Of the 42,231 people who took the test, 40,426 passed with a score of 70 or above."

So 9,417/42,231 reached the 97 mark, about 22%. So "minority-group members" passed the test at twice the rate of the total test-taking population.

--Discordiax

Auntie Analogue said...

Headline, September 2017:

BLOOMBERG AT LOSS TO EXPLAIN SKYROCKETING FIRE INSURANCE PREMIUMS

Anonymous said...

"So "minority-group members" passed the test at twice the rate of the total test-taking population."

Um, no... you have no idea what you're talking about.

Jeff said...

"I think the world would be a much better place if we all embraced cognitive elitism."

How do you account for creativity and for mental courage? Those are two amazing traits. I favor keeping our society pretty creative and pretty smart. Most really smart people are wimps, suckers and weaklings. They drive me nuts with their pretensions and their overall weak will to stand for justice and to make manly decisions.

TH said...

However, if you want to minimize the disparate impact of a hiring test, then you can only make it very easy. Making it very hard (and then hiring randomly) will only exacerbate disparate impact.

Not necessarily. If you make the test so hard that everybody must guess every question, all races will have the same average score and distribution of scores.

Michael McDaniel has written a rather sarcastic guide to gerrymandering in personnel selection.

Steve Sailer said...

If you make the test extremely hard, you reduce the white minus black passing rate differential, but you increase the white divided by black passing rate differential. And the latter is what the EEOC's Four-Fifth's Rule focuses upon.

IHTG said...

If you make the test extremely hard, you reduce the white minus black passing rate differential, but you increase the white divided by black passing rate differential. And the latter is what the EEOC's Four-Fifth's Rule focuses upon.

Steve: Wouldn't it still work, though? If you make a test REALLY difficult, the sound of the whites complaining how hard it is in the media drowns out the sound of the minorities. The sheer difficulty overshadows the disparate impact. That's what matters.

Anonymous said...

My understanding of the Fire Dept racial spoils system is Whites get the management jobs, Blacks and Hispanics get the lower level jobs. The children of the White managers get fast tracked to future manager jobs.

Since I do not have any relatives in the FDNY as a white person why should I care? Vallejo and Central Falls are examples of largely White police and fire bankrupting the largely white cities.

How about advocating privatizing the fire department?

Asset Co:
http://www.assetco.com/

rightsaidfred said...

So 9,417/42,231 reached the 97 mark, about 22%. So "minority-group members" passed the test at twice the rate of the total test-taking population.

--Discordiax


As 11:48 anonymous pointed out, this isn't accurate: roughly half the test takers were minorities of some kind (it would be interesting to see the exact makeup, but one can't expect too much information in this day and age), so it looks like non-minorities still did slightly better, but for the most part the test looks like it randomly assigned scores, thus giving us the La Griffe result.

Note well that we are closing in on the majority-minority nexus. Tsk. So much more left to do.

Also note that FDNY is 650 empty slots off their funding level. I don't hear much about their work going undone, or volunteer citizens having to fill the gap. Maybe they could cut the staffing level and give the beleaguered taxpayer some relief.

Cail Corishev said...

It sounds like the chief knows what the score is -- he's going to be handed a lot of poorly qualified people. So what will he and other departments do about it? Will they need bigger budgets so they can hire extra people, figuring 10 average firemen can do the work of 5 excellent ones? Will they just scale back their expectations, leading to more buildings destroyed and fires spreading? Or will they use the difficulty of the job to weed out the unqualified, firing them for cause or encouraging them to quit -- and if they do that, how long before they get sued for the fact that minorities quit/get fired at higher rates than whites?

It has to be a tough situation. A guy in his position can't say anything against diversity, lest he no longer hold the position. But presumably he does want to fight fires and save lives as effectively as possible, so he's going to try to find some way to hire and keep the best, while pretending that's not what he's doing.

Anonymous said...

I just hope the fire departments closest to where Garaufis lives hire most of the minorities who now "qualify" for the job.

Anonymous said...

So in response to: "Um, no... you have no idea what you're talking about." You're right, I didn't, because the Daily News f--- up, probably because they are regurgitating the BS they were spoonfed.

How did this test produced a higher NAM than white pass rate? It didn't, even though the Daily NEws wrote: "Nearly half — 42.3% — of minority-group members who took the test cleared that [97]hurdle, officials said." That didn't happen.

What happened was, 42% of the pass-97s were minorities. The pass-97s included 1859 blacks + 2131 Hispanics = 3990 NAMs/ 9417 total = 42%.

I withdraw my unvoiced suspicion that minority applicants were awarded huge bonus points for skin color, either by a blunt quota or in a "scored interview" process.

What did happen? FDNY recruited enough NAM test-takers to get enough NAM pass-97s. I wouldn't be surprised at all if the NAM pass-97 rate wasn't lower than for the 2007 test, 33% which was thrown out. OFr course, I'm relying on the Daily News numbers there too.

--Discordiax.

Cail Corishev said...

"If you make a test REALLY difficult [...]"

But they do want to hire some firemen, so they can't make the test so hard that no one passes it. (Sure, they could do that once to make a point, but then they'd have to make it easier and be right back to the same problem.) And the harder you make the test, the whiter those who pass it will be. So the only choice is to make it easy enough that everyone can pass it.

Though I have to say, I'm cynical enough to think that even if you made the test, "Write your name on this piece of paper," there would be a disparate result. After all, that requires that the person had the ability to show up at the right place and time for the test, the ambition to do so, and the literacy to write his name. That already excludes some people.

(Now I'm thinking of those excellent SNL Celebrity Jeopardy skits where "Alex" kept making the final question easier and easier for the celebs, trying to get someone to win, and they never could.)

ATBOTL said...

"Since I do not have any relatives in the FDNY as a white person why should I care?"

Because they will be coming for your job next.

This kind of retarded hyper-individualism is what's killing us.

Anonymous said...

Chalk it up as another victory for the Republicrats!

keypusher said...

Judge Garaufis must have passed a bar exam at some point in his life. I'd guess that this bar exam was at least as tough as the firefighter tests he's thrown out, its impact at least as disparate.

Title VII has been held not to apply to state bar exams, because, you see, bar examiners are not employers, even though you can't get a job as a lawyer without passing the bar exam.

And yes, of course blacks pass the bar at far lower rates than whites.

http://www.law.fsu.edu/journals/lawreview/frames/233/huntfram.html

keypusher said...

Judge Garaufis must have passed a bar exam at some point in his life. I'd guess that this bar exam was at least as tough as the firefighter tests he's thrown out, its impact at least as disparate.

Title VII has been held not to apply to state bar exams, because, you see, bar examiners are not employers, even though you can't get a job as a lawyer without passing the bar exam.

And yes, of course blacks pass the bar at far lower rates than whites.

http://www.law.fsu.edu/journals/lawreview/frames/233/huntfram.html

Brendan O said...

Wait, slow down. I would need to have the answer to a couple questions before I jumped to any conclusions on this.

First to what extent do other factors besides the test play in the NYFD hiring decisions. I've written about this here before, but most of departments where I live won't consider you for a position at the FD unless you are a time served medic. I'd be really surprised if the NYFD wasn't at least giving significant hiring preferences to its paramedics. While I personally don't like the idea of devolving the medic career into a tech requirement for FD service, it's definitely much more difficult to acquire a paramedic's license (cognitively and in every other respect) than it is to pass any fire fighter's test. EMS is essentially the little brother of fire, particularly in the cities. They have much lower pay and worse working conditions despite the fact that the service is at least as tough on its members as fire, with only a slightly lower job mortality rate.

On the same note, they are taking military service into consideration and automatically eliminating anyone with a criminal record. Assuming all of the above is in effect to some degree, the top 1% of test takers on even a joke test would still seem to be capable of yielding satisfactory hirees.

Truth said...

"This kind of retarded hyper-individualism is what's killing us"

The conservative movement was founded upon hyper-individualism there, son. Now tell me Barack, do you prefer COMMUNE-ism?

Anonymous said...

Is there a paramedic test? Has it been dumbed down? Some people might not mind stupid firemen, but how do they feel about stupid paramedics?

goatweed

pat said...

There is a reasonable case to be made for dumbing down certain civil service tests.

I went in the army after my first two years of college. In the Reception Center (that's before Basic Training) one of the sergeants called out for college graduates. No answer. So he next asked for those with two years of college or more. I - fool that I was - volunteered.

They were taking dental X-Rays. My job was to stand by the door and the recruits would spit the films out into my hand as they left the room. Lovely job.

I asked the sergeant why they needed a college graduate to be a human spitoon? He said, quite reasonably, that they had found that collge men required less instruction.

Of course, I learned never to volunteer for anything ever again, but the other lesson was that even the lowest and stupidest job is easier to teach to smarter people. This is the kind of thinking that leads to absurd intellectual requirements for dumb jobs - like firefighter.

When a few years later I was considering graduate school in psychology I learned about the San Francisco fire department. At that time the fire houses were filled with grad students who did well on standardized tests. Most fire districts in the "better" neighborhoods never had fires. So these full time civil servant firemen had a lot of free time. Some built boats or restored cars but many sought advanced degrees. The real intellectual hot spot in the city in those days was the fire department.

I knew some of these fire fighting scholars. They told me that there really was something to understanding fires. One rookie told me that an old timer told him to tear a hole in what seemed to be a perfectly good wall. To the rookie's astonishment there was a smoldering fire inside the wall. That proves that past experience counts. It doesn't prove that you need a doctorate.

I had no problem with relaxing the testings standards so as to allow more blacks into the lily white SF fire houses. I was however exercised about changing admission standards to include more women. Firemen used to have stringent physical standards to make sure they were strong enough to carry unconscious victims out of a burning building. Women couldn't pass those tests so they dumped those tests.

If I were passed out from smoke inhalation it would be fine with me for a big strong black fireman to come into the burning building and carry me out. Physical strength is a real job requirement. I wouldn't particularly care about his IQ.

Albertosaurus

Brendan O said...

"Is there a paramedic test? Has it been dumbed down? Some people might not mind stupid firemen, but how do they feel about stupid paramedics?"

Paramedic's take the national registry test which is administered by a private organization. When they pass they are certified and can generally apply for a license to practice from the state they want to work in. Before taking the test, they have to have an associate's degree or go to a paramedic's school and they have to have 2 or 3 semesters worth of prerequisites in biology and anatomy before being admitted to a school or program. It's basically works the same as nursing. There's an extra step with paramedics as you have to already be certified as an EMT and are often required to go out on a certain number of ambulance calls before you'll be admitted to paramedic's school.

The test is a computer based, adaptive exam. Most people don't find the certification test or anything else about the process to be particularly easy. I would guess that the test is protected from disparate impact suits by the fact that the group administering it is non-governmental.

Cail Corishev said...

"[T]he other lesson was that even the lowest and stupidest job is easier to teach to smarter people."

That's something most people don't realize. They know that being smart makes you capable of doing smart things like rocket science, but they don't realize that it makes you better at learning to do easy things too. If you're digging ditches, the smart guy will be able to learn to use a shovel faster than another guy.

The other day, I got a new credit card processing machine for my store. My employees knew how to use the old one without any trouble. The new one requires one more step in the middle, and it has a touch screen instead of buttons for some things. It took me about 30 seconds to be comfortable using it. It completely flabbergasted some of my people. These aren't stupid people, but when it comes to learning anything new, any difference in IQ really shows up.