On May 3, I received an email from Professor Raj Chetty of Harvard University, informing me that his famous paper on value-added assessment of teachers was being published by the American Economic Review. The paper has three authors: in addition to Chetty, the other authors include John Friedman and Jonah Rockoff, also at Harvard. When the paper was first released, it was reported on the front page of the New York Times, one of the authors discussed it on the PBS Newshour, and President Obama referred to it in his 2012 State of the Union address.
The New York Times story appeared on January 6, 2012. it began thus:
“WASHINGTON — Elementary- and middle-school teachers who help raise their students’ standardized-test scores seem to have a wide-ranging, lasting positive effect on those students’ lives beyond academics, including lower teenage-pregnancy rates and greater college matriculation and adult earnings, according to a new study that tracked 2.5 million students over 20 years.”
The reporter noted that the effect of a single “high-value” teacher was actually quite modest: “The average effect of one teacher on a single student is modest. All else equal, a student with one excellent teacher for one year between fourth and eighth grade would gain $4,600 in lifetime income, compared to a student of similar demographics who has an average teacher. The student with the excellent teacher would also be 0.5 percent more likely to attend college.” But think of the aggregate effect on an entire classroom: “Replacing a poor teacher with an average one would raise a single classroom’s lifetime earnings by about $266,000, the economists estimate. Multiply that by a career’s worth of classrooms.” President Obama cited the aggregate income gain for a classroom in his State of the Union address 18 days later.
This was the takeaway from the authors, as reported in the New York Times:
“The authors argue that school districts should use value-added measures in evaluations, and to remove the lowest performers, despite the disruption and uncertainty involved.
“The message is to fire people sooner rather than later,” Professor Friedman said.
“Professor Chetty acknowledged, “Of course there are going to be mistakes — teachers who get fired who do not deserve to get fired.” But he said that using value-added scores would lead to fewer mistakes, not more.
“Still, translating value-added scores into policy is fraught with problems. Judging teachers by their students’ test scores might encourage cheating, teaching to the test or lobbying to have certain students in class, for instance.”
The Chetty, et al, study supported VAM, which was the central feature of Race to the Top. Fire teachers sooner rather than later. One great teacher can produce lifetime gains.
Over the past few years, as more districts have implemented VAM, it has turned out to be far more complicated than the economists predicted to determine which teachers would produce great scores year after year, and which would not. Teachers were rated effective one year, ineffective the next year. Those who taught English learners, the gifted, and students with disabilities were less likely to get big gains. It turned out that VAM is affected by the composition of the classroom, since students are not randomly assigned.
But their paper continues to be the lodestar of VAM research.
Whereas it had originally appeared as a single paper published by the National Bureau of Economic Research, the editors suggested the paper was so important that it should be split into two papers and published separately. The last time this had happened was in 1971, for papers on taxation that had won two Nobel Prizes.
Here are the papers.
Professor Chetty’s email was addressed to me and Audrey Amrein-Beardsley, who has written extensively and critically about value-added assessment. In addition to her recently published book on VAM—Rethinking Value-Added Models in Education: Critical Perspectives on Tests and Assessment-Based Accountability—she writes a blog called VAMboozled that I often cite.
For the record, I have never met Raj Chetty, and I have met Beardsley once, when she interviewed me for an oral history archive.
I asked Beardsley if she would be willing to review the latest iteration of this now famous study of VAM, and she did, here on her blog.
Beardsley notes that there is a divide between econometricians, like Chetty, Friedman, and Rockoff, and educational researchers, who often feel some obligation to visit classrooms and see the effects of policies, not just analyze data from a great distance, without reference to context or something like reality.
Professor Chetty and I exchanged several emails. I asked for his permission to post our exchange. He said that he preferred that I not post his comments, which were invariably polite, but of course I was free to post my comments to him.
So here goes. This was my first response:
Dear Professor Chetty,
I certainly agree that teachers are valuable. I had some wonderful teachers
as I was growing up, also some mediocre ones, and a few really bad ones. I
went to an ordinary public school system in Houston, not an elite private
school.
I wish that this sentiment about the value of teachers was all that came
from your vast publicity machine.
Instead, we get more high-stakes testing, more test prep, more phony claims
that the work of my fourth grade or fifth-grade teacher was responsible for
my not getting pregnant when I was 15. Maybe my lifetime income was
increased by my sixth-grade teacher, though I doubt it. Funny, I was one of
eight children. We all had the same teachers, and we all turned out
differently. Some of us did well in school, others nearly flunked out. Was
it the fault of our teachers?
I know you love your celebrity–and hobnobbing with Obama and Duncan and
supporting their emphasis on testing and firing teachers sooner rather than
later—but think of the harm that you do to millions of children and their
teachers by the way you publicize your work. Do you feel good every time you
read about a teacher who is graded based on the work of children she never
taught? Or the “highly effective” teacher who was rated ineffective the next
year based on test scores? Or the precipitous decline in the number of
people who want to be teachers because of the non-stop attacks on teachers?
I don’t think your positive message is getting through. All people hear is
that you want those lousy teachers whose kids get low scores to be fired.
Now.
Diane Ravitch
On May 5, I wrote to both Raj and Audrey (we had reached a first-name basis):
Raj and Audrey,
I don’t know whether my thoughts advance or retard this informed discussion.
I look at the Chetty, etc. study as comparable to a pilot in a bomber
dropping a bomb on a city 30,000 feet below. He didn’t construct the bomb,
he doesn’t know how it hurts the people below, he can’t be held responsible
if his good intentions went wrong.
I invite you to read this blog by a teacher in Oklahoma:
http://bluecerealeducation.blogspot.com/2014/05/ms-bullens-data-rich-year.html
The odds are that he never heard of Raj Chetty. But look what Raj Chetty has
done to the quality of education, the students, and the teachers in
Oklahoma. Is this something to be proud of?
Your work–not yours alone, of course–has encouraged a technocratic
approach to education that would never be tolerated in our nation’s elite
private schools.
The pursuit of higher test scores on stupid multiple-choice standardized
tests does not improve education: it corrupts it.
Those who care deeply about humanistic education, about the life of the
mind, about deep learning, find your work–no matter how technically
perfect–utterly appalling. It drains education of joy and discovery and
makes everyone a slave to Pearson.
I would love to discuss this further with you over a glass of wine. I can’t
believe you do not understand the pernicious effects of your famous study,
featured on the first page of the New York Times, on the PBS Newshour, and
in President Obama’s State of the Union Address.
It seems to be my life work to insist that education is far, far more than a
score on a standardized test. Somehow, I suspect you agree. You are far too
intelligent not to.
Diane
Later on the same day, May 5, Raj responded, and I wrote:
Thanks, Raj,
A question and a comment.
My question: Could I publish our exchange on my blog? I get about 25,000-40,000 readers daily. But I would publish nothing without your permission.
My comment: Race to the Top has incentivized the use of VAM in most states. Your study has been cited by Obama and Duncan as evidence that they are on the right track, that it is “bad teachers,” not poverty, that cause low test scores.
Based on the real-world effects of VAM on real children and real teachers, I conclude that VAM has limited use, perhaps informative in looking at the effects of policies and programs (faithfully enacted, which they seldom are) in a school or a district, but of zero value in assessing individual teacher quality. As you must know by now, the ratings for individual teachers are unstable, and may change if a different test is used or unstable for no apparent reason at all. Teachers intuitively know that their ratings reflect the composition of the class, not their “quality” or efficacy as teachers. Even if VAM did work–and it does not–it would keep every teacher singularly focused on standardized tests, which narrow the curriculum, encourage schools and teachers to avoid the neediest students, promote test prep and cheating, and have other perverse effects.
At the end of the day, I as a mother and grandmother would not want my offspring to be enrolled in a school where standardized tests dominate teaching and learning. And that is precisely what VAM is doing to our nation’s public schools.
My third grandson enters third grade in a New York City public school next September. I hope by then that the opt out movement has grown so strong that teachers cannot be subjected to unfair and inaccurate VAMs. I will do whatever I can to encourage parents in every school district in the U.S. to keep their children home on testing day. That seems to be the only way that the giant standardized testing machine can be stopped.
Your work has been crucial in promoting standardized testing as the measure of teacher quality, even though major scholarly organizations disagree (the American Educational Research Association, the National Academy of Education, the American Statistical Association).
If you have modified your views (message: “fire teachers sooner rather than later”); if you have learned anything new since you first introduced your findings, I would love to know about it.
I repeat that I do not have the technical ability to argue algorithms with you. Your study may be technically brilliant. But its consequences for the quality of education and the lives of children and teachers have been disastrous. In its current application, it is Junk Science. Since I feel certain you don’t want to be remembered in history as the economist who sponsored Junk Science and treated children as data points, I hope you will give me reason to believe that you have rethought the conclusions of your study and provided clear warnings about the limitations and misuses of VAM.
Diane
We ended with the understanding that I would not quote his words or paraphrase them. I think I was true to that understanding.
Now, as I told him, I am not an economist, and I lack the technical proficiency to critique his paper. Maybe it will win two or three Nobel prizes. If all it says is that teachers are valuable, I agree. If it says that teacher affect eternity, I agree.
But if he really expects me to believe that my fifth-grade teacher (or was it my fourth-grade teacher) caused me to get higher test scores, and that because of her and my higher test scores, I did not get pregnant when I was 15, I think this is just plain silly.
This strikes me as the kind of study that brings huzzahs from economists for its technical precision, but is unrelated to the messiness of real life. The numbers may all add up, but there are no living, breathing students or teachers here, just data.
It is so incredibly frustrating to me to see economists and policymakers playing with the lives of children and teachers as if they were ants seen from a far distance or merely data points. I recommend to my new friend Raj a book by Yale Professor James C. Scott titled “Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed.” It changed my life. Maybe it will change his too.
I can’t wait to see Professor Chetty’s VAM score when our administration applies the same accountability regulations to colleges and universities.
He’ll be waving his hands about and protesting — “Rating COLLEGE PROFESSORS based on the test scores of their students?!? Woah, woah, WOAH! Let’s not get CRAZY here!”
Then again — if he teaches at the “right” sort of college — maybe he won’t mind.
Which he does. It’s Harvard.
YUP
The “reformers” will have to mandate frequent standardized testing in colleges and let it be known they don’t trust professors to construct, administer and score their own tests.
“fake reformers”
Indeed. Entrepreneurs and Wall Street investors are, apparently, the only people left in our society who CAN be trusted.
His hubris is astonishing…families=0, teachers=100? And yes, what will his VAM score be when the destructive forces he mindlessly unleashes gets turns to higher ed? Good point, Mr. Jones.
Do tfa “teachers” figure into VAM? Do they get “vammed” over their 2 year less or more stint at teaching in the neediest charter schools across the nation?
I cannot for the life of me understand the logic of getting the best teachers in the classroom, while practically orgasming over the tfa scabs.
I will never believe a dog is a duck, just because “they” say so, over and over.
As far as I know, VAM is not required in charter schools, since they are virtually unregulated. Competition and the free market are supposed to decide the fate of TFA “teachers” in privatized schools.
As a TFA “teacher” (why the quotes?) working in a public school (whose district had 40 unfilled positions at the start of the last school year that were never filled) where VAMs are a significant part of my evaluation, I implore you not to mix your anger over VAMs with TEACHERS (TFA or not) who are working to improve the lives of students. I am actively against VAMs, and I “cannot for the life of me” understand why you are projecting your ire onto TFA teachers.
RAJ CHETTY: “Of course there are going to be mistakes — teachers who get fired who do not deserve to get fired.”
I want to repeat that again…
RAJ CHETTY: “Of course there are going to be mistakes — teachers who get fired who do not deserve to get fired.”
And again…
RAJ CHETTY: “Of course there are going to be mistakes — teachers who get fired who do not deserve to get fired.”
And again…
RAJ CHETTY: “Of course there are going to be mistakes — teachers who get fired who do not deserve to get fired.”
And again…
RAJ CHETTY: “Of course there are going to be mistakes — teachers who get fired who do not deserve to get fired.”
to infinity………………………………………..
I think part of the problem here is the public’s general blindness to the realities of getting rehired as a teacher after you have been fired. In my state, districts MUST hire teachers at whatever salary step they were on prior to getting hired. That means that a teacher on top step (10th) is unlikely to get hired in any other district unless he or she is a rock star teacher, or the district in question is hurting for new hires. Why hire someone at top step when they can hire a newbie for half the price? The teacher in UNABLE to negotiate to accept lower pay and is therefore pretty much out of luck when it comes to finding a teaching job somewhere else.
Rhode Island has a huge number of districts for such a small state. My own has all of two high schools. If I were to lose my job due to a flawed evaluation, it would pretty much be the end of my career unless I wanted to get certified in a nearby state like Massachusetts — where I would have to start over at the bottom step, most likely after subbing for a few years.
In the business world, if you get fired due to downsizing or whatever, you can find another job somewhere else with MUCH more ease.
Their reality is what they are assuming for everyone.
Yes and you are not publicly labeled as a failure.
These are the economists who blew up the economy n 2008 and give the plutocrats justification for their right to run the country.
I’m really curious as to why Raj Chetty did not want any of his words shared verbatim, and even more curious as to why he refused to allow any paraphrasing of what he said.
The Chetty-Friedman-Rockoff study is a good example of economists –– and the mainstream press – making causal inferences based on small correlative measures. It is much ado about nothing. It’s essential finding makes little if any sense.
The essential finding of the study is this: a high-quality teacher creates “more than a quarter of a million dollars” in higher earnings per CLASSROOM of students.
If there are an average of 30 students in these urban classrooms, that amounts to $8,333 per student over a life-time of employment. If that employment career lasts 30 years, then that’s about $278 extra a year, or $23 a month, or $5.75 a week. Wow!
More curiously, the authors of the study offer up this gem in their executive summary:
“parents whose children will earn around $40,000 in their late 20s should be willing to pay $10,000 to switch from a below-average to an above-average teacher for one grade, based on the expected increase in their child’s lifetime earnings”
Does it make sense for parents to spend $10,000 in current funds to ensure that their children can make an extra $8,333 over a life-time? Given the current job market, the off-shoring of jobs, immigrant labor, corporate efforts to drive down salaries and destroy unions, the rise in poverty, the decline in the American standard of living, and the increasing lack of social mobility, and this is their recommendation?
And people keep citing the Chetty study?
Good to hear that the paper is coming out in a peer reviewed journal. There had been some concern exoressed by posters on this blog that it had not gone through that process.
Really, TE, THAT is your takeaway from all this?
The news that the paper is forthcoming in AER is the only bit of new information in this post. There was a great deal of concern expressed by some posters earlier that the research had not gone through the peer review process and those poster’s concerns can be put to rest. Here, for example was what Dr. Ravitch stated just a few months ago:
Critics quickly pointed out that the Chetty study, as it was known for its lead author Raj Chetty, had not been peer reviewed and its findings were overstated.
The post is here: https://dianeravitch.net/2014/02/03/that-chetty-study-fire-more-teachers-with-flawed-data/
When Chetty himself, one of the authors of the study, cautions that some good teachers might get the boot, then clearly, even if the results mean something, the impact of implementing policy based on those results can be reasonably expected to do more harm than good, and this study does not include study of what happens when teachers’ livelihoods are tied to VAMs. But there is accumulating anecdotal evidence that teachers deemed good by their children and parents as well, not to mention teachers who have received awards for their instruction, are receiving low VAM scores where this stuff is implemented, plus other anecdotal evidence that teacher morale is adversely impacted..
Moreover, just because an article survives the peer review process does not mean it is the final word, whatever field of research is involved. Among other issues, the evidence regarding VAMs needs to be reconciled with the evidence from several studies, as Dr. Ravitch has mentioned, that indicate a teacher’s impact on student achievement to range somewhere between 1 and 14%.
Sorry, TE, I didn’t mean to post and run, but I had to get to school to make sure I wouldn’t be causing the 8th graders I see every other day for 100 minutes (we are on a block schedule) to lose $20 a week or something ten years from now.
Peer review by economists, not educators. The economists who helped tank the economy have the audacity to claim expertise on decisions bearing on education. Diane’s analogy with dropping a bomb from 30,000 feet is on the mark, as if a little collateral damage is perfectly fine. Economists do not have to live with the consequences of their long inferential leaps through thin air. The econometric turn in education has been a disaster.
My experience here has been that it is not the methodology of economics that people object to here, but to some of the conclusions reached using that methodology.
Findings by an economist (identified as a “researcher” in Dr. Ravitch’s posting) that “The regression discontinuity estimates suggest that grade retention increases the likelihood of disciplinary incidents and suspensions in the years that follow. The findings also suggest that these adverse effects are concentrated among economically disadvantaged students.” are praised and accepted without a single doubt, yet another working paper by the very same economist using the same methodology looking at the impact of choice on student learning in the District of Colombia are condemned as being obviously wrong.
The original post is here: https://dianeravitch.net/2013/07/19/the-negative-effects-of-holding-kids-back-in-third-grade/
What I understand that Ravitch is saying about the VAM study is not that the results are necessarily inaccurate, in fact she even concedes the study might be “technically brilliant.” What I understand her to be saying, and with which I agree, is that acting on those results could produce great harm. Remember, the study is NOT looking at the impact of using VAMs to evaluate teachers/decide their professional future.
It’s a look at VAMs where no consequences have been applied. An important distinction.
And again, Chetty himself concedes that good teachers might get the ax.
Reinhart and Rogoff were also Harvard economists who published a paper used by many government austerians as justification for cutting support of social programs and safety nets. It was a UMass student Thomas Herndon who uncovered major data errors and questionable statistical techniques that undermined much of the original study.
Peer review is a good thing. But, unfortunately, peer review lags popular media. And politicians jump on the latest media report and create public policy before even considering the validity of a published study.
I am amazed a Harvard researcher would confuse correlation with causation. I’m also amazed they would attempt to extrapolate an initial effect (teaching in childhood) so far into the future. And to link teaching to suggested dependent variables like pregnancy is comical. It is junk science like this that is giving legitimate science a bad name and generating overall mistrust in the general public.
I don’t think Chetty at el confuse correlation with causation. My view of this literature is that we have reasons to believe good teaching is important to student learning independent of empirical regularity. That is certainly the opinion of most posters on this blog. Chetty et al is verifies that we see evidence of this causal relationship in the data.
Chetty is trying to establish a strong relationship between test scores and future earnings, amonsgt other response variables. Yes, there likely is an effect. But they try to quantify it (absurd) based on questionable measurement techniques (test scores), ignore lurking variables, and then project those effects into the future (risky). I always thought teen pregnancy was caused by other means and not my third grade math lesson. But, then again, I didn’t to to Harvard.
The best way to establish causation is controlled experiments. Statisticians need to take a step back, ditch the crystal ball, and be honest with the limitations of many of these predictive models. The researchers also need to use their God-given intelligence and look beyond the numbers (see my earlier post on Wald and WWII planes). The ASA has taken the right approach urging caution.
Economists do not live in a world of controlled experiments, that is why we have developed a set of statistical tools to try and tease out relationships in a world that we can not control.
Here is a good blog entry illustrating the problem that economists have. Robert Gordon is a well respected economist that argues developing country growth rates will be much lower in the future than they have been in the past. The blog entry talks about when we will be able to tell if Robert Gordon is correct.
http://growthecon.wordpress.com/2014/05/20/is-robert-gordon-right-about-u-s-growth/
“Economists do not live in a world of controlled experiments” You can say that again!. They live in a world where it is routine to lie with statistics for payement.
Harold,
What lies do you speak about? The one that says teachers matter? The one that says retention in a grade has profound impacts on students as the years pass?
With all the emphasis on bad teaching, there have been no studies to prove any of those claims or theories are true. But there was a study in the UK that discovered the number of workers who are incompetent is about the same among all professions and that includes teaching. The conclusion of that study was, the results of incompetent workers wasn’t any more significant in the education profession than it was any any other profession and kids survived and had very little loss in learning due to the few bad teachers encountered K to 12.
Therefore, why don’t we also go on a witch hunt for all the incompetent mechanics, doctors, dentists, accountants and economists. etc.?
But in the U.S., that unproven theory has had so much media coverage for so long, that now there is a vigilante hue and cry across the land from mobs of fools—similar to what took place during McCarthyism’s Red Scare—to punish all 3.3 million public school teachers for the so-called 7%, give or take a few percent, who might be incompetent.
Meanwhile, poverty continues to be ignored when study after study offers solid evidence that childhood poverty is a major force in children not learning. Even the Obama White House is planning in 2015 to push for a $75 billion dollar ten year early childhood education plan—that I’m sure the GOP will kill before it has a chance to come up for a vote—because this area is where we can gain the most in K – 12 education.
For instance, in France, early childhood education programs reach 100% of the children by age 4, and poverty was reduced from 15% to less than 7% in a few decades under that program, and France has not had its witch hunt for incompetent teacher yet.
The Chetty-Friedman-Rockoff study tends to get cited by those who endorse corporate-style education “reform.” These folks also tend to buy into the “research” of Eric Hanushek.
Hanushek is a conservative economist who rarely if ever has a positive thing to say about public schools and who’s been caught distorting data to make it support his predetermined conclusions.
Hanushek infamously discounted the Tennessee STAR study findings on small class size because the researchers– meticulous in assigning students randomly to smaller, control, and larger class sizes – hadn’t given students achievement tests BEFORE they entered kindergarten. What a tool.
Hanushek argues that if we can only fire the worst 10 percent of teachers, then test scores would soar and American economic competitiveness would miraculously be restored. And. Pigs. Will. Fly.
In the 1990s the American economy boomed. Tens of millions of jobs were created. Deficits were reduced. The federal budget was balanced multiple times and budget surpluses were generated. Wall Street and corporate America were opposed to the tax increases that allowed much of this to happen (and they pressured their primary puppets in Congress to vote against them). By the way, did anyone think to offer any credit whatsoever to the schools?
So now, after rampant deregulation, Bush-era tax cuts that were ill-advised and unaffordable, two badly-managed and unpaid for wars, the transformation of Wall Street into a taxpayer subsidized casino, massive corruption by bankers and hedge fund managers that continues mostly unpunished, and a still-broken-for-a-lot-of people economy as a result of it all, corporate flacks and charlatans and opportunists blame schools and teachers for economic stagnation. If only we test more, and hold teachers accountable, they say, then teachers can produce for students an extra $5.75 a week a decade-and-a-half or more in the future.
This line of illogic doesn’t even rise to the level of bizarre.
And don’t forget Condoleezza Rice and Joel Klein on the Council on Foreign Relations blaming teachers for putting the entire national security of the Western Hemisphere at risk.
The damage caused by arrogance and flawed thinking can be enormous. Hitler and Mao both had flawed theories and how many died and suffered from that thinking? These arrogant Harvard professors are no different.
People with flawed theories are no different from Hitler and Mao. They’re not arguably similar to them in one specific way. They’re the same.
I think that we can agree that good teaching may have strong positive impacts. That is quite different than asserting that good teaching can be measured reliably by any multiple choice test.
Good teaching can be evaluated by a good evaluation system as long as the primary goal of the evaluation system is to improve teaching. If the primary goal is to fire teachers then that system is innately corrupt.
It is unfortunate that Chetty would not give permission to quote his messages.
thank you for your blog. read it on my lunch (daily). it’s my first year teaching and i want to quit soooo bad. i wish i could write more, but i’m scared to reveal how i feel (afterall, teachers aren’t humans right?). anyways, just wanted to send you this interesting article: http://www.samefacts.com/2014/05/governance-2/the-sad-fate-of-eric-shinseki/
– public school sped teacher bk.
And yet, and yet . . . Chetty’s study confirms in a loose way an experience we or our children have all had, a pivotal teacher’s influence, one who did what great teachers do, encourage the natural interest of a student. In my case, I had that one teacher who showed a class how he wrote ten drafts before he got the final form of a poem, a real poem. I read all ten drafts pinned up on the bulletin board and said to myself: “Fruuup. I can do THAT.” And I have been ever since, that interest in writing my own poetry feeding into my degrees in English, Bachelors, Master’s, and Doctorate and making my own teaching a matter of encouraging writing wherever I found it. My son, likewise, had one teacher for one year who inspired in him an interest in computers, and allowed him to work after school in his room over the audio coupler to the Intermediate School District’s computer. He made a life in computers for himself thereafter.
Thus it seems to me, that the discussion here of the bad effects in general of the Chetty study, of its errors in dataizing people and of the misguided ways in which Obama and Duncan are trying to make good things happen in the classroom are distractions from the real point, which is: How do we get more of those seminal teachers into every school so that every child has at some point that accidental encounter with a person whose own mental activity permits the student to discover his own ability to be deeply interested in something and to learn about it.
Teachers who are not and never will be thinkers about anything of actual substance in the world SHOULD be fired, in my view. But HOW to differentiate between the truly intellectually stimulating, knowledgeable teachers and the dull time servers and do it in a fair way is the problem. VAM appears not to be working that well if superior teachers one year can show up as underperforming in a second year.
Amrein-Beardsley is in over her head trying to argue with these experts. For example, she claims that they don’t consider “how the non-random assignment of students into both schools and classrooms might have biased the effects observed.”
Actually, the first of their two papers is ENTIRELY about that very question. Amrein-Beardsley could say, at most, that they got the analysis wrong, but she just pretends that they didn’t even consider the very question that the whole paper is about.
Second, Amrein-Beardsley claims that they assume teacher quality is the same for a given teacher in every year.
But on page 2 of their paper, they say, “Our approach to estimating VA closely parallels that currently used by school districts, except in one respect. Existing value-added models typically assume that each teacher’s quality is fixed over time and thus place equal weight on test scores in all classes taught by the teacher when forecasting teacher quality. In practice, test scores from more recent classes are better predictors of current teacher quality, indicating that teacher quality fluctuates over time. We account for such “drift” in teacher quality . . . .”
“. . . trying to argue with these experts.”
Ha, ha, ha, ha, heh, heh, eh, eh, ha, ha, ha, ha, ha, je, je, je, Ha, ha, ha, ha, heh, heh, eh, eh, ha, ha, ha, ha, ha, je, je, je, Ha, ha, ha, ha, heh, heh, eh, eh, ha, ha, ha, ha, ha, je, je, je, Ha, ha, ha, ha, heh, heh, eh, eh, ha, ha, ha, ha, ha, je, je, je. . . ad infinitum.
If Chetty is an expert on education then I’m an expert in brain surgery. Hell, my vocational battery of tests when I entered college said that my interests were most like those of brain surgeons. Plus I worked in a hospital for four years. Not only that but I slept in a Holiday Inn Express last night.
“. . . test scores from more recent classes are better predictors of current teacher quality. . . ”
Well if these vaunted (at least your your eyes WT) knew anything about the COMPLETE INVALIDITY of all standardized test scores and the educational standards upon which they are based they wouldn’t be spewing this VAM bilgewater that reeks of porcine and Gallusian excrement.
Starting with crap, i.e., a standardized test score, one invariably ends up with a conclusion that is crap-VAM ratings (and SGPs). To put it in more calm terms VAM is 100% pure Grade AA Excrement.
As usual, your responses are more colorful than they are intelligent, but that probably amuses some people who don’t care about anything other than entertainment.
Let’s be clear and not take my words out of context. What you dropped off of your criticism in your first point above is key to what I I actually wrote and advanced in my critique. It’s not that “they don’t consider “how the non-random assignment of students into both schools and classrooms might have biased the effects observed.” Yes – the whole purpose of their first paper was just this, and titled accordingly. What I did write, though was that they did not “satisfactorily” consider this in their paper. Let’s not drop a key word “satisfactorily” and then be so absolutist and accusatory.
Such sloppiness will certainly tell more about your credibility in this area than the contents of your criticisms. Care is key here.
What I did mean though is that (as you may know, or not) I would not ever be able to say they got their analysis “wrong,” unless they granted me full access to their full data set, full access to all of the decision rules they made as they conducted their analyses, and full access to what they might have included in prior drafts capturing all of their technical moves…so that I could conduct a replication study (with a team of econometricians to be exact) to see in fact if what they did was “wrong.” I’m a betting woman, though, and I’d bet my good arm that I’d come up with some interesting findings, even if they were not just plain “wrong.” Most interesting here would be how levels of bias surrounding the tail ends of their distributions might demonstrate themselves to be, where the most extreme students’ test scores exist. This, my friend, is where we are in terms of the current literature on this topic.
While with this paper we can certainly file it with the other “no bias” or “very minimal bias so not to worry everyone” papers, there still exists a whole other literature Chetty et al. completely excluded from their papers that have evidence to support the contrary (e.g., bias might not be that big of a deal in the center of a normal curve but on the tail ends we see some serious problems). This is what I mean by not “satisfactorily” answering this question — as they DID NOT cite any of the other literaute in this area (except for Jesse Rothstein’s), nor did they address them, yet they claimed a final say on the matter. This is one of, if not the most hotly contested issues surrounding contemporary VAM debates. Unless they can prove the counter evidence wrong, and situating their findings in the larger literature accordingly (which is almost always common practice), my personal/professional research-based jury is still split on this matter, with the same “extreme student” issues at play. For those of you who want more information about the other pieces of literature about which I speak, please see the abovementioned blog – the literature section (vamboozled.com) with specific attention paid to the Paufler & Amrein-Beardsley, 2014 citation. While we did not “prove” anything on the counter, we definitely cited the literature more adequately and fairly, thanks in large part to our peer reviewers who were also arguably and appropriately supportive of accuracy in this area.
Lastly (as aligned to the “second” criticism above), let’s look at what I wrote, as taken directly from Chetty et al’s piece: “Assumption 1 [Stationarity]: Teacher levels of value-added as based on growth in student achievement over time follows a stationary, unchanging, constant, and consistent process. On average, “teacher quality does not vary across calendar years and [rather] depends only on the amount of time that elapses between” years.” While you (CT) might have found a place where Chetty et al added their caveaut to this assumption, this does not negate the plain and simple fact that this — Assumption 1 — is explicitly stated by them in their papers multiple times elsewhere. Further, as noted in your comment above, they say they can account for whatever they term as “drift,” regardless, and to make this assumption true even if they do not wholeheartedly believe it to be the case. Pretty fancy I say what their statistics can do on this one!
Well said, and thank you for your work!
Audrey:
1. You said, “Nor did they consider, or rather satisfactorily consider, how the non-random assignment of students into both schools and classrooms might have biased the effects observed.”
So it’s a 56-page paper in one of the most prestigious journals in the world, and all you can say is that they did not “consider, or rather satisfactorily consider,” the very topic of the paper? What you say is remarkable chutzpah no matter how it’s interpreted.
2. You do not understand what their stationarity assumption even means. What they assume here is that “mean” teacher quality is the same across years — in other words, the teacher pool as a whole is not shifting upwards or downwards across years. That may or may not be a good assumption, but it’s one that would be difficult to test in any event (how would we even know if the whole distribution of teachers shifts upwards from year to year?).
They are most certainly NOT assuming what you claim, that each teacher’s quality stays exactly the same every year. To the contrary, right up on page 2, where it’s hard to miss, they say loud and clear that the one thing that makes their VAM analysis different is that they do NOT assume each teacher’s quality is the same every year.
3. Your wish to do a replication study is commendable, although it’s not clear what would qualify you to do so, given that you don’t seem to understand what they did, other than that you’re sure it’s wrong.
Well, they acknowledge a patently obvious fact, but just how do they account for it? Why can’t you tell us?
For some reason I cannot post directly to “WT” so I’m responding here, now aware that “WT” is receiving responses to this string of posts. So as to “WT’s” most recent response to my prior post, point by point;
Point #1: Thanks for clarifying my point?!?
Point #2: Fair enough, but I will say these two conversations were inextricably tied together in the approximately 20 email exchanges Raj, Diane, and I shared, but that we promised Raj we would not share with anyone else. I’ll give you the distinction, but I won’t give you this assumption, as stated by you above, either. The assumption you paraphrase from their work follows: “the teacher pool as a whole is not shifting upwards or downwards across years.” Raj tried to make this argument, too, and also does so in similar ways in this published piece. Like Diane, I will not publish Raj’s remarks but only my response in this section of the conversation: “reading through your text surrounding Assumption 3, please note that just this section is also filled with assumptions from, like Diane said, thousands of miles up. You make assumptions in almost every line about how student/teacher sorting occurs without research based evidence in support. This, again, might be because you simply do not understanding how schools and the random sorting/assignment of students/teachers actually occurs. For example, “student sorting at an annual frequency is minimal because of the costs of changing schools?” Where is the evidence? Where and for whom is this true? Is this a blanket rule? Have you read much of the literature about school attrition and the effects of high levels of transience in highly urban schools? Or what about the assumption that “sharp changes [in the endogenous sorting of teachers] are likely driven by idiosyncratic shocks such as changes in staffing needs, maternity leaves, or the relocation of spouses?” Where is the evidence, again? What about the 2-year transition period of teachers, not to mention TFA teachers at their two-year mark? Or the third year exodus that often comes with the territory when teaching in high needs schools?” This is the kind of, apologies for this term, clumsiness that makes this study [again] and your interpretations of it still lack credibility.”
Point #3: I never said I had the full expertise to do this?!? Hence, why I carefully wrote in my prior post: “I could conduct a replication study (with a team of econometricians to be exact) to see in fact if what they did was “wrong.” Statistics are indeed different than econometrics, and never would I claim to have expertise in the latter. Again, why so sloppy?
This, again, makes me even more curious as to why the real “WT” shady isn’t standing up? Who is hiding behind (a perhaps fictitious) set of initials, or perhaps an acronym? Whoever “WT” is seems to care pretty deeply about this study, not to mention know a lot about it to cite directly from small sections of the 56 pages (something I for sure would not be able to do nor, quite frankly, would I take the time to do given I’ve already conducted two reviews of this study since 2011). Might “WT” be somebody a bit too close to this study, hence the knee-jerk, irrational reactions, that (still) lack precision and care? Everyone else (besides Harold) has fully identified themselves in this string. So what say you WT?
By the way, “WT,” you might want to stay tuned to VAMboozled.com. If you are indeed closer to this study than you are wanting to divulge, or not, you’re going to appreciate the next one. Just so you know, I have not touched it since yesterday, before this conversation started, and I will not touch before it goes live. On the flip-side, perhaps you want to stay away from VAMboozled. You might actually learn something about the realities surrounding this study…
You got me — I read the study all the way to page 2. Then I searched for a phrase that you had quoted and saw that you were misinterpreting it. Rocket science, that.
Also, it’s kind of silly to quibble with their assumption that when a particularly high value-added teacher leaves a school in, say, 2009, it is not likely to be the case that all of the good students in that school leave the school along with her. Thus, it is safe to assume that if the teachers in the same grade see a drop in total value-added the next year, it’s probably because the great teacher left rather than because kids who were better in unobserved ways suddenly and coincidentally left at the same time.
If you really think this assumption is wrong, are you saying that: 1) parents know exactly which teacher in a school is the highest value-added (and let’s remember, a high value-added teacher might not be the most ‘popular’ for any number of reasons); and 2) as soon as a high value-added teacher announces in May that she’s leaving, parents of kids who are likely to score well (above and beyond controlling for their current test scores) immediately respond by moving to a different school zone before the next school year? Really?
I mean, every study makes assumptions and someone who is a diehard ideological cynic can always quibble, but it’s hard to come up with a more reasonable assumption than this one.
Thank you. Unfortunately, I am not waiting for opt out to kick in nation-wide. I will not allow my grandchildren to attend any public school that spends months either testing or preparing students to test. Their time and future are more important than using them as Guinee Pigs so someone can find an easier way to get rid of unwanted teachers. As a Public school teacher for 43 years, I am well aware of the many reasons Administrations want to get rid of teachers and they are seldom removed because they are ineffective at explaining their curriculum to young adults. Most of the time it is because they have set the bar Too High and some student and/ or parent wants to get rid of them because the student had to work harder to get a good grade! This is not a good reason to remove a teacher, but it happens all the time. I am on the State Commission to hear these dismissal cases and this is many times the case. We just had one of the most successful AP Calculus teachers in the country “retire” early because she has been harassed for several years by a weak Administration due to a few parent complaints. She has had 100% pass rates for several years where she insists ALL of her students take the test even if she has to find outside funding to pay for their tests and most of her students are from lower income families and most are Hispanic. She usually has between 55 and 85 AP Calculus AB and BC students each year take the tests. She has also been one of the national graders hired to grade the AP tests for several years. Money and politics have corrupted the public educational system and I cannot see how this will change anytime soon.
In the second paper, w19424.pdf, the key figure for me is figure 2 where they plot earnings at age 28 by normalized VAM.
It’s quite clear from the plot that the data is non-linear – the biggest earnings are for those kids with a teacher with 0 – 0.5 normalized VAM score and it’s less on either side. However, the authors fit a straight line. It’s very poor technical skill to fit a straight line to non-linear data.
There are two reasons why the slope looks so good:
1) there is one very influential point midway between -0.5 and -1 VAM and this point I would guess causes a lot of the “drama” in the slope – if you left it out the slope would be pretty flat i.e. VAM score doesn’t say much about earnings and
2) they use summary data which means a lot of the variation is lost – if it was plotted as students earnings by student’s teacher’s VAM score than the slope would look really, really flat given the scale of the variation.
The coefficient is $350 dollars which means that the line predicts that kids with a teacher at VAM score of -1.5 earn $1050 less per annum then the kids with a teacher at VAM score at 1.5. Hardly life shattering.
“Those who care deeply about humanistic education, about the life of the
mind, about deep learning, find your work–no matter how technically
perfect–utterly appalling. It drains education of joy and discovery and
makes everyone a slave to Pearson.” …and bureaucracies, and standardization, and stanines, and charts and graphs, and monetization and commodification, and political idiots who haven’t the vaguest notion of the art of education. The appalling parade those dedicated to counting the number of angels dancing on the head of the pin goes on….
yes yes yes yes yes
The final comment of the post, “…economists… looking at ants, seen from a far distance…” is particularly valid for some of the work on education and pensions, released by the NBER, in my view.
It is of concern that the Bureau does not identify the source of its funding, at its website nor, when requested by the public. Sourcewatch cites $10,000,000 in funding, in a 15 year period, from just four foundations, including Schaife, John Olin, and Lynn and Harry Bradley. Research organizations have a responsibility to be transparent.
The Bureau’s objectivity would not be in question if :
The NBER identified its funders.
Research about the adverse effect of income inequality, gained visibility
from a paper, released by NBER. (University and think tank economists
failed to be first to produce the ground-breaking and essential Piketty paper.
The omission calls into question, their credentials for education prescriptions.)
NBER researchers published the results of a study on the adverse
effects of unrelenting criticism and resource starvation of public services.
NBER research focused on the productivity failures of the financial sector,
a fast-growing GDP segment, with grave negative consequences for the
U.S. economy.
Linda,
I think you have the wrong impression about the NBER and how academic economics works in general. If you looked at Thomas Piketty in particular, you would find that he got his Ph.D. at the London School of Economics and his first job was at MIT here in the US. He has published his work extensively in top journals, including AER where the Chetty articles are forthcoming. Here is a link to his work published in English: http://piketty.pse.ens.fr/en/publications#AR1PUBENG
And here is a link to an index of recent NBER working papers. The list is very extensive and reflects the full range of topics that economists generally think about, including the impacts of income distribution on society and societies impact on income distribution.
The link is here: http://www.nber.org/new_archive/
Quoting the blog post, “…fifth grader…higher test scores…didn’t get pregnant…”
Odd, contrived, causal permutations are not as rare in NBER research as they should be.
Here is the collection of all thoughtful and educational sayings from reliable and trusting sources:
Albert Einstein:
“I fear the day when the technology overlaps with our humanity. The world will only have a generation of idiots.”
Always remember 4 guiding principles from Buddha:
DO NOT quickly believe in the saying from:
1. People with authority, scientific knowledge, and wealth (like Mr. Duncan, Prof Chetty and Pearson)
2. People with old age, claimed to be a Wiseman
3. Any written old testaments
4. Any mystery, unfounded truth, and lack of proof of science
From His Holiness Dalai Lama:
Health
1. Drink plenty of water.
2. Eat more foods that grow on trees and plants and eat less food that is manufactured in factories
3. Live with the 3 E’s — Energy, Enthusiasm and Empathy.
Personality
1. Don’t compare your life to others’. You have no idea what their journey is all about.
2. Don’t overdo. Keep your limits
3. No one is in charge of your happiness except you
4. Realize that life is a school and you are here to learn. – Problems are simply part of the curriculum that appear and fade away like algebra class but the lessons you learn will last a lifetime
5. Smile and laugh more
Society
Each day give something good to others
Spend time with people over the age of 70 & under the age of 6
Life
Get rid of anything that isn’t useful, beautiful or joyful.
And this is what I have learned from reliable sources:
Lesson 1: from Dr. Eric S. Rabkin of University of Michigan who taught “Humanities and Social Sciences”, I have learned that mirror does not lie because the image which reflects from mirror remains as it is if a mirror’s position is in the vertical line of the up or down. However, mirror’s image will be reversed if a mirror’s position is in the horizontal line of the right or left. Likewise, life would be in order and harmony if the authority has kindness, patience and compassion, and if people know their boundary of honesty, integrity, and respect for the wisdom. Otherwise, on the right, there are bunches of manipulative greed, and on the left, there are bundles of fearful and uneducated law abiding people, together society never live in harmony.
Lesson 2: from Dr. Gautam Kaul of University of Michigan who taught “Introduction to Finance”, I have learned his golden advices that people’s life becomes suddenly uncomplicated and simple whenever they choose the right path; and that people only learn well whenever they are happy.
Lesson 3: from Dr. Walter Sinnott-Amstrong of Duke University, and Dr. Ram Neta of University of North Carolina, Chapel Hill, who teach “Think Again: How to Reason and Argue”, I have a chance to refresh the logic tables that were long forgotten in my youth’s memory. I am exciting to apply this concept of true-false through all tables of AND, OR, “IF and THEN”, “IF and ONLY IF” regarding the dualistic/yin-yang world, body-mind-spirit concept. I am so delighted to find that:
BODY will be either health (T), or sickness (F) – I call this a proposition p
MIND will be either omniscience (T), or faith (F) – I call this a proposition q
SPIRIT will be either kindness (T), or cruelty (F) – I call this a proposition r
After I try all combinations among p,q,r through all logic tables, the only ABSOLUTE TRUE will remain reliable when p, q, and r are all true. In conclusion, the authorities or leaders need to possess a healthy body, an omniscient mind, and an empathy spirit in order to lead their nation to be prosperous, compassionate, and peaceful.
Either one part of these three dimensions has a default, we have chaos because situation will be unclear, or is called a code of silence. Today, the world market is the true reflection of a mirror’s image in the horizontal line, and the true result of all default propositions [earth is depleted, leaders are ignorant, and people are full of frutration/stresses]
What are hope and dream that could remain in the hand of Higher Education’s Leaders who have conscience and courage to reinforce and to guide younger generation to reach out the better future? Back2basic.