« Courtesy, Curriculum, and a New Topic | Main | IQ Testing in Historical Perspective »

Testing and Unmistakable Links to Social Status


Dear Diane,

We’ve both now had a chance to read Charles Murray’s recent essay. I’m of two minds: to ignore it (bury it) or to take it on. You’ve settled it. But I’m not sure I’ll be happy about it a week from now. First of all, it means I’m suggesting people read his essay! Ye Gods!

He’s been a controversial figure for years in our circles—both of ours. He’s had no compunction about taking IQ as “the measure of the man” (close to the title of a wonderful attack on IQ by Stephen Jay Gould.) His scholarly book, "The Bell Curve," written with Richard J. Herrnstein was much admired and detested a few decades ago. His thesis—very roughly speaking—was that we were perpetuating a myth when we contended that all people were capable of high levels of thought, and/or that there weren’t serious differences—on average—between different groups, sorts, races, classes of people. And finally, that social policy—especially education policy—that ignored this was just asking for trouble. The gap was a gap of fact, not the fault of schooling or nurturing (although aided and abetted by both perhaps), but innate and unchangeable.

From the start of NCLB I had the fearful feeling that nothing was more likely to play into his hands than the relentless assault on the test-score gap, the data about which he had been promoting for years.

The racism of Murray’s argument (or put another more neutral way—the claim that race and intelligence were connected) has a lot of “common” sense behind it. If IQ tests measured intelligence, it was well-nigh irrefutable. The claim, after all, was common sense even before tests were invented with regard to women vs. men’s intelligence, the rich vs. the poor, and whites vs. blacks, Indians, etc. (I’ve just finished reading "Oliver Twist," and as always am amused at the fact that Oliver learns to speaks the King’s English although born in the poorhouse in which no one else speaks it. Yet it never seemed absurd to those in the story, or his readers, or me when I first read it.)

Modern psychometrics is a century-plus old, based on the development of an instrument intended not only to sort individuals but verify such long-standing assumptions. All modern tests are updated variants on the originals. And they all demonstrate, over and over again, the unmistakable social status of the society itself. Scores go up for every dollar more the family earns, for example. It places an enormous burden of self-doubt on those declared unfit by IQ. Many—such as Jews—overcame the description by the clever dodge of getting rich, and then becoming smart, too—speaking of Oliver Twist. For some white immigrant groups, it took longer.

It works to the advantage of first-borns as a whole. Women have finally managed to beat it. Asians have. So what’s wrong with African-Americans? And Native Americans? The sting, the injury is deep and enraging. And the implicit accusation, which Murray makes explicit, is not easily overcome. And year after year of headlines proclaiming these “self-evident” truths does much to vindicate Murray’s argument—or so he claims. (For many African-Americans, as you note Diane, it is equally self-evident proof of racist schools, racist teachers, curriculum, personal or systemic, or both. I buy their argument, at least in part.)

I argue, in a set of chapters in "In Schools We Trust," that most, if not all, can be explained in quite natural and equally commonsensical ways. But it requires starting off by believing, as I try to demonstrate, that tests are not what they seem. They are, I claim, a reflection of the “common sense” of a particular stratum of society at a particular time and place and of their particular form of thinking about the world. I try to show it by specific examples of specific questions and the ways test-takers respond to them. I note secondly that the very nature of the psychometric tool itself requires this to be the case. And where it doesn’t work that some fiddling sometimes makes it work, e.g. the unused items in the SAT tool as noted by Jay Rosner. (And the ways test-makers fiddled with gender differences.) I argue that “thinking like” the dominant high-IQer helps. There are ways to make such thinking easier or harder and society and schooling can contribute to doing so. It helps if you imagine you could be and would like to be them. (I note that first-born, even within the dominant white middle class, are more likely to think “like/”identify with grown-ups, for example.) Claude Steele’s “stereotype threat” theory accounts for some of it also—as a good test-taker must rely on a certain brash self-confidence that status-anxiety erodes. Related to Steele’s theory—that you not only have to “care” about doing well, but you have to have good reason to believe that your efforts will work to help rather than hurt you. And, finally, if test-prepping ever eliminated these gaps, psychometrics demands that we reinvent them since differentiation is the name of the game.

Of course, being rich and being white goes along with having both a bigger and a “richer” vocabulary (one more likely to show up on tests). Yes, of course, being rich and being white goes along with having the experiences that will help you recognize right answers from wrong ones on tests better than kids who are poor and black. And on and on. Would a test that produced results that went against this “common sense” even get off the drawing board—one in which lower-class test-takers scored higher than those from high-status families? Would even a single test question survive under such scrutiny?

It even works, as Rosner notes, on apparently neutral math test questions! One of his examples from the SAT pool is below.

If they are a reflection more of social status than intelligence or schooling, then maybe the solutions are different. Richard Rothstein has argued that we could do more to raise test scores by providing everyone with good medical care than good test-prepping.

The stakes are too big to keep avoiding this argument. The decision by the DOE in NYC to start the testing of children at age 5, the use of low test scores as a rationale for depriving kids of playtime and as a reason for increased focus on test skills and less on higher-level subject matter are powerful reasons to explore the meaning of such test differentials (aka gaps) themselves. Shall we, Diane?



If the square root of 2x is an integer, which of the following must be an integer?
a. square root of x
b. x
c. 4x
d. x squared
e. 2 times (x squared)

If the area of a square is 4 times (x squared), what is the length of a side?

a. x
b. 2x
c. 4x
d. x squared
e. 2 times (x squared)

Blacks outscored whites on the first, and whites outscored blacks on the second.


And they all demonstrate, over and over again, the unmistakable social status of the society itself. Scores go up for every dollar more the family earns, for example.

You are confusing correlation with causation. Academic achievement is correlated with socio-economic status; however the casual link between the two has not yet been established. In any event, the correlation is somewhat low (about 0.4 standard deviations).

There are three possible causal explanations for the observed correlation:

1. poverty causes low achievement (your and Rothstein's theory)

2. low achievement causes poverty

3. some third factor causes both poverty and low achievement (Murray's theory: third factor = IQ)

There are numerous counterexamples that cast much doubt on your and Rothstein's theory od causality.

* Harvard is not full of lottery winners.

* Many poor immigrants come to America and succeed academically. In fact, you can predict with a fair bit of accuracy the probability of an immigrant succeeding academically by the averge IQ for that immigrant's group in their home country.

*some children of the rich don't do well academically. Again, this correlates well with how that student's parents got rich in the first place (e.g., attainment of professional degree vs. sports/entertainment star).

* We have tried providing comprehensive medical/health services to the poor and no academic gains were forthcoming. See Headstart. See Project Follow Through in which the thousands of students in the study were given comprehensive medical, dental, health, and nutritional services as part of the experiment; the effect of these services was null.

None of this should be taken to suggest that Murray is right in his views on education.


This issue of the math problems is interesting and merits analysis.

I took a look at the Kidder and Rosner article that brings up the two math problems you mention ("How the SAT Creates Built-in Headwinds: An Educational and Legal Analysis of Educational Impact" (43 Santa Clara L. Rev. 131). I found this (note that the first math problem you gave is mentioned second here and vice versa):

"...one of the questions [the question about the side of a square--DS] is from a scored SAT and was answered correctly by 11% more Whites than African Americans. The other item [about the square root and the integer] was on the experimental SAT Math section in 1998, but was not included in a scored section of the SAT. This experimental item was answered correctly by 7% more African Americans than Whites. Is it easy to distinguish the item with a disparate impact of 11% favoring Whites from the item with an impact favoring African Americans by 7%?"

Now, it could be quite misleading to compare the disparities of the two questions. I suspect that more blacks and whites got the first (your second) question right than the other. It is straightforward and can be solved without any special thinking, so long as one knows how to calculate the area of a square. The other one requires consideration of the one instance where x is not an integer (i.e., x=1/2)--or else take the smallest possible integer (1) as the starting point, take its square root, and divide it in half. Whoever assumes x must be an integer will get the question wrong.

If it is true, as I suspect, that substantially more people got the "side of the square" question right, then the percentages may mislead. Say, for instance, 1110 whites and 1000 blacks got that question right. That's a disparity in numbers of 110. Say that, out of these, 535 blacks and 500 whites got the square root question correct--that's a disparity in numbers of 35--much closer to equal. Perhaps I am wrong, and similar numbers got both questions right. Still, the article should provide such information.

I'm not saying this to dismiss what Kidder and Rosner found. It's just hard to draw conclusions from their data when it is not presented in complete form. We would need to know how many people in each group chose the correct answer for each question, and how many chose other answers.

Then, once one had an accurate picture of the disparities, one could proceed to analyze them.

Arg, I meant "...or else take the smallest possible integer (1) as the starting point, take its square [not square root!], and divide it in half."

A bit off topic, but I greatly enjoyed today's EdWeek article on the student-teacher relationship in preschool.

"Pupil-Teacher Relationship Crucial in Preschool Learning, Study Says"


"The quality of the relationship between preschool teachers and their pupils might be more important to children’s learning than such factors as class size and teacher credentials, a new study suggests."

If the international studies are to be believed, this applies to older students as well.

Erin Johnson


Excellent points and analysis. Please contribute more.

Talk for a minute if you would about the obvious success of Asian (even Non-English speaking when they arrive) students versus other groups. Why are they so successful while other groups seem addicted to academic anemia from one generation to the next? Beyond that, the sixty-four thousand dollar question today, how could we interrupt the latter paradigm, and remedy the achievement gap?

I, too, am interested in hearing more from KDeRosa.

A few more thoughts about those SAT math problems. Far, far too much information is missing. For example:

1. Were these problems ever on the same test? It seems not. The square root one was an experimental question included on an 1998 test but not scored.

2. Assuming they were on different tests, what was the total number of test takers for each test, and the total number in each subgroup?

3. Of these, how many (and what percentage) answered each of the questions correctly? How many selected other options?

4. What is the margin of error for stats on a single test question? (Consider that a person has a 25% chance of guessing the answer correctly, since there are four options.)

5. Are these results reproducible? Do similar test problems bear similar results?

6. And so on. In other words, the more I think about it, the less convinced I am by Kidder and Rosner on this matter. The example of the two problems brings up interesting questions, but demonstrates very little, from what I can see so far.

Northeast Asian students perform about how you'd expect from their group average IQ. They perform about 3/4 of a standard deviation better in math than white students. They perform about as well as white students on verbal tasks (Second generation and higher Asian-American students are not overrepresented in fields requring verbal skills, like lawyering.)

So if you're thinking that work-ethic is the prime causal factor in Asian over-performance, you'd need an explanation as to why the over-perormance is only in math areas and not in verbal areas.

High-IQ students perform well for the simple reason that it is easier to teach them because they tolerate ambiguous instruction (i.e., the norm in education) better and they require less practice to retain what they've learned. These two factors contribute to initial academic success which breeds the motivation needed to engage in the many hours of effortful study needed to learn academic content.

This would also explain why progressive education techniques have been such a flop with the huddled masses. Progressive education is all about discovery, and inquiry, and learning in "information rich" environments which place high demands on the cognitive ability and background knowledge (also a function of cognitive ability) of students. Progressive education inadvertently rigs the system so that only the smart and (to a lesser extent) high-SES students succeed.

Deborah, Thanks for injecting some harsh reality, in the form SAT items (test questions), into your discussion. Although I’d generally prefer to engage on the larger issues, I’ll respond briefly to some specifics on the SAT for anyone interested.

Knowing how and why different groups of test-takers chose different incorrect SAT answers is often raised by academics as a matter of concern. I wish I could be more interested in that, but I’m primarily focused upon the game-determining issue of how and why 99% of pretested SAT items chosen to appear on scored sections of the test favor whites over blacks and Latinos, and, on SAT math, 99% of items favor males over females. I haven’t yet published on the latter issue. Another way to state this: if blacks, Latinos, or, on math, women, do better on a pretested item than whites (or, on math, males), the item has virtually no chance of appearing on a future scored section of the SAT.

Diana Senechal wonders whether my use of correct answering percentages may mislead on the two math items Deborah cited, and desires to know the numbers of students involved. She makes a hypothetical reference to “1110 whites and 1000 blacks” answering an item correctly. That would never happen. The white/black ratio of SAT test-takers is roughly 7:1 these days, which is why I use percentages, and not numbers, to make meaningful comparisons.

Also, Ms. Senechal wonders, “Are these results reproducible? Do similar test problems bear similar results?” When I used to teach SAT prep to kids, they’d ask, “What if the test changes and what you’re teaching us doesn’t work?” All these questions somehow miss the single most salient characteristic of the SAT – it doesn’t change. Sure, ETS will add a new question type every now and then, but the hallmark of the SAT is its consistency and comparability test-to-test. So I know for certain, having absolutely no data, that 99% of the math items on the scored section of the recent May 5th SAT favored males over females, but I wish it weren’t so.

Mr. Rosner,

I am a little startled. If my hypothetical situation (1110 whites and 1000 blacks) would never happen, then what do you mean when you say that a certain question "was answered correctly by 11% more Whites than African Americans"? Do you mean, as I inferred, that the actual number of whites who answered the question correctly was 110% that of blacks? Or do you mean, as I now take it, that the percentage of whites answering it correctly exceeded the percentage of blacks by 11?

These are quite different scenarios. If it does not trouble you to engage in specifics once more, please clarify this point for me and for others who might be interested.


Diana Senechal

My apologies for this and other errors! I meant 111 percent, not 110. Correction here:

Do you mean, as I inferred, that the actual number of whites who answered the question correctly was 111% that of blacks? Or do you mean, as I now take it, that the percentage of whites answering it correctly exceeded the percentage of blacks by 11?

Ms. Senechal, Despite my decades of experience, I'm regularly startled by the SAT. For example, I'm always amazed at the tepid response I receive whenever I describe and quantify the skewing of SAT math against women.

I appreciate your interest in all this. In response to your question: the 11% in the quote represents, as you state at the end, a subtraction of the black answering percentage from the white answering percentage. In the law review article you cite, we define this subtraction of percentages as "impact" or "disparate impact" in an effort to clarify. And, for those who want more, there are two other pairs of SAT items analyzed in the article.

Thank you very much for the clarification. Now, it seems all the more important how many students answered the square root question correctly vs. the other question. Here are a few definitions and scenarios. For the sake of simplicity, I am rounding off percentages to the nearest whole number.


Question 1, for our purposes, is the one with the side of the square, the one where the whites had the 11% impact advantage.

Question 2 is the square root/integer one, where black students had the 7% impact advantage.


Scenario A. Say there are 700 white students and 100 black students taking the test. (Of course there would be far more, but let's use this for starters.) Let's pretend, for now, that both questions are on the same test. Say 600 white students (86%) and 75 black students (75%) answered Question 1 correctly.

Now, let's say Question 2 was harder for all. Let's say 500 white students (71%) answered it correctly. So then 78 black students (78%) would have answered it correctly. One can see that the black students fared slightly better on this question than on the previous one.

Scenario B. Let's keep the figures the same for Question 1. Now suppose only 400 white students answered Question 2 correctly. That would be 57%. So then 64 black students (64% of 100) would have answered that question correctly.

In this scenario, the percentage of black students answering Question 2 correctly exceeds that of white students by 7%. Yet 11 fewer black students (15% fewer) answered this question correctly than answered the first question correctly.

In such a scenario, it seems trickier to argue that blacks scored better than whites on the second question, or that they had more of an advantage on the second question than on the first. Perhaps I am ignorant of the statistics used in test analysis. However, it seems we really need to know the total percentage of test-takers (and test-takers within each subgroup) answering each question correctly, before we can draw any conclusion about bias.

If my reasoning is flawed, I am open to correction. I don't want to draw undue attention to this detail--but I actually find it quite interesting and relevant to the larger questions.

Let me try a response, and then let's take this offline unless someone else posts here indicating interest - we're probably driving folks away from the main discussion.

Your Scenario B is closer to reality, since the second question is far more difficult than the first. Because the second question was a discarded question, ETS will likely not release the total correct answering percentage (I encourage you to try asking them). I would estimate that maybe 22% of black students and 15% of white students answered that question correctly. Your estimate of 86% of white students answering the first question correctly seems a little high to me. It may be 70%, but you're close enough for this discussion that I needn't take the time to dig out the actual number. Btw, the range of total correct answering percentages on the SAT is roughly from 92% to 10% (easier questions to harder). Each SAT question, no matter its difficulty, is "worth" the same to a total SAT score.

You posit, "It seems trickier to argue that blacks scored better than whites on the second question ..." If by "scoring better" we mean "answering correctly at a higher percentage," which is a reasonable definition for our purposes, then it's not a tricky argument - it's a tautology. It just happens that I've presented on the second question to a lot of minority audiences, and, while I haven’t done a scientific survey, I remember that every black person (many hundreds) indicated a desire to have the second question on the SAT.

I'll use male and female on math to introduce a slightly different perspective. More females than males take the SAT, but males have a higher correct answering percentage than females on 99% of the math questions, so females can take the place of blacks in your example.

Substituting females for blacks, you imply that females should take solace in the fact that on easier math questions, females answer correctly at a higher percentage than they do on harder questions, and what males are doing on each question is of secondary importance. And, if (hypothetically) there were a difficult math question on which females did better than males, it wouldn't be that important to include it on the test (not to worry, ETS doesn't). But maybe I'm not understanding what you're implying ...

Thank you for your responsiveness. I am very sorry if my questions have kept anyone from the main discussion.

I wasn't implying anything about taking solace. I was only implying that once data is invoked, the data should be made available. The specifics influence our understanding of the bigger picture.

If only 22% of blacks and 15% of whites (a ratio of 22:105, assuming a 1:7 ratio of blacks to whites) answered the second question correctly, then we need better math curriculum for all. The question is harder than the first, but not especially hard, not even for a female like me. I suspect, moreover, that the margin of error is greater at the lower end of the curve than at the higher.

So, back to the larger issues, or at least one of them. Which should come first: making the tests equitable, or improving math curriculum and instruction? I argue that the latter should take priority, for seemingly equitable tests at this stage would only mask instructional inequities.


"The quality of the relationship between preschool teachers and their pupils might be more important to children’s learning than such factors as class size and teacher credentials, a new study suggests."

I've believed this to be true all along. Not only class size and credentials but subject knowledge can also be, relatively speaking, a misleading predictor of student success. A teacher can know their subject well or better than the average bear but if they don't know how to relate to their students they can significantly diminish their potential as an educator.

This is not something that can be taught at an ed school or teachers' college either. Teachers either have these interpersonal skills or must develop them over time with their kids to maximize their potential in the classroom.


The way that our schools are now set up now with the teacher responsible for "grading" the students undermines that critical student-teacher relationship. And the "grading" is not even for the benefit of the student but for everyone else outside the classroom.

No matter how much effort that a teacher puts into developing great relationships, students will always feel graded. How does that feeling of being "graded" support a quality student-teacher relationship?

Erin Johnson

Two comments:

1) Problem #1 was easier for me to solve by plugging in numbers (1/2, 2, 4-1/2) than by algebra, while problem #2 is easily solved conceptually or by algebra as well as by plugging in #s. Not sure what that means for potential item bias...

2) Ken DeRosa's comments: "Harvard is not full of lottery winners" as well as the old saw about immigrants. Not sure how relevant the bit about Harvard is, since 13-year-olds are forbidden from playing: Harvard's also generally not full of 50- and 60-year-olds. Apart from the empirical question (which is an interesting one about longitudinal outcomes of random windfalls, not just lotteries--and watch for the self-selection effects!), Harvard is one of the places that least has to rely on standardized tests as a gateway (except as status mark[et]ing). And as many sociologists of education have noted, the relationship between immigration and education is very complicated, often depending on generational effects, and that there is more variation than one normally hears in model-minority discourses.

I've enjoyed the detailed discussion, but before we move on, shouldn't we return to Deb's big idea? Growing up a first born, White male in the middle of Pax Americanna, and being continually moved from the second lowest to the second highest elementary classes and back, I've never doubted the power of ranking. Then I discovered Jim Crow ...

Deb is addressing one of the most profound issues in American democracy; perhaps its our "original sin." Perhaps I'm willing to make more compromises than she, but whenever I do I hope I get a sick feeling in my stomach. We need precision in our discussions, but lets not allow details to overwhelm profund truthes.


Clearly our exceptionally large achievement gap is a national disgrace.

But we are not the only country to experience racism and its correspondingly negative effects on society. If you look at say the suburbs of Amsterdam, you would see substanial, overt racism towards their immigrant populations. The immigrant populations are not a small part of Dutch society as Rotterdam is poised to become a majority "minority" city.

And yet, the acheivement gap is substantially lower in the Netherlands because their school system is set up in a way that promotes quality education. The top performing school systems, such as the Dutch system, have a much smaller acheivement learning gap because their schools are set up to monitor student learning and not "seat time."

While gross injustices are difficult to eradicate, a change in our school system to focus on student learning would go a long way towards closing that acheivement gap.

Should we not focus on what we can do to close that acheivement gap instead of just bemoaning its existence?

Erin Johnson


Our exceptionally large achievement gap is a national disgrace. Worse, we face a dilemma. For our poorest kids, the challenge of meeting 21st Century challenges can not be met through incremental improvements. But our systems have not been able to “chew gum and walk at the same time,” much less produce incremental improvements. So, we’ve been tempted to “swing for the fences” during each time at bat, “ and gamble on reductionistic approaches. I don’t have any “easy answers” except to give up on “easy answers.”

I am reminded of Jesse Jackson’s dictum that you “back out of a blind alley the way you drove in.” For our poorest kids, too few adults are raising too many of our kids. So we back out by creating as many family-like relationships as possible.

You argue, “While gross injustices are difficult to eradicate, a change in our school system to focus on student learning would go a long way towards closing that acheivement gap.
1. Should we not focus on what we can do to close that acheivement gap instead of just bemoaning its existence?”

We may or may not agree. If we are talking about instructional techniques, then I don’t believe we are going a LONG WAY to close the gap. Our goal - which would be modest enough but still a huge challenge - would be to create a learning culture, a set of humane and yet challenging educational relationships. ONE WAY of building those relationships is focusing on instruction. But education, fundamentally, is a “people business.”

I suspect that there are curriculum-driven approaches that are so effective, like Core Knowledge, or relationship-driven approaches such as Oklahoma’s “Great Expectations,” that they will provide a strong counter-argument to my position. They make so much sense, given the way humans interact with each other, that those instruction-driven approaches will be more effective than my relatively pessimistic analysis would predict.

All I am saying is that it is whole human beings that learn. Let’s not get bogged down in details.


Your description of a learning culture is quite apt.

Given how dominating our school structure is on "how we do school" even those very wonderful curriculum driven approaches (Core Knowledge) and relationship-drive approaches (Great Expectations) have had very little impact on our schools as a whole because what is limiting school improvement is the school structure itself.

That is "the way we do school" and "the way we test" etc..., makes those quality relationships almost impossible.

But from the international studies, it is clear that a school structure that focuses on (and only on) student learning can greatly improve learning.

Is is easy to change our school structure to enable quality relationships between teacher and student? Not really.

But changing the structure of schools to focus on student learning is an absolutely necessary step if our children are to experience the quality learning environment that we would all wish for them.

Erin Johnson


I wholeheartedly agree that education is fundamentally a "people business," in so far as I believe that people derive meaning in their lives from relationships.

A problem with schooling in its present form is that it doesn't allow for strong relationships to form between adults and children. Why? 1) There are too many students and too few hours in the day for a teacher to spend quality, personalized time with each student. 2) The teacher is always "telling" the students what to learn, instead of asking that student what he/she wants to learn.

There is a far simpler remedy to the achievement gap than "raising standards" and "improving teaching quality."

Poor children need to interact with adult role models out in the real world.

Middle class and upper class children learn much about the world just by watching their parents interact with others, listening in on dinner parties, watching their parents read the newspaper, ect... These lessons constitute a person's "soft education" and I believe this form of learning is exponentially more powerful and influential than the traditional learning that takes place in schools.

Every child (but especially those from poorer backgrounds) needs access to adult community mentors. I love the Big Picture schools because they connect authentic learning and internships with adult mentorships.

I believe that programs like Big Brother / Big Sister are the key to turning around the achuievement gap and generational poverty in America.

It's not enough for a teacher to tell a poor kid he/she can be a doctor. This is too abstract and the teacher isn't viewed as an authentic source. Instead, the kid needs to KNOW doctors (preferably a few), needs to intern in hospitals, needs to oobserve working professionals in action, needs to hear from medical practitioners about the type of education and work ethic necessary to succeed in the field, and most of all the kid just needs someone to talk to and hang out with.

Until we start offering young people this type of "soft education," I believe the achievement gap will never truly close.

Case in point, the KIPP schools are great at closing the academic achievement gap for lower income students, but the middle class kid with a solid B average is going to continue to have the advantage if he's comfortable interacting with adults in a variety of social settings - cultivated through a lifetime of experiences watching successful adult role models interact.

One more point, I think this type of intensive real-world learning and adult mentoring needs to take place earlier rather than later - certainly by the time the child reaches the ninth grade, and preferably even sooner.


Your assessment of why schools are doing such a poor job is quite correct, "in its present form is that it doesn't allow for strong relationships to form between adults and children."

But this is not because "..there are too many students and too few hours in the day for a teacher to spend quality, personalized time with each student. 2) The teacher is always "telling" the students what to learn, instead of asking that student what he/she wants to learn."

The relationships between teachers and students (in our very ineffective school system) are highly strained because of the insistence of our school system in demanding that teachers grade/evaluate their students.

It is that antagonistic role that hinders every teacher (no matter how caring or nice) in establishing a trusting, quality relationship with their students. So no matter how much a teacher may want to give to their students, students do not trust that their teachers have their best interest in mind.

If teachers truly did have their students' best interest in mind, would they not only give them grades that would serve them well in the external world (that would be A's)?

While "real-world" learning certainly has it's appeal in some quarters, it is the quality, trusting relationship between teacher and student that allows quality learning to blossom.

Erin Johnson

I agree and disagree with almost everyone here. (I pretty much wholeheartedly disagree with Mr. Rosner, from what I can tell, but that's a different matter. I still appreciate his willingness to engage in the questions to some degree.)

John Thompson--yes, it is whole human beings that learn, and teaching is a human activity in the highest sense of the word. But there's another side of it too. Through various angles (such as individual subjects) we gain insights into the whole. We need the angles--what William Torrey Harris called the "windows of the soul."

Matt, you make an excellent point about the importance of adult role models and mentors. Such "soft" education, though, should supplement, not replace, "hard" education.

Erin, I agree with you that we should focus on student learning (as well as subject matter). I am still skeptical about your ideas regarding evaluation. I do not think evaluation has to be harmful, or that it necessarily undermines trust between a teacher and a student. It can be refreshing for a student to know a teacher's honest assessment and opinion. We all want to know: "Is this good?" And if a teacher does not abuse this power of evaluation, it can be enormously helpful.

My poetry professor in college--a wise and brilliant man--refused to tell us whether we had the makings of poets. He said it was impossible to tell at that stage. Most of the time he focused on what we could learn from great poets, and what we were doing well and not so well. Yet with all his constructive advice, he did not entirely eschew evaluation. He would tell us, if asked, whether he thought a given poem was good or not. Those assessments helped immensely in their own way.

Sherman Dorn, I'm glad someone else took interest in those math problems! I solved problem 1 by asking, "Is x necessarily an integer?" When I arrived at the exception x=1/2, I found only one of the options would be correct for that value. So I suppose it was a combination of reasoning and plugging in numbers.

I, too, have trouble seeing where bias might come into play with this question nd still have lots of trouble believing there's any demonstrable bias when such a low percentage of test-takers (roughly 22% of blacks and 15% of whites) got the question right. Through sheer guessing, one has a 20% chance of selecting the right answer. At any rate, there's no excuse for such a low percentage! What are the schools teaching?

The devil, they say, is in the details. So is the divine. Mediocrity is in the muddle.

"Scores go up for every dollar more the family earns, for example. "

Only after sorting for race. As has been established in many studies, poor white kids do better than wealthy African Americans.

One thing that no one has mentioned in the comparison of those SAT questions is the explanation that explains it best--that low ability students are more likely to guess. The properties question is much more difficult. Students who randomly guess would have a 20% chance of getting it right. A good student who tries to figure the problem out and fails will probably fall for one of the "trap" answers, and be *more* likely to get the problem wrong than would occur with a random guess.

That's a far more likely reason that the "black success" questions are eliminated, and it is the College Board's reason for doing so--which is why not mentioning it is absurd. It's not some plot to reduce their scores. It's that the questions they do better on are too difficult. If random guessing gives better results than working the problem, it's a test question that has to go. I've read the research and I know the argument against the guessing tactic. I don't agree with it, but that's a different fight. It's simply irresponsible not to mention the testmaker's reasoning, and I'm surprised Deborah didn't mention it.

Finally, I have spent a lot of time working with low income minority students in test prep (you can see two years results at thosewhocan.us), and I can tell you a few things:

1) Low ability students of any race are not "doing better" on hard problems because they approach problems differently. They are guessing.

2) All low ability students regardless of race, with few exceptions (say 10%) underperform compared to abilities, at least on ACT and SAT tests. In most cases, their lower performance is due to timing requirements and unfamiliarity with the basic nature of standardized tests.

3) Good test preparation that focuses on content instruction will dramatically improve low income minority test scores. It does not involve "teaching them how the test questions work" or making students feel scared that "the test will change". I know Jay Rosner worked for Princeton Review, but I didn't think they were that irresponsible.


There is a huge difference between teachers giving feedback to students on how to improve their work and evaluating the student for the external world.

And as for the quality of feedback that teachers give, should teachers tell their students that they like/dislike their material or give them suggestions on how to improve?

Certainly, for those students who already know how to do something, a positive "This is a wonderful essay/poem" can be very nice and ego boosting.

But what about those students who have not learned yet? A teacher telling them, "This is not good" does not make them feel empowered and most importantly does not give them a path forward so that they can learn how to do a wonderful job. Isn't it the job of the teacher to teach students how to do something well or to think about something in a new way?

Would you rather have your students incorporate your feedback into their work or know that you liked/disliked it?

Studies have shown that when students are given grades, grades with feedback or feedback alone on their work, students only look at the feedback when there is not a grade attached. So what is more important, the feedback/path towards improvement or the grade?

Additionally, when external evaluations are used to evaluate student learning, there is a dramatic increase in learning with a 1-year advantage by 8th grade!

Given a choice, should we keep teachers "grading" students or use external evaluations to empower our students to learn substantially more/better?

Erin Johnson

Sherman, the infirmities in some of my sweeping generalizations are certainly no greater than the infirmities in Deb's sweeping generalization re SES and student achievement. Had I known you were going to nitpick I would have used moe preceise language.

I do find find it odd that you, in Accountability Frankenstein, and Deb, in this post, rely on Gould to dismiss out of hand the IQ/student achievement correlation (which is at least as high as the SES acheivement correlation) and completely ignore all the research that is counter to Gould (and others) in reaching your conclusions. That seems to be confirmation bias.

Comments are now closed for this post.


Most Viewed on Education Week



Recent Comments