Not surprisingly, the debate surrounding the “leveling” of classes at DHS is quite heated and involved.  Pit a bunch of highly-educated and motivated parents against a school board and you can expect nothing less.

In contrast to yesterday’s posting detailing the reasons and research supporting the elimination of “tracking”, Cherie sent along two detailed messages that oppose the move and challenge the research provided by Principal McKain-Fernandez.  (Both can be read in the continuation at your leisure)

I’m not foolish enough to expound further on this subject from a research perspective.  I’ll let the self-proclaimed Ph.D.s fight this out.  I’ll just say from personal experience there are definite merits on both sides of this argument.  Classes with higher-achieving students were always more interesting and engaging than those without.  But at the same time, classes populated with non-honors students were really uninspiring, soul-sucking experiences.

So, what’s the compromise?

Dear Lauri,

I’m writing this in response to your April 2008 Bulldog News. In that issue you describe
a planned move from leveled to heterogeneous classes in grades 10-12 based upon the
positive experience in this year’s 9th grade. You invite comment and question. As you
know, I have a child in the ninth grade experience as well as a senior who went through
the old system. I question the interpretation of results from this year’s “freshmen
experience” presented. The “results” cited are an anecdote from a single (new) teacher
followed by reference to the research of Carol Dweck at Stanford.

As a research scientist, experienced in experimental design and analysis, I know that one
of the great dangers in science is to over or incorrectly interpret result of a given study. I
agree that Dr. Dweck’s research into “Mindsets” effect on learning is significant and
applicable to education in Decatur and beyond. However, her research does not address
structure of classes. That research focuses on the attitude of administration, teachers, and
ultimately the learner. She and Richard Lavoie, who is also referenced, show that labels
such as “smart”, “gifted”, “challenged” and the like can actually be limiting when they
are applied to a state of being, rather than doing. Likewise, they and others show that
learners who go on to excel are those who believe that their “actions” are linked to their
success and rewards rather than an innate property such as “smart” or “gifted”.  It is
important to realize that this learning attitude is not an inherent function of class structure
and my observation is that the “freshman experience” has missed that fundamental point.

If we truly want to support each individual toward realizing their potential to be the best
they can be, then we do have to have an environment that encourages that. A good body
of research that addresses this is that of “self-efficacy”
(http://www.des.emory.edu/mfp/self-efficacy.html) I fully support changing any “label”
on classes, however I see a some significant problems with the current plan moving
toward taking away the “honors” systems.
1) “Honors” is not an entitlement program – it is an achievement program.
a. I know that a tremendous motivator for my daughter Ellen, who does not
consider herself “gifted” academically, has been the idea that if she
worked hard, made good grades, and overcame personal challenges, that
she would be able to access small classes with like minded learners.
b. The converse is that she also knew that if she didn’t perform well, her
opportunity for change was diminished accordingly.
2) The heterogeneous placement of this year’s freshman system takes away
achievement-based incentives and rewards complacency.
a. I know many student who were excited about high school, and have
become bitter in response to the freshman experience.
b. Teachers in these mixed classes do NOT have the resources to provide for
advanced students. We have first hand experience of this.
c. There is no criteria for getting interested students in and uninterested out
3) Honors coursework helps those with college aspirations
a. This is true regardless of race, sex, or any other orientation – It provides a
platform that is more like the “real” world of achievement-based
academics.
4) I know of no marker that has been established to objectively evaluate success or
failure of any “change” to the current system.
a. Anecdotes saying that some people think this is or is not working are
insufficient.
b. Data collection should be done across the broad body of student faculty
parents and evaluated with and eye toward statistical significance in study
design.
c. Change for the sake of change is only randomly effective
5) A high percentage of students involved in honors coursework graduate from
Decatur High and get placed in higher education programs
a. Why is there action to remove a system that is shown to work as a
proposed solution for something that doesn’t?
b. Why not model other classes after what works with the honor system?
6) The external research cited as the support for the change appears to have very
little relationship to actions taken
a. This gets back to #4 – how will you know if your change is working for
the better or worse?
b. Do you have other evidence you aren’t sharing?

I recommend the administration instead consider a formal program where honor students
may act as study mentors for other students. This would be based on the idea of self-
efficacy where effort is recognized and rewarded. Such a program may achieve the
unifying result we would all like, and provide productive challenge to those of all
performance levels.

My request is that these above points be address directly in your public meeting.

My daughter, Ellen, requests the opportunity to speak at the meeting and give her
perspective as a ninth grade student. She has prepared an articulate presentation of less
than 3 minutes.

With regard,

James Nettles, PhD.
Assistant Professor, Department of Pediatrics,
Emory University School of Medicine
Father of 1 Decatur High freshman, and 1 Decatur High senior

———————————————

Review of Literature on Ability Grouping Julie Hotchkiss, Ph.D.

5/1/2008

– 1 –

To prepare this review of the literature, I selected the most recent papers I could find that offered an assessment of the impact of heterogeneous vs. homogeneous class structure on student performance outcomes. I limited the review to studies that considered performance of middle or secondary school students and appeared to offer some rigor in their analysis. The exception is the review of Carol Dweck’s book, which does not actually address the issue at hand, but has been used as justification for the changes at DHS. These studies use the terms homogeneous vs. heterogeneous, grouping vs. non-grouping, and tracking vs. non-tracking. I was careful to select papers for which all of these term pairs could be used synonymously.

1) The most important factor in student learning is not the structure of the classroom it is the curriculum, motivation, challenge, etc. that students receive in the classroom. Praising talent and focusing on encouraging a growth mindset, as opposed to a fixed mindset are included in

the positive characteristics of instruction at all levels. Other research found that heterogeneity without concerted modification addressing the unique needs of each student will doom students of all skill levels to failure.

Carol Dweck, Mindset: The New Psychology of Success (Random House, 2006).

The basic message of this book is that people who have a growth mindset will accomplish more, perform better, etc.

than people with a fixed mindset. In it’s application to the educational setting, promoting or teaching a growth mindset amounts to praising ability, not talent, in order to motivate students to work harder and achieve more. This message, which has applications beyond secondary education, has been used by others as an argument against grouping students

by ability.

Grouping students by ability is not praising talent; it is an efficient use of limited resources (including students’ time).

Teaching that intelligence is something learned and developed should be a goal for any level of class; the method of doing this will likely differ given the natural abilities of students. Grouping students by ability does not relieve any teacher from the obligation of motivating students to achieve their fullest potential. Carol Dweck’s book advocates a paradigm, not a specific class structure. There is no test in this book about whether grouping or not grouping students by ability facilitates or hinders a growth mindset. In fact, the author concludes from one of the studies she cites, “It didn’t matter whether students started the year in the high- or low-ability groups. Both groups [in which teachers practiced a growth mindset] ended the year way up high” (p. 66).

Yehezkel Dar and Nura Resh, “Classroom Intellectual Composition and Academic Achievement,” American

Educational Research Journal (Fall 1986): 357-374.

The goal of this research was to investigate the relative importance of ethnic, socioeconomic, and intellectual heterogeneity in determining performance on standardized tests for groups of Israel students in 8th through 12th grade.

In terms of performance, it was found that the intellectual component of the student composition outweighs both ethnic and socioeconomic components and that classroom intellectual level is a more important determinant of individual performance than the variance in intellectual ability. The implication is that low-ability students will benefit from the higher intellectual level of heterogeneous classes and that high-ability students will suffer. The analysis also finds that the gains to low-ability students outweigh the losses to high-ability students, but high-ability students still suffer.

Carol Ann Tomlinson et al., “Differentiating Instruction in Response to Student Readiness, Interest, and Learning

Profile in Academically Diverse Classrooms,” Journal for the Education of the Gifted, (Winter 2003): 119-145.

This article provides an extensive review of the literature on how effective heterogeneous classrooms are in delivering academically responsive instruction. The article also offers a prescription about how a classroom can deliver effective differentiated instruction. The primary conclusion from the review is that, “…most teachers still do little to adjust their instruction in ways that effectively reach out to academically diverse populations.”

Review of Literature on Ability Grouping Julie Hotchkiss, Ph.D.

5/1/2008

– 2 –

The main lesson from this article is that classroom heterogeneity without concerted modification addressing the unique needs of each student will result in poorer performance among students of all skill levels. There are many dimensions on which students will differ as they enter the classroom: readiness, interest, learning profile (or, a combination of learning style, intelligence preference, gender, and culture).

Noreen M. Webb et al., “Short Circuits or Superconductors? Effects of Group composition on High-Achieving Students’

Science Assessment Performance,” American Educational Research Journal (Winter 2002): 943-989.

This study investigates the performance primarily of high-ability students in a science assessment. Some students were in homogeneous and some were in heterogeneous ability classrooms. Performance of low-ability students was also investigated.

Low ability students were found to perform better in heterogeneous settings, whereas some high ability students performed better and some worse in heterogeneous settings. The performance of high-ability students depended on the social interaction and learning environment of the heterogeneous group. The differences between performance of highability students in hetero- and homogeneous environments were tied to characteristics of the students, existing

relationships between members of the group, and history of classroom interaction. The main point is that it is the environment of the classroom and what students bring to the classroom that seems to matter more than merely the hetero- or homogeneous structure of the class. The focus then becomes on the ability of the teacher of a heterogeneous classroom to foster cooperation and learning.

A very clear implication identified by the authors is, “…heterogeneous groups should not be considered as a single, undifferentiated category of group composition. It is more informative to view heterogeneous groups as highperforming or low-performing according to the groups’ dynamics.” And, that the performance of high-ability students will suffer in poorly functioning heterogeneous groups.

2) There is also evidence, from the most rigorous analysis reviewed, that the performance of both high-ability and low-ability students suffers in heterogeneous groups, relative to performance in homogeneous group settings.

David Figlio and Marianne Page, “School Choice and the Distributional Effects of Ability

Tracking: Does Separation Increase Inequality?” Journal of Urban Economics 51(3) (2002).

This article does a careful job of correcting several of the empirical problems that make the results in the education literature suspect. Namely, the researchers account for the potential endogeneity of the placement of students into a specific tracks and for the fact that the existence of grouping may determine school choice, changing the distribution of student ability itself. The authors make use of the National Educational Longitudinal Study, which produces a nationally representative sample of 8th, 10th, and 12th graders. The outcome measure is performance on the itemized response theory (IRT) math test score. The authors compare student outcomes across schools that group and schools that do not group, based on performance of the students themselves–not comparing based on their track designation.

This provides for a cleaner measure of impact across a true distribution of ability, rather than a tracking-designated determination of ability.

The researchers conclude, “When we address these issues we find no evidence that low-ability students are harmed by being grouped together and conclude that the trend away from tracking is misguided. In fact, we find that tracking programs may be associated with test score gains for students in the bottom third of the initial test score distribution.

We conclude that the move to end tracking may harm the very students that it is intended to help.”

Michael Fertig, “Educational Production, Endogenous Peer Group Formation and Class Composition — Evidence from the PISA 2000 Study,” IZA Discussion Paper #714 (February 2003).

This paper investigates the impact of achievement heterogeneity on student (15-16 year olds) performance on standardized reading test. The level of aggregation is the school level, so application to the classroom setting may not

Review of Literature on Ability Grouping Julie Hotchkiss, Ph.D.

5/1/2008

– 3 –

be direct. The analysis controls for other potentially important performance determinants, including class size, parental education level, student-teacher ratios, physical condition of school, and family structure.

The analysis concludes, “…the higher the heterogeneity of achievement in a student’s school, the lower is the individual performance.” An increase of 10% in the coefficient of variation in student test performance (excluding the own student’s score) results in a decline of 17 points in the student’s reading score (the mean reading score is about 500, so this decline is about 3% off the mean).

3) Raising the expectations and acceleration of instruction of all group levels improves outcomes of all students, low and high ability. Grouping students into heterogeneous ability groups was found to be beneficial to all levels when the level of instruction was at the highest level — raising all boats. In this case, it wasn’t the heterogeneity that improved performance; it was, again, the quality of instruction.

Carol Corbett Burris et al., “Math Acceleration for All,” Educational Leadership (February 2004): 68-71.

also see

Carol Corbett Burris et al., “Accelerating Mathematics Achievement Using Heterogeneous Grouping,” American

Educational Research Journal (Spring 2006): 105-136.

This article describes a program adopted by a school district in New York in 1995 in which accelerated math wasoffered to all students in middle school. The program culminated with all 8th grade students completing algebra.

The point of the article is not that class structure is important, but that expectations and opportunities for all students can be successfully elevated. Again, the focus is on type and quality of instruction, not class structure. Indeed, the authors state, “…all students can reach high levels of achievement if they receive high-quality curriculum and instruction.”

Another important feature of the success of this program was that it took years to implement, gradually increasing enrollment in the accelerated math classes over several years.

A panel analysis of the impact of accelerating math for all students found that the percentages of all student groups (by race, poverty, initial achievement level) taking math courses beyond algebra II in high school increase significantly.

Notably, initially identified high achievers were more likely to take even the highest level math classes. Performance of this post-acceleration cohort of high achievers on AP exams was higher than performance of the pre-acceleration cohort.

No indication of what other controls were used in the analysis.

4) If the mechanism for grouping students is not effective in resulting in truly homogeneous ability groups, then the gains from grouping are lost. Again, effort needs to be exerted where the problem is — effective grouping.

Maureen T. Hallinan, “Ability Grouping and Student Learning,” Brookings Papers on Education Policy (2003): 95-124.

also see

Maureen T. Hallinan, “Tracking: From Theory to Practice” Sociology of Education (April 1994).

The main results from this paper are that classes designed to group students by ability rarely succeed in doing so; there is a significant degree of heterogeneity remaining in classes that are grouped by ability. The result is that students do not reap the expected advantage of having curriculum geared toward their skill level and learning styles, because teachers still face significant heterogeneity in student skill. The source of this failure of ability grouping to achieve the expected outcome is that the mechanism that assigns students to groups is flawed — there are too many socioeconomic factors that come into play in deciding ability group.

A second result is that any student placed in a higher-ability group improves his/her performance on tests, and the performance of any student placed in a lower-ability group performs worse. The researchers trace this outcome to the fact that students in low-ability groups are given less challenging material and fewer opportunities to learn.

Review of Literature on Ability Grouping Julie Hotchkiss, Ph.D.

5/1/2008

– 4 –

The implication, then, is that students should be assigned to the highest level group possible without discouraging the student’s efforts. A second implication is that the conditions for learning across all groups should resembled those typically found in higher-ability classes, “If learning opportunities can be made independent of ability group level, then teachers are free to create more homogeneous groups to better gear instruction to the learning level of the students.”

The key is to get the groups as homogeneous as possible, however, and the reason we see failures in ability-grouped environments is because not enough effort and objectivity is used to correctly form the groups.

5) Change can not be made in a vacuum. One analysis identified that eliminating grouping may result in a school losing its high-ability students to the private sector, but gain lowerability students from families wealthy enough to have chosen private school before. The net result from eliminating grouping is a movement toward the lower end of the ability distribution and a loss of school enrollment of families able to afford private schooling.

Dennis Epple et al.,”Ability Tracking, School Competition, and the Distribution of Educational Benefits,” NBER
Working Paper #7854 (August 2000).

This theoretical analysis identifies a potential unintended consequence for the school that decides to do away with tracking students, or grouping students based on ability. The researchers investigate how the presence of a policy of tracking affects the ability composition of the school as the policy of tracking changes the attractiveness of private school opportunities for certain students.

The researchers conclude that, “some degree of tracking can both increase the size of the public school sector and its average quality [in terms of student performance],” but it does so at the cost of modifying the socioeconomic distribution of students within the school. Specifically, public schools are found to lose a greater proportion of higherability students by de-tracking, but gain more wealthy, lower-ability students who would have otherwise gone private.

Empirical confirmation of this theoretical prediction is found in Figlio and Page (see #2 above). They find that the addition of a tracking program decreases the proportion of students in the low end of the socioeconomic distribution, students who are also more often found in the lower end of the ability distribution.