U.S. Department of Education: Value-Added Not Good for Evaluating Schools and Principals

Just this month, the Institute of Education Sciences (IES) wing of the U.S. Department of Education released a report about using value-added models (VAMs) for measuring school principals’ performance. The article conducted by researchers at Mathematica Policy Research and titled “Can Student Test Scores Provide Useful Measures of School Principals’ Performance?” can be found online here, with my summary of the study findings highlighted next and herein.

Before the passage of the Every Student Succeeds Act (ESSA), 40 states had written into their state statutes, as incentivized by the federal government, to use growth in student achievement growth for annual principal evaluation purposes. More states had written growth/value-added models (VAMs) for teacher evaluation purposes, which we have covered extensively via this blog, but this pertains only to school and/or principal evaluation purposes. Now since the passage of ESSA, and the reduction in the federal government’s control over state-level policies, states now have much more liberty to more freely decide whether to continue using student achievement growth for either purposes. This paper is positioned within this reasoning, and more specifically to help states decide whether or to what extent they might (or might not) continue to move forward with using growth/VAMs for school and principal evaluation purposes.

Researchers, more specifically, assessed (1) reliability – or the consistency or stability of these ratings over time, which is important “because only stable parts of a rating have the potential to contain information about principals’ future performance; unstable parts reflect only transient aspects of their performance;” and (2) one form of multiple evidences of validity – the predictive validity of these principal-level measures, with predictive validity defined as “the extent to which ratings from these measures accurately reflect principals’ contributions to student achievement in future years.” In short, “A measure could have high predictive validity only if [emphasis added] it was highly stable between consecutive years [i.e., reliability]…and its stable part was strongly related to principals’ contributions to student achievement” over time (i.e., predictive validity).

Researchers used principal-level value-added (unadjusted and adjusted for prior achievement and other potentially biasing demographic variables) to more directly examine “the extent to which student achievement growth at a school differed from average growth statewide for students with similar prior achievement and background characteristics.” Also important to note is that the data they used to examine school-level value-added came from Pennsylvania, which is one of a handful of states that uses the popular and proprietary (and controversial) Education Value-Added Assessment System (EVAAS) statewide.

Here are the researchers’ key findings, taken directly from the study’s summary (again, for more information see the full manuscript here).

  • The two performance measures in this study that did not account for students’ past achievement—average achievement and adjusted average achievement—provided no information for predicting principals’ contributions to student achievement in the following year.
  • The two performance measures in this study that accounted for students’ past achievement—school value-added and adjusted school value-added—provided, at most, a small amount of information for predicting principals’ contributions to student achievement in the following year. This was due to instability and inaccuracy in the stable parts.
  • Averaging performance measures across multiple recent years did not improve their accuracy for predicting principals’ contributions to student achievement in the following year. In simpler terms, a principal’s average rating over three years did not predict his or her future contributions more accurately than did a rating from the most recent year only. This is more of a statistical finding than one that has direct implications for policy and practice (except for silly states who might, despite findings like those presented in this study, decide that they can use one year to do this not at all well instead of three years to do this not at all well).

Their bottom line? “…no available measures of principal [/school] performance have yet been shown to accurately identify principals [/schools] who will contribute successfully to student outcomes in future years,” especially if based on students’ test scores, although the researchers also assert that “no research has ever determined whether non-test measures, such as measures of principals’ leadership practices, [have successfully or accurately] predict[ed] their future contributions” either.

The researchers follow-up with a highly cautionary note: “the value-added measures will make plenty of mistakes when trying to identify principals [/schools] who will contribute effectively or ineffectively to student achievement in future years. Therefore, states and districts should exercise caution when using these measures to make major decisions about principals. Given the inaccuracy of the test-based measures, state and district leaders and researchers should also make every effort to identify nontest measures that can predict principals’ future contributions to student outcomes [instead].”

Citation: Chiang, H., McCullough, M., Lipscomb, S., & Gill, B. (2016). Can student test scores provide useful measures of school principals’ performance? Washington DC: U.S. Department of Education, Institute of Education Sciences. Retrieved from http://ies.ed.gov/ncee/pubs/2016002/pdf/2016002.pdf

New Empirical Evidence: Students’ “Persistent Economic Disadvantage” More Likely to Bias Value-Added Estimates

The National Bureau of Economic Research (NBER) recently released a circulated but not-yet internally or externally reviewed study titled “The Gap within the Gap: Using Longitudinal Data to Understand Income Differences in Student Achievement.” Note that we have covered NBER studies such as this in the past in this blog, so in all fairness and like I have noted in the past, this paper should also be critically consumed, as well as my interpretations of the authors’ findings.

Nevertheless, this study is authored by Katherine Michelmore — Assistant Professor of Public Administration and International Affairs at Syracuse University, and Susan Dynarski — Professor of Public Policy, Education, and Economics at the University of Michigan, and this study is entirely relevant to value-added models (VAMs). Hence, below I cover their key highlights and takeaways, as I see them. I should note up front, however, that the authors did not directly examine how the new measure of economic disadvantage that they introduce (see below) actually affects calculations of teacher-level value-added. Rather, they motivate their analyses by saying that calculating teacher value-added is one application of their analyses.

The background to their study is as follows: “Gaps in educational achievement between high- and low-income children are growing” (p. 1), but the data that are used to capture “high- and low-income” in the state of Michigan (i.e., the state in which their study took place) and many if not most other states throughout the US, capture “income” demographics in very rudimentary, blunt, and often binary ways (i.e., “yes” for students who are eligible to receive federally funded free-or-reduced lunches and “no” for the ineligible).

Consequently, in this study the authors “leverage[d] the longitudinal structure of these data sets to develop a new measure of persistent economic disadvantage” (p. 1), all the while defining “persistent economic disadvantage” by the extent to which students were “eligible for subsidized meals in every grade since kindergarten” (p. 8). Students “who [were] never eligible for subsidized meals during those grades [were] defined as never [being economically] disadvantaged” (p. 8), and students who were eligible for subsidized meals for variable years were defined as “transitorily disadvantaged” (p. 8). This all runs counter, however, to the binary codes typically used, again, across the nation.

Appropriately, then, their goal (among other things) was to see how a new measure they constructed to better measure and capture “persistent economic disadvantage” might help when calculating teacher-level value-added. They accordingly argue (among other things) that, perhaps, not accounting for persistent disadvantage might subsequently cause more biased value-added estimates “against teachers of [and perhaps schools educating] persistently disadvantaged children” (p. 3). This, of course, also depends on how persistently disadvantaged students are (non)randomly assigned to teachers.

With statistics like the following as also reported in their report: “Students [in Michigan] [persistently] disadvantaged by 8th grade were six times more likely to be black and four times more likely to be Hispanic, compared to those who were never disadvantaged,” their assertions speak volumes not only to the importance of their findings for educational policy, but also for the teachers and schools still being evaluated using value-added scores and the researchers investigating, criticizing, promoting, or even trying to make these models better (if that is possible). In short, though, teachers who are disproportionately teaching in urban areas with more students akin to their equally disadvantaged peers, might realize relatively more biased value-added estimates as a result.

For value-added purposes, then, it is clear that the assumptions that controlling for student disadvantage by using such basal indicators of current economic disadvantage is overly simplistic, and just using test scores to also count for this economic disadvantage (i.e., as promoted in most versions of the Education Value-Added Assessment System (EVAAS)) is likely worse. More specifically, the assumption that economic disadvantage also does not impact some students more than others over time, or over the period of data being used to capture value-added (typically 3-5 years of students’ test score data), is also highly susceptible. “[T]hat children who are persistently disadvantaged perform worse than those who are disadvantaged in only some grades” (p. 14) also violates another fundamental assumption that teachers’ effects are consistent over time for similar students who learn at more or less consistent rates over time, regardless of these and other demographics.

The bottom line here, then, is that the indicator that should be used instead of our currently used proxies for current economic disadvantage is the number of grades students spend in economic disadvantage. If the value-added indicator does not effectively account for the “negative, nearly linear relationship between [students’ test] scores and the number of grades spent in economic disadvantage” (p. 18), while controlling for other student demographics and school fixed effects, value-added estimates will likely be (even) more biased against teachers who teach these students as a result.

Otherwise, teachers who teach students with persistent economic disadvantages will likely have it worse (i.e., in terms of bias) than teachers who teach students with current economic disadvantages, teachers who teach students with economically disadvantaged in their current or past histories will have it worse than teachers who teach students without (m)any prior economic disadvantages, and so on.

Citation: Michelmore, K., & Dynarski, S. (2016). The gap within the gap: Using longitudinal data to understand income differences in student achievement. Cambridge, MA: National Bureau of Economic Research (NBER). Retrieved from http://www.nber.org/papers/w22474

New Mexico Lawsuit Update

As you all likely recall, the American Federation of Teachers (AFT), joined by the Albuquerque Teachers Federation (ATF), last fall, filed a “Lawsuit in New Mexico Challenging [the] State’s Teacher Evaluation System.” Plaintiffs charged that the state’s teacher evaluation system, imposed on the state in 2012 by the state’s current Public Education Department (PED) Secretary Hanna Skandera (with value-added counting for 50% of teachers’ evaluation scores), was unfair, error-ridden, spurious, harming teachers, and depriving students of high-quality educators, among other claims (see the actual lawsuit here). Again, I’m serving as the expert witness on the side of the plaintiffs in this suit.

As you all likely also recall, in December of 2015, State District Judge David K. Thomson granted a preliminary injunction preventing consequences from being attached to the state’s teacher evaluation data. More specifically, Judge Thomson ruled that the state could proceed with “developing” and “improving” its teacher evaluation system, but the state was not to make any consequential decisions about New Mexico’s teachers using the data the state collected until the state (and/or others external to the state) could evidence to the court during another trial (initially set for April 2016, then postponed to October 2016, and likely to be postponed again) that the system is reliable, valid, fair, uniform, and the like (see prior post on this ruling here).

Well, many of you have (since these prior posts) written requesting updates regarding this lawsuit, and here is one as released jointly by the AFT and ATF. This accurately captures the current and ongoing situation:

September 23, 2016

Many of you will remember the classic Christmas program, Rudolph the Red Nose Reindeer, and how the terrible and menacing abominable snowman became harmless once his teeth were removed. This is how you should view the PED evaluation you recently received – a harmless abominable snowman.  

The math is still wrong, the methodology deeply flawed, but the preliminary injunction achieved by our union, removed the teeth from PED’s evaluations, and so there is no reason to worry. As explained below, we will continue to fight these evaluations and will not rest until the PED institutes an evaluation system that is fair, meaningful, and consistently applied.

For all of you, who just got arbitrarily labeled by the PED in your summative evaluations, just remember, like the abominable snowman, these labels have no teeth, and your career is safe.

2014-2015 Evaluations

These evaluations, as you know, were the subject of our lawsuit filed in 2014. As a result of the Court’s order, the preliminary injunction, no negative consequences can result from your value-added scores.

In an effort to comply with the Court’s order, the PED announced in May it would be issuing new regulations.  This did not happen, and it did not happen in June, in July, in August, or in September. The bottom line is the PED still has not issued new regulations – though it still promises that those regulations are coming soon. So much for accountability.

The trial on the old regulations, scheduled for October 24, has been postponed based upon the PED’s repetitive assertions that new regulations would be issued.

In addition, we have repeatedly asked the PED to provide their data, which they finally did, however it lacked the codebook necessary to meaningfully interpret the data. We view this as yet another stall tactic.

Soon, we will petition the Court for an order compelling PED to produce the documents it promised months ago. Our union’s lawyers and expert witnesses will use this data to critically analyze the PED’s claims and methodology … again.

2015-2016 Evaluations

Even though the PED has condensed the number of ways an educator can be evaluated in a false attempt to satisfy the Courts, the fact remains that value-added models are based on false math and highly inaccurate data. In addition to the PED’s information we have requested for the 2014-2015 evaluations, we have requested all data associated with the current 2015-2016 evaluations.

If our experts determine the summative evaluation scores are again, “based on fundamentally, and irreparably, flawed methodology which is further plagued by consistent and appalling data errors,” we will also challenge the 2015-2016 evaluations. If the PED ever releases new regulations, and we determine that they violate statute (again), we will challenge those regulations, as well.

Rest assured our union will not stop challenging the PED until we are satisfied they have adopted an evaluation system that is respectful of students and educators. We will keep you updated as we learn more information, including the release of new regulations and the rescheduled trial date.

In Solidarity,

Stephanie Ly                                   Ellen Bernstein
President, AFT NM                         President, ATF

New Mexico Is “At It Again”

“A Concerned New Mexico Parent” sent me yet another blog entry for you all to stay apprised of the ongoing “situation” in New Mexico and the continuous escapades of the New Mexico Public Education Department (NMPED). See “A Concerned New Mexico Parent’s” prior posts here, here, and here, but in this one (s)he writes what follows:

Well, the NMPED is at it again.

They just released the teacher evaluation results for the 2015-2016 school year. And, the report and media press releases are a something.

Readers of this blog are familiar with my earlier documentation of the myriad varieties of scoring formulas used by New Mexico to evaluate its teachers. If I recall, I found something like 200 variations in scoring formulas [see his/her prior post on this here with an actual variation count at n=217].

However, a recent article published in the Albuquerque Journal indicates that, now according to the NMPED, “only three types of test scores are [being] used in the calculation: Partnership for Assessment of Readiness for College and Careers [PARCC], end-of-course exams, and the [state’s new] Istation literacy test.” [Recall from another article released last January that New Mexico’s Secretary of Education Hanna Skandera is also the head of the governing board for the PARCC test].

Further, the Albuquerque Journal article author reports that the “PED also altered the way it classifies teachers, dropping from 107 options to three. Previously, the system incorporated many combinations of criteria such as a teacher’s years in the classroom and the type of standardized test they administer.”

The new state-wide evaluation plan is also available in more detail here. Although I should also add that there has been no published notification of the radical changes in this plan. It was just simply and quietly posted on NMPED’s public website.

Important to note, though, is that for Group B teachers (all levels), the many variations documented previously have all been replaced by end-of-course (EOC) exams. Also note that for Group A teachers (all levels) the percentage assigned to the PARCC test has been reduced from 50% to 35%. (Oh, how the mighty have fallen …). The remaining 15% of the Group A score is to be composed of EOC exam scores.

There are only two small problems with this NMPED simplification.

First, in many districts, no EOC exams were given to Group B teachers in the 2015-2016 school year, and none were given in the previous year either. Any EOC scores that might exist were from a solitary administration of EOC exams three years previously.

Second, for Group A teachers whose scores formerly relied solely on the PARCC test for 50% of their score, no EOC exams were ever given.

Thus, NMPED has replaced their policy of evaluating teachers on the basis of students they don’t teach to this new policy of evaluating teachers on the basis of tests they never administered!

Well done, NMPED (not…)

Luckily, NMPED still cannot make any consequential decisions based on these data, again, until NMPED proves to the court that the consequential decisions that they would still very much like to make (e.g., employment, advancement and licensure decisions) are backed by research evidence. I know, interesting concept…

One Score and Seven Policy Iterations Ago…

I just read what might be one of the best articles I’ve read in a long time on using test scores to measure teacher effectiveness, and why this is such a bad idea. Not surprisingly, unfortunately, this article was written 20 years ago (i.e., 1986) by – Edward Haertel, National Academy of Education member and recently retired Professor at Stanford University. If the name sounds familiar, it should as Professor Emeritus Haertel is one of the best on the topic of, and history behind VAMs (see prior posts about his related scholarship here, here, and here). To access the full article, please scroll to the reference at the bottom of this post.

Heartel wrote this article when at the time policymakers were, like they still are now, trying to hold teachers accountable for their students’ learning as measured on states’ standardized test scores. Although this article deals with minimum competency tests, which were in policy fashion at the time, about seven policy iterations ago, the contents of the article still have much relevance given where we are today — investing in “new and improved” Common Core tests and still riding on unsinkable beliefs that this is the way to reform the schools that have been in despair and (still) in need of major repair since 20+ years ago.

Here are some of the points I found of most “value:”

  • On isolating teacher effects: “Inferring teacher competence from test scores requires the isolation of teaching effects from other major influences on student test performance,” while “the task is to support an interpretation of student test performance as reflecting teacher competence by providing evidence against plausible rival hypotheses or interpretation.” While “student achievement depends on multiple factors, many of which are out of the teacher’s control,” and many of which cannot and likely never will be able to be “controlled.” In terms of home supports, “students enjoy varying levels of out-of-school support for learning. Not only may parental support and expectations influence student motivation and effort, but some parents may share directly in the task of instruction itself, reading with children, for example, or assisting them with homework.” In terms of school supports, “[s]choolwide learning climate refers to the host of factors that make a school more than a collection of self-contained classrooms. Where the principal is a strong instructional leader; where schoolwide policies on attendance, drug use, and discipline are consistently enforced; where the dominant peer culture is achievement-oriented; and where the school is actively supported by parents and the community.” This, all, makes isolating the teacher effect nearly if not wholly impossible.
  • On the difficulties with defining the teacher effect: “Does it include homework? Does it include self-directed study initiated by the student? How about tutoring by a parent or an older sister or brother? For present purposes, instruction logically refers to whatever the teacher being evaluated is responsible for, but there are degrees of responsibility, and it is often shared. If a teacher informs parents of a student’s learning difficulties and they arrange for private tutoring, is the teacher responsible for the student’s improvement? Suppose the teacher merely gives the student low marks, the student informs her parents, and they arrange for a tutor? Should teachers be credited with inspiring a student’s independent study of school subjects? There is no time to dwell on these difficulties; others lie ahead. Recognizing that some ambiguity remains, it may suffice to define instruction as any learning activity directed by the teacher, including homework….The question also must be confronted of what knowledge counts as achievement. The math teacher who digresses into lectures on beekeeping may be effective in communicating information, but for purposes of teacher evaluation the learning outcomes will not match those of a colleague who sticks to quadratic equations.” Much if not all of this cannot and likely never will be able to be “controlled” or “factored” in or our, as well.
  • On standardized tests: The best of standardized tests will (likely) always be too imperfect and not up to the teacher evaluation task, no matter the extent to which they are pitched as “new and improved.” While it might appear that these “problem[s] could be solved with better tests,” they cannot. Ultimately, all that these tests provide is “a sample of student performance. The inference that this performance reflects educational achievement [not to mention teacher effectiveness] is probabilistic [emphasis added], and is only justified under certain conditions.” Likewise, these tests “measure only a subset of important learning objectives, and if teachers are rated on their students’ attainment of just those outcomes, instruction of unmeasured objectives [is also] slighted.” Like it was then as it still is today, “it has become a commonplace that standardized student achievement tests are ill-suited for teacher evaluation.”
  • On the multiple choice formats of such tests: “[A] multiple-choice item remains a recognition task, in which the problem is to find the best of a small number of predetermined alternatives and the cri- teria for comparing the alternatives are well defined. The nonacademic situations where school learning is ultimately ap- plied rarely present problems in this neat, closed form. Discovery and definition of the problem itself and production of a variety of solutions are called for, not selection among a set of fixed alternatives.”
  • On students and the scores they are to contribute to the teacher evaluation formula: “Students varying in their readiness to profit from instruction are said to differ in aptitude. Not only general cognitive abilities, but relevant prior instruction, motivation, and specific inter- actions of these and other learner characteristics with features of the curriculum and instruction will affect academic growth.” In other words, one cannot simply assume all students will learn or grow at the same rate with the same teacher. Rather, they will learn at different rates given their aptitudes, their “readiness to profit from instruction,” the teachers’ instruction, and sometimes despite the teachers’ instruction or what the teacher teaches.
  • And on the formative nature of such tests, as it was then: “Teachers rarely consult standardized test results except, perhaps, for initial grouping or placement of students, and they believe that the tests are of more value to school or district administrators than to themselves.”

Sound familiar?

Reference: Haertel, E. (1986). The valid use of student performance measures for teacher evaluation. Educational Evaluation and Policy Analysis, 8(1), 45-60.

The Late Stephen Jay Gould on IQ Testing (with Implications for Testing Today)

One of my doctoral students sent me a YouTube video I feel compelled to share with you all. It is an interview with one of my all time favorite and most admired academics — Stephen Jay Gould. Gould, who passed away at age 60 from cancer, was a paleontologist, evolutionary biologist, and scientist who spent most of his academic career at Harvard. He was “one of the most influential and widely read writers of popular science of his generation,” and he was also the author of one of my favorite books of all time: The Mismeasure of Man (1981).

In The Mismeasure of Man Gould examined the history of psychometrics and the history of intelligence testing (e.g., the methods of nineteenth century craniometry, or the physical measures of peoples’ skulls to “objectively” capture their intelligence). Gould examined psychological testing and the uses of all sorts of tests and measurements to inform decisions (which is still, as we know, uber-relevant today) as well as “inform” biological determinism (i.e., “the view that “social and economic differences between human groups—primarily races, classes, and sexes—arise from inherited, inborn distinctions and that society, in this sense, is an accurate reflection of biology). Gould also examined in this book the general use of mathematics and “objective” numbers writ large to measure pretty much anything, as well as to measure and evidence predetermined sets of conclusions. This book is, as I mentioned, one of the best. I highly recommend it to all.

In this seven-minute video, you can get a sense of what this book is all about, as also so relevant to that which we continue to believe or not believe about tests and what they really are or are not worth. Thanks, again, to my doctoral student for finding this as this is a treasure not to be buried, especially given Gould’s 2002 passing.

Five “Indisputable” Reasons Why VAMs are Good?

Just this week, in Education Week — the field’s leading national newspaper covering K–12 education — a blogger by the name of Matthew Lynch published a piece explaining his “Five Indisputable [emphasis added] Reasons Why You Should Be Implementing Value-Added Assessment.”

I’m going to try to stay aboveboard with my critique of this piece, as best I can, as by the title alone you all can infer there are certainly pieces (mainly five) to be seriously criticized about the author’s indisputable take on value-added (and by default value-added models (VAMs)). I examine each of these assertions below, but I will say overall and before we begin, that pretty much everything that is included in this piece is hardly palatable, and tolerable considering that Education Week published it, and by publishing it they quasi-endorsed it, even if in an independent blog post that they likely at minimum reviewed, then made public.

First, the five assertions, along with a simple response per assertion:

1. Value-added assessment moves the focus from statistics and demographics to asking of essential questions such as, “How well are students progressing?”

In theory, yes – this is generally true (see also my response about the demographics piece replicated in assertion #3 below). The problem here, though, as we all should know by now, is that once we move away from the theory in support of value-added, this theory more or less crumbles. The majority of the research on this topic explains and evidences the reasons why. Is value-added better than what “we” did before, however, while measuring student achievement once per year without taking growth over time into consideration? Perhaps, but if it worked as intended, for sure!

2. Value-added assessment focuses on student growth, which allows teachers and students to be recognized for their improvement. This measurement applies equally to high-performing and advantaged students and under-performing or disadvantaged students.

Indeed, the focus is on growth (see my response about growth in assertion #1 above). What the author of this post does not understand, however, is that his latter conclusion is likely THE most controversial issue surrounding value-added, and on this all topical researchers likely agree. In fact, authors of the most recent review of what is actually called “bias” in value-added estimates, as published in the peer-reviewed Economics Education Review (see a pre-publication version of this manuscript here), concluded that because of potential bias (i.e., “This measurement [does not apply] equally to high-performing and advantaged students and under-performing or disadvantaged students“), that all value-added modelers should control for as many student-level (and other) demographic variables to help to minimize this potential, also given the extent to which multiple authors’ evidence of bias varies wildly (from negligible to considerable).

3. Value-added assessment provides results that are tied to teacher effectiveness, not student demographics; this is a much more fair accountability measure.

See my comment immediately above, with general emphasis added to this overly simplistic take on the extent to which VAMs yield “fair” estimates, free from the biasing effects (never to always) caused by such demographics. My “fairest” interpretation of the current albeit controversial research surrounding this particular issue is that bias does not exist across teacher-level estimates, but it certainly occurs when teachers are non-randomly assigned highly homogenous sets of students who are gifted, who are English Language Learners (ELLs), who are enrolled in special education programs, who disproportionately represent racial minority groups, who disproportionately come from lower socioeconomic backgrounds, and who have been retained in grade prior.

4. Value-added assessment is not a stand-alone solution, but it does provide rich data that helps educators make data-driven decisions.

This is entirely false. There is no research evidence, still to date, that teachers use these data to make instructional decisions. Accordingly, no research is linked to or cited here (as well as elsewhere). Now, if the author is talking about naive “educators,” in general, who make consequential decisions as based on poor (i.e., the oppostie of “rich”) data, this assertion would be true. This “truth,” in fact, is at the core of the lawsuits ongoing across the nation regarding this matter (see, for example, here), with consequences ranging from tagging a teacher’s file for receiving a low value-added score to teacher termination.

5. Value-added assessment assumes that teachers matter and recognizes that a good teacher can facilitate student improvement. Perhaps we have only value-added assessment to thank for “assuming” [sic] this. Enough said…

Or not…

Lastly, the author professes to be a “professor,” pretty much all over the place (see, again, here), although he is currently an associate professor. There is a difference, and folks who respect the difference typically make the distinction explicit and known, especially in an academic setting or context. See also here, however, given his expertise (or the lack thereof) in value-added or VAMs, about what he writes here as “indisputable.”

Perhaps most important here, though, is that his falsely inflated professional title implies, especially to a naive or uncritical public, that what he has to say, again without any research support, demands some kind of credibility and respect. Unfortunately, this is just not the case; hence, we are again reminded of the need for general readers to be critical in their consumption of such pieces. I would have thought Education Week would have played a larger role than this, rather than just putting this stuff “out there,” even if for simple debate or discussion.

Another Oldie but Still Very Relevant Goodie, by McCaffrey et al.

I recently re-read an article in full that is now 10 years old, or 10 years out, as published in 2004 and, as per the words of the authors, before VAM approaches were “widely adopted in formal state or district accountability systems.” Unfortunately, I consistently find it interesting, particularly in terms of the research on VAMs, to re-explore/re-discover what we actually knew 10 years ago about VAMs, as most of the time, this serves as a reminder of how things, most of the time, have not changed.

The article, “Models for Value-Added Modeling of Teacher Effects,” is authored by Daniel McCaffrey (Educational Testing Service [ETS] Scientist, and still a “big name” in VAM research), J. R. Lockwood (RAND Corporation Scientists),  Daniel Koretz (Professor at Harvard), Thomas Louis (Professor at Johns Hopkins), and Laura Hamilton (RAND Corporation Scientist).

At the point at which the authors wrote this article, besides the aforementioned data and data base issues, were issues with “multiple measures on the same student and multiple teachers instructing each student” as “[c]lass groupings of students change annually, and students are taught by a different teacher each year.” Authors, more specifically, questioned “whether VAM really does remove the effects of factors such as prior performance and [students’] socio-economic status, and thereby provide[s] a more accurate indicator of teacher effectiveness.”

The assertions they advanced, accordingly and as relevant to these questions, follow:

  • Across different types of VAMs, given different types of approaches to control for some of the above (e.g., bias), teachers’ contribution to total variability in test scores (as per value-added gains) ranged from 3% to 20%. That is, teachers can realistically only be held accountable for 3% to 20% of the variance in test scores using VAMs, while the other 80% to 97% of the variance (stil) comes from influences outside of the teacher’s control. A similar statistic (i.e., 1% to 14%) was similarly and recently highlighted in the recent position statement on VAMs released by the American Statistical Association.
  • Most VAMs focus exclusively on scores from standardized assessments, although I will take this one-step further now, noting that all VAMs now focus exclusively on large-scale standardized tests. This I evidenced in a recent paper I published here: Putting growth and value-added models on the map: A national overview).
  • VAMs introduce bias when missing test scores are not missing completely at random. The missing at random assumption, however, runs across most VAMs because without it, data missingness would be pragmatically insolvable, especially “given the large proportion of missing data in many achievement databases and known differences between students with complete and incomplete test data.” The really only solution here is to use “implicit imputation of values for unobserved gains using the observed scores” which is “followed by estimation of teacher effect[s] using the means of both the imputed and observe gains [together].”
  • Bias “[still] is one of the most difficult issues arising from the use of VAMs to estimate school or teacher effects…[and]…the inclusion of student level covariates is not necessarily the solution to [this] bias.” In other words, “Controlling for student-level covariates alone is not sufficient to remove the effects of [students’] background [or demographic] characteristics.” There is a reason why bias is still such a highly contested issue when it comes to VAMs (see a recent post about this here).
  • All (or now most) commonly-used VAMs assume that teachers’ (and prior teachers’) effects persist undiminished over time. This assumption “is not empirically or theoretically justified,” either, yet it persists.

These authors’ overall conclusion, again from 10 years ago but one that in many ways still stands? VAMs “will often be too imprecise to support some of [its] desired inferences” and uses including, for example, making low- and high-stakes decisions about teacher effects as produced via VAMs. “[O]btaining sufficiently precise estimates of teacher effects to support ranking [and such decisions] is likely to [forever] be a challenge.”

VAMs Are Never “Accurate, Reliable, and Valid”

The Educational Researcher (ER) journal is the highly esteemed, flagship journal of the American Educational Research Association. It may sound familiar in that what I view to be many of the best research articles published about value-added models (VAMs) were published in ER (see my full reading list on this topic here), but as more specific to this post, the recent “AERA Statement on Use of Value-Added Models (VAM) for the Evaluation of Educators and Educator Preparation Programs” was also published in this journal (see also a prior post about this position statement here).

After this position statement was published, however, many critiqued AERA and the authors of this piece for going too easy on VAMs, as well as VAM proponents and users, and for not taking a firmer stance against VAMs given the current research. The lightest of the critiques, for example, as authored by Brookings Institution affiliate Michael Hansen and University of Washington Bothell’s Dan Goldhaber was highlighted here, after which Boston College’s Dr. Henry Braun responded also here. Some even believed this response to also be too, let’s say, collegial or symbiotic.

Just this month, however, ER released a critique of this same position statement, as authored by Steven Klees, a Professor at the University of Maryland. Klees wrote, essentially, that the AERA Statement “only alludes to the principal problem with [VAMs]…misspecification.” To isolate the contributions of teachers to student learning is not only “very difficult,” but “it is impossible—even if all the technical requirements in the [AERA] Statement [see here] are met.”

Rather, Klees wrote, “[f]or proper specification of any form of regression analysis…All confounding variables must be in the equation, all must be measured correctly, and the correct functional form must be used. As the 40-year literature on input-output functions that use student test scores as the dependent variable make clear, we never even come close to meeting these conditions…[Hence, simply] adding relevant variables to the model, changing how you measure them, or using alternative functional forms will always yield significant differences in the rank ordering of teachers’…contributions.”

Therefore, Klees argues “that with any VAM process that made its data available to competent researchers, those researchers would find that reasonable alternative specifications would yield major differences in rank ordering. Misclassification is not simply a ‘significant risk’— major misclassification is rampant and inherent in the use of VAM.”
Klees concludes: “The bottom line is that regardless of technical sophistication, the use of VAM is never [and, perhaps never will be] ‘accurate, reliable, and valid’ and will never yield ‘rigorously supported inferences” as expected and desired.
***
Citation: Klees, S. J. (2016). VAMs Are Never “Accurate, Reliable, and Valid.” Educational Researcher, 45(4), 267. doi: 10.3102/0013189X16651081

The Danielson Framework: Evidence of Un/Warranted Use

The US Department of Education’s statistics, research, and evaluation arm — the Institute of Education Sciences — recently released a study (here) about the validity of the Danielson Framework for Teaching‘s observational ratings as used for 713 teachers, with some minor adaptations (see box 1 on page 1), in the second largest school district in Nevada — Washoe County School District (Reno). This district is to use these data, along with student growth ratings, to inform decisions about teachers’ tenure, retention, and pay-for-performance system, in compliance with the state’s still current teacher evaluation system. The study was authored by researchers out of the Regional Educational Laboratory (REL) West at WestEd — a nonpartisan, nonprofit research, development, and service organization.

As many of you know, principals throughout many districts throughout the US, as per the Danielson Framework, use a four-point rating scale to rate teachers on 22 teaching components meant to measure four different dimensions or “constructs” of teaching.
In this study, researchers found that principals did not discriminate as much among the individual four constructs and 22 components (i.e., the four domains were not statistically distinct from one another and the ratings of the 22 components seemed to measure the same or universal cohesive trait). Accordingly, principals did discriminate among the teachers they observed to be more generally effective and highly effective (i.e., the universal trait of overall “effectiveness”), as captured by the two highest categories on the scale. Hence, analyses support the use of the overall scale versus the sub-components or items in and of themselves. Put differently, and In the authors’ words, “the analysis does not support interpreting the four domain scores [or indicators] as measurements of distinct aspects of teaching; instead, the analysis supports using a single rating, such as the average over all [sic] components of the system to summarize teacher effectiveness” (p. 12).
In addition, principals also (still) rarely identified teachers as minimally effective or ineffective, with approximately 10% of ratings falling into these of the lowest two of the four categories on the Danielson scale. This was also true across all but one of the 22 aforementioned Danielson components (see Figures 1-4, p. 7-8); see also Figure 5, p. 9).
I emphasize the word “still” in that this negative skew — what would be an illustrated distribution of, in this case, the proportion of teachers receiving all scores, whereby the mass of the distribution would be concentrated toward the right side of the figure — is one of the main reasons we as a nation became increasingly focused on “more objective” indicators of teacher effectiveness, focused on teachers’ direct impacts on student learning and achievement via value-added measures (VAMs). Via “The Widget Effect” report (here), authors argued that it was more or less impossible to have so many teachers perform at such high levels, especially given the extent to which students in other industrialized nations were outscoring students in the US on international exams. Thereafter, US policymakers who got a hold of this report, among others, used it to make advancements towards, and research-based arguments for, “new and improved” teacher evaluation systems with key components being the “more objective” VAMs.

In addition, and as directly related to VAMs, in this study researchers also found that each rating from each of the four domains, as well as the average of all ratings, “correlated positively with student learning [gains, as derived via the Nevada Growth
Model, as based on the Student Growth Percentiles (SGP) model; for more information about the SGP model see here and here; see also p. 6 of this report here], in reading and in math, as would be expected if the ratings measured teacher effectiveness in promoting student learning” (p. i). Of course, this would only be expected if one agrees that the VAM estimate is the core indicator around which all other such indicators should revolve, but I digress…

Anyhow, researchers found that by calculating standard correlation coefficients between teachers’ growth scores and the four Danielson domain scores, that “in all but one case” [i.e., the correlation coefficient between Domain 4 and growth in reading], said correlations were positive and statistically significant. Indeed this is true, although the correlations they observed, as aligned with what is increasingly becoming a saturated finding in the literature (see similar findings about the Marzano observational framework here; see similar findings from other studies here, here, and here; see also other studies as cited by authors of this study on p. 13-14 here), is that the magnitude and practical significance of these correlations are “very weak” (e.g., r = .18) to “moderate” (e.g., r = .45, .46, and .48). See their Table 2 (p. 13) with all relevant correlation coefficients illustrated below.

Screen Shot 2016-06-02 at 11.24.09 AM

Regardless, “[w]hile th[is] study takes place in one school district, the findings may be of interest to districts and states that are using or considering using the Danielson Framework” (p. i), especially those that intend to use this particular instrument for summative and sometimes consequential purposes, in that the Framework’s factor structure does not hold up, especially if to be used for summative and consequential purposes, unless, possibly, used as a generalized discriminator. With that too, however, evidence of validity is still quite weak to support further generalized inferences and decisions.

So, those of you in states, districts, and schools, do make these findings known, especially if this framework is being used for similar purposes without such evidence in support of such.

Citation: Lash, A., Tran, L., & Huang, M. (2016). Examining the validity of ratings
from a classroom observation instrument for use in a district’s teacher evaluation system

REL 2016–135). Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center for Education Evaluation and Regional Assistance, Regional Educational Laboratory West. Retrieved from http://ies.ed.gov/ncee/edlabs/regions/west/pdf/REL_2016135.pdf