Difficulties When Combining Multiple Teacher Evaluation Measures

A new study about multiple “Approaches for Combining Multiple Measures of Teacher Performance,” with special attention paid to reliability, validity, and policy, was recently published in the American Educational Research Association (AERA) sponsored and highly-esteemed Educational Evaluation and Policy Analysis journal. You can find the free and full version of this study here.

In this study authors José Felipe Martínez – Associate Professor at the University of California, Los Angeles, Jonathan Schweig – at the RAND Corporation, and Pete Goldschmidt – Associate Professor at California State University, Northridge and creator of the value-added model (VAM) at legal issue in the state of New Mexico (see, for example, here), set out to help practitioners “combine multiple measures of complex [teacher evaluation] constructs into composite indicators of performance…[using]…various conjunctive, disjunctive (or complementary), and weighted (or compensatory) models” (p. 738). Multiple measures in this study include teachers’ VAM estimates, observational scores, and student survey results.

While authors ultimately suggest that “[a]ccuracy and consistency are greatest if composites are constructed to maximize reliability,” perhaps more importantly, especially for practitioners, authors note that “accuracy varies across models and cut-scores and that models with similar accuracy may yield different teacher classifications.”

This, of course, has huge implications for teacher evaluation systems as based upon multiple measures in that “accuracy” means “validity” and “valid” decisions cannot be made as based on “invalid” or “inaccurate” data that can so arbitrarily change. In other words, what this means is that likely never will a decision about a teacher being this or that actually mean this or that. In fact, this or that might be close, not so close, or entirely wrong, which is a pretty big deal when the measures combined are assumed to function otherwise. This is especially interesting, again and as stated prior, that the third author on this piece – Pete Goldschmidt – is the person consulting with the state of New Mexico. Again, this is the state that is still trying to move forward with the attachment of consequences to teachers’ multiple evaluation measures, as assumed (by the state but not the state’s consultant?) to be accurate and correct (see, for example, here).

Indeed, this is a highly inexact and imperfect social science.

Authors also found that “policy weights yield[ed] more reliable composites than optimal prediction [i.e., empirical] weights” (p. 750). In addition, “[e]mpirically derived weights may or may not align with important theoretical and policy rationales” (p. 750); hence, the authors collectively referred others to use theory and policy when combining measures, while also noting that doing so would (a) still yield overall estimates that would “change from year to year as new crops of teachers and potentially measures are incorporated” (p. 750) and (b) likely “produce divergent inferences and judgments about individual teachers (p. 751). Authors, therefore, concluded that “this in turn highlights the need for a stricter measurement validity framework guiding the development, use, and monitoring of teacher evaluation systems” (p. 751), given all of this also makes the social science arbitrary, which is also a legal issue in and of itself, as also quasi noted.

Now, while I will admit that those who are (perhaps unwisely) devoted to the (in many ways forced) combining of these measures (despite what low reliability indicators already mean for validity, as unaddressed in this piece) might find some value in this piece (e.g., how conjunctive and disjunctive models vary, how principal component, unit weight, policy weight, optimal prediction approaches vary), I will also note that forcing the fit of such multiple measures in such ways, especially without a thorough background in and understanding of reliability and validity and what reliability means for validity (i.e., with rather high levels of reliability required before any valid inferences and especially high-stakes decisions can be made) is certainly unwise.

If high-stakes decisions are not to be attached, such nettlesome (but still necessary) educational measurement issues are of less importance. But any positive (e.g., merit pay) or negative (e.g., performance improvement plan) consequence that comes about without adequate reliability and validity should certainly cause pause, if not a justifiable grievance as based on the evidence provided herein, called for herein, and required pretty much every time such a decision is to be made (and before it is made).

Citation: Martinez, J. F., Schweig, J., & Goldschmidt, P. (2016). Approaches for combining multiple measures of teacher performance: Reliability, validity, and implications for evaluation policy. Educational Evaluation and Policy Analysis, 38(4), 738–756. doi: 10.3102/0162373716666166 Retrieved from http://journals.sagepub.com/doi/pdf/10.3102/0162373716666166

Note: New Mexico’s data were not used for analytical purposes in this study, unless any districts in New Mexico participated in the Bill & Melinda Gates Foundation’s Measures of Effective Teaching (MET) study yielding the data used for analytical purposes herein.

NCTQ on States’ Teacher Evaluation Systems’ Failures

The controversial National Council on Teacher Quality (NCTQ) — created by the conservative Thomas B. Fordham Institute and funded (in part) by the Bill & Melinda Gates Foundation as “part of a coalition for ‘a better orchestrated agenda’ for accountability, choice, and using test scores to drive the evaluation of teachers” (see here; see also other instances of controversy here and here) — recently issued yet another report about state’s teacher evaluation systems titled: “Running in Place: How New Teacher Evaluations Fail to Live Up to Promises.” See a related blog post in Education Week about this report here. See also a related blog post about NCTQ’s prior large-scale (and also slanted) study — “State of the States 2015: Evaluating Teaching, Leading and Learning” — here. Like I did in that post, I summarize this study below.

From the abstract: Authors of this report find that “within the 30 states that [still] require student learning measures to be at least a significant factor in teacher evaluations, state guidance and rules in most states allow teachers to be rated effective even if they receive low scores on the student learning component of the evaluation.” They add in the full report that in many states “a high score on an evaluation’s observation and [other] non-student growth components [can] result in a teacher earning near or at the minimum number of points needed to earn an effective rating. As a result, a low score on the student growth component of the evaluation is sufficient in several states to push a teacher over the minimum number of points needed to earn a summative effective rating. This essentially diminishes any real influence the student growth component has on the summative evaluation rating” (p. 3-4).

The first assumption surrounding the authors’ main tenets they make explicit: that “[u]nfortunately, [the] policy transformation [that began with the publication of the “Widget Effect” report in 2009] has not resulted in drastic alterations in outcomes” (p. 2). This is because, “[in] effect…states have been running in place” (p. 2) and not using teachers’ primarily test-based indicators for high-stakes decision-making. Hence, “evaluation results continue to look much like they did…back in 2009” (p. 2). The authors then, albeit ahistorically, ask, “How could so much effort to change state laws result in so little actual change?” (p. 2). Yet they don’t realize (or care to realize) that this is because we have almost 40 years of evidence that really any type of test-based, educational accountability policies and initiatives have never yield their intended consequences (i.e., increased student achievement on national and international indicators). Rather, the authors argue, that “most states’ evaluation laws fated these systems to status quo results long before” they really had a chance (p. 2).

The authors’ second assumption they imply: that the two most often used teacher evaluation indicators (i.e., the growth or value-added and observational measures) should be highly correlated, which many argue they should be IF in fact they are measuring general teacher effectiveness. But the more fundamental assumption here is that if the student learning (i.e., test based) indicators do not correlate with the observational indicators, the latter MUST be wrong, biased, distorted, and accordingly less trustworthy and the like. They add that “teachers and students are not well served when a teacher is rated effective or higher even though her [sic] students have not made sufficient gains in their learning over the course of a school year” (p. 4). Accordingly, they add that “evaluations should require that a teacher is rated well on both the student growth measures and the professional practice component (e.g., observations, student surveys, etc.) in order to be rated effective” (p. 4). Hence, also in this report the authors put forth recommendations for how states might address this challenge. See these recommendations forthcoming, as also related to a new phenomenon my students and I are studying called artificial inflation.

Artificial inflation is a term I recently coined to represent what is/was happening in Houston, and elsewhere (e.g., Tennessee), when district leaders (e.g., superintendents) mandate or force principals and other teacher effectiveness appraisers or evaluators to align their observational ratings of teachers’ effectiveness with teachers’ value-added scores, with the latter being (sometimes relentlessly) considered the “objective measure” around which all other measures (e.g., subjective observational measures) should revolve, or align. Hence, the push is to conflate the latter “subjective” measure to match the former “objective” measure, even if the process of artificial conflation causes both indicators to become invalid. As per my affidavit from the still ongoing lawsuit in Houston (see here), “[t]o purposefully and systematically endorse the engineering and distortion of the perceptible ‘subjective’ indicator, using the perceptibly ‘objective’ indicator as a keystone of truth and consequence, is more than arbitrary, capricious, and remiss…not to mention in violation of the educational measurement field’s “Standards for Educational and Psychological Testing.”

Nonetheless…

Here is one important figure, taken out of context in some ways on purpose (e.g., as the text surrounding this particular figure is ironically, subjectively used to define what the NCTQ defines as as indicators or progress, or regress).

Near Figure 1 (p. 1) the authors note that “as of January 2017, there has been little evidence of a large-scale reversal of states’ formal evaluation policies. In fact, only four states (Alaska, Mississippi, North Carolina, and Oklahoma) have reversed course on factoring student learning into a teacher’s evaluation rating” (p. 3). While this reversal of four is not illustrated in their accompanying figure, see also a prior post about what other states, beyond just these four states of dishonorable mention, have done to “reverse” the “course” (p. 3) here. While the authors shame all states for minimizing teachers’ test-based ratings before these systems had a chance, as also ignorant to what they cite as “a robust body of research” (without references or citations here, and few elsewhere in a set of footnotes), they add that it remains an unknown as to “why state educational agencies put forth regulations or guidance that would allow teachers to be rated effective without meeting their student growth goals” (p. 4). Many of us know that this was often done to counter the unreliable and invalid results often yielded via the “objective” test-based sides of things that the NCTQ continues to advance.

Otherwise, here are also some important descriptive findings:

  • Thirty states require measures of student academic growth to be at least a significant factor within teacher evaluations; another 10 states require some student growth, and 11 states do not require any objective measures of student growth (p. 5).
  • With only [emphasis added] two exceptions, in the 30 states where student
    growth is at least a significant factor in teacher evaluations, state
    rules or guidance effectively allow teachers who have not met student
    growth goals to still receive a summative rating of at least effective (p. 5).
  • In 18 [of these 30] states, state educational agency regulations and/or guidance
    explicitly permit teachers to earn a summative rating of effective even after earning a less-than-effective score on the student learning portion of their evaluations…these regulations meet the letter of the law while still allowing teachers with low ratings on
    student growth measures to be rated effective or higher (p. 5). In Colorado, for example…a teacher can earn a rating of highly effective with a score of just 1 for student growth (which the state classifies as “less than expected”) in conjunction with a top professional practice score (p. 4).
  • Ten states do not specifically address whether a teacher who has not met student growth goals may be rated as effective or higher. These states neither specifically allow nor specifically disallow such a scenario, but by failing to provide guidance to prevent such an occurrence, they enable it to exist (p. 6).
  • Only two of the 30 states (Indiana and Kentucky) make it impossible for a teacher who has not been found effective at increasing student learning to receive a summative rating of effective (p. 6).

Finally, here are some of their important recommendations, as related to all of the above, and to create more meaningful teacher evaluation systems. So they argue, states should:

  • Establish policies that preclude teachers from earning a label of effective if they are found ineffective at increasing student learning (p. 12).
  • Track the results of discrete components within evaluation systems, both statewide and districtwide. In districts where student growth measures and observation measures are significantly out of alignment, states should reevaluate their systems and/or offer districts technical assistance (p. 12). ][That is, states should possibly promote artificial inflation as we have observed elsewhere. The authors add that] to ensure that evaluation ratings better reflect teacher performance, states should [more specifically] track the results of each evaluation measure to pinpoint where misalignment between components, such as between student learning and observation measures, exists. Where major components within an evaluation system are significantly misaligned, states should examine their systems and offer districts technical assistance where needed, whether through observation training or examining student growth models or calculations (p. 12-13). [Tennessee, for example,] publishes this information so that it is transparent and publicly available to guide actions by key stakeholders and point the way to needed reforms (p. 13).

See also state-by-state reports in the appendices of the full report, in case your state was one of the state’s that responded or, rather, “recognized the factual accuracy of this analysis.”

Citation: Walsh, K., Joseph, N., Lakis, K., & Lubell, S. (2017). Running in place: How new teacher evaluations fail to live up to promises. Washington DC: National Council on Teacher Quality (NCTQ). Retrieved from http://www.nctq.org/dmsView/Final_Evaluation_Paper

Last Saturday Night Live’s VAM-Related Skit

For those of you who may have missed it last Saturday, Melissa McCarthy portrayed Sean Spicer — President Trump’s new White House Press Secretary and Communications Director — in one of the funniest of a very funny set of skits recently released on Saturday Night Live. You can watch the full video, compliments of YouTube, here:

In one of the sections of the skit, though, “Spicer” introduces “Betsy DeVos” — portrayed by Kate McKinnon and also just today confirmed as President Trump’s Secretary of Education — to answer some very simple questions about today’s public schools which she, well, very simply could not answer. See this section of the clip starting at about 6:00 (of the above 8:00 minute total skit).

In short, “the man” reporter asks “DeVos” how she values “growth versus proficiency in [sic] measuring progress in students.” Literally at a loss of words, “DeVos” responds that she really doesn’t “know anything about school.” She rambles on, until “Spicer” pushes her off of the stage 40-or-so seconds later.

Humor set aside, this was the one question Saturday Night Live writers wrote into this skit, which reminds us that what we know more generally as the purpose of VAMs is still alive and well in our educational rhetoric as well as popular culture. As background, this question apparently came from Minnesota Sen. Al Franken’s prior, albeit similar question during DeVos’s confirmation hearing.

Notwithstanding, Steve Snyder – the editorial director of The 74 — an (allegedly) non-partisan, honest, and fact-based backed by Editor-in-Chief Campbell Brown (see prior posts about this news site here and here) — took the opportunity to write a “featured” piece about this section of the script (see here). The purpose of the piece was, as the title illustrates, to help us “understand” the skit, as well as it’s important meaning for all of “us.”

Snyder notes that Saturday Night Live writers, with their humor, might have consequently (and perhaps mistakenly) “made their viewers just a little more knowledgeable about how their child’s school works,” or rather should work, as “[g]rowth vs. proficiency is a key concept in the world of education research.” Thereafter, Snyder falsely asserts that more than 2/3rds of educational researchers agree that VAMs are a good way to measure school quality. If you visit the actual statistic cited in this piece, however, as “non-partison, honest, and fact-based” that it is supposed to be, you would find (here) that this 2/3rds consists of 57% of responding American Education Finance Association (AEFA) members, and AEFA members alone, who are certainly not representative of “educational researchers” as claimed.

Regardless, Snyder asks: “Why are researchers…so in favor of [these] growth measures?” Because this disciplinary subset does not represent educational researchers writ large, but only a subset, Snyder.

As it is with politics today, many educational researchers who define themselves as aligned with the disciplines of educational finance or educational econometricians are substantively more in favor of VAMs than those who align more with the more general disciplines of educational research and educational measurement, methods, and statistics, in general. While this is somewhat of a sweeping generalization, which is not wise as I also argue and also acknowledge in this piece, there is certainly more to be said here about the validity of the inferences drawn here, and (too) often driven via the “media” like The 74.

The bottom line is to question and critically consume everything, and everyone who feels qualified to write about particular things without enough expertise in most everything, including in this case good and professional journalism, this area of educational research, and what it means to make valid inferences and then responsibly share them out with the public.