Also Last Thursday in Nevada: The “Top Ten” Research-Based Reasons Why Large-Scale, Standardized Tests Should Not Be Used to Evaluate Teachers

ShareTweet about this on TwitterShare on Facebook72Email this to someoneShare on Google+1Share on LinkedIn2Share on Reddit0

Last Thursday was a BIG day in terms of value-added models (VAMs). For those of you who missed it, US Magistrate Judge Smith ruled — in Houston Federation of Teachers (HFT) et al. v. Houston Independent School District (HISD) — that Houston teacher plaintiffs’ have legitimate claims regarding how their EVAAS value-added estimates, as used (and abused) in HISD, was a violation of their Fourteenth Amendment due process protections (i.e., no state or in this case organization shall deprive any person of life, liberty, or property, without due process). See post here: “A Big Victory in Court in Houston.” On the same day, “we” won another court case — Texas State Teachers Association v. Texas Education Agency —  on which The Honorable Lora J. Livingston ruled that the state was to remove all student growth requirements from all state-level teacher evaluation systems. In other words, and in the name of increased local control, teachers throughout Texas will no longer be required to be evaluated using their students’ test scores. See prior post here: “Another Big Victory in Court in Texas.”

Also last Thursday (it was a BIG day, like I said), I testified, again, regarding a similar provision (hopefully) being passed in the state of Nevada. As per a prior post here, Nevada’s “Democratic lawmakers are trying to eliminate — or at least reduce — the role [students’] standardized tests play in evaluations of teachers, saying educators are being unfairly judged on factors outside of their control.” More specifically, as per AB320 the state would eliminate statewide, standardized test results as a mandated teacher evaluation measure but allow local assessments to account for 20% of a teacher’s total evaluation. AB320 is still in work session. It has the votes in committee and on the floor, thus far.

The National Center on Teacher Quality (NCTQ), unsurprisingly (see here and here), submitted (unsurprising) testimony against AB320 that can be read here, and I submitted testimony (I think, quite effectively 😉 ) refuting their “research-based” testimony, and also making explicit what I termed “The “Top Ten” Research-Based Reasons Why Large-Scale, Standardized Tests Should Not Be Used to Evaluate Teachers” here. I have also pasted my submission below, in case anybody wants to forward/share any of my main points with others, especially others in similar positions looking to impact state or local educational policies in similar ways.

*****

May 4, 2017

Dear Assemblywoman Miller:

Re: The “Top Ten” Research-Based Reasons Why Large-Scale, Standardized Tests Should Not Be Used to Evaluate Teachers

While I understand that the National Council on Teacher Quality (NCTQ) submitted a letter expressing their opposition against Assembly Bill (AB) 320, it should be officially noted that, counter to that which the NCTQ wrote into its “research-based” letter,[1] the American Statistical Association (ASA), the American Educational Research Association (AERA), the National Academy of Education (NAE), and other large-scale, highly esteemed, professional educational and educational research/measurement associations disagree with the assertions the NCTQ put forth. Indeed, the NCTQ is not a nonpartisan research and policy organization as claimed, but one of only a small handful of partisan operations still in existence and still pushing forward what is increasingly becoming dismissed as America’s ideal teacher evaluation systems (e.g., announced today, Texas dropped their policy requirement that standardized test scores be used to evaluate teachers; Connecticut moved in the same policy direction last month).

Accordingly, these aforementioned and highly esteemed organizations have all released statements cautioning all against the use of students’ large-scale, state-level standardized tests to evaluate teachers, primarily, for the following research-based reasons, that I have limited to ten for obvious purposes:

  1. The ASA evidenced that teacher effects correlate with only 1-14% of the variance in their students’ large-scale standardized test scores. This means that the other 86%-99% of the variance is due to factors outside of any teacher’s control (e.g., out-of-school and student-level variables). That teachers’ effects, as measured by large-scaled standardized tests (and not including other teacher effects that cannot be measured using large-scaled standardized tests), account for such little variance makes using them to evaluate teachers wholly irrational and unreasonable.
  1. Large-scale standardized tests have always been, and continue to be, developed to assess levels of student achievement, but not levels of growth in achievement over time, and definitely not growth in achievement that can be attributed back to a teacher (i.e., in terms of his/her effects). Put differently, these tests were never designed to estimate teachers’ effects; hence, using them in this regard is also psychometrically invalid and indefensible.
  1. Large-scale standardized tests, when used to evaluate teachers, often yield unreliable or inconsistent results. Teachers who should be (more or less) consistently effective are, accordingly, being classified in sometimes highly inconsistent ways year-to-year. As per the current research, a teacher evaluated using large-scale standardized test scores as effective one year has a 25% to 65% chance of being classified as ineffective the following year(s), and vice versa. This makes the probability of a teacher being identified as effective, as based on students’ large-scale test scores, no different than the flip of a coin (i.e., random).
  1. The estimates derived via teachers’ students’ large-scale standardized test scores are also invalid. Very limited evidence exists to support that teachers whose students’ yield high- large-scale standardized tests scores are also effective using at least one other correlated criterion (e.g., teacher observational scores, student satisfaction survey data), and vice versa. That these “multiple measures” don’t map onto each other, also given the error prevalent in all of the “multiple measures” being used, decreases the degree to which all measures, students’ test scores included, can yield valid inferences about teachers’ effects.
  1. Large-scale standardized tests are often biased when used to measure teachers’ purported effects over time. More specifically, test-based estimates for teachers who teach inordinate proportions of English Language Learners (ELLs), special education students, students who receive free or reduced lunches, students retained in grade, and gifted students are often evaluated not as per their true effects but group effects that bias their estimates upwards or downwards given these mediating factors. The same thing holds true with teachers who teach English/language arts versus mathematics, in that mathematics teachers typically yield more positive test-based effects (which defies logic and commonsense).
  1. Related, large-scale standardized tests estimates are fraught with measurement errors that negate their usefulness. These errors are caused by inordinate amounts of inaccurate and missing data that cannot be replaced or disregarded; student variables that cannot be statistically “controlled for;” current and prior teachers’ effects on the same tests that also prevent their use for making determinations about single teachers’ effects; and the like.
  1. Using large-scale standardized tests to evaluate teachers is unfair. Issues of fairness arise when these test-based indicators impact some teachers more than others, sometimes in consequential ways. Typically, as is true across the nation, only teachers of mathematics and English/language arts in certain grade levels (e.g., grades 3-8 and once in high school) can be measured or held accountable using students’ large-scale test scores. Across the nation, this leaves approximately 60-70% of teachers as test-based ineligible.
  1. Large-scale standardized test-based estimates are typically of very little formative or instructional value. Related, no research to date evidences that using tests for said purposes has improved teachers’ instruction or student achievement as a result. As per UCLA Professor Emeritus James Popham: The farther the test moves away from the classroom level (e.g., a test developed and used at the state level) the worst the test gets in terms of its instructional value and its potential to help promote change within teachers’ classrooms.
  1. Large-scale standardized test scores are being used inappropriately to make consequential decisions, although they do not have the reliability, validity, fairness, etc. to satisfy that for which they are increasingly being used, especially at the teacher-level. This is becoming increasingly recognized by US court systems as well (e.g., in New York and New Mexico).
  1. The unintended consequences of such test score use for teacher evaluation purposes are continuously going unrecognized (e.g., by states that pass such policies, and that states should acknowledge in advance of adapting such policies), given research has evidenced, for example, that teachers are choosing not to teach certain types of students whom they deem as the most likely to hinder their potentials positive effects. Principals are also stacking teachers’ classes to make sure certain teachers are more likely to demonstrate positive effects, or vice versa, to protect or penalize certain teachers, respectively. Teachers are leaving/refusing assignments to grades in which test-based estimates matter most, and some are leaving teaching altogether out of discontent or in professional protest.

[1] Note that the two studies the NCTQ used to substantiate their “research-based” letter would not support the claims included. For example, their statement that “According to the best-available research, teacher evaluation systems that assign between 33 and 50 percent of the available weight to student growth ‘achieve more consistency, avoid the risk of encouraging too narrow a focus on any one aspect of teaching, and can support a broader range of learning objectives than measured by a single test’ is false. First, the actual “best-available” research comes from over 10 years of peer-reviewed publications on this topic, including over 500 peer-reviewed articles. Second, what the authors of the Measures of Effective Teaching (MET) Studies found was that the percentages to be assigned to student test scores were arbitrary at best, because their attempts to empirically determine such a percentage failed. This face the authors also made explicit in their report; that is, they also noted that the percentages they suggested were not empirically supported.

ShareTweet about this on TwitterShare on Facebook72Email this to someoneShare on Google+1Share on LinkedIn2Share on Reddit0

5 thoughts on “Also Last Thursday in Nevada: The “Top Ten” Research-Based Reasons Why Large-Scale, Standardized Tests Should Not Be Used to Evaluate Teachers

  1. Thank you again for your efforts. If this testing nonsense is not stopped I will retire next year after 22 years at the same poor title 1 school. I will miss the kids. This week we had our tests, the SBAC. Half of my students just hit submit without answering the questions. They are not just data points and their scores are not indicative of their abilities. As a Nevada teacher I thank you. If AB 320 passes I fear our governor will veto it. None the less, you have helped put momentum on our side, the stage is set to go to court.

  2. If the pending legislation is signed by the governor, Florida will discontinue use of their invalid standardized test scores as part of teacher evaluations.

  3. I join in thanking you and especially for calling out the MET study with deeply flawed data that keeps getting recycled in the peer reviewed Educational Researcher. As an oldtimer working on behalf of arts education, I am also amazed at the overall distortion of educational research caused by the ease of access to data from tests of math and reading, at the expense of attention to so much else.

  4. The title of the article would have been even more accurate if it had ended at
    The “Top Ten” Research-Based Reasons Why Large-Scale, Standardized Tests Should Not Be Used!
    Except for the most surface level acquisition of facts, learning can not be measured, although it certainly can be assessed. Why do people believe that assigning a number to something makes it “more true”–even when that number is meaningless?

Leave a Reply

Your email address will not be published. Required fields are marked *