The Role of Assessment in a Learning Culture LORRIE A. SHEPARD Educational Researcher, Vol. 29, No. 7, pp. 4-14 T his article is about classroom assessment--not the kind of assessments used to give grades or to satisfy the ac- countability demands of an external authority, but rather the kind of assessment that can be used as a part of instruction to support and enhance learning. On this topic, I am especially interested in engaging the very large num- ber of educational researchers who participate, in one way or another, in teacher education. The transformation of as- sessment practices cannot be accomplished in separate tests and measurement courses, but rather should be a central concern in teaching methods courses. The article is organized in three parts. I present, first, an historical framework highlighting the key tenets of social efficiency curricula, behaviorist learning theories, and "sci- entific measurement." Next, I offer a contrasting social- constructivisbconceptual framework that blends key ideas from cognitive, constructivist, and sociocultural theories. In the third part, I elaborate on the ways that assessment prac- tices should change to be consistent with and support social- constructivist pedagogy. The impetus for m y development of an historical frame- work was the observation by Beth Graue (1993) that "as- sessment and instruction are often conceived as curiously separate in both time and purpose" (p. 291, emphasis added). As Graue notes, the measurement approach to classroom assessment, "exemplified by standardized tests and teacher-made emulations of those tests," presents a barrier to the implementation of more constructivist ap- proaches to instruction. To understand the origins of Graue's picture of separa- tion and to help explain its continuing power over present- day practice, I drew the chronology in F ~ I . A longer- term span of history helps us see that those measurement perspectives, now felt to be incompatible with instruction, came from an earlier, highly consistent theoretical frame- work (on the left) in which conceptions of "scientific mea- surement" were closely aligned with traditional curricula and beliefs about learning. To the right is an emergent, con- structivist paradigm in which teachers' close assessment of students' understandings, feedback from peers, and stu- dent self-assessments would be a central part of the social processes that mediate the development of intellectual abil- ities, construction of knowledge, and formation of students' identities. The best way to understand dissonant current practices, shown in the middle of the figure, is to realize that instruction (at least in its ideal form) is drawn from the emergent paradigm, while testing is held over from the past. Historical Perspectives: Curriculum, Psychology, and Measurement The historical framework I present here is familiar to you. Yet, it is important to remind ourselves where traditional views of testing came from and to appreciate how tightly entwined these views of testing are with past models of cur- riculum and instruction--because dominant theories of the past continue to operate as the default framework affecting and driving current practices and perspectives. Belief sys- tems of teachers, parents, and policymakers derive from these old theories. A more elaborated version of the paradigm that has pre- dominated throughout the 20th century can be shown as a set of interlocking circles ~ . The central ideas of so- cial efficiency and scientific management in the curriculum cj.rcle were closely linked, respectively, to hereditarian the- 0ties of individual differences and to associationist and be- haviorist learning theories. These psychological theories were, in turn, served by scientific measurement of ability and achievement. In the early 1900s, the social efficiency movement grew out of the belief that science could be used to solve the prob- lems of industrialization and urbanization. According to so- cial efficiency theory, modern principles of scientific man- agement, intended to maximize the efficiency of factories, could be applied with equal success to schools. This meant taking F. W. Toy.lo_L's example of a detailed analysis of the movements performed by expert bricklayers and applying similar analyses to every vocation for which students were being prepared (Kleibard, 1995). Then, given the new asso- ciationist or connectionist psychology with its emphasis on fundamental building blocks, every step would have to be taught specifically. Precise standards of measurement were required to ensure that each skill was mastered at the de- sired level. And because it was not possible to teach every student the skills of every vocation, scientific measures of ability were also needed to predict one's future role in life and thereby determine who was best suited for each en- deavor. For John Franklin Bobbitt, a leader in the social ef- ficiency movement, a primary goal of curriculum design was the elimination of waste (1912), and it was wasteful to teach people things they would never use. Bobbitt's most telling principle was that each individual should be educated LORRIE A. SHEPARD is a professor in the School of Education, Campus Box 249, University of Colorado at Boulder, Boulder, CO 80309-0249. Her area of specialization is psychometrics and the uses and misuses of tests in educational settings. This article was presented as the presidential address at the 2000 AERA Annual Meeting in New Orleans, LA. 4 EDUCATIONAL RESEARCHER

: \ I Reformed Vision i of Curriculum z ,,,- ",,, ,,,.-" .... ~..__ + /" ,,, / \ :~ d ~ / Cognitive & [ \ \ ! l I * ��� i I i [ Instruction ~ , Constructtvlst ~ ~ Classroom i ~ ~ Learning ~ ] Assessment l ',, ,, ',, T eo.es y / ��� j " ��� ��� 20th Century Dominant Paradigm ~ Dissolution of Old Paradigm: New Views of (circa 1900s - 2000+) Instruction/Old Views of Testing (circa 1980s - 2000+) Emergent Paradigm (circa 1990s - 2000+) FIG U RE 1. An historical overview illustrating how changing conceptions of curriculum, learning theory, and measurement explain the current incompatibility between new views of instruction and traditional views of testing. "according to his capabilities." These views led to a highly differentiated curriculum and a largely utilitarian one that disdained academic subjects for any but college prepara- tory students. Alongside these curriculum theories, Edward Thomdike's (1922) associationism and the behaviorism o--6-f-Hull (1943), Skinner (1938, 1954) and Gagne (1965) conceived of learn- ing as the accumulation of stimulus-response associations. The following quotation from B:.F. Skinner is illustrative: The whole process of becoming competent in any field must be divided into a very large number of very small steps, and reinforcement must be contingent upon the ac- complishment of each step. This solution to the problem of creating a complex repertoire of behavior also solves the problem of maintaining the behavior in strength . . . . By making each successive step as small as possible, the frequency of reinforcement can be raised to a maximum, while the possibly aversive consequences of being wrong are reduced to a minimum. (Skinner, 1954, p. 94) Note that this viewpoint promotes a theory of motivation as well as one of cognitive development. Several key assumptions of the behavioristic model had consequences for ensuing conceptualizations of teaching and testing: 1. Learning occurs by accumulating atomized bits of knowledge 2. Learning is tightly sequenced and hierarchical 3. Transfer is limited, so each objective must be explicitly taught 4. Tests should be used frequently to ensure mastery be- fore proceeding to the next objective 5. Tests are isomorphic with learning (tests = learning) 6. Motivation is external and based on positive rein- forcement of many small steps. It is no coincidence that Thorndike was both the origina- tor of associationist learning t'ne'6fy-~ffad the "father" of "sci- entific measurement," a name given him by Ayers in 1918. Thorndike and his students fostered the deve'6q-6-~n-bnt and dominance of the "objective" test, which has been the single most striking feature of achievement testing in the United States from the beginning of the century to the present day. Recognizing the common paternity of behaviorist learning theory and objective testing helps us to understand the con- tinued intellectual kinship between one-skill-at-a-time test items and instructional practices aimed at mastery of con- sti~uent elements. ~Looking at any collection of tests from early in the cen- tury, as shown in F,F,F~ure 3, one is immediately struck by how much the questions emphasized rote recall. To be fair, at the time, this was not a distortion of subject matter caused by the adoption of objective-item formats. One hun- dred years ago, various recall, completion, matching, and multiple-choice test types, along with some essay ques- tions, fit closely with what was deemed important to learn. However, once curriculum became encapsulated and rep- resented by these types of items, it is reasonable to say that these formats locked in a particular and outdated concep- tion of subject matter. The dominance of objective tests in classroom practice has affected more than the form of subject-matter knowl- edge. It has also shaped beliefs about the nature of evidence and principles of fairness. In a recent assessment project, for example, both teachers and researchers were surprised to find that despite our shared enthusiasm for developing al- ternatives to standardized tests we nonetheless operated from different assumptions about how "standardized" as- sessments needed to be in classrooms. More surprising still, it was teachers who held beliefs more consistent with traditional principles of scientific measurement. From the perspective of our teacher colleagues, assessment needed to be an official event, separate from instruction (Bliem & Davinroy, 1997). To ensure fairness, teachers believed that assessments had to be uniformly administered, so they were reluctant to conduct more intensive individualized assess- ments with only below-grade-level readers��� Because of the belief that assessments had to be targeted to a specific in- structional goal, teachers felt more comfortable using two separate assessments for separate goals, "running records" to assess fluency and written summaries to assess compre- hension rather than, say, asking students to retell the gist of a story in conjunction with running records. Most signifi- OCTOBER 2000 5