The performance of students from different racial or ethnic subgroups and of students receiving bilIngual (Spanish and English) or monolingual (English only) instruction in mathematics was studied using students from schools in the QUASAR (Qualitative Und-rstanding Amplifying Student Achievement and Reasoning) project, a mathematics education reform project supporting innovative instruction for middle school students in economically disadvantaged communities. Patterns of ethnic distribution vary across QUASAR cites, and linguistic diversity was found at many QUASAR sites. Data were from administration of the QUASAR Cognitive Assessment Instrument (QCAI) to approximately 1,000 students in grades 6, 7, and 8 at 2 QUASAR project schools, one of which provided bilingual classes for its predominantly Latino population. Results indicate that the instructional programs of both schools provided similar educational opportunities in mathematics, with parallel gains for African American and Caucasian students. Results also indicate that high quality mathematics instruction can be made available to students in bilingual classes. Evidence also supports the validity of the QCAI. (Contains 5 figures and 40 references.) (SLD) Reproductions supplied by EDRS are the best that can be made from the ori,-inal document. ***** U S DEPARTMENT OF EDUCATION Office ot Educalfonal Reseatce and Improvement rs EDUCATIONAL RESOURCES INFORMATION CENTER (ERIC) dys document has peen ,eproduced as ,ece.yea trorn Ine person Of 0,0111.fiaffOn of fgfnalong ft C\ ncnot changes have been reade to treprove Cr) eOroductton ouattly Potnts 01 vtew or optotons Mated to trosdoto ment do nOt neCessanly represent off tctat Utvl OE RI postbon potcy Ethnic And Linguistic Subgroup Performance Gains I PERMISSION TO REPRODUCE THIS MATERIAL HAS BEEN GRANTED BY 5oznA)4 (JciAi&TO THE EDUCATIONAL RESOURCES INFORMATION CENTER IERICI An Examination of the Perfo:.Liance Gains of Culturally and Linguistically Diverse Students on a Mathematics Performance Assessment Within the QUASAR Project Suzanne Lane, Edward A. Silver, and Ning Wang University of Pittsburgh Running Head: ETHNIC AND LINGUISTIC SUBGROUP PERFORMANCE GAINS Preparation of this paper was supported by a grant from the Ford Foundation (grant number 890-0572) for the QUASAR project. Any opinions expressed herein are those of the authors and do not necessarily represent the views of the Ford Foundation. This paper was presented in the symposium, “Examining Equity Issues Embedded Within Mathematics Instructional Reform: Findings From the QUASAR Project” at the annual meeting of the American Educational Research Association , San Francisco. CA, April 1995. DRAFT March 1995 DRAFT Ethnic And Linguistic Subgroup Performance Gains 2 An Examination of the Performance Gains of Culturally and Linguistically Diverse Students on a Mathematics Performance Assessment Within the QUASAR Project Currently, considerable effort is being devoted to the reform of precollege education in many academic subject areas. In the area of mathematics, for example, reports from the National Academy of Sciences (National Research Council, 1989) and the National Council of Teachers of Mathematics (1989, 1991), have captured the attention of many educational practitioners and policy makers. The reports specify for mathematics education a set of goals and principles, usually referred to as standards, and they provide descriptions of desired mathematical proficiency, with respect to reasoning, problem solving, communication, and conceptual understanding. Moreover, the reports also indicate the expectation that mathematical proficiency should and can be attained by all students (Silver, 1994). Despite the current optimism that all students can learn mathematics, it is the case that a pattern of achievement differences in mathematics and many other areas has been found for students in racial or ethnic minority subgroups (O’Connor, 1989). In his review of findings related to mathematics achievement by racial and ethnic subgroups on the National Assessment of Educational Progress (NAEP), Secada (1992) concluded: The general picture of racial and/or ethnic disparities in mathematics achievement that comes from the NAEP data is that Whites perform much better in mathematics than do Hispanics who, in turn, achieve slightly better than do African Americans. (p. 628) Although the size of the performance gap between racial or ethnic subgroups narrowed during the 1970s and 1980s (Jones, !984), gains appear to have been due primarily to increased performance on the parts of the assessment that measure bp..ic knowledge and skills; far less change has been detected for tasks assessing the more complex forms of knowledge and proficiency that are emphasized in the current reform. DRAFT March 1995 3 DRAFT Ethnic And Linguistic Subgroup Performance Gains 3 Moreover, a substantiai gap still remains in the average mathematics performance on NAEP by members of racial or ethnic subgroups (Mullis, Dossey, Cambell, Gentile, O’Sullivan, & Latham, 1994). And Secada (1992) also notes several studies that have found a fairly direct relationship between language proficiency and performance on mathematics achievement tests. For example, De Avila (1988) found significant correlations between English language proficiency and CTBS mathematics achievement test scores for fourth, fifth, and sixth grade students. In addition to findings that indicate a mathematics achievement gap between racial or ethnic subgroups and between groups differing in language proficiency, other findings suggest that the gap often increases as students progress through school. As Secada (1992) notes in his summary of research in this area, “.. achievement disparities, which are great to begin with, increase over time as students grow older” (p. 628). Support for this claim of a widening gap over time and grade level is supported not only by the NAEP data at grades 4, 8 and 12, but also by some other studies using cross-sectional analyses. For example, Gross (1988) reported that the achievement gap between White or Asian students and non-White or non-Asian students attending elementary schcols in a large . suburban district were small in grade I but progressively larger for each succeeding grade level. In general, the findings related to achievement gaps in mathematics and other academic subjects have been obtained through the use of tests utilizing multiple-choice item formats. Some proponents of educational reform have argued that the use of alternate forms of assessment, usually referred to as “authentic” or performance assessments, could yield different patterns of results. This argument is often tied to a notion of assessment-driven instruction. According to this view, the use of performance assessments aimed at more complex types of knowledge and proficiency is likely to lead teachers to alter their instruction, so as to promote good performance by their students (e.g., Resnick & Resnick, 1992). Increased instructional attention to higher-levi I DRAFT March 1995 DRAFT Ethnic And Linguistic Subgroup Performance Gains 4 cognitive goals would then equalize opportunity to perform well, thereby leading to a narrowing of any existing achievement gap. Another reason for optimism that the use of performance assessments will lead to narrowing of the achievement gap is that these tasks could be more accessible to diverse populations of students. In contrast to decontextualized, multiple-choice test items, performance assessment tasks can allow for diverse approaches and solutions, thereby “tapping a wide range of talents, a variety of life experiences, and multiple ways of knowing” (Darling-Hammond, 1995, p. 99).1 Despite the opiimism of advocates of performance assessments that differences in performance on these tasks among ethnic, racial, and linguistic subgroups would be narrower than those observed on multiple-choice tests, early evidence suggests that the performance differences are about the same regardless of item type (Baker, O’Neil, & Linn, 1991; Dunbar, 1987; Dunbar, Koretz, & Hoover, 1991; Feinberg, 1990; Linn, Baker, & Dunbar, 1991). For example, Linn et al. (1991) indicated that score differences for African-American and Caucasian students on written essays on the NAEP were about the same size as those found on multiple-choice reading tests. Of course, these results are not surprising, since the existence of performance differences is considered by many to be largely a consequence of unequal access to quality curriculum and instruction (Barr & Dreeben, 1983; College Entrance Examination Board, 1985; Darling-Hammond & Snyder, 1991; Oakes, 1990). According to this view, it cannot be expected that the form of assessment will have a major impact on the quality of performance unless the quality of instruction improves for minority students, since the persistent pattern of performance 1 It is worth noting that not everyone agrees with these arguments. For example, several commentators have pointed to weaknesses in the logical necessity and/or practical feasibility of the argument for assessment-driven instructional reform (e.g., Silver, 1992; Winfield, 1995). Also, because evaluation of performance assessments typically involves subjective judgment. several commentators have pointed to their susceptibility to scoring bias that could disadvantage students from diverse backgrounds (e.g.. O’Connor, 1989; Winfield, 1995). Moreover, it can be argued that performance assessments, which tend to involve more use of language than do multiple-choice test items are likely to be especially problematic toi students whose native language is not the one used in the assessment. For example, in their review of research on mathematics learning by bilingual learners, Wong Fillmore and Valadez (1986) discuss the special challenges that tasks given in verbal formats present to non-native speakers of English when they a