ساخت و اعتباربخشی انگاره نابرابری آموزشی در آزمون‌های سرنوشت‌ساز: مطالعه موردی آزمون‌های بسندگی زبان دکتری

Different theoretical approaches have envisaged different goals for education. Educational equality and equity are among these objectives that are long discussed in many dimensions of education including need analysis, curriculum development, syllabus design and teaching. However, the issue of equity and equality in educational evaluation and assessment have been marginal issues in validity of test development, usage and interpretation. Language education has not been an exception in this regard. Review of literature reveals that inequity and inequality in language tests have brought about diverse educational and non-education effect in educational and private lives of language learners. Therefore, the current research was an early and innovative attempt to investigate this issue in high-stake tests in Iranian context and develop and validate a model of educational inequality in proficiency tests obligated for graduation of PhD candidates in different disciplines of humanities. To this aim a multi-phase/multi-method design was selected which entailed three different phases of structural equation modeling, structural interpretation, and scenario-development. Structural equation modeling was utilized to extract factors and variables involved in the construct of language testing inequality in the country while structural interpretation was employed to shed light on the nature and behavior of the identified and validated variables and the way they can be manipulated for achieving best educational outcomes. Also, scenario development was utilized for developing the most compatible education scenario for removing the language testing inequality in the context of the research. The results of structural equation modeling revealed that through documentary research, systematic literature review and semi-structured interviews with 7 English language professor (panel of experts) selected from different university of the country through convenient sampling, 33 early variables were identified in the construct of test inequality in the tentative theoretical model. To validate this early tentative model the identified variables were rendered into a Likert-type questionnaire and administered to an adequate number of participants in different occasion for pilot study, reliability estimation, construct validation through exploratory factor analysis and construct fit indices in confirmatory factor analysis. The results of Cronbach alpha estimation revealed that the early construct is acceptable. Also the results of exploratory factor analysis showed that, based on Eigen value, four factors are involved in the construct of the tests inequality including inequality resulted from construct invalidity (containing seven variables emphasizing issues of cut-score, adoption of a product view toward language testing, inattentiveness to language productive skills, role of chance, test-retest unreliability, lack of concurrent validity, and inconsistency between obtained score and language proficiency), inequality resulted from content invalidity(containing four variables emphasizing inconsistency between test content and education content, inattentiveness to test takers’ needs and interests, inconsistency between test content and test context, and inconsistency between test content and educational background of PhD candidates), inequality resulted from consequential invalidity (containing four variables emphasizing learners anxiety, burdened economical expenses, negative effect on self-perception and engendering negative washback), and inequality resulted from test implication and use (containing seven variables emphasizing powerful stakes of tests for decision making purposes, inattentiveness to critical issues, bias toward some disciplines against others, and lack of equal access to tests’ materials, sources and news). Besides, the results of structural interpretation implemented in crossed-impacts matrix revealed that the educational system of tests’ inequality in Iranian context is a stable system and its modification is a cumbersome practice for educational managers for most of the variables were distributed in critical, environmental, exceptional and excluded, and output variables with a scant number of them scatted around areas of bi-dimensional variables the matrix of direct impacts. Finally, to establish a relationship between findings of the research and real practice of educational management in the country, a third phase was carried out using matrix of balanced-effects in scenario-wizard. It was revealed that the most compatible and ideal scenario with Iranian educational management context - for removing inequality in the aforesaid high-stake tests - entails three main favorable axes including educational goal-setting (Weight 3), need analysis (Weight 3) and glocalization (Weight 2). The key factors associated with these axes in the corrective strategy obtained are “matching proficiency tests’ goals with educational goals in PhD course”, “conducting a real analysis of learners educational, professional and consequent needs”, and “observing global standards, theoretical evidences and local settings while developing tests” respectively. These findings were consistent with a bulk of studies conducted on test validity, test consequences, educational deficiencies at local and global scales. These findings have epistemological, methodological and practical implications. From an epistemological perspectives it developed a model of testing inequality in Iranian context that is supported by global standards and theories and at the same time meets the local needs of a great number of Iranian language learners outside language education discipline. From the methodological viewpoint, the study has implications for researchers to not to terminate their studies in structural equation modeling and blend it with more practical insights of structural interpretation and strategy development to fill the gap between research and practice. From the pragmatic viewpoint, the strategy developed based on the validated data fed into MicMac and Scenario Wizard has real pragmatic implications for education managers. This study had a number of limitations in selection of participants especially for consequential factor analysis. Future researchers are suggested to include more participants in different phases of the study.

Evidence suggests that commercialized western and non-indigenous models are fundamentally different from our educational system. This difference is important in the macrostructures of education such as educational policies because it changes the direction of political, economic and identity interests of the Iranian-Islamic society and also causes the non-native worldview and ideological system to be the uppermost document in all educational designs. Use education and training. The prescriptive model for educational justice is no exception to this rule and does not respond to our local needs. Based on this, it is necessary to build and validate an appropriate model of knowledge in educational testing in the country. Apart from the basis used in education and apart from the approaches to educational equality, the inclusive person as a human being is central. Paying attention to biological and educational needs and inclusive interests in the test-making stage is also an emphasized principle that seeks the support of both moral and educational schools. In some moral-educational schools, the main goal of education is to develop the learner in psycho-motor dimensions and systems and bring it to the entrance of science (self-sufficiency), which should also be taken into account in the preparation of tests.