Interpreter certification has been largely ignored by the literature in Interpreting Studies. What little research that does exist on the topic is understandably elementary, discussing concerns that are general in nature or describing the development of individual certification instruments. No research has as yet examined the psychometric properties of certification instruments, despite the urgings of legal precedent, the weight of professional opinion, and the requirements of professional standards. The present study has attempted to bridge this gap in two ways. First, to gain some sense of the current state of affairs in interpreter testing, a validation study of an existing certification test was conducted. Particular attention was paid to information about the test's constructs. Second, in a bid to improve construct validity, a new certification test was constructed and subsequently validated. Constructs in the new test were based on a discursive model of the interpreting process. Validation of the two tests revealed some interesting findings. Constructs in the existing test bore only a weak relationship to the interpreting profession. Also, the constructs did not prove to be useful in predicting relationships among the existing test scores. It was expected that three separate constructs would emerge from the data; instead, empirical evidence suggested that the data were unidimensional in nature. In contrast, constructs in the new test were strongly linked to the interpreting profession. They also proved to effectively predict relationships among the new test scores. It was expected that three separate constructs would emerge from the data, and empirical evidence suggested that this was, in fact, the case. The results of the present study indicate that there is room for improvement in the current state of interpreter testing, and that a test instrument based on discursive constructs is likely to demonstrate acceptable levels of validity. There is obviously a need for a great deal of future research on the psychometric aspects of interpreter certification, and it is strongly recommended that this research adopt a discursive approach.

While most large-scale assessments are designed for one intended use, in many instances, a single administration of an assessment is used for one or more uses in addition to the intended use. I suggest the term multiple-use to refer to this practice. The purpose of this dissertation is to consider the implications of the multiple-use of large-scale assessments for the process of validation. I begin by discussing multiple-use from a theoretical perspective, identifying the challenges this practice creates for the process of validation. I maintain that multiple-use increases the stakes associated with an assessment and argue that where multiple-use is known to occur, the separate validation of each use may not be adequate since interactions between multiple-uses may take place. To build on this theoretical discussion, I report on an in-depth case study of one instance of multiple-use. Using questionnaire data, document analysis, school-level case studies and interviews with test development personnel, I describe the multiple-use of the Education Quality and Accountability Office (EQAO) Grade 9 Assessment of Mathematics which is administered in Ontario, Canada. I identify the pattern of uses for this assessment and focus on two multiple-uses: the use of the assessment for accountability and the use of the assessment by teachers as part of students' grades. Evidence of interactions between the two uses is provided and the limitations of applying Kane's (2006) argument-based model of validation given these interactions are discussed. I demonstrate how the concepts of boundary objects and boundary encounters, situated within sociocultural theory, contribute to the process of validation for this assessment and may be beneficial for the validation of other instances of multiple-use. A number of ways of rethinking the process of validation to better address the multiple-use of large-scale assessments are suggested and areas for further research are identified.

Multivariate meta-analysis methods typically assume the dependence of effect sizes. One type of experimental-design study that generates dependent effect sizes is the multiple-endpoint study. While the generalized least squares (GLS) approach requires the sample covariance between outcomes within studies to deal with the dependence of the effect sizes, the univariate three-level approach does not require the sample covariance to analyze such multivariate effect-size data. Considering that it is rare that primary studies report the sample covariance, if the two approaches produce the same estimates and corresponding standard errors, the univariate three-level model approach could be an alternative to the GLS approach. The main purpose of this dissertation was to compare these two approaches under the random-effects model for synthesizing standardized mean differences in multiple-endpoints experimental designs using a simulation study. Two data sets were generated under the random-effects model: one set with two outcomes and the other set with five outcomes. The simulation study in this dissertation found that the univariate three-level model yielded the appropriate parameter estimates and their standard errors corresponding to those in the multivariate meta-analysis using the GLS approach. / A Dissertation submitted to the Department of Educational Psychology and Learning Systems in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Fall Semester, 2014. / November 6, 2014. / effect sizes, gls, meta-analysis, multilevel, multivariate / Includes bibliographical references. / Betsy Jane Becker, Professor Directing Dissertation; Fred Huffer, University Representative; Insu Paek, Committee Member; Yanyun Yang, Committee Member.

The dissertation explored the efficacy of using a POMDP to select and apply appropriate instruction. POMDPs are a tool for planning: selecting a sequence of actions that will lead to an optimal outcome. RTI is an approach to instruction, where teachers craft individual plans for students based on the results of screening test. The goal is to determine whether the plans crafted by a POMDP model in a RTI setting offer advantages over the current practice that uses simple cut score methods. Two simulated data sets were used to compare the two approaches; the model had a single latent reading construct and two observed reading measures: Phoneme Segmentation Fluency (PSF) for phonological awareness and Nonsense Word Fluency (NWF) for phonics. The simulation studies evaluated the POMDPs forecasts of the students' end-of-year reading performance, and the studies compared how the students were placed into instructional groups using the two approaches. The POMDP-RTI model forecasted the students PSF and NWF scores for the last time period based on their scores in the previous time period as well as a forecast standard deviation. In the study, 91% of PSF scores and 94% of NWF scores fell within two standard deviations. The assignment to tiers was very different (after the initial time block) with just over half the students assigned differently under the two models at the last time point. The growth was better under the POMDP-RTI approach with a difference in mean reading ability of .49 on a standardized scale. The gain is because the POMDP model can take into account past observations and instructional history in its forecasts. / A Dissertation submitted to the Department of Educational Psychology and Learning Systems in partial fulfillment of the Doctor of Philosophy. / Spring Semester 2016. / March 31, 2016. / MCMC, POMDP, POMDP-RTI, RTI / Includes bibliographical references. / Russell G. Almond, Professor Directing Dissertation; Young-Suk Kim, University Representative; Betsy J. Becker, Committee Member; Insu Paek, Committee Member.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.