English Language Testing (ELTiS)
ELTiS is now online!
ELTiS Score Interpretation Guidelines
As part of an ongoing analysis of test use, CSIET member organizations and Ballard & Tighe have been monitoring ELTiS score patterns since the introduction of the test. Patterns have shown that students from some geographical areas that traditionally had high pass rates on the SLEP (and whose students had gone on to successful exchange years in the U.S.) were experiencing lower pass rates on ELTiS. Differences between the content of the two tests also pointed to an emerging need to explain the meaning of ELTiS scores independently of SLEP, a discontinued test. To address these issues, two studies were conducted. One was a new analysis of the data from the original ELTiS-SLEP equating study, and the other was a standard setting with the TESOL English language proficiency standards. These guidelines summarize the results of the two studies and supersede previously published ELTiS score interpretation guidelines.
Using ELTiS Proficiency Levels to Guide Score Interpretation
The proficiency levels in the table below are based on a standard setting study between ELTiS and the TESOL proficiency levels. In our latest study, a panel of experts representing CSIET member organizations undertook a rigorous process of analyzing the content and difficulty of ELTiS items relative to the TESOL proficiency levels. The experts also helped refine and finalize the ELTiS proficiency level descriptors below. CSIET and Ballard & Tighe encourage stakeholders to communicate with one another regarding the formation of their own cut scores and acceptance policies.
Using SLEP to Guide Score Interpretation
The table below is based on new analysis of score distributions from the original ELTiS-SLEP Equating Study. The new analysis emphasizes information from students who took the entire ELTiS and SLEP tests during the study, rather than just one part of the SLEP. Since the entire test is longer than a one-skill test, the scores for this subset of student are are more reliable, and this improves the confidence we can place in the results of the equating. Correlation between the two sets of scores has improved from .73 to .82. A wider range of statistical techniques for estimating the nature of the relationship, including confidence and error, were also used in this analysis to ensure that no further changes will be needed. While we are confident that this new information will help improve screening, we advise all stakeholders to begin to rely more exclusively on your own accumulating experience with ELTiS and the scores themselves (as well as the content-based descriptions of the meaning of the scores) rather than correlations to a discontinued test.