 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
CREST | Research
IES awards College researchers $4.8 million/5 years to create a new doctoral training program for educational scientists
CIRGE | Research
CIRGE has been busy: its 5-Years Out study drew wide attention and it has hosted three int'l workshops on PhD education...
Research
Projects
Project Title: Validity Studies for the Washington Assessment of Student Learning (WASL)
COE Investigator: Catherine S. Taylor
Funding Source: Office of the Superintendent of Public Instruction, State of Washington
Start/End Dates: Ongoing
The Office of the Superintendent of Public Instruction has funded research on the new state assessment, the Washington Assessment of Student Learning (WASL). The assessment, which is designed to measure the state’s Essential Academic Learning Requirements (EALRs or curriculum standards) for grades K- 4, 5-7, and 8-10, is unique in that it is composed of multiple choice, shortanswer, and extended-response items. Given the relatively new use of shortanswer and extended response items in large-scale reading and mathematics tests, the WASL has provided a fruitful area of applied measurement research. In addition, unlike national standardized tests, each year’s WASL test forms are composed of different items, with the exception of a set of items that link test forms from one year to the next. Some of the critical research studies included in the work are: examination of scale drift over 5 years of equating, examination of the validity of test-level scores and ‘objective-level’ scores from the tests, examination of differential item functioning based on ethnicity and gender, and examination of the dimensionality of the open-ended items (in other words, does performance on a short-answer or extended response mathematics item require skills that are shared with other tested subjects, such as writing). Results of these studies have provided strong support for the validity of WASL scores; however, there have been some surprising results. In particular, data trends suggest that when comparing white and non-white students with the same total scores, non-white students tend to get their scores from open-ended items while white students tend to get their scores from multiple-choice items. Similarly, when comparing boys and girls with the same total scores, girls tend to get their scores from open-ended items while boys tend to get their scores from multiplechoice items. These results have compelling implications for the future of largescale test development since equity and fairness are important goals in largescale testing.
Projects
