ABSTRACT
The 2002 fi eld study demonstrated that a computer-based test of communicative academic language profi ciency could be successfully delivered on a small scale (3,000 test takers) in a low-stakes environment. The next step in the developmental process was to fi nalize the test blueprint. This required an evaluation of whether the proposed test design was robust enough to be sustained on a much larger scale (700,000 to 900,000 test takers per year) in a high-stakes environment. A review of the proposed test design and the results of the 2002 fi eld study raised a number of issues that needed resolution.
