Based on the field study forms, this study found that it was not possible to statistically specify more
than the current three levels for the listening section (200, 300 and 400) and four levels for the reading
section (150, 250, 350 and 450) and at the same time to identify enough anchor items to discriminate
between examinees at adjacent levels. The more levels that were specified, the fewer the number of
items that could be anchored at a level. While this study identified as many anchor items as possible,
the distance between levels had to be expanded. Thus, the final levels are some distance apart and
examinees scoring between levels are instructed to review statements for both the level below and
the level above. In July and August of 2006, ETS performed a statistical analysis based on several