## Score Comparability, Composite SAT to Beta Test

Despite the different content of the SAT V+M and the Beta Test, the high correlation between the two tests (r = 0.83, N=100 after two scores discarded) suggests that it is possible to develop a meaningful relationship between the two sets of scores for this particular population of test takers. The conversion method used here consists of setting scores equal that correspond to the same percentile rank (equipercentile).

Regression (as shown in the scatter diagram trend line) is not used as a conversion method. Equating transformations need to be symmetric. Regression of Test Y on X is, in general, different from the regression of X on Y.

Two scores were discarded before performing the conversion:

1. SAT V+M = 1485, BT = 13.75
2. SAT V+M = 1490, BT = 18.00

Note: This comparability graph does not touch upon the issue of test reliability. I.e., conversions near the middle of the graph are likely to have less uncertainty than conversions near the top or bottom.