Developing practical achievement tests for use at the primary-grade level is a difficult task. Some problems encountered appear to be resolved by using verbally administered yes-no tests. But such tests are criticized as having a low reliability because they offer only two choices. Two modifications of the yes-no test have been proposed to increase reliability. One is the "matched-pairs" technique, in which every "yes" item has a matching item to be answered "no". Both items must be answered correctly for either to be counted. The second technique of the all-no test, an attempt to counter the children's proclivity to answer "yes" even when the answer is not known. Some 200 first grade children were administered an economics test, in which all three techniques were used. The test scores indicated that the all-no test had the greatest reliability, but it was less valid than the matched-pairs test. Thus, the matched-pairs test would be the best way to construct the yes-no type of achievement tests. Another article by the same authors (see PS 001 819) also deals with the subject. (WD)