To compare the means, reliability coefficients, and concurrent validity coefficients for four tests, two with and two without optional items; and to determine if pupils could select items that would improve their scores, 210 seventh grade pupils were first administered the seventh grade spelling subtest of the Iowa Tests of Basic Skills. They were then given one of four experimental tests based on this subtest: (1) 24 items whose difficulty indices ranged closest to 50%; (2) 24 randomly selected items; (3) all 48 items of the subtest with pupils directed to answer all 48, but to select 24 items to be scored; and (4) all 48 items with pupils directed to select and answer only 24 items. Working time was the same for the four tests. No statistically significant differences occurred among the means or concurrent validity of coefficients of the four tests, although test number two was significantly less reliable than either of the two optional-item tests. Pupils taking the third test tended to select less difficult items more frequently than those taking test number four. (Author/JMC)