Levels of independent variable(s) are often mixed within each block of trials rather than each level being presented in a separate block in research of various types. Two experiments involving tasks of a visual nature were conducted to demonstrate that such mixing can easily obscure the effects of principal interest. A target circle was projected parafoveally to the left or right of fixation. The target position was flanked by a nontarget or by no figure. One group of eight subjects received two blocks of trials, with nontarget types mixed within each block, and nontargets appearing to the foveal side of the target position in one block, but to the peripheral side in the other. Analysis of variance indicated a difference in performance between the blocks, with no difference between nontarget types. Another group of eight subjects received nontargets blocked by both position and type, with both variables emerging as significant. The importance of using unmixed designs to determine the effect of different levels of the independent variable(s) was emphasized. (Author)