38 subjects were administered the Omnibus Personality Inventory and 91 subjects the Personal Orientation Inventory under instructions to respond to each item with one, both, or neither response alternative. Under these instructions the keyed and nonkeyed scores for some scales are not highly correlated. This finding is contrary to the forced high negative correlations resulting from the typical instructions to answer all items with only one alternative. Implications were drawn for interpreting scores and for the assessment of change.