The assertion that "higher order" thinking skills can be improved by learning to program computers is not a new one. The idea endures even though the empirical evidence over the years has been mixed at best. In fact, there is no reason to expect that all programming courses will have identical, or even similar, effects. Such courses typically differ more by the languages in which they are taught than by anything else, and rarely do they explicitly address higher level instructional goals. To properly assess the extent of transfer, or any other learning, empirical measures must be criterion-referenced to specific curriculum objectives. This paper describes the results from three field studies. In two of them, ninth graders who learned structured programming methods using the"Karel the Robot" teaching language performed considerably better on a series of expository writing tasks than did students in the studies' control groups. In the third study, students who began their introductory programming methods course with Karel performed substantially better on difficult structured programming tasks using Pascal. (Author/SWC)