An incomplete repeated measures design experiment was done to test people's ability to solve problems that varied in difficulty at three levels (easy, moderate, and hard). The researcher used all possible orders to balance practice effects. Results indicated that performance on the hard problems was better when they were preceded by a moderate difficulty problem than when they were preceded by an easy problem. Which of the following is a possible explanation of this finding?
A. negative transfer
B. differential transfer
C. progressive error
D. nonlinear practice effects