Children struggle to conduct controlled tests, even with explicit instruction (Chen & Klahr, 1999). How learners approachmultivariable tasks can be affected by task goals; a scientist uncovers causal regularities whereas an engineer produceseffects (Klahr, et al., 2011). This study investigated whether science vs. engineering goals presented in a narrative picturebook influenced childrens ability to conduct a controlled test.Six-to-8-year-olds (N=72) were first pre-tested on their ability to design a controlled test of a variable predicting how fara ball travels down to a ramp. Children were then read a picture book that contained a science (conduct controlled test) orengineering (create faster ramp) goal. Next, they completed an identical post-test and transfer-test with two new variables.Childrens ability to design a controlled test improved significantly from pre- to post-test (p=.008) and marginally frompre-test to transfer (p=.067) in both conditions, suggesting that children learned from both goals.