Thanks for the question! I assume you're talking about this project: http://www.sciencebuddies.org/science-f ... ml#summary
(Candy Confusion on Science Buddies).
To help answer some of your questions, an experimental study involves specifically modifying/testing one parameter and looking for desired results. A correlational study, on the other hand, seeks to find relationships between two variables (usually in a population) without specifically exposing one group of people to a modified parameter. While Candy Confusion may not be a "true scientific experiment" in that sense, it is still a valuable project to do to understand and apply the scientific method.
A manipulative variable is what you change to observe its effects on something else, the dependent variable. The manipulative variable could be considered as age of the child in this study (as you're looking to see what effect age has on distinguishing candy from medicine), but to double-check, I'd point you to the Science Buddies resource pages. There is information on the scientific method and these terms that may help you greatly!