My approach for answering these sorts of questions is to kind of list the utmost important information pieces, then put them all together. From the reading passage we have.

1) Particular educational program. (whenever I see words like "particular", "unique", etc. It's like a warning signal to me)
2) 3-5 years old participants.
3) Conclusion: similar programs for "all" children 3-5 yo will increase their chances of success.

The two pillars holding the author's conclusion are points 1 and 2. We can certainly choose point 2 as our target to weaken the argument, say: "children aged 3 to 5 have a very short attention span and won't remember much of what they are being taught in the upcoming future".
Regarding point 1, the author seems too confident about his/her educational program. Now, how can we attack such a scheme. When we talk about research studies, there are some main critical aspects researchers must care for: methodology and data processing. The methodology can be faulty by not following standard procedures or using seemingly advantageous mechanisms; while data might be corrupted unwillingly as a result of poor data management (Sorry if I'm using a bit of outside knowledge here). To wrap this up, given the answer choices it seems that B and C are the most likely possible correct answers. B is quite compelling since if children in this age group display widely varying cognitive abilities, it would be hard to conclude your study can apply to all children in this group without extensive proof that your program accounts for all that variance. C, however, does hit directly the author's argument supporting stone, his study. By averting the study's data was corrupted, we are rendering the study inconclusive. To me, this is a debacle, a quandary, a hard to contradict dilemma for the research study's author.