Editorial: In a compelling preponderance of clinical trials,

Dear akashb106
First of all, be very careful with precise phrasing. (B) doesn't say the results were "self-recorded" but rather "self-reported". That's a HUGE difference. If the data were "self-recorded", that would meet each participant were writing down all the data him or herself, and of course, many people would flake and forget to write stuff down, and the resultant collection of data would be entirely useless from a scientific point of view. That's not what (B) is saying.
Mood is "self-reported" in the sense that the only way a researcher has of determining a subject's mood is by asking them, verbally, "how do you feel?" or "what mood do you have?", and getting verbal information in response. In some cases, self-reported information is not particularly reliable, especially if the subject has some motivation to lie or stretch the truth --- consider asking the questions "are you honest?", "have you ever broken the law?", "do most people like you?", "have you ever hurt someone on purpose?". With questions such as that, people may consciously lie, or they may just have an unrealistic picture of themselves, and thus the self-reported response might not be reliable. For example, if I ran a research project, and asked 200 people in a city "have you ever committed a felony?", and they all said "no", it would be a pathetically gullible conclusion for me to say --- well, none of them were lawbreakers! In such a research project, the problem of "self-reported" data is huge.
By contrast, in this case, people take a pill for a few weeks, and then we ask them, "do you still feel depressed?" or "has the pill you were taking helped your depression?" Yes, maybe one person would lie, but there would not be any systematic reason to expect most of the people to lie in the same direction. Because there's no reason to expect that (a) people will not have a consistent motivation to misrepresent the truth, and (b) people will clearly know whether their own depression got better or not, then there's no reason to conclude that "self-reported" data would pose a liability in any way. Furthermore, with something such as depression, it's unclear what measure you would use instead of self-report. It's not as if there's a particular brain scan they can do to prove someone is depressed. There's really nothing that can be objectively measured and analyzed. The only way anyone finds out that a person is depressed is by talking to him. This is fundamentally why (B) is wrong.

Understanding why (C) is correct depends on understanding the details of the placebo effect. It might be worthwhile to find a psychology textbook and read a detailed description of this phenomenon. Sugar pills are standard placebos --- they are perfectly safe, and they have absolutely no medicinal value. They don't cure anything. Nobody thinks that sugar pills really are effective in treating anything --- that's precisely the flaw of the argument, and it would be completely ridiculous if an educated person seriously took this position. The placebo effect is all about the ability of the body to generate healing effects purely based on belief and trust in a medical figure, apart from any medicinal value of anything you ingest. In this experiment, the folks in the "control group" received sugar pills, placebos, but the doctor giving it to them didn't say, "this is a sugar pill" --- rather, the doctor said, "This pill will help your depression", and the subject, hearing that assurance from a medical profession, experienced healing effects in her body purely because of the trust she felt in the doctor. The belief, the trust, caused the improvement, and the content of the pills, sugar, was 100% irrelevant. This is the placebo effect. This is a standard effect, widespread in medical testing, and very well documented. Again, this is a difficult idea to grasp from a brief explanation: you may well have to read articles about the placebo effect before you fully understand it.
As you may imagine, the placebo effect poses great challenges to medical research --- if some people get better when you give them almost anything, how can you tell that a medicine actually works? In practice, we can conclude that a medicine works only if it significantly outperforms the placebo effect ---- say, 40% of folks who get the placebo get better, and 90% of folks who get the medicine get better --- that would be a medicine that is accepted as valid because it significantly outperformed the placebo effect. Accommodating this issue involves a experimental design called "double-blind" experiments. You can find all of this explained in detail in a good psychology text or a good statistics book.

Theoretically, you don't need know outside knowledge to answer GMAT CR, but you need to know the basics of how the world works. It is conceivable that, like this problem, a problem on the real GMAT could spell everything out but, in a way, expect you to have familiarity with the basics of the placebo effect. Arguably, it has implication for how advertising works, and in that sense, it may be highly relevant to what you need to know for your career some day. I would argue: it something every educated person should understand.

Does all this make sense?
Mike