While facebook has access to a vast amount of data, there are also a few disadvantages of doing research on their dataset:

1. You only have data of people capable of buying a computer and having internet access. And then having the time and willingness to use facebook. This places a bais in the information and obtained research.

2. The profile of users don't reflect who they truly are. It represent what they want to be, or at least how they want to be percieved by others. Admittingly this problem is also present when doing surveys.

It certainly provides a lot more information and would aid social scientist, but it's not perfect. Then again they passively collect a huge pile of information too (e.g. using their like buttons all over the internet).

>2. The profile of users don't reflect who they truly are. It represent what they want to be, or at least how they want to be percieved by others. Admittingly this problem is also present when doing surveys.

Do you realize that you are saying what you think people are?

Which is which? who they really are and what they really want to be. Is Clark Kent real or Superman real? This is why sociologists have day jobs.

While I agree with 2, I'm not convinced by 1--where can you get data that is unbiased without FORCING random sets of people to participate in a study in the way you want them to? Even FDA-mandated experiments are biased by the fact that it requires volunteers.

The published social science they have produced so far is unimpressive, though. I attended a talk by a Facebook Data Group employee a little while ago, and it was a bit sad.