>This could be a long conversation, but basically I agree with your view that social sciences deal with immensely complex and variables -filled topics.

It won't be long, since I haven't found any disagreement here.

>My minor was in macroeconomics, which was quite interesting, since we got to use a lot of math mixed with behavioral science in order to understand the past and predict the future, With math in the mix we got much closer to making good predictions at least for the near future, but it still remained an inexact science with a lot of "human variables." You don't really need to look further than the stock market to see that perception drives economy as much as any underlying "hard" fiscal data.

In a weird twist, I have found out (no exact proof, of course, just an observation which can't be conclusively proven nor disproved) is that human behavior depends on the presence of an observer, which makes for very interesting effects when trying to get some hard data or to introduce change and then expect a certain outcome. Take laws, for instance - they are supposed to rectify a behavior which appears in a given society, but as soon as they are enacted, they become part of the environment and the whole motivational landscape that caused the behavior they are trying to rectify looks different now - and leads to all sorts of outcomes, not necessarily the desired ones.

Prohibition was one great example of that - an attempt to instill puritan values into the populace gave rise to the distilleries and organized crime.

You can't even get data. Take polls - if done by phone, they'll never reach the people who won't bother (perhaps because of previous experience with polls? - another influence), people without landlines (cell-only is becoming quite frequent; homelessness too, and there are many unlisted numbers), people who simply aren't at home, who are UI (not user interface, but enjoy their daily session of violent behavior toward chemical compounds) or just have their own rhythm of sleep. If done by a webpage, worse: it excludes everyone who is not connected, then everyone who never heard of the website, then everyone who stumbled upon it and didn't feel like clicking. If done by email, it's spam. If done on foot, there are unreachable areas of towns - the communities which are fenced in or out, where it's not healthy to enter without bodyguards, or where bodyguards won't let anyone in. IOW, there's not a single point where pollsters can get a decent vertical slice to get a representative sample. They get only those they can catch.