One of the classic logic traps that befuddles conspiracy theorists of all sorts is the test of Falsifiability.   Loosely stated, falsifiability is used to place the burden upon the proponent of a theory to come up with a test that would / could demonstrate that theory to be false. If there's no conceivable test / data that could prove the hypothesis false, then we discard the hypothesis as simply being unprescriptive.  Not necessarily untrue, but merely so what?.

For example, one theory could be that plants need sunlight to keep their leaves green.   A different theory would be that little invisible fairies hover around the leaves and keep them green.  The sunlight theory is falsifiable because one could readily conceive of a test to disprove the link between sunlight & green leaves (for ex., remove sunlight from the plant and see if the leaves stay green).   The Fairy proponent, however, is burdened with coming up with a test that conclusively removes magical influence from the leaves.  

No matter how much our shaman believes it, without a test that could conceivably falsify it, the scientific community will throw aside the Fairy theory as simply "potentially true" but far from prescriptive and thus not part of the scientific canon relative to competing explanations.

The important results of Falsifiability are orthogonally related to another logical fallacy - correlation implying causation.  2 sets of correlated data points (for ex., folks who purchase a Britney Spears CD & occurrence of acne) can be shown to NOT carry causality if a falsifiable test to detangle the 2 can be constructed (for ex., If Britney causes acne, then were there cases of acne before Britney was on the scene?  The data to disprove this theory can be readily conceived - for ex., kids w/ acne in my own High School pictures.)

Falsifiability can be applied to politics whenever someone proposes a conspiracy theory. 

For example,  a friend recently forward me a blog post by Julius Civitatus alleging that terror alerts were created by Bush to improve his approval ratings.   The post includes the chart -

 

The post includes the following gems of analysis -

...We finally built this timeline of terror alerts and how they relate to the news headlines of the days immediately prior to that very alert. I think it's very easy to see a pattern recurring

...There are few things that are quite evident from the chart:

- Whenever his ratings dip, there's a new terror alert.

- Every terror alert is followed by a slight uptick of Bush approval ratings.

- Whenever there are many unfavorable headlines, there's another alert or announcement (distraction effect).

Now let's leave aside the most obvious methodological error here -- e.g. if anything, the graph implies the opposite of what the author intends - the fewer terror reports issued, the higher the approval rating -- there's still a rather fundamental problem of falsifiability of Julius' hypothesis.

We'll assume for a second that some, most, or even all of these terror alerts were in fact politically motivated wag the dog thrusts by Karl Rove and the Bush administration.   Given this hypothesis, what sort of counter data could conceivably convince us that these terror threats were, in fact real?  How do you demonstrate a terror alert that has no a priori connection to a Bush approval rating?

The primary way is to reveal all the data used to construct an alert and make a judgement about whether that data - modulo Bush intentions - was truly enough to call for the alert.   However, Bush clearly can't reveal it because it would compromise investigative sources & methods.   The proponent of this theory almost certainly doesn't have direct access to this data himself.   A second reason falsifying data can't be generated is intrinsic to the issuance of the terror alert.   The alert itself can serve as a deterrent to the act & thus there's even a systemic reason for the most successful terror alerts themselves to be met with... nothing. 

Now, there ARE ways to construct something like the author's theory in a manner that's falsifiable.  But this block of implied "correlation / causation" data is NOT how it's done.   If a conspiracy to manipulate the public via terror alerts is really happening, then the best way to directly prove this would be a single "leak" pertaining to a specific alert.  In our age of Richard Clarke's, James Wilson's, Wesley Clark's, Imperial Hubris and others, there is clearly an existence proof that an "insider" can / will / and does have a powerful incentive to break from the coalition and report such a conspiracy if it were the case.  

In lieu of falsifiable tests for this conspiracy theory, we're left with merely our own a priori suspicions about Bush's motives - Julius has clearly revealed that he believes the Bush Approval Fairy is behind the alerts...

---