Quantum Eraser Experiment says nothing? So, we start with a standard double slit experiment interference pattern.

Then we add a QWP to determine which way information.

But here is where my first problem arises. Aren't we just ignoring data to produce this pattern? The interference pattern is still happening we're just blocking it at the QWP. Look at the number of hits between image one and two. Where'd all that data go?
So, then we add a polarizer to the other stream to 'erase' the which way information and restore the interference pattern.

But look at the numbers again! All we've done is remove more data to return the interference pattern just on a smaller scale.
My question is this: How is this in any way demonstrating quantum properties?
This looks like how to do tricks with statistics using polarizers on your data. What does this have to do with entanglement?
All images taken from: http://laser.physics.sunysb.edu/~amarch/eraser/
Edit:
A quick summary of the quantum eraser experiment:


*

*A quantum entangled pair is split and sent to two detectors (A and B) and a coincidence circuit. 

*B is a double slit experiment before the circuit. (produces image one) 

*Opposite quarter wave plates are put in front of each slit to restrict only one type of polarization on the entangle particle from getting through the slit. This is now "marked path" and produces image two.

*A polarizer is put in front of A to restrict some more particles. This is now "erased" and produces image three.


It's possible my interpretation of what a polarizer is doing is different than yours so if there's any confusion check the page I linked or the wiki for Quantum Eraser experiment.
Edit 2: To restate the problem:
Does the interference pattern created in the double slit experiment contain the non-interference pattern when which-way information is present and further contain the interference pattern when which-way information is "erased"?
Or, are these completely different data sets?
 A: Your interpretation is essentially correct, although the specific numbers you keep referencing are a red herring. (More to the point, the number of counts is just proportional to how long you run each experiment; it's the proportions that count.) Despite those failings in your analysis, the conclusion is mostly correct: the quantum eraser experiment is ultimately (only) about the possible correlations between the different observables that come out.
This therefore means that, with respect to your main question,

This looks like how to do tricks with statistics using polarizers on your data. What does this have to do with entanglement?

the statistics has everything to do with entanglement. Entanglement is purely a kind of correlation between the measurement outcomes of the two entangled systems, and it is not some form of magical communication between those systems. (Or, at least: any description in terms of communication is interpretation-dependent and unsupported by the actual quantum theory.)
That doesn't mean it isn't mysterious, because Bell's theorem tells us that the correlations derived from entanglement are still stronger than any that we can derive from classical (i.e. local hidden variable) theories, unless we allowed the systems to communicate superluminally. It's even weirder than that, because entanglement provably cannot be used to communicate faster than light, so if there really are little green men communicating to orchestrate measurement results, they conspire to make this power unavailable to us. (And, even worse, the no-FTL-communication requirement is too weak: there are conceivable systems that obey it, known as PR boxes, that are provably 'more correlated' than QM, so they are unphysical, which then tells you that the no-communication requirement is not strong enough on its own.)
Does that seem weird? Then welcome to quantum mechanics!
