Obviously, there are similarities on some issue and differences on others. I am not denying that current union members are more likely to vote Democratic either. I do not think I can get into the debate of whether attitudes precede union membership or the other way around without thorough familiarity with the nature of the data you quote (which I do not have) - but I do not know how you can decide this issue without longitudinal data.
You observed a relationship between union membership and Democratic vote in a subset of the population and you are claiming that this relationship should also hold outside that subset. All I am saying that it may or may not. To show that it does you need to demonstrate a causal connection between membership and voting that goes beyond statistical correlations - one possibility is unions providing political education for its members, except that they seldom do that nowadays, especially in comparison to the massive propaganda deluge coming from the other side. So in the absence of identifiable causal connection between union membership per se (rather than its correlates!) and voting - all we have is statistical correlations that may or may not hold outside the population set from which they were derived. Consequently, expanding union membership may or may not turn out more Democratic votes.
Again, I am not trying to prove you wrong - I am both pro union and vote Democratic. All I am saying is that in all likelihood it will take much more than expanding union membership by a few percentage points or, for that matter, getting a Democrat to the White House to end corporate hegemony and turn things around a bit.
Wojtek