The 2008 exit polls showed Obama receiving 59% of the vote among union households, and 51% among non-union households. While certainly an advantage, that doesn't seem like a huge difference, at least not enough to justify the Republican party's all-out political war on unions.
But if we filter out some more variables we can get a clearer picture of the link between unions and voting behavior.
In the last post I showed you the strong relationship between education and support for Obama among white voters older than 30 (we can't look in depth at minority respondents because too many of them aren't actually minorities). These numbers come from the aggregated data of the summer DailyKos/SEIU/PPP polls. A white voter over 30 with a postgraduate degree is ~55-65% more likely to vote for Obama than a white voter over 30 with a high school degree only. That's a huge difference, and it's pretty impressive.
But somebody who's white, over 30, with only a high school degree, and who lives in a union household is ~50-90% more likely to support Obama than someone with the same demographics who does not live in a union household. In other words, a union's (probably) better than a graduate degree, as far as a marker of political behavior.
We don't have all that many respondents from union households - only 1285 - so what I'm going to do is combine respondents from two education levels, high school and some college, in order to look at the effect of gender and geography.
The question remains, are Democrats more likely to go for union jobs, or do unions make workers more Democratic? Like most things in life, it's probably not black-and-white, of course. We do know that public sector employees are far more likely to be in a union (37%) compared to private sector employees (7%). Are voters with Democratic leanings more likely to choose a career in the public sector? It's certainly plausible.
What we can say for sure is living in a union household is related to being far more likely to vote Democratic. This is not exactly news, but the differences are far larger when we drill down than the exit polls can show us.