Would You Tax Soda to Pay For Health Care?

Via Matt Yglesias and Ezra Klein, I see that the Kaiser Foundation has some interesting figures on public support for using sumptuary taxes to pay for health care. In particular, lots of respondents were upbeat about with this question: "In order to help pay for health care reform and provide coverage for more of the uninsured, would you favor or oppose increasing taxes on items that are thought to be unhealthy, such as soda, alcohol, junk food, and cigarettes?" Kaiser found:

pigouvian taxes for healthcare.jpg
Matt writes: "I think part of the reason this gets a relatively good result is that it's framed in the context of revenue." And part of the reason this gets a relatively good response is that the question lumps fairly disparate activities -- cigarettes and candy? -- together. Both suspicions are in fact confirmed by the next couple of slides:

paying for health care.jpg
tax arguments kaiser.jpg
But I'm interested in why support flags as the questions move from cigarettes, to alcohol, to snack foods, to soft drinks. I understand why support waxes and wanes based on how progressively or regressively the proposal is framed. But why would you favor taxing snack foods but not soft drinks? Or cigarettes but not snack foods?

You might say that cigarettes and alcohol have negative externalities: I have to breathe your second hand smoke, you have to avoid my drunken staggering, etc. But if you wanted to be a little punctilious about it, you could make the argument that junk food has externalities, too: You are more likely to be overweight when surrounded by people who are overweight. (The connection is though to be causal and not just correlative.) So why the difference?