I think it is a bit misleading - the title should read "Top Twenty five Most Republican/Democrat leaning Cities" because this is what they actually measured, albeit I do appreciate Baltimore being on the liberal list - after all it is a fairly liberal city - but then most large cities are.
I think their tabulations would look somewhat different if they considered other measures, such as union membership, political participation in progressive/conservative organizations, support/opposition to gay marriages (and equal rights more generally), , support/opposition to national health care system, support/opposition to proportional representation, support/opposition to public social safety measures - to name the most important issue.
Bean counting is notoriously inadequate in capturing political sentiments because it is confined to the options confined to the status quo and cannot predict how people will react if these issues/political candidates are framed in a different way. French sociologist Alain Touraine recognized that long time ago, and employed very different method to study political social movement - which he called "sociological intervention." This was basically a kind of focus group approach in which the interviewer would challenge the participants' views (or participants' would challenge each other's views). The purpose of this exercise is to find out what is the actual range of opinions, how people react to different framing of the issues, and what framing can produce most consensus.
Wojtek