I think there is such a thing as neutral reporting on candidate activities. For example, a source could report every rally held by any candidate, and say where and when and add a transcript of their remarks.
There is certainly a kind of inaccurate reporting that favors one candidate over another. I think Keith Olbermann is much liklier to report an embarrassing story about McCain than about Obama (I get this impression from having watched him often because I enjoy him).
Troy McClure, I disagree with your implication that “better” is generally hard to define for a cadidate. Certainly in some cases it’s hard, because ending the war quickly seems like a good thing to some of us and a bad thing to others of us. But candidates may be caught lying, or selling their influence, or confusing important objective facts, or contradicting themselves, or failing to deliver on past promises. Doing these is always “worse” than not doing them.
I ask the OP with an answer already in view. I think that news agents can mislead the electorate by systematically favoring one candidate over another, and that this is harmful, and that some examples of doing this are objectively testable and some are not. It would be nice if this kind of misleading would be deplored and reduced. And, a news agent can show more pleasing stories about one candidate and less pleasing ones about another because the candidates deserve these. Some of the ways of doing this are testable and some aren’t. Finally, these effects ought to have names, and I think the name “bias” is a good name for the first one, the “misleading” one, and not a good name for the second one, the “pleasing” one. This is consistent with my understanding of “bias” in scientific discussion. But studies I see about “media bias” seem to apply the term “bias” to both of these; thus the posting.