Ruy Teixeira argues that it is more appropriate to think of swing voters as those who are persuadable, not necessarily just independent:
For an individual voter to qualify as a swing voter, the relevant criterion that needs to be fulfilled is persuadability. And that’s not a quality that’s exclusive only to those who are completely undecided, or who are only weakly committed to a candidate. Even those who are moderately committed can be persuaded to deepen their commitment. And the deepening of an existing affiliation with a candidate can be just as significant, both statistically and electorally speaking, as attracting mild commitment from someone who had previously been mildly committed to another candidate.
The important factor is not where voters’ inclinations started out, but the fact that their inclinations were changed at all. The act of persuading a swing voter has traditionally been thought of as moving a given voter from more likely to vote against a given candidate to more likely to vote for him—say from 55 percent likely to vote against to 55 percent likely to vote for. But it could also mean moving that voter from somewhat likely to vote for a candidate to very likely to support that candidate (say from 55 percent likelihood to 65 percent)—or, for that matter, from very likely to almost certain (65 percent to 75 percent). All three of these examples are mathematically equivalent—and it makes sense to think of them all as swing voters.
He recommends the following book if you want more info.
