Oddly, some journalists and Democrats seem confused about why the presidential race is so competitive at this point given Donald Trump’s liabilities as a nominee.

Voters’ antipathy to Hillary Clinton obviously is a good part of the answer, but just as important is the paucity of true swing voters.

Big swings in polls get the big headlines because they imply big changes in a race. But over the past three-and-a-half decades, I’ve grown skeptical of big swings in polls at this late date, and I’ve seen only modest changes in voting behavior in presidential elections.

It isn’t that big swings can’t happen during a campaign. It’s that big swings are rare because there are so few swing voters in the electorate and all swing voters rarely move in the same direction.

A little more than a month ago Gallup found 38 percent of respondents saying they were political Independents, while only 31 percent said they were Democrats and 27 percent said they were Republicans.

But once respondents who “leaned” to one party or the other are allocated, only 11 percent of Americans called themselves true Independents (and some of them probably are closet partisans). This same distribution has been found by the Pew Research Center.

An NBC News/Wall Street Journal survey conducted after the Democratic convention (July 31-Aug. 3) showed 33 percent of respondents were independents, while 13 percent were true independents.

Dramatic events – such as a national party convention or a heavily watched televised debate  – can produce significant swings in the polls, and a bad economy, presidential scandal or unpopular war can produce substantial electoral swings, such as when normally Republican Indiana voted for Barack Obama in 2008.

But unless there is dramatic news, you should be very skeptical about any survey showing a dramatic shift in voters’ attitudes over a few weeks. Remember, not only are survey results subject to a margin of error, but 5 percent of surveys produce results outside that margin.

If you look at actual presidential election results, you may be surprised to see how little swing there is when all the votes are counted.

In Ohio, the Democratic nominee during the last four presidential elections has won 46.5 percent of the vote, 48.7 percent, 51.5 percent and 50.7 percent. That is a spread of 5 points.

In Florida, the Democratic nominee in those four contests attracted 48.8 percent of the vote, 47.1 percent, 50 percent and 51 percent – a spread of 3.9 points.

In Pennsylvania, the Democratic nominee has drawn 50.6 percent, 50.9 percent, 54.5 percent and 52 percent. The spread: only 3.9 points.

Of course, there is a big difference between winning 48.7 percent of the vote and 50.7 percent, and in some states even a small change in voting behavior can flip a state from one party to the other. But in other states there are so few swing voters that even a competitive state, such as Pennsylvania, is difficult to flip.

Some states have had larger swings during an election or two (e.g., West Virginia), but the national swing over the last four elections has been very modest.

Al Gore drew 48.4 percent of the vote in 2000, while John Kerry drew an almost identical 48.3 percent four years later. Barack Obama drew 52.9 percent in 2008 and a slightly less 51.1 percent four years later. The swing from the most votes to the least – a mere 4.6 points in four elections the two parties split.

Only three states switched partisan columns between 2000 and 2004 – New Hampshire went from Republican to Democratic, while Iowa and New Mexico switched in the other direction.

In 2008, at the height of voters’ dissatisfaction with the Bush administration, 9 states flipped partisan allegiance: Iowa and New Mexico switched back to the Democratic column, and seven others flipped to Obama: Florida, Ohio, Virginia, Colorado, Nevada, North Carolina and Indiana.

Four years later, only two states flipped – Indiana and North Carolina switched back to their previously “normal” vote backing the Republican.

To be sure, huge differences in enthusiasm and turnout can affect an election’s outcome, as can dramatic events that define the entire election or the nominees.

But in years when there is no enthusiasm gap between the parties and base party turnout is comparable, only a handful of voters in a handful of states decide election outcomes in very competitive states. And since many of those voters decide late (unlike partisans who decide early), some contests look closer early than they will be.

So, the next time you see media reports that Florida or Ohio is “close” or that Pennsylvania is “tightening,” you should not be surprised. They are narrowly divided states where it is difficult for one party to blow out the other in a competitive federal election.

But being “close” isn’t the same thing as saying a state is a “toss up” or both candidates have the same chance of winning it.

And because of that, the Clinton-Trump race can be both “competitive” and even “close” in a state and nationally, but at the same time clearly favor one nominee – in this case Clinton.

Stuart Rothenberg writes about the politics of the presidential and congressional races.