What decided the 2000 election? A few hanging chads? The Supreme Court? Or was it Fox News?

A new working paper argues that former President George W. Bush’s popular vote total would have been 1.6 percentage points lower in his race against former Vice President Al Gore if Fox had not launched four years earlier. The paper provides new evidence that Fox and MSNBC have a real influence on how their audiences are likely to vote.

The fact that Republicans are more likely to watch Fox and Democrats to watch MSNBC is a chicken-and-egg problem. To be sure, people prefer to watch anchors and commentators whose views they already agree with, but do the channels actually make their viewers more liberal or conservative as well?

To solve this riddle, the researchers, Emory University’s Gregory Martin and Stanford University’s Ali Yurukoglu, took advantage of a surprising pattern among cable subscribers: People are more likely to watch any station with a lower channel number.

As Martin explained, that’s probably because the oldest and most popular channels, like ESPN, usually have lower numbers. Viewers watching those channels might flip through a few others on their remotes during a commercial break, but they won’t stray too far.

Fox’s and MSNBC’s numbers are more or less random across the country, and in towns where MSNBC has a lower a number, cable subscribers tend to be more liberal — even compared to people who get their television through a satellite dish.

These viewers watch a few more minutes of MSNBC a week on average, but not because they agree with the hosts’ politics. They’re watching MSNBC because they’re more conveniently placed in the line-up. The same is true of Fox.

Martin and Yurukoglu found that watching four more minutes of Fox a week makes you 0.9 percentage points more likely rto vote Republican, while watching MSNBC for four more minutes makes you 0.7 percentage points more likely to vote Democratic.

Matthew Gentzkow, an economist at the University of Chicago and an expert on bias in the media, called the paper “exciting” and “extremely clever.”

Other researchers have analyzed Fox’s effect on voting, he said, but the new study examines data from more channels over a longer period of time, with more detailed data.

The paper raises immediate practical questions for Time Warner Cable and Comcast, which have proposed a merger.

Federal regulators have long denied mergers in media markets to prevent any one person or company from acquiring too much control over Americans’ opinions and to be sure that people hear a variety of points of view. This paper suggests they had good reasons for doing so. To the extent that people only watch news anchors whose views they already agreed with, large media companies reach a broader audience and make more money by offering opposing views. But people will watch what’s available, even if they don’t always agree, Martin and Yurukoglu found. Over time, they’ll find themselves persuaded by what they hear, which will make them less interested in listening to other ideas.

These days, most people confront a range of opinions online and on television, in contrast to three decades ago, when almost everyone watched the one of three nightly news broadcasts. Media companies that are looking to merge will have to find ways to protect that diversity. When Comcast acquired NBCUniversal a few years ago, for example, regulators required Comcast to assign Bloomberg a number close to CNBC’s, so that viewers would not have to hunt for an outlet that was a competitor of Comcast’s new subsidiary in financial news. Comcast didn’t live up to the agreement, the Federal Communications Commission later found.

“If we maintain some diversity, then we can maintain some optimism that things will wash out, but if everything is concentrated in the hands of one owner, that’s going to be a problem,” Martin said.