Political scientists Kevin Arceneaux, Martin Johnson, René Lindstädt and Ryan J. Vander Wielen studied how members of Congress voted during the rollout of Fox News, 1997-2002. In particular, they measured how likely members of Congress were to vote with their party. What they suspected is that the “pull” of Fox News would grow stronger as the election approached — when members are most attentive to their constituencies and when parties often allow members greater discretion in how they vote.
And this is exactly what they found. As the election drew near, Republicans in districts with Fox News became more likely to vote with their party, and Republicans in districts without Fox News less likely to vote with their party. Democrats, however, behaved the opposite. Democrats in districts with Fox News became less likely to vote with other Democrats. Here’s a graph to illustrate (click to enlarge):
As the graph suggests, party loyalty was high no matter what. But Arceneaux and colleagues argue that even these small shifts could have netted the Republicans enough votes — about 26-27 — to affect actual lawmaking:
On average, the Republican coalition size with Fox News is 26.59 votes larger than without Fox News, which is statistically discernible from zero at 95% confidence. Therefore, while the gap in predicted probabilities across non-Fox News and Fox News members at election time may at first glance appear quite small at the individual level, this difference in predicted probabilities has considerable implications for Republican coalition sizes when aggregated to the entire membership. The advantages reaped by the Republican party are of particular substantive interest when one considers that, during the period of analysis (1997-2002), the Republicans won 26.4% of all party votes by 20 or fewer votes and 10.3% by 10 or fewer votes.
It is an open question, of course, whether this effect persisted after 2002, especially given that MSNBC eventually began to provide a left-leaning counterweight to Fox News. Regardless, Fox News appeared to affect how members of Congress behaved, especially when they thought voters were watching.
[Update: See also related research from Joshua Clinton and Ted Enamorado.]