Wednesday, November 16, 2011

Slant Quotient

"Like everyone else you were born into bondage. Into a prison that you cannot taste or see or touch. A prison for your mind."
--Morpheus (The Matrix)

Previously, we examined Groseclose's (2011) political quotient (PQ) as the basic building block for measuring the degree of liberal bias in the media. Because it measures the extent to which the roll call votes of Congresspeople align with the agenda of the liberal interest group Americans for Democratic Action (ADA), the PQ cannot be directly determined for media outlets.

Groseclose needed a way to link the bias exhibited by politicians to the bias exhibited by media outlets. He did so by using citations to think tanks. To give their positions more legitimacy and credibility, politicians embed think tank citations when giving speeches. Journalists have the same need for legitimacy and credibility, so they embed think tank citations in their stories.

Think tanks are a useful data source because they span the politican spectrum from far Left to far Right. And because politicians as well as journalists are likely to support their points of view with 'authoritative' sources that espouse similar views (courtesy of confirmation bias/selective reasoning), it is likely that those on the Left will prefer citing Leftist think tanks and those on the Right will prefer citing Rightist think tanks (although that is not a direct hypothesis put forth by Groseclose & Milyo 2005).

Groseclose and Milyo's (2005) analytical method basically proceeded as follows. First, scan the Congressional record over a period of time for instances where members of Congress cited a prominent think tank or policy groups (using an authoritative source, the researchers had established a list of about 200 such organizations). Next, record the PQ of that member of Congress.

The stories of media outlets over a similar period of time were also scanned in search of citations to the same list of think tanks.

In Table 1 of their paper, Groseclose and Milyo (2005) list the 50 most cited think tanks by media outlets, along with the average PQ of Congresspeople who cited the same think tank and the number of times those Congresspeople cited the source was cited. When scanning the list, one takeaway is that the average PQ of Congressional citations matches conventional wisdom about the ideological positions of various think tanks. For example, thinks tanks broadly considered to lean 'conservative' (e.g., Heritage Foundation, Christian Coalition, National Right to Life Foundation) were associated with low PQs, while think tanks broadly considered to lean 'liberal' (e.g., NAACP, Sierra Club, Common Cause) were associated with high PQs.

It is also worth noting that among the top ten think tanks cited by the media, four of them were assoicated with PQs above 60 while only one was associated with a PQ below 40. I calculate the average PQ associated w/ the top ten cited sources to be 55.8. In other words, the most popular think tank sources used by media outlets lean Left.

Another preliminary indicator of media bias comes by comparing the PQs of the top ten media-cited think tanks with the top ten Congress-cited think tanks. Four of the top ten think tanks cited had a PQ less than 40, and the average PQ of the top 10 sources by Congress was by my calcs 47.3 (altho also by my calcs the difference between the two means was not significant using a simple two-sided t-test (p=.22)). The findings suggest that the media is more liberal than Congress, although the differences captured by this rudimentary analysis are not conclusive.

It is after this rudimentary phase that the two PhDs earn their pay. To conduct a more thorough, generalizable analysis, Groseclose and Milyo (2005) essentially use a series of regression models to estimate the degree of liberal bias in the media. The methods are beyond our scope here, but the basic thought process is: "I know that there is a significant relationship between a Congressperson's PQ and the extent to which that person cites various think tanks. So if I have data on the extent to which a particular media outlet cites various think tanks, then I should be able to estimate a PQ for that media source."

Groseclose (2011) calls this estimate the Slant Quotient (SQ). The higher the SQ, the more liberal the media outlet. An SQ of zero means that a media outlet sounds about as conservative as a speech by Michelle Bachman. An SQ of 100 means that the outlet sounds about as liberal as a speech by Nancy Pelosi. An SQ of 50 means that the outlet is perfectly centrist, displaying the same level of liberal bias as the typical American voter.

Here are SQs of variou media outlets. We'll discuss more in the future, but it should be apparent from a cursory scan of the data the media is biased Left, and significantly so.


Groseclose, T. 2011. Left turn: How liberal media bias distorts the American mind. New York: St Martin's Press.

Groseclose, T. & Milyo, J. 2005. A measure of media bias. Quarterly Journal of Economics, 120(4): 1191-1237.


dgeorge12358 said...

Jed Leland: Bernstein, these men who are now with the Inquirer, who were with the Chronicle until yesterday...

Jed Leland: Bernstein, Bernstein, these men who were with the Chronicle, weren't they just as devoted to the Chronicle policies as they are now to our policies?

Bernstein: Sure they are just like anybody else. They got work to do, they do it. Only they happen to be the best men in the business.

Jed Leland: Do we stand for the same things the Chronicle stands for, Mr. Bernstein?

Bernstein: Certainly not. Listen, Mr. Kane will change them to his kind of newspapermen in a week.

Jed Leland: There's always a chance, of course, that they will change Mr. Kane without his knowing it.

~Citizen Kane

fordmw said...

That is a fantastic quote. Great find!