Blogs, Participation and Polarization

by Henry Farrell on July 1, 2008

Eric Lawrence, John Sides and I have just finished writing a paper which looks at the first decent dataset that allows us to figure out what blog readers look like. This isn’t a final version (there are comments from Eszter and a couple of other readers that we want to incorporate – further comments and criticisms welcome), but it is just about fit for wider human consumption. The paper is “available at “: SSRN (if you’re signed up with them, we’d love you to download it from there cos it’ll bump up our hit count), and at “”: if that’s more convenient. So what do we find?

We were interested in two questions, both of which stem from normative debates in political science and political theory. One was whether blogs make it more likely that people will get access to points of view other than their own. Many deliberation theorists argue that this is a good thing. The other is whether blogs affect people’s likelihood of participating in politics – again regarded by many theorists as a good thing, for obvious reasons. Unfortunately, our data doesn’t allow us to make causal assertions – but it does point to some very striking patterns of correlation.

First – blog readers seem to exhibit strong homophily. That is to say, they overwhelmingly choose blogs that are written by people who are roughly in accordance with their political views. Left wingers read left wing blogs, right wingers read right wing blogs, and very few people read _both_ left wing and right wing blogs. Those few people who read both left wing and right wing blogs are considerably more likely to be left wing themselves; interpret this as you like. Furthermore, blog readers are politically very polarized. They tend to clump around either the ‘strong liberal’ or the ‘strong conservative’ pole; there aren’t many blog readers in the center. This contrasts with consumers of various TV news channels, as the figure below illustrates. All of this suggests that blog readership is unlikely to be associated with the kinds of deliberative exchange between different points of view that some political theorists would like to see.

Second – blog readers are much more likely than non blog readers to engage in politics (through voting, giving money to candidates etc). Not only that, but left wing blog readers are significantly more likely than right wing blog readers to participate in politics. You could interpret this as evidence of more general depression among conservatives etc, but our best guess is that this is in large part the result of the netroots effect. Having a strong political movement which is pushing readers to make donations etc is likely to have real consequences. Obviously, we would like to have more data before we could make a really good case that our guess is correct.

So whether you like political blogs will depend to some extent on whether you prefer deliberation across party lines to participation, or vice versa. Personally (at least as regards political efficacy in the current era), I’m on the vice versa side, but we leave this question deliberately open, as people from different perspectives may disagree &c &c.

Update: John Sides posts more on this paper “here”:”:



mq 07.01.08 at 5:31 pm

I think this stuff is going to change over time. The blog world took shape at an extremely politically polarized time, and the polarization was still there in 2006. As far-right ideology becomes less salient in American politics there will be more tendency for a new center to form.

There’s a real problem in separating the ideological characteristics of the historical period from the inherent properties of the medium. Both are active — obviously the near-infinite number of blogs permits you to slice an audience more thinly than a few news stations — but I really think you’re looking at an unusual moment.


markus 07.01.08 at 5:43 pm

Tahnk you for that.
However, at first glace it strikes me that you seem to be talking about blogs and blog readers when you should be talking about blog readers in 2006.
Overall polarisation, the current fortunes of both sides and the issues of the day likely have a considerable effect on blog reading behaviour. I mean, I can’t be the only one who dropped a bunch of conservative blogs when they became apologists for torture and rendition. (Another crucial factor from personal experience is time available for reading.)


Ingrid Robeyns 07.01.08 at 6:06 pm

Since I know I won’t have time to read the paper any time soon I’ll just ask my question…. do you only look at blogreaders based in the US, or around the world, and if the latter, than do you observe any differences between readers situated in different countries- regions – continents?


daniel 07.01.08 at 6:30 pm

I have to disagree with mq. There is little pressure to present any non- or less- polarized blog. The demise of Slate’s Blog Report(nee the Daou Report) is case in point. A central locale for posts from both sides, and it was not viable.

I feel that the impetus for the formation for a new center will likely come only after a major restructuring of the social landscape of this country. On the order of (and as likely to occur — say once every 100 years) as a third party.


Mark S. 07.01.08 at 6:49 pm

Very interesting. Not sure if the data is out there, but I think it would add a lot to the paper to chart this over time. Could also make for a very cool 3D chart with the third axis being time.


James Joyner 07.01.08 at 7:12 pm

Actual link to Sides’ post: <a href=”I’m clearly an outlier, a right-of-center blogger who not only reads from both sides of the aisle but likely reads disproportionate from the other side. “>LINK

Interesting study.


chiasmus 07.01.08 at 7:14 pm

Interesting work. My quibble is with the finding of homophily. On p. 11 you note that that the survey gave people a limited space to report the number of blogs they read, you speculate that people might choose only to report those that reflected their ideological orientation, and then you wave this away by saying that “There are likely few blog readers affected by the character limit”. I’m not sure I’m convinced by this.

1. It’s not just the physical character limit that is at issue; people may just put down a couple of blogs rather than taking the time to think of every single one they read.

2. I think you may underestimate the number of people who read a large number of blogs–as web-based RSS readers like Google Reader become more commonly used, it becomes easier to follow lots of blogs. There are about 40 blogs that I read pretty regularly, of which a few are right-leaning. But I would likely have been categorized as a political “carnivore” in your research. My habits are probably pretty atypical, but maybe not.

A related question: maybe I missed it, but I didn’t see any information in the paper about what the modal or median number of blogs read per person was. That would obviously have some bearing on my points above.


noen 07.01.08 at 7:21 pm

It’s sort of interesting to note what media outlets are not represented.

As far-right ideology becomes less salient in American politics

One can hope I suppose, but I’m not holding my breath. Personally, I expect things to get far worse.


John Quiggin 07.01.08 at 8:22 pm

What would cross-party deliberation involving Republicans look like?

Looking at Australia, there’s a large group of centrist, centre-left and leftist blogs that engage in reasonable deliberation on a range of issues. As here, there’s also a group of rightwing blogs (mostly US-oriented) with which it makes little sense to engage.


Henry 07.01.08 at 8:30 pm

Markus – there is comparable data for 2007 (this was a panel study), and one of my co-authors took a quick look; as far as I remember, he didn’t find any differences significant enough to warrant further investigation, although I would have to check.

Ingrid – this is all data on US based readers, for better or worse.

James – got grief from my co-authors about this too – fixed (I guess the address didn’t copy into my clipboard as it should).


Tom Maguire 07.01.08 at 8:42 pm

Not only that, but left wing blog readers are significantly more likely than right wing blog readers to participate in politics. You could interpret this as evidence of more general depression among conservatives etc, but our best guess is that this is in large part the result of the netroots effect. Having a strong political movement which is pushing readers to make donations etc is likely to have real consequences.

1. All righties (yes, ALL!) want to be Rush Limbaugh; lefties want to organize and remake the world;

2. It was far from obvious to the RNC and the ‘powers that be’ on the right that, from 2002-2006 the Republican approach was failing and they needed to reach out to blogdom; it was painfully evident to the world that the Dem Party stood to benefit from a new approach.

Consequently, there are many righties who haven’t raised a dime for the Reps, don’t have any plans to do so, and don’t feel like miserable failures; on the left, everyone who is anyone seems to raise cash for something or someone.

I would guess that by 2012 Rep on-line fundraising will be notable, although who will be pushing it I have no idea.


Tom Maguire 07.01.08 at 9:03 pm

Those few people who read both left wing and right wing blogs are considerably more likely to be left wing themselves; interpret this as you like.

Well, you have the data – make the call!

From what I can see in the chart on p. 33, the most popular righty blogs are Drudge, Little Green Footballs, Michelle Malkin, NewsMax, Powerline, Town Hall, and Real Clear Politics.

I would be shocked to learn that lefties are frequenting LGF, Malkin, NewsMax, or most of the others. However, Drudge and Real Clear Politics are pretty useful aggregators that even a self-respecting lefty could visit (Drudge for its news-pushing value, RCP for its polls).

So if the data shows that cross-over lefties are hitting Drudge and RCP, color me unimpressed.

Sorry if that is covered in more detail in the text or charts; I couldn’t find it.


Henry 07.01.08 at 9:20 pm

Tom – we show the ideological profile of Drudge readers in Figure 6 – it is very heavily skewed towards the right. We haven’t done the same for Real Clear Politics; it would not surprise me if it was less skewed. But it seems to me to be pretty obvious that we are going to see some sites on the right being more attractive to lefties than others.


Henry 07.01.08 at 9:22 pm

And unfortunately, having the data doesn’t mean that we can make the call – evidence of correlation not telling us that much about causation (we’re pretty careful throughout to emphasize that our data is insufficient to get at underlying causal relationships).


John Emerson 07.01.08 at 9:39 pm

What jumps out at me from that is something completely different.

Conservatives have a network to watch: Fox.

Strong liberals, who are one of the biggest demographics, if not the biggest, have nothing to watch. None of the non-Fox networks is liberal; they’re all centrist at best, and most of them are as likely to veer right as they are to veer left. (I’d just barely describe PBS as center-left, and maybe MSNBC.)

The way I’d describe things is that on the internet people get what they’re looking for, whereas in broadcast media conservatives and centrists get what they’re looking for but liberals don’t. Liberals “get access to points of view other than their own” because their point of view is simply suppressed.

Liberal broadcast media, to my knowledge, consists of Keith Olbermann and Bill Moyers and a few others at PBS. Yet if you look at the graph, the two most liberal segments are the two biggest.


John Emerson 07.01.08 at 9:44 pm

The data back up my gut feeling that the internet is simply satisfying a need (for liberal journalism) that the mainstream media have failed or refused to fill.

That’s “polarization” only in the sense that we now have two poles, instead of monopolar right wing domination. The right pole has been strong for a long time.


Hemlock for Gadflies 07.01.08 at 10:03 pm

Liberals read conservative blogs so that we have stuff to make fun of.


Lee Sigelman 07.01.08 at 10:06 pm

In re your comment #14: I wish you were more careful throughout to avoid saying “data is.” Teaching you English is going to be a major undertaking.


Henry 07.01.08 at 10:38 pm

Tom – as a PS – I went back to check, and we code Real Clear Politics as non-partisan (my memory is that the basis of our decision was that even if a few of the people involved are definitely on the right, it doesn’t itself get into the partisan back-and-forth). This decision (as others) can maybe be challenged, but seems reasonable to me. In any event, RCP doesn’t affect our results on ideological cross-over readership.

Lee – according to the “Oxford English Dictionary”: folks, you are fighting for a lost cause here, and the standard Murican usage seems to be singularizing. But far be it from me to teach you how to speak what is ostensibly your own quaint dialect of the English language…


Bozak 07.01.08 at 11:08 pm

Not only that, but left wing blog readers are significantly more likely than right wing blog readers to participate in politics. You could interpret this as evidence of more general depression among conservatives etc, but our best guess is that this is in large part the result of the netroots effect.

I believe one of the reasons is age. The young and old which make up a large segment of the Democrats online have more time available. Check out daily kos yearly konvention pics to see the racial and age breakdown of daily kos.
Republicans are working or driving to work. Thus explaining the success of AM talk radio.


John Quiggin 07.02.08 at 12:01 am

“Data” is a mass noun, not a plural count noun, so “is” is correct. If it weren’t such constructions as database, data processing and so on would be ungrammatical.

The Latin singular from which “data” was derived (“datum”) has been replaced by “data point” (or “observation”) except in a few special cases.


synova 07.02.08 at 2:09 am

When I’ve commented on left-wing blogs, more often than not, my comments are deleted. So I don’t bother. Why read them? I may want to explain why I disagree in comments (and considering that my politics are a social-liberal capitalist libertarian sort I’m hardly channeling Rush or Savage) but I’m simply not allowed.

Left-wingers may read right-wing blogs more than vice versa because they’re tolerated better by those right-wing blogs.

The authors may find it interesting that the most vigorous exchange of widely opposing political viewpoints on the internet may be in MMORPG chat. ;-)


hinky 07.02.08 at 3:23 am

The reason many lefties “visit” conservative blogs is simple. Most are trolls.

“As far-right ideology becomes less salient in American politics there will be more tendency for a new center to form.”

Translation: As soon as all of those people with whom I disagree are marginalized or go away the prevailing political view will be in accordance with mine. Then it will be more centered.

Posted by John Emerson · July 1st, 2008 at 9:39 pm

Anyone who thinks MSNBC is “center left” is on an alternate plane of reality. Just what the hell would it take to please you? A station that shows 24 hour footage of President Bush being burned in effigy while a crawler at the bottom lists every single negative occurrence the world over as a consequence of man-made “climate change”


noen 07.02.08 at 4:05 am

Synova makes an excellent point. There is a lot going on in the gaming world. It’s probably bigger than the blogosphere. But I don’t know how you would track that activity or categorize it.


Tom Maguire 07.02.08 at 4:33 am

Tom – we show the ideological profile of Drudge readers in Figure 6 – it is very heavily skewed towards the right.

Thanks, I had seen that but it doesn’t quite get to my question, which is this – are there specific right wing blogs which are favored by the left readers and which are driving that result (lefties read more right blogs than vice versa)?

There are very few omnivores from either left or right (8% of respondents, if my eyes don’t deceive me on Fig. 3, p. 33)), so it wouldn’t take many lefty readers of Drudge to get the result in question. And if the white dot in Fig 5 represents the media, the median Drudge reader is quite different from any of the other five blogs depicted.

Well – from a different angle, I am sure I could find righties who would say something like “Why do I need to read lefty blogs when I am paying $40 a month for the dead Tree Times?”. A righty can find mainstream lefty outlets easily; a lefty who prefers not to abide Fox, Rush, or Sean is probably going to have to check some blogs to see what the right is ranting about.


mw 07.02.08 at 4:56 am

“As already noted, we find that there is a general tradeoff between deliberation and participation among blog readers, an increasingly important group of highly politically aware individuals. Blog readers are more likely to participate in politics than non-blog readers, but also very likely to read only blogs whose political leanings accord with their own.”

Interesting read. Of course, as a political blogger, there is nothing I find more interesting than a study of, you know, political bloggers.

I guess I am among the 6% omnivores as I do read blogs from both the right and left. However, I agree with the markus and mq points that our source data could be highly time and context dependent. Since I advocate for divided government on my blog, in 2006 I spent more time flocking with left of center blogs who at the time were big fans of the divided government meme. Now, not so much. So I find myself spending more time on right-of-center blogs, who are in the midst of an epiphany and are beginning to see the wisdom of voting to maintain divided government into 2009. Funny how that works.

I’ll also note this historic parallel to the authors’ conclusions, linking increased participation and polarization as going hand in hand:

“The period from 1840 to 1890 has been labeled “the party period” and “the golden age of parties” because the major political parties (Democrats and Whigs until the mid-1850s, then Democrats and Republicans) were the strongest they have been in American history. Party leaders used patronage and campaign practices that aroused partisan enthusiasm to gain wide membership and keep them loyal and active. It worked. Voter turnout during this period was the highest in American history: between 70 and 80 percent for presidential elections and sometimes higher in state and local contests.”

Different self-selection and organizing principles, but the more things change…


Eric 07.02.08 at 6:53 am

first off, seeing a paper like this is absolutely great. it’s about time we saw some quantitative studies of blog readers with this sort of data, not least because I’ve been working on studying blog readers, albeit from a qualitative perspective.

I want to ask a question about what constitutes participation. does participation imply only voting; does it including campaigning, calling a representative, or making a contribution; or could it consist of public debate and discourse? perhaps I don’t understand because I’m outside the political science discipline, but I don’t see a definition of participation jumping out of the paper (granted, I’ve not read it thoroughly, yet). McKenna and Pole (2004 – Do Blogs Matter?) argue that, due to its status as a form of public discourse, political blogging itself actually counts as political participation. I’m curious how this notion of blogging as participation fits into the argument you’re making here that “blog readers are highly likely to participate” (p. 22).


John Emerson 07.02.08 at 11:53 am

Hinky, you’re a moron and don’t have any imagination either. I’d be pleased by a left-liberal news program with a few democratic socialists of the kind dominant in Europe. That’s how American liberals think, and we would call a station of that type “liberal”.

As it is, Olbermann and Maddow are liberal, Chuck Todd is a neutral numbers guy, Dan Abrams and Dick Gregory are weak neutral centrists who sometimes do good reporting, and Chris Matthews is a Clinton-hating centrist freak with weird sexual obsessions. I guess if Maddow came on full-time and replaced Matthews, I’d be satisfied. (Remember, about 20% of Americans are more liberal than Obama or Clinton).

The reason you’re confused is that you have the weird conservative habit of classifying everyone other than Bush loyalists and movement conservatives as “liberal”. Dick Gregory counts as liberal because he’s reported honestly about Bush a few times. People are even accusing Scotty McClellan of being liberal now, because he told the truth.

On top of that, the quote you ascribe to me isn’t mine, but mqs.

Basically, I hope for a situation within which movement Republicans have been marginalized by us the way we have been marginalized by them for the last several decades. The new political debate would be between moderate Republicans and conservative Democrats on the one side, and liberal Democrats on the other, with moderate Democrats the swing vote.

In politics someone always loses. Since 1980 the losers have been liberal Democrats. I’d just like to see the movement conservatives be the losers for a few decades.


Tim Worstall 07.02.08 at 1:21 pm

“Those few people who read both left wing and right wing blogs are considerably more likely to be left wing themselves;”

It’s the research that throws up the unexpected results that’s interesting. I find that I’m considerably likely to be left wing, something that’s certainly unexpected.


Henry 07.02.08 at 3:47 pm

A very late response to chiasmus at 7. The issue of the 64 character limit is less consequential than it sounds. The average list was 23 characters long; only 7% of responses were scraping up against the limits. So I think that there _is_ likely some degree of underestimation here – but that it isn’t that consequential. Here, people like chiasmus (and me) who read lots of blogs would appear to be the exception rather than the rule.


dfreelon 07.02.08 at 4:16 pm

You may want to compare your results to those of an early survey of political blog users by Johnson & Kaye:

Thomas J Johnson; Barbara K Kaye
Journalism and Mass Communication Quarterly; Autumn 2004; 81, 3

Always great to see new scholarly work on these issues, especially in political science.


Eric L 07.02.08 at 4:39 pm

Just to elaborate on Henry @30, some respondents just stop typing at the cutoff, but other respondents find ways around the 64 character limit, primarily through creative spelling and abbreviating. We didn’t calculate what proportion of respondents typed “The Huffington Post,” e.g., but I doubt that more than half did.


not even an mba 07.02.08 at 5:10 pm

Thanks for the elegant presentation of the Overton Window. Imagine a world where ABC is considered left-biased and Fox News is closer to centrist than CNN.


TLB 07.03.08 at 6:07 am

1. Almost all top bloggers aren’t really “conservative/liberal” so much as GOP/Dem. Almost all of them are more or less party hacks who, when the chips are down, do what those higher in their party tell them to do (either directly or through proxies). Maybe I missed it, but it seems that the paper is attributing to ideology what can be better explained by the suck-up effect.

2. Almost all top bloggers are also establishment hacks. See, for instance, the reactions of various Dem bloggers to the NAFTASuperhighway.

3. Many top bloggers are willing to suck up to the MSM even as they criticize it. See, for example, all those who tripped over themselves to go blog at CNN in 11/06.

4. I’ve had comments deleted and even edited without notice at at least two dozen blogs; do a search at my site to see what those bloggers were afraid of.

5. A partial explanation for the polarization is due to many bloggers regarding those who disagree with them as some form of non-human beings. That might come from the fellow commenters, and in some cases from the hosts themselves.

6. Some bloggers have a habit of getting things wrong (e.g., ThinkProgress), and many have a habit of just ripping and posting without doing any research or much thinking.

7. Compare comments #19, #52, and #71 from me with the post and the other comments:

8. Thank gosh I run sites, not blogs.


Steven Johnson 07.03.08 at 1:08 pm

Henry, really interesting study, and I’ve just had a chance to skim this morning so forgive me if I’m missing something…

It seems to me that the proper “old media” control for this should not be general news sources with a slight ideological bent, but rather openly partisan periodicals: The Nation, Harper’s, Weekly Standard, National Review, etc. (Or Limbaugh on the radio.) That’s where political junkies *used* to get their fix, and now we’re shifting over to the blogosphere. I’ve always assumed that those worlds were significantly more insular than even a polarized blogosphere, because it’s so much easier to pop over to an opposing viewpoint on the web than it is to, say, subscribe to the National Review. (Crosslinking may be rare, but it’s much more common than it was in the age of magazines, in that they didn’t have links!)

To me, the most accurate study would look at specific articles/posts read by people of different political persuasions. For people on the left who subscribe to the Nation and Mother Jones, how many read an article from National Review in the past month? And then for regular readers of DailyKos, how many read a post from Drudge or Powerline in the same period? My hunch would be that there would be much more crossover in the blogosphere than in old media, but it’s just that — a hunch!

Is there a reason why you did those “violin” charts comparing more “objective” news sources, and didn’t look at more explicitly partisan ones? That’s the apples-to-apples comparison, in my mind…

Anyhow, great stuff and a real contribution to this important debate…



Henry 07.03.08 at 4:50 pm


Thanks for your critique, which is I think a very plausible argument (and one which we may end up talking through in a footnote). I agree that there is likely more back-and-forth going on between blogs than between the traditional political opinion magazines.This is something we discussed when doing the research. The problem is, of course, that the dataset we were using didn’t have the right kind of data to do this (it also didn’t have data on newspaper readership, which would also have been very interesting to see). That’s why we did the charts that we did – if we had had more data, we would have loved to have done more comparisons.

To my knowledge, there hasn’t been any statistical research done on readers of these magazines, in part because the readership is small enough that you are unlikely to find enough readers for statistically significant results in the kinds of polls that political scientists do. You could, of course, draw respondents from subscription lists or the like with the cooperation of the magazines’ publishers, although you would then have problems, obviously, in drawing comparisons between the readers and the general population.


b-psycho 07.03.08 at 6:12 pm

I think one thing that may be of further interest in the context of blogs & polarization is where the people that don’t quite fit the traditional left-right spectrum tend to go. There are people that are either social conservatives with a populist streak on economics or civil libertarians with anti-state fiscal views, and I do wonder what wins out in their cases due to the filing at the edges.

For a quick example: I myself am a libertarian, a radical one at that. Yet other than some blogs of fellow radical libertarians, I mostly frequent liberal sites, and my intake of “conservatives” these days consists of Andrew Sullivan & Daniel Larison.


Britt 07.06.08 at 2:35 am

Problem is, a right winger who reads the Post every day and then logs on to NRO has just covered both sides of the issue. The mistake I think many on the left make is thinking the media is a neutral source. Reading both the New York Times and the LA Times is not covering both sides of the street.

I’m a right of center person, and the fact is that I read lefty blogs for entertainment, not for information. Reading a DailyKos diary called “Barack Obama: Human or Deity?” is just plain fun. For serious leftist thought, I go to the editorial page of a major newspaper. For serious conservative thought, I go to National Review. For rants and paranoia, over to the Democratic Underground and Free Republic (I know, not blogs per se).

That’s what I think your study is missing. Because while many conservatives may not read liberal blogs, we do watch NBC/ABC/CBS/CNN/MSNBC, listen to NPR, and read the NYT/WaPo/LAT/ChiTrib/TIME/Newsweek. Who really gets their opinion’s challenged on a daily basis? The dextrosphere and sinistrasphere are different. One serves as media watchdog more then anything else, and the other serves as fundraiser and soapbox.

Comments on this entry are closed.