1 Introduction

The vision of the Internet as a common space stands in stark contrast to the enclosed and controlled space of intellectual property which is being constructed by existing (primarily corporate) interests. One of these common spaces is the so-called “blogosphere” of blogs. These online journals and columns have been touted as solutions to the concentrated ownership, limited diversity, and commercial and political pressures affecting mainstream media. But do the millions of blogs of the blogosphere offer a viable alternative for news and analysis, or do the discussions in blogs take place only between those of like mind?

In this study, I attempt to address the role of blogs by examining the ways in which a sample of 120 blog posts deal with a single controversial political issue: same-sex marriage. Using content analysis, I have recorded a number of features of these posts and the Web pages they link to, in order to address the following questions:

I have also analyzed three blog posts in detail to illustrate and expand on the results provided by content analysis.

2 About Blogs

Wikipedia (as of 18 February 2005) defines a blog as “a web application which contains periodic posts on a common webpage”. These posts are like articles; the blog itself like a newspaper, with posts typically listed in reverse chronological order. Within this definition, blogs are diverse: a private diary, a mainstream newspaper, or a sequence of photographs could all qualify. Furthermore, blogs vary in their form: some are simply lists of links, while others include multi-page articles.

Beyond their chronological organization, several features of blogs distinguish them from other media. First is the use of hyperlinks: blogs often link to articles that they are written in response to, or which provide more information about a subject. In turn, some blogs provide “trackbacks” – reverse links to posts which have commented on the current post. For those that lack this facility, blog search engines, such as Technorati (www.technorati.com) are capable of searching for posts linking to a particular page.

Second, many blogs allow comments, which allow readers to respond to a post. These are then displayed along with the post, either on the same page or on a linked page.

Third, many blogs support syndication technology1. This allows subscribers to be notified automatically when a blog is updated with new or changed posts, and to view the text of those posts stripped of extraneous formatting (such as the graphics and menus typical of Web pages). It also provides the additional information used by blog search engines to keep up-to-date with changing blogs. The rapid feedback provided by syndication and search engines makes it possible for bloggers to have back-and-forth conversations in a brief period of time.

Finally, the technology for creating blogs requires little technical know-how. Free commercial packages and online services make creating a blog as easy as using online email. These packages include standard support for features such as trackbacks, comments, and syndication.

3 Background

Effective news media are considered an essential part of a democratic society. But there is a sense that the mainstream media are failing to fulfill that role. Robert McChesney, for example, proposes three criteria for democracy: no vast differences in inequality, a sense of community, and an effective system of political communication. He argues that “the commercial basis of U.S. mainstream media . . . undermines all three” (McChesney, 1997, pp. 5-6). Hackett and Zhao (1998) criticize the media for failing to provide a “space for civil public discourse” (p. 188).

For their advocates, blogs offer a remedy for many of the failings of the mainstream media. The proponents argue that blogs create that space for discourse, engaging readers in the process of constructing news and analysis. For them, this is a new, more democratic medium. Bowman and Willis (2003, p. 9) consider blogs part of a wider movement towards what they call “participatory journalism”, whose bottom-up structure contrasts with traditional media. They suggest that Nicholas Negroponte's (1995, pp. 152-154) prediction of a personalized newspaper, which he called “The Daily Me” is being realized, but in a collective form they term “The Daily We” (Bowman & Willis, 2003, p. 7).

McChesney (1997, pp. 30-34) and Hackett & Zhao (1998, pp. 190-200) have expressed skepticism of the ability of the Internet to escape commercial domination. Feenberg & Bakardjieva (2004, pp. 1-2) suggest an alternative to the commercial model of online communication, which they term the community model. A number of characteristics of blogs do distinguish them from commercial media. Unlike mass media, the ability to publish is distributed rather than centralized, and readers can respond through comments and links. While Frost (2003, pp. 21-22) has cited good grounds for arguing that the Web fulfills Innis's (1991) criteria for a space-based medium, blogs have a much stronger sense of time and history than commercial sites online – for example, articles are always dated and archived, and blogger etiquette frowns on changing existing posts without noting the changes (I have expanded on blogs' use of time elsewhere – see Glass, 2004). Stephen Downes (White, 2005) has pointed to the importance of this temporal aspect, explaining that he returns to topics repeatedly over time in his blog postings, with different posts presenting different perspectives on an issue.

But if bloggers communities, might those communities isolate themselves from each other and from the wider society? The suspicion that personal forms of Internet-based communication will isolate citizens from the public sphere has not gone away. Manuel Castells has written of his concerns about “networked individualism” (2001, p. 131). Cass Sunstein (Johnson & Bimber, 2004, p. 250) fears that Internet debate leads to polarization and extremism, something Rogers Putnam calls “cyberbalkanization” (Castells, 2001, p. 132).

Balkanization would not be surprising: Rogers and Kincaid's (1981) analysis of personal networks has found that most interpersonal connections are homophilous, which is to say most people communicate with people similar themselves. Yet the “information-exchange potential of dyadic communication is related to the degree of heterophily between the transceivers” (Rogers and Kincaid, 1981, p. 142). In other words, new information often comes from people who are different.

Wilhelm has studied political debate on Usenet online bulletin boards (Wilhelm, 2000, pp. 86-104). He applied content analysis to determine (among other things), “to what extent is there in-group homogeneity of political opinion on Usenet newsgroups?” Wilhelm's study examined high-level properties of Usenet posts, such as homogeneity with the overall tone of the newsgroup, whether the post replied to a previous post, and the number of words in a message. He found that within a newsgroup, over 70 percent of the posts were homophilic – “demonstrating either strong or moderate support for the dominant position on a political topic or candidate”.

More recently, Adamic and Glance (2005) studied the degree to which left- and right-wing blogs linked to each other during the 2004 U.S. election. Examining over 14,000 blogs, they found that “liberals and conservatives linking primarily within their separate communities, with far fewer cross-links exchanged between them” (2005, p. 14).

Yet newsgroups are centralized, blogs are not. And the Adamic and Glance study was focused on the two camps in U.S. politics, not the wider blogosphere. Their sample was deliberately composed only of popular political blogs that fell either into the liberal or conservative camps; moderates and other political positions were explicitly excluded (2005, p. 3). They did not address the possible middle ground that might exist in the less popular blogs, or in those that were not explicitly political or politically aligned. Might a different analysis obtain a different result?

4 Method

I chose same-sex marriage because of its polarization: if there is a middle ground here, there is more likely to be one in less contentious issues. I am personally in favor of same-sex marriage, but I have endeavored to be even-handed in this study; my goal is to draw out the strengths and weaknesses of blogs. I believe they do offer an valuable alternative to the failures of existing media, but I am concerned that the promises of previous technologies have gone unfulfilled. I want to understand the strengths, but also point out their flaws in hopes that these can be remedied.

Same-sex marriage provided the fulcrum for understanding; content analysis provided the means to efficiently and non-intrusively examine a relatively large sample of blogs. As Wilhelm (2000, p. 91) points out, content analysis allows for examination of messages in context: “it is not necessary to know who the participants are, from what walk of life they come, or with what political parties they are affiliated.” On the other hand, content analysis captures only one part of the discourse: it provides few clues about the behavior of readers or how they construct meaning from what they read. Also, content analysis is unable to determine causality (see Strauss, 1998, p. 125). If blogs are “walled gardens”, content analysis cannot explain whether this is a cause or effect of political isolation in the larger society, or even whether there is a connection at all.

Categorizing the two obvious stances – for and against – is not sufficient to capture more ambiguous discourse. Following Gamson's 1992 work on action frames in political talk, I chose to examine frames (positions and arguments) within individual blog posts. These can capture more nuanced positions, but they also provide a measure of overlap between the arguments and points of view in linked pages.

In addition, I have selected three posts for closer textual analysis in order to characterize some of the features of a typical article, and to capture some of the diversity that does not show up in the quantitative measures of content analysis.

4.1 Sampling & Data Collection

The unit of analysis for this study is not the blog, it is the individual blog post and the neighborhood of other resources and comments linked to the post. This is necessary for two reasons: first, blogs can contain thousands of articles, making it impractical to analyze an entire blog; second, not all blogs do contain thousands of articles, and so are difficult to compare. Each post contains content, but also what I will call the post neighborhood – comments added by readers of the post, and the other web resources, such as media stories and other blog posts – “articles” – that it links to.

The samples were selected by searching for “gay marriage” on Technorati, a blog search engine (www.technorati.com). The choice of search phrase is not neutral; some people see “gay marriage” as pejorative compared to “same-sex marriage”. I have picked this phrase because it is more widespread: as of 23 February 2005, Google lists approximately 1.7 million incidences of gay marriage in Google vs 0.9 million incidences of “same-sex marriage”; Technorati's results are even more pronounced. The biases inherent in the term are less relevant because same-sex marriage itself is not the subject of this study, only an issue to focus discourse.

The search for “gay marriage” was performed ten times it Technorati, once each for a random date in January or February 2005. The random selection of dates is intended to minimize the effect of any particular political event. Each of these ten “blocks” consists of the first twelve search results (starting at a random point on that date) matching the following criteria:

For search results which did not qualify, URLs and the reasons for disqualification were recorded but no further analysis was performed.

4.2 Articles

The recording unit is the article, which corresponds to a blog post, a set of reader comments, a media story, or another web page. Each post in the primary sample (resulting from the Technorati search) was a recorded as a single article. In addition, I recorded the post neighborhood, or secondary samples – that is, other content connected to the post. This included comments written within one month of the post. Furthermore, up to five outbound hyperlinks were treated as additional articles. These only qualified if they contained at least one frame, and if their content appeared to be static (e.g. frequently changing media or blog home pages were excluded). If a post had more than 5 qualifying links, the first 5 were taken, although the last link was considered first (sometimes the last link has special significance). I only followed the first link from an article; further links, even to a second page of the same story, were ignored.

For each article, a number of fields were recorded, including: its URL, the date given on the post, the country of origin, the gender of the blogger, the political orientation of the blogger (left or right), the stance regarding same-sex marriage (for or against), the prominence of same-sex marriage in the post, the tone of the post (e.g. polemic, diary, reportage), the mode of address (e.g. community, public), the number of outbound links, the number of reader comments, the prominence of block quotes and their sources, a measure of post length, and a list of frames. Not all fields were relevant to all articles: for example, most do not apply to a set of comments. The protocol is detailed in Appendix 1.

This information was based on the text of each article. For blog posts, newspaper stories, and similar articles, surrounding boilerplate such as menus, banners, advertising, graphics etc. was ignored. Text in the form of images within the article body was considered part of the article text. The surrounding content is excluded for three reasons: 1) it is often irrelevant; 2) it is almost always automatically generated, so it reflects no intent on the part of the author; 3) many readers ignore may ignore it, and it is not present in the syndication feeds through which many readers view blogs.

4.3 Frames

One of the most important, but also one of the most complex items of information I recorded for each post was frames. Following Gamson, and Krippendorff's description of contingency analysis (1980, pp. 114-115), I recorded only the presence or absence of frames in individual posts. Counting might give a more precise measure of the weight a post gives to a particular frame, but is much more time consuming and poses difficult questions about reliability and the appropriate unit of measurement.

Gamson's analysis was concerned with the types of frame present, e.g. action frames, weak vs. strong frames, etc. I am not concerned with the issues per se, but rather with the distribution and commonality of these frames between blog posts. I have endeavored to choose the frames according to three criteria:

  1. The set of frames should cover the subject. If a post argues for or against same-sex marriage, at least one frame should be present.

  2. Frames should be as distinct as possible from each other, partly to improve coding reliability, but equally importantly so that they reflect genuine divides is the politics of the issue and hence provide a good measure of common ground between posts.

  3. To aid in #2, frames should emerge from the data rather than be imposed. These are likely to be more clearly defined for a polarized topic, like same-sex marriage.

  4. They should be relatively small in number to improve coding reliability and efficiency. On the other hand, Krippendorff (1980, p. 115) describes the practice of measuring many attributes (in this case frames), then clustering the resulting data. I found in practice that a larger number of frames were easier to distinguish if they corresponded to patterns in the data (as in #3), but that combining them into logical groups increased reliability and made the size of the set more manageable.

A precis of each frame follows; full details are in the protocol in Appendix 1. Each frame has an implicit stance on same-sex marriage, which is noted in parentheses.

5 Results

The primary sample consists of 120 blog posts obtained from Technorati. These link to 87 articles with frames in the secondary sample, and 18 of them have comments with frames. I examined a total of 613 articles. Some were excluded for lack of frames, some were inadvertently analyzed but rejected because they exceed the limit of 12 posts per sample block, some were broken or incorrect links, and some were duplicates or repeat occurrences of blogs already in the sample.

An intra-reader test of reliability for a random sample of ten blogs from the primary sample (applied approximately two weeks after data gathering was complete) found that most recorded data was reliable, but there were problems with the measurement of frames. Reliability measures are for the combined values used for analysis, not of the raw values recorded (see Appendix 1 for explanations of the difference for each measure).

Table 1 - Reliability





10 / 10


9 / 10


10 / 10


10 / 10

7 / 10 for raw values (due to the fuzzy distinction between polemic, analyis, and rant, as noted in Appendix 1)

mode of address

9 / 10


27 / 31 frames

for posts, the number was 6 / 10 (i.e., the 10 posts had 31 frames total, of which 4 had one frame too many or too few)

5.1 Demographics & News Context

5.1.1 American Dominance

I chose the topic of same-sex marriage partly in hopes of obtaining data from Canadian bloggers. The subject was topical in the national media at that time, while in the U.S. election, in which same-sex marriage had been an issue, was over. Despite this, the sample is overwhelmingly American: I was able to confirm that 73% (87) of the bloggers were in the U.S, while only 8% (9) were in Canada (one U.S. blogger was a Canadian, another was originally from India). I was able to identify the location of only two others – one in Ecuador, one in Spain – both of which were American. The proportion of Canadian to U.S. bloggers mirrors the relative populations on the two countries. The lack of reports from elsewhere might be explained by two factors beyond the anglocentrism of the Web: first, the search phrase was in English; second, same-sex marriage is a current issue in the United States and Canada. As a result, this study provides no evidence for international diversity online one way or another.

5.1.2 A Slow News Period

Two political events are referenced repeatedly in the sample, and are central to several posts. The first was a ruling on 4 February by the a New York State judge who, in his ruling on an action by a gay-rights group, found a ban on same-sex marriage in New York unconstitutional. The second was a statement by President Bush in an interview with the Washington Post on 15 January, in which he said that there was no urgent need to pursue constitutional change to prevent same-sex marriage. A number of other stories are mentioned more than once, including pope John Paul's claim that same-sex marriage was an “ideology of evil”, and several childrens' television programs in the U.S. which depicted or were seen to depict same-sex couples. Overall however, no particular stories dominate the blogs.

5.1.3 Mostly Men

The sample is dominated by men. Of the authors of the 120 posts in the primary sample, 25% (30) are female and 53% (63) are male. The gender of the remaining 22% (27) is not clear, although three of these seem likely to be female. It is possible that women would choose not to identify their gender online, but it is unlikely that would account for more than a fraction of the gap.

5.1.4 Liberals

A startling trend in the sample is the preponderance of left-wing views. In regards to same-sex marriage the ratio of for to against is 3-to-1: 48% (57) of the posts are clearly in favor but only 15% (18) clearly against. A further 8% (9) appear to be in favor, and 2% (2) more against; 2% (2) have explicitly mixed views on the issue. The position of the bloggers in the remaining posts is unclear.

The evidence for political alignment is weaker, but a similar pattern holds: 29% (35) are left aligned, while only 14% (17) are right aligned. A further 9% (11) appear likely to be left, while just 1 other is likely to be right. Two are explicit centrists.

Given these two statistics, there can be little doubt that the left outnumbers the right in the sample. Yet blogging is not portrayed as a left-leaning activity. Adamic and Glance (2004, p. 3) found that their sample was nearly balanced (676 liberal blogs vs. 659 conservative). One possible explanation is the different sampling technique. Adamic and Glance explicitly sought out lists of popular left- and right-wing bloggers. This would tend to bias towards equal populations of each, and could fail to represent a larger number of blogs on one end or the other of the political spectrum, especially if the majority of those were less popular. It would also fail to capture the political opinions of those whose blogs were not concerned mainly with politics.

Chris Anderson (2004) proposed a theory of the “long tail”, which emphasizes the impact of less popular media. For example, he states that over half of Amazon's sales are from books outside their top 130,000 titles. If blogs follow a similar pattern, then it is not sufficient to examine the most popular ones in order to gain a clear picture of the blogosphere.

There are a number of reasons why the left, and particularly same-sex marriage advocates, might be better represented in the blogosphere. These include the tendency of users of newer technologies to be younger, better-educated, urban, and middle class. Several of these correlations are supported by a Pew Forum poll (Pew Forum, 2003).

5.2 Communities of Discourse: Mode of Address

Mode of address gives an indication of who a blog author believes the audience is. This measurement is somewhat loosely-defined, as it is derived largely from implicit assumptions in posts (see Appendix 1). Such evidence is present in the majority of posts in the primary sample:

Table 2 - Mode of Address

mode of address:












One of the posts in the community category was a hate-filled diatribe addressed to gays. All others strongly implied membership of the blogger in the community.

In this sample, most of the blogs take part in community discourse rather than public discourse. However, those communities are not necessarily defined by the issue of same-sex marriage. Furthermore, the actual audience may not match the blogger's expectations.

5.3 Linking to Opponents: Walled Gardens

One way of measuring the degree to which blogs connect with different view points is to examine hyperlinks: do they link to other opinions? For this sample, there are two answers. At the crude level of whether blogs links to opponents, the data is limited but points to no. But at the more nuanced level of of frames, the picture is different: many of the posts in the sample do connect, in hyperlinks or comments, with different opinions and arguments.

First, a proviso: although the majority of the posts in the primary sample do contain links, fewer than half (42%, or 51 of 120) contain links. These 51 posts link to a total of 87 articles (containing frames) in the secondary sample, and only a fraction (20%, or 20 of 120) have comments with frames. Furthermore, trackbacks with links are so rare in the sample (there were only seven) that I excluded them from analysis. This is disappointing, as they are one of the features that sets the medium apart (Bowman & Willis compare them to “an editorial page of commentary on the Web, automatically generated to appear alongside a story” (2003, p. 23)).

Now for the granular measure, whether blogs link to articles with opposed attitudes to same-sex marriage. In the following correspondence table, the stance of the source blog from the primary sample is on the left, the stance of the linked article in the secondary sample is along the top. Each cell contains a number of links between a source and destination with the given stances. For example, of the 21 posts in the primary sample which favor of gay marriage, and which contained links to articles with at least one frame and a stance of for, neutral (or indeterminate) or against, 13 linked to articles which were also in favor.

Table 3 - Stance vs Linked Stance




# of source posts
















# of target posts




Articles with ambiguous stances (for? or against?) have been excluded; they make little difference, as all but two of them link to neutral targets. Note also that each blog can have more than one link, so the sum of the columns may exceed the total. Here is a graphical representation of the same relationships:

Figure 1 - Stance vs Linked Stance

TThe data is too limited to draw general conclusions (especially given that only 8 of the primary sample are against), but this does follow the pattern found by Adamic and Glance: bloggers are much more likely to link to those with whom they agree than with those with whom they disagree. The overall ratio of links in agreement to links opposed is 16-to-5, or about 76%. This is close to Wilhelm's finding of 70 percent homogeneity for Usenet posts, but lower than the result in Adamic & Glance's study, which found that 91% of the links from a community remain within that community. In addition to reflecting the intention of the blogger, this also influences the experience of readers of the blog who follow links.

The numbers also reveal a pronounced tendency of the posts in the sample to link to neutral articles. This corresponds to a high rate of linking to mainstream media stories (whose tone is primarily objective). I will explore this further in the section on the relationship to the mainstream media.

5.4 Extra Frames from Links and Comments

While the comparison of stance suggests that links to opponents are rare, that is a very coarse measure of debate. Worse, it provides no useful information about those posts whose stance is indeterminate or neutral, but which made up a significant portion of the sample. Here I will instead examine frames – in particular, whether posts in the sample are connected (via links or comments) to “extra” frames, that is, frames not present in the source post. This reveals that blogs in the sample play a minor role compared to mainstream media, comments, and other linked articles.

For the measure of extra frames to be relevant, the post in question must deal with same-sex marriage in some detail. Therefore, this portion of the analysis only includes the 90 posts in the primary sample for which the focus on same-sex marriage is central (it is the subject of the post) or related (it is dealt with in some detail), and exclude those for which it is incidental.

The following chart illustrates the number of posts from the 90 in the subsample meeting certain criteria. Values in the “All” column indicate the number of posts with links or comments, regardless of whether or not those links or comments contain any frames (this data is incomplete because only links with frames were analyzed in detail). The “Frame” column contains the number of links or comments containing at least one frame; the “Extra Frame” records the number of links or comments containing at least one extra frame. Each row further qualifies the count according to the nature of the linked article. For example, of all 90 blogs, 57 contain links, 46 contain at least one link to an article with a frame, and 32 contain at least one link to an article with an extra frame.

Table 4 - Number of primary posts (out of subsample of 90 concerned with same-sex marriage) having a link or comment, a link or comment containing a frame, or a link or comment containing an extra frame.



Extra Frame

link or comment

81% (73)

60% (54)

43% (39)


63% (57)

51% (46)

36% (32)

comment or non-media link

43% (39)

31% (28)

non-media link

32% (29)

22% (20)

comment or blog link

30% (27)

21% (19)

media link

28% (25)

17% (15)

non-media non-blog link

20% (18)

14% (13)


36% (32)

20% (18)

12% (11)

blog link

17% (15)

10% (9)

Note that the “blog” category includes links to the source blog; there are three such cases which appear in the sample, and all contain extra frames. Here is the same information as percentages (of the 90 posts in the relevant/central subsample):

Figure 2 - Frames & Extra Frames

This shows that, in the sample, most (81%) of the blogs discussing same-sex marriage contain links or comments – in fact, most contained links or comments relevant to the subject (the 60% containing frames). However, the number with links or comments adding extra frames is less than half – 43%. If only links are considered, this number drops to 36%. This is a fairly large percentage: over a third of blogs provided access to other resources with alternative frames on the subject.

There are four basic categories of other resource: comments, links to mainstream media articles, links to blogs, and links to other resources (these included alternative media, religious sites, NGOs, etc.). Of these, mainstream media provide more additional frames, followed by other articles, then comments, and finally blogs. However, the disparities are not huge: the source providing the most extra frames, the media, provides slightly more than half again as many as blogs.

There are a few caveats with this analysis. First, not all frames are equal; certainly this is apparent on a per-frame basis, which I will detail later. Second, there are more against frames (7) than for or neutral frames (5). So articles with many extra frames may be more likely to be opposed to same-sex marriage. Third, several media stories were broken or required payment, so are not included in the sample. Fourth, some frames counted as being part of a post are actually in quotes from other sources. As the next section shows, that suggests that the media's contribution of extra frames is underrepresented in this analysis.

5.5 Echoing the Mainstream Media

The prominence of links and quotes from mainstream media sources in the sample is striking. Of the 51 posts in the primary sample with links to sources with frames, 53% (27) have links to mainstream media sources. This compares with 27% (14) which link to other blogs. The phenomenon is even more pronounced for excerpts from other sites: of the 52 posts containing block quotes, 63% (33) contain block quotes from media sources, while only 20% (18) include block quotes from other blogs. Figure 3 shows the dominance of links to media sources.

Figure 3 - Proportion of posts with links or quotes to particular sources, as a percentage of all posts with quotes or links.

For comparison, Adamic and Glance (2004, pp. 8, 10) found that the top 40 left- and right-wing blogs in their sample linked to mainstream media “about once every other post”. However, they had comparatively more links to mainstream media: the ratio of blog to media links was 28%, compared to 52% in my sample. Note, however, that I measured links differently, as I excluded those that linked to articles lacking frames.

The tone of the stories reveals a similar strong connection to the mainstream media. Nearly a fifth of the blogs are simply passing on the news: the tone of 18% (22) is reportage, meaning that the blogger did not include any analysis or commentary. Of the 33 posts that did quote the mainstream media, fully 61% (20) consist either mostly or entirely of quotes.

In other words, over twice as many posts in the sample are reacting to or commenting on mainstream media stories as to other blog posts.

5.6 Frames in Blogs and the Media

Blogs and the mainstream media do not give the same prominence to the same frames. Because of their structure, and the regime of objectivity described by Hackett and Zhao (1998), it is unsurprising that blogs would present information and argue differently. As I have shown, however, through quotes and links the media form an important foundation for many of the posts in the sample. If blogs have something else to offer, it could show up in a comparison of frames used in blogs and the media.

The following correspondence table shows the occurrence of frames in the media and in blogs in the primary sample with for and against stances (see the Method section above or Appendix 1 for an explanation of each frame). The bottom row shows the frame counts for all blogs in the primary sample.

Table 5 - Frame Occurrence
































































































Here is the same information in graphical form, sorted by the rate of occurrence of each frame in the media:

Figure 4 - Frame Occurrence

The bottom (media) portion of each bar forms a curve decreasing to the right. Where the bars as a whole go against this trend – where they spike – indicates an frame which is more prominent in the blogs – for and/or against – than in the media. The chief examples of this are RELIGION and MORALITY. There are two other points of particular interest: LOVE and PRIVACY. LOVE, for example, receives nearly identical coverage from both the for and against camps, but is virtually ignored by the mainstream media. PRIVACY receives some coverage from both camps, but is completely ignored by the media. I must caution that these sample sizes are very low, particularly in the case of against, which has only three posts showing the LOVE frame. What is most interesting, if this could be shown to be representative, is that on these two issues – and possible in the case of RELIGION and MORALITY also – the two camps seem to be reacting to each other's views more than the media are.

There are areas where the for and against camps are do not cover each other's positions. The against posts, for instance, do not address BENEFITS, CONSISTENCY, or COMMITMENT at all, and give short shrift to EQUALITY and PRIVACY. Similarly, the for camp does not address the SLIPPERY SLOPE argument, and largely ignores the EQUALITY (Q) position as well.

There are several important caveats with this data. Again, the sample size is quite small – especially for the against blogs, of which there were only 18. The numbers should be most reliable for the frames with the most references, but for those with very few (e.g. SLIPPERY SLOPE), they should be taken with a large grain of salt.

Second, not all media stories were completely analyzed: the newspapers in question required payment, so only an abstract was used for recording. As I have mentioned above, this does however represent accurately what most readers would see.

Third, the media stories are not a random sample: they are the stories linked to by posts in the primary sample. Primarily these are news reports with an objective tone. Subjective media stories, such as columns, editorials, investigative reports, and letters to the editor are rare. On the one hand this makes them very different in tone from the blog posts; on the other, because these articles are chosen by the bloggers they may be more likely to frame the issue similarly.

Finally, as mentioned before, many of the frames present in the blogs are actually from excerpts from media articles in the form of block quotes. This means that in many cases, frames from a media article also appear in the post that references it.

These last two points may serve to make the frames used by the blogs more similar to those used by the media articles than might be the case in a different sample of media stories. This is appropriate for studying the relationships between bloggers and the media they reference, but it also serves to further emphasize that these results should not be be taken as a comparison between this sample and media coverage during the period of the study.

5.7 Diversity: Adding to the Conversation

The mainstream media follows a number of practices which serve to reduce the diversity of reporting. For example, practising objectivity, they typically favor of straight reporting of facts and quotes, and balance between sides of an issue. Opinions and analysis are excluded or separated from objective news. As Hackett & Zhao (1998) make clear, true objectivity is unattainable: facts must still be selected and are seldom themselves truly objective; neither are specialists and spokespeople interviewed by the media. Furthermore, the sides of an issue may not be equivalent, something which is further concealed by exclusion of analysis, etc.

The media articles in the sample clearly demonstrate the practice of objectivity: of the 33 mainstream media articles in the secondary sample, 76% (25) present an objective tone and only 2% (2) take a position on the issue of same-sex marriage. The dominance of mainstream media sources in the sample suggests that most blogs are not constructing alternative discourses. But content analysis forces all measurements into categories; those that fall outside the categories are not represented. If the dominant discourses are homophilic, the middle ground is more likely to appear in posts that are more diverse and less suited to content analysis. In the course of my recording, I noted approximately ten blogs that seemed to me to have something to offer which failed to fit into quantitative categories. These range from relegious exegesis to racist anti-semic hate to personal experiences and mixed views about same-sex marriage. These diverse posts present viewpoints and analysis not present in the mainstream media. What follows are analyses of three of these: one post chosen for being statistically average, and two divergent ones.

5.7.1 A Walled Garden: Sample S10-21

Although no one post is representative of the majority of posts in the sample, S10-21 is fairly typical. Statically, it has numerous features in common with many other blogs (see Appendix 2 for the text of the post). It is written by an American. The subject is one of the main news stories surrounding same-sex marriage in the sample, namely the Bush administration's deemphasizing of the constitutional amendment. The article quotes mainstream news sources extensively, is of moderate length, and it presents a political opinion. The main aspects of the article captured by content analysis were the stance of the poster (in favor of same-sex marriage) and the use of two of the most common frames, RELIGION and GOVERNMENT. Both of these frames are neutral, which is interesting in itself: the author does not aim to argue for same-sex marriage. In this, and in its isolation from opposing points of view, it is not uncommon.

The article leads with a headline – “'Red' Christians Get Their Pig In A Poke, Promise Payback On Plans To Gut Social Security”. Following this, it links to the L.A. Times and excerpts four paragraphs, immediately positioning itself as reaction to a mainstream story. The article quotes five more paragraphs from the New York Times, so that media quotes make up the majority of the article text. The link to the LA Times is broken; the New York Times link leads only to an abstract – a fee is required to view the full article.

The thrust of the article is an analysis of the political situation of the Bush administration (the article is somewhat exceptional in that it takes a sarcastic rhetorical tone). The mode of address shifts: some comments are about Karl Rove (a political advisor in the Bush administration): “How does it it feel to a 'Red' Christian to be played for a fool? Uncle Karl wonders as he chuckles”. Others are addressed directly to him: “What a fine mess, Karl”. Rove is set up as an opponent and a representative of the Republican administration. The article is clearly addressed to left-aligned readers who, like the poster, are anti-Republican and at least tolerant of same-sex marriage. This symbolic use of Rove also indicates that the audience is assumed to follow U.S. politics closely; they are implicitly American.

Although same-sex marriage is central to the article, it is not actually the subject of analysis. It is instead one of two political issues presented as the rock and the hard place that have trapped Rove, and therefore the Bush administration. The article exults in the situation, concluding: “Let mine adversaries be clothed with shame, and let them cover themselves in their own confusion, as with a mantle (Ps 109:29)”. This use of Biblical quotation both begins and ends the commentary by the blogger, and supports the article's attack on “'Red' Christians”.

This is not the only sarcastic use of quotes: the words “compassionate conservative” are also quoted (which echoes the common use of quotes around “marriage” by opponents). Similar devices include renaming the Defense of Marriage Amendment (DOMA) to the Hate Amendment, and of the Social Security reform proposal the Wealth Transfer To Wall Street Act. These are the two issues which the poster claims have forced Rove into admitting dishonesty. That dishonesty is implicit in the name changes, the use of quotes, the suggestion that “'Red' Christians” have been “played for a fool” by Rove, that he will “blow [his] . . . cover”, and that his “duck will start quacking big time”. The “'Red' Christians” are further portrayed as “sheep” who could be told to “blindly follow [Rove] off a cliff” if their pastors tell them to.

In sum, this article exists in its own walled garden, aimed only at those who already agree with the poster. It is an exultant report of the misfortunes of opponents, who would likely be offended by its tone and unmoved by its arguments. The five links to like-minded blogs further emphasize this community of thought, as does the single trackback from another blogger who agrees with the poster. Only one additional frame (albeit TRADITION, against) appears among the linked blogs.

5.7.2 Rational Analysis: Sample S3-14

Sample S3-14 is one of a handful of analyses in the sample which take an intensely rational or academic approach to the issue. Other similar posts largely argued based on religious exegesis and principle; this one combines philosophical and lexicological theory with mathematics and context (history, law, and personal experience) to explain how definitions are constructed to have certain meanings, and how this applies to the definition of the word “marriage”. The direction of the argument – in favor of allowing same-sex marriage – is clear from the outset, but the article restricts itself to evidence and logic until the final paragraph.

The article seizes on a few of the elements of the argument against same-sex marriage and analyzes them in detail. These include the frequent use of quotation marks around the word “marriage” when it is between two people of the same sex and “the idea that defining marriage to include same-sex couples would be like defining the word 'dog' to include 'cat'. The argument is relentless, bypassing other issues in the debate, such as religious values, morality, or the impact on society at large. The objective of the article is to make a single point, and to make it clearly. The logical progression shows that the definition of marriage is constructed, that it can be changed, and finally, in the last sentence, that the author believes that it should be changed. Here is an abbreviated version:

“[Richard] Robinson [a philosopher] uses stipulative definition to 'mean the explicit and self-conscious setting up of the meaning-relation between some word and some object, the act of assigning an object to a name (or a name to an object)” . . . the goal [for a mathematician] is not to arrive at a certain Truth, but rather to figure out which definition will be most useful in communicating certain concepts . . . It seems to me that when a definition is put into a statute it is also stipulative . . . So when I hear the analogy about defining a “dog” to include a “cat”, I am undisturbed. The law does things like that all the time. . . . In 1985 the Supreme Court declared unanimously that Long Island was not legally an island . . . Maggie Gallagher says that marriage was not and cannot be created merely by law. Even if that were true, we are concerned with what the legal definition of marriage should be, and that most certainly is created by the law.

Like other similar academic arguments on both sides, this posts contributes something that is not present in the media (this article is 1200 words, beyond what most papers would print), and is rare in this sample. Although the statistics show that this article has both for and against frames, and that extra frames are present in reader comments and in a link to a poll, the strength of the argument does not appear in the content analysis. It does not isolate itself from its opposition, but accepts it and addresses the arguments clearly. It does not demand agreement, but argues for it. The author concludes:

I am sympathetic to those that are reluctant to embrace same-sex marriage because to them that is not what marriage is. Change is scary, and it will take some time for people to get used to the idea of two people of the same sex being married. I will, however, continue to push for this change because I think it is unjust to do otherwise and I believe more and more people will come to accept it before long.

By applying philosophy rather than politics, by addressing opponents, and by expressing sympathy for them, this post has found a valuable middle ground not present in mainstream media discourses (neither, if this sample is representative, is it common online) about same-sex marriage.

5.7.3 Human Empathy: Sample S5-27

Sample S5-27 does something quite different. Where S10-21 creates walls, and S3-14 applies reason, S5-27 creates empathy. It is not unusual in that it is a post with a diary tone which only mentions same-sex marriage in passing. But it demonstrates perhaps more clearly than any other post in the sample how a diary tone can serve to humanize people, even if they disagree.

The title of the post is “A theme for Sunday”. It is by a 26 year old man living in New Orleans who aspires to be a writer and recently started a new blog. This post explains some of his reasons, and his expectations for his audience. He says changed to a blog from another service because:

No one except the people I had written about would read it. It would just be a resovoir of feelings that I might as well keep to myself. . . . I believe that Blog's with their infinite cybertext and unlimited audince, is a more perfect fit . . . if I wrote these kind of things in BP [the other service] it would lose some of it's integrity because BP is mostly to perpetuate meeting people, where as the blog is pretty much confined to people who already know how much my friends and family mean to me so it's not as if I'm trying to impress them with anything. [sic]

The next sentence, the only mention of same-sex marriage in the article, references to the MORALITY and RELIGION frames:

It would be kind of like me running for public office and talking about how I'm anti Gay marriage because it's against God's law...I may actually feel that way but it could be percieved as though I'm saying it just to gain favor of a certain voter audience.

But the real subject of the post, and what makes the writer so human, is his “theme for Sunday”: his love for his younger sister. He writes about his childhood with her as though it were a memory of a mythical golden age. He “used to brush her hair in her sleep. Sometimes she would be irritated and turn her head so I would brush the other side. . . . For a long time whenever I would lean to kiss her cheek in her sleep she would smile. Not so many years later I would play the same game except she would frown.” As he mentioned before, he expects this will be read by his family and friends, and says he is “a pretty different person around my family than I am around the rest of the world.” He claims he is cynical, yet he concludes:

Whenever I think of how big a screw up I am I think about myself in relation to Saia [his sister] and how much of a triumph she is. And I don't feel like a screw up. I feel like I'm part of some special achievement.

I think he sells himself short. I also think that it's writing like this that has the potential to foster respect and break down the barriers between political opponents: this man is no “'Red' Christian” sheep. Hossein Derakhshan (2005) made a similar point in a presentation about blogging in Iran when he suggested that the voices of 70,000 Iranian bloggers would make it much more difficult for the United States to invade, because the voices of the people in Iran would build sympathy in the outside world. Mainstream sources run “human interest” stories, but they lack the power of a man's honest love for his sister.

The final point about this article is that the mention of same-sex marriage is only incidental. I encountered it essentially by accident because it happened to contain the phrase “gay marriage”. The same sort of happy accident could lead other readers to unexpected articles and viewpoints, and so serve to bridge communities.

6 Limitations & Further Investigation

A strength of content analysis is also a weakness. Focused on the transmission model of communication – what is said, to whom, why, and with what effect (Krippendorff, 1980, p. 34), it examines messages without reference to the context of the recipients. This makes it unable to take account of the recipient's construction of meaning.

Content analysis can tell us much about the author of the message, because her or his role in the production of the message is complete. This disregards, of course, the transactional or cybernetic nature of communication: authors write for an audience, and are interested in the effect their communication has on that audience. Blogs provide many opportunities for such feedback through comments and trackbacks. Another method – such as participant observation or focus groups – would help to make the link between the structure and content of blogs and how readers use these to construct meaning.

The previous sample, S5-27, for example, believes he is only being read by his friends and family. Is he? An examination of the content of blog posts provides only scraps of information about the behavior and experience of blog readers, and even less about how they interpret what they read. How often do they follow links, and what kinds of links do they follow? How do they find posts in the first place? Are they, like me, likely to stumble upon something unexpected? Do they seek out different points of view? I restricted myself to one dimension of the blogosphere: same-sex marriage. But readers may participate in multiple dimensions, multiple communities. Seemingly unrelated subjects may connect what appears to be isolated.

One way that content analysis can take this into account is through its operationalization of time in a sequence of communications. I wrote above about the importance of time in blogs: there is often frequent feedback between reader and writer in the form of comments and other blog posts referencing a previous post. Time would be one way to pursue the interaction between reader and writer. More helpful still would be to apply other methods, such as interviews or ethnographic studies, to bloggers and their audiences.

This study only examined a single issue: same-sex marriage. As a result, the data reveals at least as much about the discourse around same-sex marriage as about blogs as a medium. The obvious way to address this is to perform similar analyses to other subjects – less contentious, less political, etc. – and compare to discover the influence of the common medium. Even for this topic, the sample of 120 posts produced minimal data for some measures, such as the frequency of links between posts with opposing viewpoints.

The other startling result – the dominance of mainstream media stories – requires more study also. The media articles included in the study were linked to by blogs, they were not a random sample. A comparison with a more representative set of media stories could either confirm these results (especially regarding the use of frames) or suggest a pattern in the stories used by bloggers.

7 Conclusion

It is impossible to generalize from the case of same-sex marriage to the role of blogs overall, or even from this sample to the issue of same-sex marriage online. However, there are four definite patterns in the sample.

First, the majority of posts in the sample are addressing a particularly community, not the public at large. Those with links relevant to same-sex marriage link to those with similar views three times as often as to those opposed. They are walled gardens.

Second, the mainstream media have a dominant influence. Over half of links are to and nearly two thirds of quotes are from them. Other blogs, by comparison, are seldom linked to or quoted. The blogs in the sample are acting more as an extension of the maintsream media than an alternative to it.

Third, the characteristics of blogs as a media – in particular, hyperlinks and reader comments – do contribute to their ability of the posts in the sample to provide additional perspectives: two-fifths of the sample have a link or comment which provides at least one additional frame.

Fourth, some of posts in the sample provide views and analysis not available in mainstream media. They do offer an alternative. However, they are a small minority.

The majority of posts in this sample is indeed balkanized, dependent on mainstream media which ironically provides a kind of neutral middle ground between sides in the debate. Alternative discourses are rare, but valuable. The question remains, however, what is the experience of readers?

8 Results Online

I have placed the results of my content analysis online. Please see www.geof.net/research/2005/05/13/balkan-blogs/

9 Appendix 1 – Protocol

9.1 Sample Classification

Samples are not the only units of content recorded: some posts or pages which result from the Technorati search that produces the sample fail to qualify for one reason or another (see Relationship, below). The first piece of information recorded is whether a recorded page is part of the sample. The classifications are as follows (letter case is significant):

Only information for S records is complete; s records may have missing information depending on whether a particular field is relevant for that type of resource. Information for t and blank records may be incomplete or inaccurate.

9.2 ID

Every sample is given a unique ID , e.g. 9-4. The first number is the block number, the second is a sequential number within the block. All records from a Technorati search for a specific day share the same block number; the links resulting from that search are numbered sequentially.

Secondary samples have numbers beginning like 9-4L1. The L indicates that this is a link from the primary sample, and the 1 is a unique number to differentiate this record from other links. A C indicates a comment, a T a trackback.

9.3 Type

This is the type of resource, as follows:

Some of these categories were further simplified into larger groupings for analysis, as follows:

9.4 Link Type

This is the relationship between a secondary resource and the primary resource:

9.5 Relationship

For primary and secondary samples this is meaningless, but for records excluded from the sample the following codes indicate why. The IC code also occurs for mainstream media articles included in the sample:

9.6 URL

The URL to the resources. For most comments this is blank; they are available from the primary sample page instead.

9.7 Title, Blog Title

These are the title of the post or page and of the blog or site. These are not included in the analysis, but make finding specific articles easier.

9.8 Recording Date

The last date that the recorded information was updated. This may or may not be accurate, but it useful for record-keeping, especially if the web page later becomes inaccessible.

9.9 Date

This is the date associated with the article itself, usually the date the article was posted, but it could be the date it was last updated depending on the practice of the blogger or web page author.

.10 Country

This is the country of residence of the poster. In a couple of cases where someone of one nationality is in another country, I have also recorded nationality. This information may be found in the post, elsewhere on the page, in an About page for the blog or in the name of a media source. It may be implicit from the content of a post; where this does not necessarily indicate that the person lives in a place a question mark is appended (e.g., a detailed discussion of local news may certainly suggest that someone is American, but is not proof, so country is recorded as “us?”).

9.11 Sex

This is the gender of the author of a blog post. It is not relevant for non-blog pages. This may be determined by the content of the post, by an about page for the blog, from a photo of the poster, from the poster's first name if that name is unisex, or from comments referring to the poster.

9.12 Alignment

This is the political alignment of the author of the resource, one of “left”, “right”, “neutral” or ”?”, or in rare cases “center” if the author states this explicitly. Explicit support for a political party or philosophy is evidence of alignment, as is a clear rejection of the opposing philosophy or party. A pattern of political opinions may also constitute sufficient evidence, but one or two is insufficient. Opinions about homosexuality and gay marriage are disregarded. If the evidence is strongly suggestive but inconclusive, “right?” or “left?” is recorded.

9.13 Stance

This is the author's stance on the issue, in this case gay marriage. Possible stances are “for”, “against”, “neutral” / “?” (either may be recorded, but should be considered equivalent), or “mixed” (for a conflicted post; for analysis these are folded into neutral). Support for one stance or opposition for the other must be voiced by the author. A position in quoted material only if the author cites agreement or disagreement with that the relevant argument. Comments about homosexuality itself are not sufficient to indicate support or opposition.

For analysis, stances are combined as follows, unless otherwise specified:

  • neutral replaces mixed, ?, for? and against?

9.14 Mode of Address

The values for this category emerged from the data. Any indication of an address to a specific group is used to flag mode of address. Because of the diversity and ambiguity of the groups addressed, I divided them into three clearly defined categories:

9.14.1 Community

This mode of address indicates that addresses a group of people shared values, i.e. a group that is homophilic in some way. The subcategories included:

  • a clear reference to an undefined community (e.g., “We all knew this was coming”)

  • Catholics, fellow Christians

  • friends/peers or online friends

  • lawyers or legal professionals

  • fellow liberals

  • fellow conservatives

  • fellow same-sex marriage advocates

  • fellow same-sex marriage opponents

  • the extreme right

  • gay people

  • fellow gays

9.14.2 Public

There is no address to a particular group. Two other sub-categories are also included here, one is address to fellow Americans, the other to fellow politicos – i.e., followers of politics where there is no further indication of homophily. Where the mode of address was indeterminate, I considered it public.

9.14.3 Tone

The tone of an resource is the overall style of the writing. In most cases, this corresponds to the purpose of the author, e.g. to inform, argue for a position, etc. This tone is not considered particularly in relation to same sex marriage.

Where a resource has more than one tone, only the overall tone is recorded. The introduction and especially the conclusion provide stronger evidence for tone than the body of the article: if they are in agreement, then they determine the tone.

Several of these categories are very difficult to distinguish. For analysis, tones are combined as follows:

  • reportage replaces objective

  • opinion replaces polemic, rant, and analysis

9.14.4 Objective

The article uses the forms of objectivity to avoid showing bias or opinion for or against any political position, including gay marriage. This does not mean that the author does not have a position, simply that an explicit reading of the text does not reveal it. Most media articles fall into this category, as do blog posts which quote media without comment. Analysis may be present, but it cannot be dominant. All articles recorded as objective could also be recorded as reportage.

9.14.5 Reportage

The purpose of reportage is to inform. Objective reportage is one form; this description is used for reportage which may also add an opinion or other thought that would not be considered objective.

9.14.6 Polemic

The purpose of a polemic is to make a point. The article is primarily aimed at making an argument, usually political. In order to support this argument, the article may provide evidence or analysis, but it is the conclusion(s) and/or opinions which matter.

9.14.7 Rant

Like a polemic, a rant aims to make a point or express an opinion. Unlike a polemic, it does not use analysis, evidence, or reason to convince. By this definition, a simple opinion, such as “marriage is about love”, would be considered a rant.

9.14.8 Analysis

The focus of analysis is understanding, both by the reader and by the writer. The article examines an issue, but there is no firm political position on the issue. There may be arguments for different sides, uncertainty about the conclusion, or mixed thoughts. Of course, an article can argue for a nuanced or mixed conclusion. If this is the thrust of the argument, then it would be a polemic. An objective media story could fall into this category if it is focused on analysis rather than reporting facts.

9.14.9 Diary

The article is a description or narrative of personal experience. This is somewhat like an analysis in that it does not advance a political position, but unlike an analysis it revolves around subjective experience, not reasoning or external sources. “I” and “me” are common keywords which may indicated a diary tone.

9.14.10 Humor / Sarcasm

The article is presented as humor or sarcasm. Uses of humor or sarcasm are insufficient for this tone to apply, but if that is the overall tone of the article then the article falls into this category, even if it would otherwise be considered a rant or polemic.

9.14.11 Strategy

A strategy article focuses on how to achieve a political goal, almost always in the context of a discussion with like-minded people. Unlike an analysis, it focuses on the means to achieve a goal, not the objective itself, which is assumed.

9.14.12 Announcement / Press Release

An announcement is like reportage, except rather than passing on third-party information it is released by the originator (or a participant, e.g. a party to a lawsuit).

9.14.13 Ramble

An article with no coherent presentation is a ramble. If it takes a political position, it would instead be a polemic. If it discussed personal experience, it would be a a diary; reasoning would make it an analysis. Answers to a quiz or collections of unrelated quotes or web sites also fall into this category.

9.14.14 Comments / Letters to the Editor

Letters to the editor etc. fall into this category, because their tones may be diverse.

9.14.15 Interview

An interview is a more specific form of reportage.

9.14.16 Legal

This is a form of analysis in the context of law, with legal expertise.

9.14.17 Law

This is the text of an actual statute or legal document. Unlike a legal article, it is often the product of political reasoning rather than legal reasoning. In a sense it is like an announcement, except that law carries legal force – it is an action.

9.15 Focus

This is the relationship between the resource an the issue of same sex marriage:

  • central – same sex marriage is the topic of the article

  • related – the article goes into some depth about same sex marriage or a related topic, but it is not itself the main topic

  • incidental – same sex marriage is only mentioned in passing


This is a number indicating how dominant quotes are in a blog post, as follows:

  • 0: no quotes are present

  • 1: there are one or more inline quotes, but no block quotes

  • 2: block quotes take up less than half of the post

  • 3: block quotes take up the majority of the post

  • 4: the entire post (headline excluded) consists of one or more block quotes

A block quote is an HTML block quote or the visual equivalent: an entire paragraph quotation. This may be displayed indented, on a separate background, or simply surrounded by quotation marks.

There is likely to be a much larger error between 0 (no quotes) and 1 (non-block quotes). Analysis only deals with block quotes; these are both treated as an absence of block quotes.

9.17 Quote Sources

This is a list of the types of resources which have been quoted. These values are the same as are recorded under type.

9.18 Link Count

This is the number of outbound links from the body of the article. Boilerplate links at the bottom (e.g. to the author, comments, trackbacks, the blog home page, etc.) are not counted. Links are counted even if they are broken. URLs are not included if they are not clickable links.

9.19 Trackback Count

This is a count of trackbacks to a blog post, as determined by a Technorati search. This proved to be unreliable and was excluded from analysis.

9.20 Comment Count

This is the number of comments made about the article within the first 30 days after it was posted.

9.21 Length

The length of a post is based on a number of “screens”, using a standardized browser window size. I chose a window 42 ems high by 120 exes wide, which is approximately 35 lines using a standard font size and spacing. Measurement begins not at the top of a web page, but at the first line of content beneath the title of a blog or article (this is to account for varying font sizes for titles). Font size, margins, embedded images etc. can all change how many words fit on a screen, so this measure is far from precise.

The length measurement actually takes on one of the following values:

  • 0 – the article is no more than three sentences long

  • 1 – the article is up ½ a screen long

  • 2 – the article is up to 1 screen long

  • >2 – the article is up to n – 1 screens long, where n is the measurement

For calculating mean lengths, a measure of 0 is treated as 0.2 screens, 1 is treated as 0.5 screens, and all other measures are reduced by one.

9.22 Error Estimate

Each article includes an error estimate. This is to take into account the problems of discovering frames and other characteristics of extremely long articles, or long articles with little relevant content. A value of 0 indicates that I examined the entire article; a 1 indicates that I scanned the whole article and searched for the terms “marriage” and “sex” (to capture “gay marriage” and “same-sex marriage”) within the article and examined surrounding content, and 2 indicates that I did not examine the entire article (in which case I noted why not and how much I did read).

9.23 Frames

Following Gamson (1992, p. 216), I have attempted to reduce bias in the definition of coding frames by providing actual example quotes from advocates of each position, culled from blog posts. Quotes with an asterisk (*) following the quote are from the sample; those without are from elsewhere. Blogs vary greatly in writing style, so the style of these quotes is varied and cannot be consistent with all of the samples in the study. Not all frames include such quotes. I have included keywords for some frames; these are words which, if present, strongly suggest the presence of that frame.

Most frames correspond to a stance – either for or against same-sex marriage; this is indicated in parentheses. The frame is only present if it is referenced in this sense, but that need not be the overall position of the article: the author could be arguing that this frame is invalid or irrelevant. For example, an argument that gay people should not be able to adopt children would not fall into frame B1. On the other hand, an argument that same-sex marriage would be good because it would provide pension benefits, but that this is not enough to justify it, would fall into frame B1.

Debates about same-sex marriage often center on homosexuality itself. This leads to a problem: those opposed to homosexuality are almost certainly also opposed to same-sex marriage, but those who reject homophobia may not in fact be in favor of same-sex marriage. This study is about same-sex marriage, not attitudes towards homosexuality in general. I therefore accept homophobic or anti-gay remarks as matching some of these frames, but gay-friendly remarks are insufficient to be considered support for the issue. Furthermore, I have found a greater diversity of frames opposing marriage to those supporting it. The results of this study are likely to be weighted towards anti-homosexual points of view. I must emphasize that they are not an accurate representation of attitudes towards homosexuals in blogs.

9.23.1 B – BENEFITS / LAW (for)
  • B1 – “Gay and lesbian people have families, and their families should have legal protection”, “the right to have partner rights”; “Gay couples find life increasingly difficult. Among other things, they are legally prohibited against making critical decisions for each other”*. The article mentions support for gay marriage on the basis that it provides access to social benefits (such as pensions), custody suits, the right to adopt children, legal remedies, etc.

9.23.2 C – COMMITMENT (for)
  • C1 – Support for gay marriage enhances commitment. “By allowing gays to marry, we will be encouraging life-long commitments, monogamy, and two-parent families”; “Marriage is the public expression of lifelong commitment”; “You couldn't help but be moved by the commitments and marriages that people have entered into.”* Keyword: commitment.

  • C2 – Same-sex marriage is good for the community or society at large. “Marriage enhances the welfare of the community”, “greater recognition of gay and lesbian families . . . can only lead to strengthening of the family and humankind”*, “SSM would make society more stable”*.

  • C3 – Same-sex marriage is good for children because it provides for stable families. “Society is better served by maximizing the number of children who grow up in loving, stable homes”; “they [same-sex marriage opponents] would rather have a child suffer in a vicious foster care system, than be placed in a loving home with gay parents”*, “she thought . . . the ideal environment for children to grow up is a two-parent household with a mother and father . . . this is a slap on the face to all the kids in foster care”*.

9.23.3 D – DEMOCRACY (against)
  • D1 – This is an issue that should should be decided by politicians or elected officials, not by courts or unelected officials. Criticism of activist judges also falls into this category. “'Activist judges' . . . attempting to impose their views 'against the will of the people”* (in the context of judgements for same-sex marriage), “judges insist on forcing their arbitrary will upon the people”*. Keywords: activist courts, activist judges.

  • D2 – Same-sex marriage goes against the will of the majority. Argument against same-sex marriage on the basis that gays are a minority also qualifies. “It’s [not] what the people want”, “When such a [judicial] decision [in favor of same-sex marriage] is contrary to to strong public opinion, its undemocratic character is unmistakable”*, “it [a ban on same-sex marriage] allows democracy to work.”*

9.23.4 E – EQUALITY (for)
  • E1 – An argument appealing to equal rights: “everyone should have equal rights”; “The case for allowing gays to marry begins with equality, pure and simple.”, civil unions (instead of full marriage) are “unconstitutional, discriminatory”*. References to civil rights, equal rights, and discrimination also qualify. Note that this must be in reference to same-sex marriage itself; support for equal rights for homosexuals is not sufficient unless it connects with marriage. Keywords: equality, equal rights, civil rights, human rights, discrimination.

  • E2 – The argument that minority rights to not exist at the whim of the majority, or that rights are not a matter of democracy. A gay marriage ammendment is “all about raw majority rule with no concern for minority protections”* or “pure majoritarianism”*.

9.23.5 Q – EQUALITY (against)
  • Q1 – The argument that “Gay marriage threatens religious freedom” or otherwise threatens the rights or freedoms of those with religious beliefs (e.g. to refuse to marry homosexuals). “Christian leaders . . . fear . . . [gay marriage] legislation would threaten the rights of the church”*, “Christians must face the full wrath of the militant homosexual lobby for infringing on their self created right to be above all reproach”*.

  • Q2 – Same-sex marriage grants homosexuals special rights, or in some way exceeds equal rights. “Gay marriage, I am against that. Special rights, I am against that.” Keyword: “special rights”.

9.23.6 F – FAMILY (against)
  • F1 – Same-ex marriage threatens the family, or even society at large: “If gays can get married, that will weaken traditional marriages.”; “protecting marriage”; “[a minister], under the guise of 'protecting marriage', wants to forcibly dissolve their [same-sex] union”*. An emphasis on the preservation or saving of “traditional” marriage from same-sex marriage or from homosexuals.

  • F2 – Same-sex marriage is bad for children. “It has a lot of negative influences on kids”; “Children should be raised by a man and a woman, not by two men or two women.”

  • F3 – Same-sex marriage is selfish. “It's all about the selfish and I, I, I”

9.23.7 G GOVERNMENT (neutral)
  • G1 – Any mention of “civil unions” or other mechanisms to grant the rights of marriage without being called “marriage”.

  • G2 – Any discussion of how same-sex marriage might be implemented or forbidden by government. This includes discussion of the level of government (state/federal etc.), constitutional issues, legal or political mechanics of instituting or implementing gay marriage. Keyword: defense of marriage act.

  • G3 – The suggestion that same-sex marriage should be put to a vote or referendum.

9.23.8 H CONSISTENCY (for)

All of these frames must refer to same-sex marriage, not just homosexuality.

  • H1 – Suggestions that same-sex marriage cannot weaken marriage more than it already is. “Heterosexual couples have turned marriage into a disposable commitment”; “look at the ease with which people get married and divorced in this country. That’s a far bigger threat to the sanctity of marriage than any number of gay marriages could ever be.”, “gay couples . . . are every bit as capable of making that loving commitment as straight couples are”, “different family structures in this country”.

  • H2 – Any reference to the argument that same-sex marriage is justified as an extension of historical changes that have made society more liberal. “We have interracial marriage now, we have Catholics marrying Protestants.”

  • H3 – Any reference to the argument that same-sex marriage cannot be rejected on the basis of historical precedents about which we have become less liberal. “Didn’t most of the Old Testament patriarchs have multiple wives?”

9.23.9 L – LOVE (for)
  • L1 – Any connection between same-sex marriage and love. “It's about love”; it's about companionship. Keyword: love.

9.23.10 M – MORALITY (against)

These argument apply equally to same-sex marriage and to homosexuality on its own: if homosexuality is wrong, then same-sex marriage is implicitly also wrong.

  • M1 – Any reference to same-sex marriage or homosexuality being sinful or contrary to religious doctrine, writings, or beliefs. “It's a terrible sin” (includes gay marriage or simple homosexuality). Keywords: sin, (heterosexual) marriage as sacrament.

  • M2 – Any reference to non-religious moral opposition to same-sex marriage: “two men [or women] living together is not right.”

9.23.11 N NO NEED (against)

This frame was discarded due to its rarity.

  • N1 – Homosexuals can be satisfied without marriage. “A gay couple can stay committed and faithful to one another, just as if they were married, so there’s no need to grant legal recognition.”

  • N2 – The prohibition against same-sex marriage is by definition not prejudice because they are permitted to marry the opposite sex, or because it applies equally to straight people being forbidden to marry the same sex. “Gays choose to be with same-sex partners. If they chose to be with opposite-sex partners, they could get married.”

9.23.12 P PRIVACY
  • P1 – Marriage is private and does not have any negative effects on anyone else. “Live and let live”; “What will our marriage and happiness do to affect anyone else?”

  • P2 – It's not the business of the state, including separation of church and state: “church and state are meant to be kept separate”; “government doesn't belong in the bedrooms of the nation”*.

  • P3 – Homosexuals are adults and should be permitted to make their own decisions: “consenting adults”.

9.23.13 R – RELIGION
  • R1 – Any connection between religion and the issue of same-sex marriage (not simply homosexuality).

9.23.14 Y – REPULSIVE

This frame was discarded for lack of evidence.

  • Y1 Homosexuality (or same-sex marriage) is it's disgusting or repulsive.

  • S1 – Same-sex marriage will lead to allowing other undesirable activities, especially polygamy. “Once we allow gays to marry, it will open the door to bestiality, pedophilia, necrophilia, and worse.”; “What about polygamy? What will prevent that door from opening?” Keywords: polygamy, bestiality.

  • T1 – Same-sex marriage contradicts the (correct) definition of marriage: “marriage is between a man and a woman”. This must be presented as the correct definition of marriage, not simply a restatement of existing law or practice.

  • T2 – Homosexuals marriage will not lead to children: “Gays don’t need to get married because they can’t have children.”; “men and women were designed for each other for procreation . . . Procreation ensures that our society survives”, “if all the people on the Earth would turn that way, the people in the world would die out”.

  • T3 – Reference to tradition or traditional marriage. Keywords: tradition, traditional marriage.

  • T4 – Any reference to gay marriage or simply homosexuality being an illness or not being natural.

10 Bibliography

Adamic, L. & Glance, N. (2005 March 4) The Political Blogosphere and the 2004 U.S. Election: Divided They Blog. Retrieved March 8, 2005 from http://www.blogpulse.com/papers/2005/AdamicGlanceBlogWWW.pdf

Anderson, C. (2004 October) The Long Tail. Wired Magazine 12.10. Retrieved April 20, 2005 from http://www.wired.com/wired/archive/12.10/tail.html

Castells, M. (2001) The Internet Galaxy. Oxford: Oxford University Press.

Derakhshan, H. (2005 February 17) The New Blogosphere and Civil Society in Iran [Seminar]. Simon Fraser University.

Feenberg, A. & Bakardjieva, M. (2004) Consumers or Citizens? The Online Community Debate. Community in the Digital Age: Philosophy and Practice. US: Rowman & Littlefield.

Frost, C. (2003) How Prometheus is Bound: Appling the Innis method of Communication Analysis to the Internet. Canadian Journal of Communication (28:1) (pp. 9-24).

Gamson, W. A. (1992) Talking Politics. US: Cambridge University Press.

Glass, G. (2005) Time and the Internet Commons. Retrieved May 13, 2005 from http://www.geof.net/research/2004/11/29/hacker-time.html

Hackett, R. A. & Zhao, Y. (1998) Sustaining Democracy? Journalism and the the Politics of Objectivity. Toronto: Garamond Press.

Innis, H. (1991) The Bias of Communication. Canada: University of Toronto Press.

Johnson, D. & Bimber, B. (2004) “The Internet and Political Transformation Revisited”. Community in the Digital Age. Ed. Andrew Feenberg and Darin Barney. US: Rowman & Littlefield.

Krippendorff, K. (1980) Content Analysis. US: Sage Publications.

McChesney, R. W. (1997) Corporate Media and the Threat to Democracy. New York: Seven Stories Press.

Negroponte, N. (1995) Being Digital. New York: Alfred A. Knopf.

Pew Forum. (2003 November 18). Gay Marriage. Religious Beliefs Underpin Opposition to Homosexuality. Retrieved May 11, 2005 from http://pewforum.org/docs/index.php?DocID=39#4

Rogers, E. M. & Kincaid, D. L. (1995) Communication Networks. New York: The Free Press.

Rogers, E. M. (1985) Diffusion of Innovation. New York: The Free Press.

Strauss, A. L. & Corbin, J. (1998) Basics of Qualitative Research. US: Sage Publications.

White, N. (2005) Stephen Downes' Talk at Northern Voice. Full Circle Associates Online Interaction & Community Blog. Retrieved February 19, 2005 from http://www.fullcirc.com/weblog/2005/02/stephen-downes-talk-at-northern-voice.htm

Wikipedia. (2005 February 18) “Weblog”. Retrieved 18 February, 2005 from http://en.wikipedia.org/w/index.php?title=Weblog&oldid=10389559

Wilhelm, A. G. (2000) Democracy in the Digital Age. New York: Routledge.

1The name for one such format, RSS, has almost become a generic name for the technology.