Extracting Communication Networks - Bail 2012

[jamesallenevans]

Post questions here for:

Bail, Christopher A. 2012. “The fringe effect: Civil society organizations and the evolution of media discourse about Islam since the September 11th attacks.” American Sociological Review 77(6):855-879.

[lkcao]

Both methods and theoretical reflection of this paper are great. However I still have several questions: (1), Bail use the anti-plagirism software to measure the distance between organizational narratives and mainstream media narratives. Is this the best method? I guess we can have several different expressions for the same idea, and these expressions cannot be compared directly. (2), Should we call the original anti-Muslim organizations (in the first period) "fringe"? Although their attitudes are quite different from mainstream in the first period, obviously they have more social relationships/media influence/and maybe financial resouces since these resources are often related to the first two. This seems more like a story about "how do some powerful people manipulate the mass". Can we just take it as a choice of intepretation?

[wunicoleshuhui]

Bail mentioned that he calculated closeness centrality from a relational database of interlocking directorates. How exactly is this done in practice, and how do we guarantee that there is minimal human error in manual coding?

[cindychu]

This article did great social science theory driven text analysis. However, in this study, the author also did a lot of qualitative coding by hand, but did not mention the reliability between coders and concrete and specific schemes for qualitative coding. Even for the emotion coding one, Christopher changed from text to video, which could be suspected of ‘cooking’ the data. also, the emotion index extracted from video might be biased by many factors, for example different reporters and channels for different press release as well.

I am also wondering how the plagiarism software works and is it suitable for this corpus? since the author did not find anything interesting from the software, it could be the results of impropriate techniques of this program for this corpus.

[heathercchen]

I didn't get what the author tries to emphasize in this article. I understand the theoretical framework that he wants to explain and I totally agree that such a theme can be further dug into if provided with high-quality quantitative data. However, I do not understand the theoretical construction process in this article, to be clear, why the author chooses certain quantitative variables to represent the phenomena. To make my question specific, I am very confused about the Negative Binomial Regression analysis. As far as I know, regression analysis requires strong statistical hypotheses. What is most important is that there can be no reverse causality, which implies we must rule out the possibility that these discursive organizations are gradually influencing the media and catching public attention so that they can have more opportunities to organize all kinds of events. Hence I am wondering the validity of regression analysis in this paper.

[deblnia]

Echoing @cindychu and @clk16, I was also concerned about the use (or maybe the mention) of the anti-plagerism software. I'm not sure if it was just unsuitable for this corpus, of if "plagerism" can be theoretically equivalent to "expressing the same idea". (One is dishonest, the other is merely derivative.) How should we think about thoughtfully incorporating others APIs/software into our work?

[di-Tong]

I'm curious about if there are any further studies with different designs that generalize the fringe effect occurring within other discursive fields transitioning out of unsettled times. How does the fringe effect plays out differently under different contexts?

[luxin-tian]

As is mentioned by @cindychu, I am also curious about the validation of hand-coded data, it is likely that there would be bias introduced by coders and researchers. Furthermore, I think it is also necessary to more carefully examine the algorithm underlying plagiarism detection software to decide whether this can be used to measure the similarity of ideas.

[meowtiann]

It's a great article but I also have doubts about what is exactly mainstream? Is it the influence, the civil society organizations, the board members, or the arguments they adopt?

[alakira]

Like many of you mentioned earlier, I am concerned about the way the author compares each press releases. They primarily use plagiarism detection software that compresses words in a thesaurus and compares word occurrences. This is a very simple way to determine similarity and seems to depend heavily on the quality of the thesaurus. How does adopting a different method change the result?

[VivianQian19]

Bail’s article is a very informative read. She differentiate two types of civil society organizations, mainstream vs fringe organizations where the latter are those that use the least popular frames (858). She argues that fringe organizations over time significantly shaped the main strain media discourse about Muslims after the September 11 attacks (870). She also shows that this result stands when state organizations were included in the sample. Her use of video format of press conference to triangulate her qualitative coding results of emotional displays of fear and anger is illuminating. However, I’m not entirely sure how she “triangulates” the two types of data. For instance, if her results in qualitative coding and the video coding are different, how does that impact the final coding result?