I have done a pass and resolved most comments, still left some that have explanations rather than questions.
I also fixed various small issues with LaTEX code, and rearranged figure 2 into three separate figures since we have some space left. I reorganized section 6 using the \paragraph command, so that now sections 4, 5 and 6 all follow the same scheme of some introductory test, followed by four paragraphs dedicated to the four reduction techniques.
I will give all @icqe22_authors a few hours to, like @Jan said, holler, and then submit this afternoon. Good work everyone.
I’ve read through it again and it reads really well. I’ve made just a few small grammatical edits. One thing, though: As a number of us have worked on the editing, there is a mix of UK and US English spelling. Does that matter? If so, I can quickly go through the text and change -ise to -ize, etc.
About the title: should we be tongue-in-cheek and call it Lévi-Strauss in the network. Reducing networks of ethnographic codes co-occurrence in anthropology? It’s a bit unfair towards Turner and Simmel, but if it is taken with a bit of humour it can work.
Ok, I changed my mind. We can always change the title during the review process.
Dear authors,
We received your submission to ICQE22 (International Conference on
Quantitative Ethnography 2022):
Authors : Alberto Cottica, Veronica Davidov, Magdalena Góralska, Jan Kubik, Guy Melançon, Bruno Pinaud and Wojciech Szymánski
Title : Reducing networks of ethnographic codes co-occurrence in anthropology
Number : 9471
Hello @icqe22_authors. good news: paper accepted. I also like the reviews, listed below.
Next steps:
I register for the conf.
Get in touch with Applied Network Science. @melancon, @bpinaud, do you have an opinion on this?
Submission of the final paper and consent-to-publish form, by Sept 5th.
The final reviews are meta-reviews completed by a member of the ICQE22 Program Committee. Please consider the reviewers’ suggestions when finalizing your paper for the conference proceedings.
This submission examined four reduction techniques for networks of co-occurrences, focusing on “the affinity of each of the four techniques with a prominent method of analysis associated in turn with an identifiable school of thought in sociology or anthropology” (p. 2). These analyses were conducted with the aim of supporting the rigor and transparency of the choices researchers make when integrating qualitative and quantitative methods.
To begin with, I should note that data reduction techniques are beyond my areas of expertise. Accordingly, my review will not focus on the soundness of the techniques themselves, but rather on the connections offered between such techniques and related anthropological schools of thought. With respect to this aim, I believe this submission offers important and insightful contributions. This is the case as the authors go beyond examining the mathematical properties of each technique and offer important insights with respect to the type of theoretical lens each technique supports. Moreover, the authors do not pit these positions against each other, but rather highlight their relative contributions, and the types of insights they are likely to provide.
One caveat I wish to note is that the example data set offered by the authors, as well as the decisions made in the analysis process, were not always clearly elaborated. For instance, it was not clear enough how their choice of the “catholic church” code as an entry code shaped their later analyses. Nevertheless, I think this submission could lead to important and timely conversations concerning the interplay of analysis technique and underlying theories.
This paper makes a strong case for network reduction techniques and demonstrates their application in anthropology. The authors argue that dense networks are difficult to interpret. Although network reduction provides an aide to address the issue, researchers must be aware of the risks of losing information while choosing reduction techniques. The technical soundness and presentation of this work is strong and should be included in the conference. However, I urge the authors to make the connection to QE more explicit. How might network reduction impact participatory QE? What biases might be introduce in choosing to reduce networks as it relates to issues of justice, equity, diversity and inclusion?
This is an ambitious and refreshingly original piece of work, which was appreciated by both reviewers. The main issues raised by both reviewers relate to the clarity of the presentation and example. Bear in mind that the conference audience will not likely be familiar with the anthropological theories or the empirical example being discussed. As such, it would be very helpful to walk us through both much more slowly, elaborating the key ideas as you progress. Both reviewers and I had a hard time following the argument as presented (though I certainly am intrigued and want to learn more).
I thought that the connections to QE are very clear, though note that the second reviewer would like for you to make them more explicit. (I’m not sure why the reviewer asked about connections to participatory QE – this seems unnecessary to me.)
@Nica, @Jan, @Richard I was about to dive into a final rewrite of the ICQE22 paper, but now I see that you, and not I, are best qualified for making the changes that the reviewers suggest. The money quote is here:
Bear in mind that the conference audience will not likely be familiar with the anthropological theories or the empirical example being discussed. As such, it would be very helpful to walk us through both much more slowly, elaborating the key ideas as you progress.
It is surprising that the attendees of a conference called ICQE22 would not be familiar with these anthropological theories, but… would you be up to re-read with this in mind, and try to improve the legibility of the paper by this audience? Keep the page limit in mind, we need to be economical.
The thing that I can do myself is add a sentence that responds to this concern of Reviewer 1:
For instance, it was not clear enough how their choice of the “catholic church” code as an entry code shaped their later analyses.
To which the answer is: given the importance of the Catholic Church in any analysis of Polish populism, we expected that code to be still present in the maximal , for all techniques and for fairly high values of the respective reduction parameters. In fact, our paper already says that
We selected this particular code in the expectation that the Catholic Church would be fairly central in any ethnographic study of populism in Poland.
We are already at the page limit. I am now doing a sweep to see if i can save some space by compacting what we already have. I will let you know when I’m done.
I tried to expound a bit. It was a little unclear to me what they want because, like you, I am a little puzzled by what the reviewers imagine the audience will not know, but I tried to write things out a bit more and not assume the level of background knowledge I previously did. So not major changes but a bit.
Because I am not familiar with Overleaf, I could not figure out how to highlight. Right now the parts that I added (I tried to do minimal additions since I know we are constrained in terms of word count/page count) are in italics. I added two sources:
Des Chene, Mary. 1996. Symbolic Anthropology. In Encyclopedia of Cultural Anthropology. David Levinson and Melvin Ember eds. Pp. 1274-1278. New York: Henry Holt.
and
Spencer, Jonathan. 1996. Symbolic Anthropology. In Encyclopedia of Social and Cultural Anthropology. Alan Barnard and Jonathan Spencer ed. Pp. 535-539. London and New York: Routledge.
Hi, the submission is done. I used @Nica’s additions and reviewed, trying to simplify the language and shorten the sentences here and there. We only have a few hours to review and resubmit, I am afraid (deadline is today at 23.59 PDT).
A question: there is a sentence I no longer understand the meaning of. I think it comes from @Jan, and it is in Section 5, at the end of the description of the association depth technique:
It is important to note that the decision to stop data collection upon reaching what is deemed sufficient saturation, a common strategy in qualitative research, would skew the results of this method.
I no longer understand the reasoning, I would have loved to explain it.
I didn’t author it but I think what is meant is – as we were discussing in the ethnographers meeting on Friday, the more interviews there are, the more rich the visualization data is, especially for the association breadth, and presumably that means there can be more varied, different reductions for breadth. Sufficient saturation, when it comes to thematic saturation (how exhaustive the “inventory” of topics is in a finite corpus) hinges on a 5% threshold (number of identified concepts across sampling iterations changes by 5% or under 5%). But since we are interested in nuance and difference, in addition to cohesion, and since (I think?) we are interested in outliers because identifying them as outliers can be valuable data in cultural analysis (and the depth and breadth reductions practiced together make sure that the “zealot” possibility does not lead to incorrect conclusions about the relevance of a specific concept), the more data we have, the richer the visualization will be.
I did not know this! OK, but then it’s a no. Reason: we are dealing here with heavily reduced networks, where only 10-20% of the codes and 10-5% of the edges are preserved. In most cases, the reduced network before saturation and the the reduced network after saturation will be quite similar.
@alberto and @Nica and All, the fragment removed is not necessary. It is too cryptic, I agree. The point (that we do not need to make in this paper) is that “saturation” is never 100% (maybe after a perfectly representative, large sample), so any new interview may change the “edgorama” somewhat, even if incrementally. Something like that.