3. Clustering & Topic Modeling to Discover Higher-Order Patterns of Meaning - [E4] Heiberger, RH., Munoz-Najar Galvez, S., & McFarland, DA.

[lkcao]

Post questions here for this week's exemplary readings:

4. Heiberger, RH., Munoz-Najar Galvez, S., & McFarland, DA. (2021). “Facets of Specialization and Its Relation to Career Success: An Analysis of U.S. Sociology, 1980 to 2015.” American Sociological Review, 86(6), 1164–1192.

[XiaotongCui]

I find this research particularly intriguing, particularly the findings related to novelty and productivity. The revelation that productivity can have a negative impact when consistency is low is quite surprising. Does this imply that "wildly publish papers in many areas than it make things sucks?

Furthermore, the approach to measuring novelty, as stated in the paper, is noteworthy. The use of "Our measure of novelty (Hypothesis 2) captures the extent to which a dissertation introduces rare or unique combinations of topics with respect to prior dissertations" raises an interesting point. However, one might argue that a unique combination may not necessarily equate to novelty, as it could simply signify that "this is now worth studying." Nevertheless, I appreciate the effort to quantify the concept of novelty, considering its inherently vague nature.

[Dededon]

Authors of this paper performs a structural topic modelling (STM) on the abstract of sociology PhD dissertations, to study the question of how specialized the field of sociology is. Also, they got a smart way to measure the career outcomes of the PhD students by looking at whether they become supervisors in the future. The visualization is beautiful: the author have shown a network between co-referenced topics within sociology to give us a sight of "territorialization" within the field. My questions for this paper are as follows:

1. The event history analysis seems little disappointing: all of the author's identified variables except focus are significant, and only two interactional terms in model (5) is not significant. The reason that focus is not significant in both the independent variable and the interaction term is maybe focus on some topic is not good, or it could also because that focus is not a well-constructed independent variable.
2. Also, one of the author's findings (the students with consistency in publication track have higher chance to become a supervisor) is maybe statistically grounded, but qualitatively questionable. It might also have interactional effects that whether the student have secured tenured positions and professorships in their career outcomes.

[Vindmn1234]

This study presents a comprehensive and methodologically sophisticated approach to understanding the impact of dissertation specialization on academic career trajectories in sociology. The use of Structural Topic Modeling (STM) to analyze large-scale academic data is innovative and a highlight. For the authors, I'd like to ask: Given the sophisticated use of STM and quantitative analysis in this study, how to account for the nuanced and often subjective nature of academic work and career progress? Specifically, how to ensure that the quantitative measures used accurately capture the complexity and depth of academic specialization and its impact on career trajectories? Additionally, what are your thoughts on extending this research methodology to other disciplines or academic cultures outside the U.S. to assess its broader applicability and potential insights?

[YucanLei]

This is an interesting paper, however, I want to know how does this research account for the personal background of the students? We know that the research interest as well as the approach of the sociologist students can be easily influenced by the background of this student. I don't think I read much analysis about this factor.

[muhua-h]

I am slightly concerned about the operationalization of 'sociology advisor' in the context of this paper, especially given the statement that " only around 3 percent of all PhDs in our sample end up as a primary advisor for doctoral students later in their career", which is significantly lower than the tenure track rate in sociology according to the report (https://www.asanet.org/wp-content/uploads/savvy/research/NewDoctorates.pdf ). Why?

As sociology is a broad and interdisciplinary field, I would not be surprised that some Ph.D. graduates would stay in academia and become faculty in adjacent fields (e.g., organizational behavior) in business school. My speculation is that the narrow definition of sociology does not capture all possible academic routes in sociology PhD's placement. I would like to hear more about the cross-disciplinary placement results.

[QIXIN-LIN]

This study provides valuable insights into research strategies in sociology. However, it leaves us curious about how these strategies compare with those in humanities and STEM fields. Are the trends observed in sociology - like specialization and consistency - also prevalent in these other areas? Furthermore, how do education-focused disciplines, often considered more niche, align or diverge in their research approaches compared to broader fields like sociology?

[erikaz1]

I appreciate the author's use of "time to advisorship" as a proxy for achieving career success in academia. Though I didn't think this variable made sense at first, looking back, it makes sense that a PhD student, after investing many years into achieving the degree, focusing on a narrow area of study and then going into academia, could be considered successful if they remained in academia long enough to take on the role of mentoring other PhD candidates (and further stepping into the role as gatekeepers).

I was intrigued by the study trends (quantitative models/methods rose in popularity over the last 20 years) and the statement that "it seems theory, in contrast, has become an auxiliary tool for empirical graduate research and less of a thought style in its own right" (1180). While this is just Heiberger et al.'s interpretation of their findings, perhaps they are alluding to the idea that it is easier for a subfield to agree on/appreciate methods but not theory/findings ("clash of theories" mentioned in the introduction).

[erikaz1]

I am slightly concerned about the operationalization of 'sociology advisor' in the context of this paper, especially given the statement that " only around 3 percent of all PhDs in our sample end up as a primary advisor for doctoral students later in their career", which is significantly lower than the tenure track rate in sociology according to the report (https://www.asanet.org/wp-content/uploads/savvy/research/NewDoctorates.pdf ). Why?

As sociology is a broad and interdisciplinary field, I would not be surprised that some Ph.D. graduates would stay in academia and become faculty in adjacent fields (e.g., organizational behavior) in business school. My speculation is that the narrow definition of sociology does not capture all possible academic routes in sociology PhD's placement. I would like to hear more about the cross-disciplinary placement results.

Notes 4 suggests that they had to do some digging around on department websites (it's not clear) to confirm advisorship status. A lot of data could have been missed in this process. I'd like to know what's going on here too.

[Marugannwg]

I'm more into the methodology present here. I feel the researchers developed a very informative dataset with more potential presented in the paper:

• there is time information, can extrapolate popular sociology topics migration across time
• various theme information and how those themes may combine and co-exist in the paper (forming a network of topics)
• authors' information and career track in the future
• I wonder if the thematic information (e.g., the keywords that pinpoint the topic of dissertations) can be considered as labeling for further explorations.

[beilrz]

I think this is a very interesting paper. One of my concern is, to what extent, their modeling approach capture the themes of the paper. I think the majority of the themes they found from topic modeling are substantive, rather than methodological. This also raise the concern of their definition of "novelty ", which mostly likely means including two different substantive focus in the thesis, according to their method. This may ignore methodological innovation on existing topic. Another question in general when unsupervised approaches are appropriate for research: an issue I often found that the discrepancy between the result of unsupervised models and the interpretation of such result by researcher. This paper utilized "expert opinion" to address this, but it certainly does not rule out alternative explanations.

[ethanjkoz]

I think the authors were very thoughtful in their measure of occupational/academic success. They convincingly reason that measuring which sociology students end up as sociology advisors is a conservative proxy for a more general discussion of success in the field than measures like awards and pay. Digging into the online supplements, they mention that finding out the right number of topics is important part of the process. Understanding that there is no canonical approach for such decisions, what are the pros and cons to the methods these authors incorporated? What other ways of deciding k are there?

[HamsterradYC]

Structural Topic Modeling (STM) as a text analysis tool may have limitations in capturing the nuanced changes and developments in sociological research themes. Can the use of STM accurately reflect the complex thematic structures in sociology papers without leading to oversimplification or misunderstanding?

[yunfeiavawang]

This paper successfully answered the question: How do sociology students garner recognition from niche field audiences through specialization? I am curious about the concept of "degree of novelty" -- I think the operationalization of this concept leans to represent the diversity of the dissertation topics. An emerging combination of topics could indicate novelty, but just as a part of it. Novelty can also occur when new phenomena are detected, new methods are developed, or new theories are elucidated.

[icarlous]

How do the strategies of specialization, including targeted topics, focus, novelty, and consistency in research output, influence the career trajectories of sociology PhD graduates in the context of changing academic and societal landscapes from 1980 to 2015? I am interested in how cross-disciplinary investigation in this strategy would be like.