Open smiklin opened 5 years ago
Thanks for thought-provoking insight in gerrymandering, super interesting topic. As a political science student, I have read several Prof. King's article before, but mainly in Chinese Politics. Super delighted that you can come to Chicago and deliver a workshop on your newest finding.
US politicians have the history of utilizing gerrymandering to control the House or Senate, among which one representative example is that Texas Republicans redistricted in Texas in 2003 aimed only at partisan advantage. This research provides a new insight into district division from compactness perspective which is so fascinating. I have two questions on this finding:
Another influencing factor in constituency division is population flow which is measured by US Census every ten years. Some parties have use population flow as an excuse for gerrymandering ( i.e. Texas in 2003 as well as UMNO in 2016 in Malaysia). In your opinion, which factor might have a greater impeachment on democracy and fairness on different parties?
As you mentioned, there are huge discrepancies between the definition of compactness between your methodology and existing methodology? Hence, are there any incline in terms of political parties, geographical distribution etc.? For example, is it more likely Republicans would regard some places in Texas as compactness to secure their party interest?
Thanks again for coming to Chicago!
Thanks for the great presentation! The measures of compactness seem really interesting and effective, and I am wondering what is the possibility of actually implementing the developed measures? What are some of the limitations and struggles that can happen in actually putting it in to practice?
Thank you very much for your presentation.
As not much of an expert in political science, your papers were hard to read but gave me a lot to think about. In Katz, King, and Rosenblatt (2019) article, I was wondering what happens if the payoff of a particular party is not to win the most seats, but rather to create diversion in the current, mostly bipartisan, political system. I am specifically thinking about a minor party trying to enter the campaign environment, with their payoff not necessarily driven by the V or the statewide average district vote as they sort of already realize that they are not going to get much votes anyways. In that case, the payoff might depend on, say, the variance of the district votes that they yield (a wild guess). How might this (rather unlikely) scenario fit in the story?
Thank you for the presentation. I am deeply impressed by the authors' delicate design for a new measurement, which is of great importance for both political application and academic research. My question is that when developing an idea that is quite different from mainstream or existing literature, what is the most important thing to consider? Also, for young scholars, it may be unrealistic to expect sufficient support to conduct massive surveys or experiments to generate a groundbreaking new concept. Do you have any suggestions for us about making innovations in social science?
Thank you very much for your presentation! Politics is always a fascinating topic to delve into especially when combined with computational methods and statistical models. To convincingly define the originially-ill-defined concept is very exciting, what about applying these methods to other topics of political sciences, like predicting the results of the elections or exploring other factors that may contribute significantly to those results? What do you think of the role of statistical analyses in prediction and even their potential impact on elections?
Thanks for the presentations! I have one question about the compactness paper. I would imagine eyeball criteria works pretty well with judges and the only reason to quantify compactness is for research use (please correct me if I'm wrong). In that case, do you think the absolute value of compactness index more important than the ranking itself? And if we feed the training model with enough districts from all over the state, can it output an index in digits?
Thanks in advance for your presentation. You conclude Theoretical Foundations and Empirical Evaluations of Partisan Fairness in District-Based Democracies by mentioning the challenge of formalizing concepts that "adjudicate trade-offs between partisan fairness and other goals."
What are some specific factors one would have to address when trying to balance partisan fairness and racial fairness within the framework that you discussed?
Thank you so much for the talk in advance! The way you measure compactness, the idea you talk about fairness are both really inspiring. As you redefine the method of measuring the compactness, I believe it will definitely have influence in academic field, but will it be implemented to the real statistically analysis? What do you think is the current obstacle to put this method into practice?
Thank you for your presentation! My question is when using computational methods in social sciences, sometimes it is difficult to define a criterion to measure the performance of models. That is, sometimes it is difficult to create or choose a good predicted variable to detect whether the model performs well. Could you elaborate more about this? Thank you!
Thanks for the presentations! I am really interested in the computational part of the Kaufman et al. (2018) paper. What is your advice when dealing with big geographical data like this? Did you use any method(say, GPU or parallel computing) to accelerate your computation? What are the difficulties of the computing part and how did you cope with them?
Thank you for your presentation. I am interested in the statistical and computational method you used in election. But political science is a new field to me. I am curious how your results and methods could be extend to other field? And is there any application in reality?
Thank you for your presentation. Although not familiar with topics on political science, I am really impressed by the modeling and computational methods implemented. In the validation section, it seems like you focus on samples who are highly-educated (public officials and judges from many jurisdictions, as well as redistricting, consultants and expert witnesses, law professors, law students, graduate students, undergraduates, ordinary citizens, and Mechanical Turk workers). Will the result be biased in this case? Also a question in more general perspective: is it possible that such method could be applied to other fields in social science such as economics?
Thank you so much to share us this novel research! The method that you used to measure people's multidimensional understanding of compactness is quite interesting and I think it is more like a psychological survey? But I am concerned about whether the ranking process for respondents is somewhat complicated and thus would bring some system error to the result?
Your work is very interesting. You mention that your work can be extended to include minority and racial fairness as well. Could you please discuss more on the work that has been progressing along this direction or the way forward to look at these questions?
Thanks for coming and your presentation! It is good to know that people have common sense on some major features of compactness, and the method seems to turn some loose ideas into thinkable and measurable. But how can we transform the agreement on the compactness of geometry into the context of political districts? Thank you again!
Thank you for this work Gary - it’s super cool reading it. I’m not a Political Science student or have much experience in this field, but the papers are fun to read and get the point across well.
I have two questions:
I loved the idea of attempting to quantify or demonstrate the ‘more than we can tell’ nature of things which we humans do. I think that while sometimes it can be used to draw conclusions which might be off the mark (in my opinion anyway, like with the Wang paper), in this case you do a good job with it. I wonder though… how can we be careful with reading too much into it, even when everything appears to look statistically significant? it makes sense with gerrymandering - but how do we measure the very ‘more than we can tell’ nature of things? Either way, soliciting opinions from a wide variety of folks, relevant or otherwise, and seeing how they match up does seem like a good idea.
How much do you focus on communicating these ideas and thoughts to people who can take action on these things? When we computational social scientists (including us students in training!) are doing research, I do believe there are certain normative considerations we must take. Our research has consequences and communicating this to the right audience is super important. Any tips on how you’ve done this in the past? How much of your own research is based on normative reasons?
Cheers and thank you once again!
Thanks for the presentation! What factors do you think were responsible for the delay in unifying the literature on 'partisan fairness' given the developments of statistical methods?
Thank you for such fascinating papers and for sharing your research with us. Measurement of abstract or hard-to-define concepts is very tricky but, when done correctly, can lead to many insightful results. I think you and the coauthors have demonstrated to us that developing a fairly good measurement is not an impossible task. Not only did I learn the technique, but I feel that I learned how a good measurement paper is supposed to be.
Speaking of measuring compactness, I am wondering what you think about incorporating other aspects such as a population's partisan identity into the measurement. More specifically, I am thinking about cases where natural features such as deserts and rivers affect drawing a district line which could be misunderstood or suspected as gerrymandering when, in fact, the strange geographic shape is a result of a natural boundary. Is this a possibility that you considered? Perhaps this is not so much of a problem in the United States, but I might imagine a state like Alaska having this kind of problem.
Another tangential question on measurement. I am very much interested in measuring polarization beyond the US context of partisanship. Do you have any general thoughts/ideas about measuring polarization?
Thank you for coming and taking time to share your research. I also want to second to @wanitchayap 's concern regarding cases where n=100 seems a bit arbitrary. Apart from this, I would like to hear your idea on how technologies would change political life in the future, in established democracies and authoritarian countries as well.
Thanks for the presentations. It's been impressive to see how you measure an ill-defined concept. I think your approach is quite inspiring given the fact that social science is inundated with those 'you-know-only-when-you-see' concepts. I'm wondering about how do you think that qualitative and quantitative methods should assist each other to give a formal representation of ill-defined concepts.
Thank you for sharing your work! In the paper, you mention "the claim of an objective standard, measured on a single dimension, can only be supported if most educated people evaluated a district’s compactness in the same way". I am wondering about the role of education in shaping our judgement and perception about something. What if most people viewed something the same way but fundamentally inaccurately because of what they learned to be true?
Thank you for providing such an interesting paper to read. As we can see from the research finding that we can use a statistical model to estimate more conceptual ideas. Is it better to use it to replace the qualitative method of studying conceptual ideas or it will serve as an assistant for these? Thank you!
The Katz et. al. (2019) paper was a fascinating read; thank you for that and your contributions to the field. I have two questions:
'...as Markov independence such that an election does not change the electoral system that generates vote proportions (after conditioning on X (voter demographics)).'
Isn't this an exceptionally strong assumption to make? Gerrymandering and the introduction of voter id laws to increase the cost of turnout are examples where the results of an election do change the electoral system and its outcomes. You discuss the implications of the assumption -- with the points about the incoherence of the votes-seats curve -- but could you elaborate on the justification of the assumption?
Thanks for your presentation! This is a really interesting topic to me. The law system is actually very different from the U.S. and other countries (like China). In the U.S. people are more likely to use case law and in China we use "regulation law". I wonder the reason why the law in U.C. offers few precise definitions is because they are designed to use in a more practical way?
Thank you for your presentation! It is really an interesting topic. In your article District Compactness, you converted qualitative surveys into statistical indices, which is very thrilling. I am wondering whether we can apply this method to other social science topics to reduce the cost of searching for exactly fitted dataset and worrying about measurement error of choosing indices.
Thank you for the fascinating paper! The effort on quantifying seemingly subject and unquantifiable notions/concepts is inspiring. My question is, would you consider your methodology as assigning a tentative operational definition to a concept, which, due to the limitations of operational definitions, may fail to contain all its extensions?
Thank you very much! To convincingly define the originially-ill-defined concept is very exciting, what about applying these methods to other topics of political sciences, like predicting the results of the elections or exploring other factors that may contribute significantly to those results? What do you think of the role of statistical analyses in prediction and even their potential impact on elections?
Thanks for presenting. I am interested in district-based democracy itself. I wonder what's its advantages? Since moving is common nowadays, and polls can be completed online easily, I doubt whether it's worthy to have district-based polls rather than city-based ones except for deciding something within that district.
Thank you for presenting such interesting topic. The intent of measuring the ill-defined concepts is astounding. I am wondering whether you can give us more examples of quantifying "what we know when we see"? Also, I am interested in whether your model can also be apply to other social science study? Is there any limitation on applying the model?
@Yilun0221 I asked Gary about this over lunch, and he shared that he felt the most "fair" approach would be to enforce the symmetric property that if, at the level of the state (or country), X% of the populous voted for one party, then X proportion of the seats should go to that party (and the same for both parties). The U.S. Supreme Court is reticent to weigh in on this because the issue is so political, but the reasoning is sound...if party membership is not influenced or changed by districting.
If it is (less likely in the U.S. than in other countries with many-party systems), then the reasoning becomes less sound because the districts would then influence which party people chose to endorse or join.
Comment below with questions or thoughts about the reading for this week's workshop.
Please make your comments by Wednesday 11:59 PM, and upvote at least five of your peers' comments on Thursday prior to the workshop. You need to use 'thumbs-up' for your reactions to count towards 'top comments,' but you can use other emojis on top of the thumbs up.