Hi all,
This week at the Applied Statistics workshop we will be welcoming M. Daniele Paserman, a Professor of Economics at Boston University. He will be presenting work entitled "Gender Differences in Cooperative Environments? Evidence from the U.S. Congress." Please find the abstract below and on the website. You can find the paper here: http://www.nber.org/papers/w22488
We will meet in CGIS Knafel Room 354 at noon and lunch will be provided.
Best,
Pam
Title: "Gender Differences in Cooperative Environments? Evidence from the U.S. Congress" (joint with Stefano Gagliarducci)
Abstract: This paper uses data on bill sponsorship and cosponsorship in the U.S. House of Representatives to estimate gender differences in cooperative behavior. We employ a number of econometric methodologies to address the potential selection of female representatives into electoral districts with distinct preferences for cooperativeness, including regression discontinuity and matching. After accounting for selection, we find that among Democrats there is no significant gender gap in the number of cosponsors recruited, but women-sponsored bills tend to have fewer cosponsors from the opposite party. On the other hand, we find robust evidence that Republican women recruit more cosponsors and attract more bipartisan support on the bills that they sponsor. This is particularly true on bills that address issues more relevant for women, over which female Republicans have possibly preferences that are closer to those of Democrats. We interpret these results as evidence that cooperation is mostly driven by a commonality of interest, rather than gender per se.
Hi all,
This week at the Applied Statistisc workshop we will be welcoming In Song Kim, an Assistant Professor of Political Science at MIT. He will be presenting work entitled "When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Longitudinal Data?" Please find the abstract below and on the website. The most recent version of the paper can be found here: http://web.mit.edu/insong/www/pdf/FEmatch.pdf
We will meet in CGIS Knafel Room 354 at noon and lunch will be provided.
Best,
Pam
Title:
When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Longitudinal Data?
Abstract:
Many social scientists use linear fixed effects regression models
for causal inference with longitudinal data to account for
unobserved time-invariant confounders. We show that these models
require two additional causal assumptions, which are not necessary
under an alternative selection-on-observables approach.
Specifically, the models assume that past treatments do not directly
influence current outcome, and past outcomes do not directly affect
current treatment. The assumed absence of causal relationships
between past outcomes and current treatment may also invalidate some
applications of before-and-after and difference-in-differences
designs. Furthermore, we propose a new matching framework to
further understand and improve one-way and two-way fixed effects
regression estimators by relaxing the linearity assumption. Our
analysis highlights a key trade-off --- the ability of fixed effects
regression models to adjust for unobserved time-invariant
confounders comes at the expense of dynamic causal relationships
between treatment and outcome.
Hi everyone,
This week at the Applied Statistics workshop we will be welcoming Tyler VanderWeele, a Professor of Epidemiology at the Harvard T.H. Chan School of Public Health. He will be presenting work entitled "Religion and Health: An Assessment of Causality, Interaction, Feedback, and Mechanisms." Please find the abstract below and on the website.
We will meet in CGIS Knafel Room 354 at noon and lunch will be provided.
Best,
Pam
Title: Religion and health: an assessment of causality, interaction, feedback, and mechanisms
Abstract: A large literature has suggested that religious service attendance is associated with better mental and physical health. Two major questions that have emerged from these studies are: (i) is the relationship causal? and (ii) if so, what are the mechanisms? We present analyses using data from the Nurses Health Study, with repeated measures of religious service attendance, health outcomes, and time-varying confounders, to address these questions with respect to mortality, depression, and suicide as outcomes. Marginal structural models are employed to address issues of potential feedback and reverse causation; methods from causal mediation analysis are used to assess mechanisms; sensitivity analysis techniques are used to assess the robustness of conclusions to potential unmeasured confounding. Potential interaction between service attendance and either race or Protestant vs. Catholic affiliation are assessed. Discussion is given to the relation of these results to the observations in Durkheim's work Suicide, potential implication for social policy, and the needs and challenges for future research in this area.
Hi all,
This week at the Applied Statistics workshop we will be welcoming Solédad Prillaman, a Ph.D. student at Harvard University. She will be presenting work entitled "Strength in Numbers: How Women's Networks Close India's Political Gender Gap." Please find the abstract below and on the website.
We will meet in CGIS Knafel Room 354 at noon and lunch will be provided.
Best,
Pam
Title: Strength in Numbers: How Women's Networks Close India's Political Gender Gap
Abstract: In India there persists a striking gender gap in political participation and representation, despite several decades of targeted policy interventions. Women's political participation is important not only on normative grounds of inclusion, but because we know that when women do participate, politics changes. I present a theoretical model of political behavior in rural India which argues that women's lack of political participation is the result of coordinated political behavior in the household. I then argue that women's access to economic networks of other women is one channel through which we can see a shift towards a gender-inclusive equilibrium, even when resource allocations, social norms, and household dynamics would suggest otherwise. I test this potential channel for women's political empowerment using a geographic regression discontinuity design with pair-matched villages to identify the impact of a program aimed at mobilizing women into small credit collectives. Original survey data from 7,770 women and men demonstrates that women who participated in this network intervention were signicantly more active in local politics - women's attendance at local public meetings is estimated to double. I show evidence of three possible mechanisms underlying this network effect: (1) increased capacity for collective action, (2) information transfers, and (3) civic skills and confidence. I confirm with qualitative interview data. I show income to be uncorrelated with political participation. These findings have implications for larger studies of political participation and importantly help to fill the gap in our understanding of gendered political behavior.
Hi everyone,
This week at the Applied Statistics Workshop we will be welcoming Tina Eliassi-Rad, an Associate Professor of Computer Science at Northeastern University. She will be presenting work entitled "The Reasonable Effectiveness of Roles in Complex Networks." Please find the abstract below and on the website.
We will meet in CGIS Knafel Room 354 at noon and lunch will be provided. See you all there!
Best,
Pam
Title: The Reasonable Effectiveness of Roles in Complex Networks
Abstract: Given a network, how can we automatically discover roles (or functions) of nodes? Roles compactly represent structural behaviors of nodes and generalize across various networks. Examples of roles include "clique-members," "periphery-nodes," "bridges," etc. Are there good features that we can extract for nodes that indicate role-membership? How are roles different from communities and from equivalences (from sociology)? What are the applications in which these discovered roles can be effectively used? In this talk, we address these questions, provide unsupervised and supervised algorithms for role discovery, and discuss why roles are so effective in many applications from transfer learning to re-identification to anomaly detection to mining time-evolving networks and multi-relational graphs.