Macrosystems ecology: Understanding ecological patterns and processes at continental scales

Macrosystems ecology is the study of diverse ecological phenomena at the scale of regions to continents and their interactions with phenomena at other scales. This emerging subdiscipline addresses ecological questions and environmental problems at these broad scales. Here, we describe this new field, show how it relates to modern ecological study, and highlight opportunities that stem from taking a macrosystems perspective. We present a hierarchical framework for investigating macrosystems at any level of ecological organization and in relation to broader and finer scales. Building on well-established theory and concepts from other subdisciplines of ecology, we identify feedbacks, linkages among distant regions, and interactions that cross scales of space and time as the most likely sources of unexpected and novel behaviors in macrosystems. We present three examples that highlight the importance of this multiscaled systems perspective for understanding the ecology of regions to continents. © The Ecological Society of America.

The ecocultural context and child behavior problems: A qualitative analysis in rural Nepal.

Commonly used paradigms for studying child psychopathology emphasize individual-level factors and often neglect the role of context in shaping risk and protective factors among children, families, and communities. To address this gap, we evaluated influences of ecocultural contextual factors on definitions, development of, and responses to child behavior problems and examined how contextual knowledge can inform culturally responsive interventions. We drew on Super and Harkness' "developmental niche" framework to evaluate the influences of physical and social settings, childcare customs and practices, and parental ethnotheories on the definitions, development of, and responses to child behavior problems in a community in rural Nepal. Data were collected between February and October 2014 through in-depth interviews with a purposive sampling strategy targeting parents (N = 10), teachers (N = 6), and community leaders (N = 8) familiar with child-rearing. Results were supplemented by focus group discussions with children (N = 9) and teachers (N = 8), pile-sort interviews with mothers (N = 8) of school-aged children, and direct observations in homes, schools, and community spaces. Behavior problems were largely defined in light of parents' socialization goals and role expectations for children. Certain physical settings and times were seen to carry greater risk for problematic behavior when children were unsupervised. Parents and other adults attempted to mitigate behavior problems by supervising them and their social interactions, providing for their physical needs, educating them, and through a shared verbal reminding strategy (samjhaune). The findings of our study illustrate the transactional nature of behavior problem development that involves context-specific goals, roles, and concerns that are likely to affect adults' interpretations and responses to children's behavior. Ultimately, employing a developmental niche framework will elucidate setting-specific risk and protective factors for culturally compelling intervention strategies.

Chaucer and the Disconsolations of Philosophy: Boethius, Agency, and Literary Form in Late Medieval Literature

This study argues that Chaucer's poetry belongs to a far-reaching conversation about the forms of consolation (philosophical, theological, and poetic) that are available to human persons. Chaucer's entry point to this conversation was Boethius's Consolation of Philosophy, a sixth-century dialogue that tried to show how the Stoic ideals of autonomy and self-possession are not simply normative for human beings but remain within the grasp of every individual. Drawing on biblical commentary, consolation literature, and political theory, this study contends that Chaucer's interrogation of the moral and intellectual ideals of the Consolation took the form of philosophical disconsolations: scenes of profound poetic rupture in which a character, sometimes even Chaucer himself, turns to philosophy for solace and yet fails to be consoled. Indeed, philosophy itself becomes a source of despair. In staging these disconsolations, I contend that Chaucer asks his readers to consider the moral dimensions of the aspirations internal to ancient philosophy and the assumptions about the self that must be true if its insights are to console and instruct. For Chaucer, the self must be seen as a gift that flowers through reciprocity (both human and divine) and not as an object to be disciplined and regulated.

Chapter one focuses on the Consolation of Philosophy. I argue that recent attempts to characterize Chaucer's relationship to this text as skeptical fail to engage the Consolation on its own terms. The allegory of Lady Philosophy's revelation to a disconsolate Boethius enables philosophy to become both an agent and an object of inquiry. I argue that Boethius's initial skepticism about the pretentions of philosophy is in part what Philosophy's therapies are meant to respond to. The pressures that Chaucer's poetry exerts on the ideals of autonomy and self-possession sharpen one of the major absences of the Consolation: viz., the unanswered question of whether Philosophy's therapies have actually consoled Boethius. Chapter two considers one of the Consolation's fascinating and paradoxical afterlives: Robert Holcot's Postilla super librum sapientiae (1340-43). I argue that Holcot's Stoic conception of wisdom, a conception he explicitly links with Boethius's Consolation, relies on a model of agency that is strikingly similar to the powers of self-knowledge that Philosophy argues Boethius to posses. Chapter three examines Chaucer's fullest exploration of the Boethian model of selfhood and his ultimate rejection of it in Troilus and Criseyde. The poem, which Chaucer called his "tragedy," belonged to a genre of classical writing he knew of only from Philosophy's brief mention of it in the Consolation. Chaucer appropriates the genre to explore and recover mourning as a meaningful act. In Chapter four, I turn to Dante and the House of Fame to consider Chaucer's self-reflections about his ambitions as a poet and the demands of truth-telling.

Bayesian Inference in Large-scale Problems

Many modern applications fall into the category of "large-scale" statistical problems, in which both the number of observations n and the number of features or parameters p may be large. Many existing methods focus on point estimation, despite the continued relevance of uncertainty quantification in the sciences, where the number of parameters to estimate often exceeds the sample size, despite huge increases in the value of n typically seen in many fields. Thus, the tendency in some areas of industry to dispense with traditional statistical analysis on the basis that "n=all" is of little relevance outside of certain narrow applications. The main result of the Big Data revolution in most fields has instead been to make computation much harder without reducing the importance of uncertainty quantification. Bayesian methods excel at uncertainty quantification, but often scale poorly relative to alternatives. This conflict between the statistical advantages of Bayesian procedures and their substantial computational disadvantages is perhaps the greatest challenge facing modern Bayesian statistics, and is the primary motivation for the work presented here.

Two general strategies for scaling Bayesian inference are considered. The first is the development of methods that lend themselves to faster computation, and the second is design and characterization of computational algorithms that scale better in n or p. In the first instance, the focus is on joint inference outside of the standard problem of multivariate continuous data that has been a major focus of previous theoretical work in this area. In the second area, we pursue strategies for improving the speed of Markov chain Monte Carlo algorithms, and characterizing their performance in large-scale settings. Throughout, the focus is on rigorous theoretical evaluation combined with empirical demonstrations of performance and concordance with the theory.

One topic we consider is modeling the joint distribution of multivariate categorical data, often summarized in a contingency table. Contingency table analysis routinely relies on log-linear models, with latent structure analysis providing a common alternative. Latent structure models lead to a reduced rank tensor factorization of the probability mass function for multivariate categorical data, while log-linear models achieve dimensionality reduction through sparsity. Little is known about the relationship between these notions of dimensionality reduction in the two paradigms. In Chapter 2, we derive several results relating the support of a log-linear model to nonnegative ranks of the associated probability tensor. Motivated by these findings, we propose a new collapsed Tucker class of tensor decompositions, which bridge existing PARAFAC and Tucker decompositions, providing a more flexible framework for parsimoniously characterizing multivariate categorical data. Taking a Bayesian approach to inference, we illustrate empirical advantages of the new decompositions.

Latent class models for the joint distribution of multivariate categorical, such as the PARAFAC decomposition, data play an important role in the analysis of population structure. In this context, the number of latent classes is interpreted as the number of genetically distinct subpopulations of an organism, an important factor in the analysis of evolutionary processes and conservation status. Existing methods focus on point estimates of the number of subpopulations, and lack robust uncertainty quantification. Moreover, whether the number of latent classes in these models is even an identified parameter is an open question. In Chapter 3, we show that when the model is properly specified, the correct number of subpopulations can be recovered almost surely. We then propose an alternative method for estimating the number of latent subpopulations that provides good quantification of uncertainty, and provide a simple procedure for verifying that the proposed method is consistent for the number of subpopulations. The performance of the model in estimating the number of subpopulations and other common population structure inference problems is assessed in simulations and a real data application.

In contingency table analysis, sparse data is frequently encountered for even modest numbers of variables, resulting in non-existence of maximum likelihood estimates. A common solution is to obtain regularized estimates of the parameters of a log-linear model. Bayesian methods provide a coherent approach to regularization, but are often computationally intensive. Conjugate priors ease computational demands, but the conjugate Diaconis--Ylvisaker priors for the parameters of log-linear models do not give rise to closed form credible regions, complicating posterior inference. In Chapter 4 we derive the optimal Gaussian approximation to the posterior for log-linear models with Diaconis--Ylvisaker priors, and provide convergence rate and finite-sample bounds for the Kullback-Leibler divergence between the exact posterior and the optimal Gaussian approximation. We demonstrate empirically in simulations and a real data application that the approximation is highly accurate, even in relatively small samples. The proposed approximation provides a computationally scalable and principled approach to regularized estimation and approximate Bayesian inference for log-linear models.

Another challenging and somewhat non-standard joint modeling problem is inference on tail dependence in stochastic processes. In applications where extreme dependence is of interest, data are almost always time-indexed. Existing methods for inference and modeling in this setting often cluster extreme events or choose window sizes with the goal of preserving temporal information. In Chapter 5, we propose an alternative paradigm for inference on tail dependence in stochastic processes with arbitrary temporal dependence structure in the extremes, based on the idea that the information on strength of tail dependence and the temporal structure in this dependence are both encoded in waiting times between exceedances of high thresholds. We construct a class of time-indexed stochastic processes with tail dependence obtained by endowing the support points in de Haan's spectral representation of max-stable processes with velocities and lifetimes. We extend Smith's model to these max-stable velocity processes and obtain the distribution of waiting times between extreme events at multiple locations. Motivated by this result, a new definition of tail dependence is proposed that is a function of the distribution of waiting times between threshold exceedances, and an inferential framework is constructed for estimating the strength of extremal dependence and quantifying uncertainty in this paradigm. The method is applied to climatological, financial, and electrophysiology data.

The remainder of this thesis focuses on posterior computation by Markov chain Monte Carlo. The Markov Chain Monte Carlo method is the dominant paradigm for posterior computation in Bayesian analysis. It has long been common to control computation time by making approximations to the Markov transition kernel. Comparatively little attention has been paid to convergence and estimation error in these approximating Markov Chains. In Chapter 6, we propose a framework for assessing when to use approximations in MCMC algorithms, and how much error in the transition kernel should be tolerated to obtain optimal estimation performance with respect to a specified loss function and computational budget. The results require only ergodicity of the exact kernel and control of the kernel approximation accuracy. The theoretical framework is applied to approximations based on random subsets of data, low-rank approximations of Gaussian processes, and a novel approximating Markov chain for discrete mixture models.

Data augmentation Gibbs samplers are arguably the most popular class of algorithm for approximately sampling from the posterior distribution for the parameters of generalized linear models. The truncated Normal and Polya-Gamma data augmentation samplers are standard examples for probit and logit links, respectively. Motivated by an important problem in quantitative advertising, in Chapter 7 we consider the application of these algorithms to modeling rare events. We show that when the sample size is large but the observed number of successes is small, these data augmentation samplers mix very slowly, with a spectral gap that converges to zero at a rate at least proportional to the reciprocal of the square root of the sample size up to a log factor. In simulation studies, moderate sample sizes result in high autocorrelations and small effective sample sizes. Similar empirical results are observed for related data augmentation samplers for multinomial logit and probit models. When applied to a real quantitative advertising dataset, the data augmentation samplers mix very poorly. Conversely, Hamiltonian Monte Carlo and a type of independence chain Metropolis algorithm show good mixing on the same dataset.

The Politics of Gender Socialization

This manuscript is comprised of three papers that examine the far-reaching and often invisible political outcomes of gender role socialization in the United States. These papers focus primarily on two areas: political confidence amongst girls and women, and the effects of gender on survey measurement and data quality.

Chapter one focuses on political confidence, and the likelihood that women will run for political office. Women continue to be underrepresented at all levels of political leadership, and their lack of political ambition, relative to men, has been identified as a primary cause. In this paper, I explore the relationship between an individual's masculinity and femininity and her development of political ambition. Using original survey data from the 2012 Cooperative Congressional Election Study (CCES), I first empirically demonstrate that gender (masculinity/femininity) and sex (male/female) are unique elements of identity and, moreover, are both independently related to political ambition. I then explore the relevance of gender for the study of candidate emergence, testing whether and how masculinity and femininity might be related to political ambition are supported empirically. While the results suggest that masculinity is positively associated with the development of political ambition, the relationship between femininity and candidate emergence seems to be more complicated and not what prevailing stereotypes might lead us to expect. Moreover, while the relationship between masculinity and political ambition is the same for men and women, the relationship between femininity and political ambition is very different for women than it is for men. This study suggests that gender role socialization is highly related with both men's and women's desire to seek positions of political leadership.

Chapter two continues this exploration of gendered differences in the development of political ambition, this time exploring how social attractiveness and gendered perceptions of political leadership impact the desire to hold political office.Women are persistently underrepresented as candidates for public office and remain underrepresented at all levels of government in the United States. Previous literature suggests that the gendered ambition gap, gender socialization, insufficient recruitment, media scrutiny, family responsibilities, modern campaign strategies, and political opportunity structures all contribute to the gender imbalance in pools of officeholders and candidates. To explain women's reticence to run, scholars have offered explanations addressing structural, institutional, and individual-level factors that deter women from becoming candidates, especially for high positions in the U.S. government. This paper examines a previously unexplored factor: how dating and socialized norms of sexual attraction affect political ambition. This study investigates whether young, single, and heterosexual women's desire for male attention and fear of being perceived as unattractive or "too ambitious" present obstacles to running for office. The results of these experiments suggest that social expectations about gender, attraction and sexuality, and political office-holding may contribute to women's reticence to pursue political leadership. Chapter two is a co-authored work and represents the joint efforts of Laura Lazarus Frankel, Shauna Shames, and Nadia Farjood.

Chapter 3 bridges survey methodology and gender socialization, focusing on how interviewer sex affects survey measurement and data quality. Specifically, this paper examines whether and how matching interviewer and respondent sex affects panel attrition--respondents dropping out of the study after participating in the first wave. While the majority of research on interviewer effects suggests that matching interviewer and respondent characteristics (homophily) yields higher quality data, little work has examined whether this pattern holds true in the area of panel attrition. Using paradata from the General Social Survey (GSS), I explore this question. My analysis reveals that, despite its broader positive effects on data quality, matching interviewer and respondent sex increases likelihood to attrit. Interestingly, this phenomenon only emerges amongst male respondents. However, while assigning female interviewers to male respondents decreases their propensity to attrit, it also increases the likelihood of biased responses on gender related items. These conflicting outcomes represent a tradeoff for scholars and survey researchers, requiring careful consideration of mode, content, and study goals when designing surveys and/or analyzing survey data. The implications of these patterns and areas for further research are discussed.

Together, these papers illustrate two ways that gender norms are related to political outcomes: they contribute to patterns of candidate emergence and affect the measurement of political attitudes and behaviors.