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.

A Study of How Economic Attitudes Are Shaped by Environmental Shocks and Life Experiences

Social attitudes, attitudes toward financial risk and attitudes toward deferred gratification are thought to influence many important economic decisions over the life-course. In economic theory, these attitudes are key components in diverse models of behavior, including collective action, saving and investment decisions and occupational choice. The relevance of these attitudes have been confirmed empirically. Yet, the factors that influence them are not well understood. This research evaluates how these attitudes are affected by large disruptive events, namely, a natural disaster and a civil conflict, and also by an individual-specific life event, namely, having children.

By implementing rigorous empirical strategies drawing on rich longitudinal datasets, this research project advances our understanding of how life experiences shape these attitudes. Moreover, compelling evidence is provided that the observed changes in attitudes are likely to reflect changes in preferences given that they are not driven just by changes in financial circumstances. Therefore the findings of this research project also contribute to the discussion of whether preferences are really fixed, a usual assumption in economics.

In the first chapter, I study how altruistic and trusting attitudes are affected by exposure to the 2004 Indian Ocean tsunami as long as ten years after the disaster occurred. Establishing a causal relationship between natural disasters and attitudes presents several challenges as endogenous exposure and sample selection can confound the analysis. I take on these challenges by exploiting plausibly exogenous variation in exposure to the tsunami and by relying on a longitudinal dataset representative of the pre-tsunami population in two districts of Aceh, Indonesia. The sample is drawn from the Study of the Tsunami Aftermath and Recovery (STAR), a survey with data collected both before and after the disaster and especially designed to identify the impact of the tsunami. The altruistic and trusting attitudes of the respondents are measured by their behavior in the dictator and trust games. I find that witnessing closely the damage caused by the tsunami but without suffering severe economic damage oneself increases altruistic and trusting behavior, particularly towards individuals from tsunami affected communities. Having suffered severe economic damage has no impact on altruistic behavior but may have increased trusting behavior. These effects do not seem to be caused by the consequences of the tsunami on people’s financial situation. Instead they are consistent with how experiences of loss and solidarity may have shaped social attitudes by affecting empathy and perceptions of who is deserving of aid and trust.

In the second chapter, co-authored with Ryan Brown, Duncan Thomas and Andrea Velasquez, we investigate how attitudes toward financial risk are affected by elevated levels of insecurity and uncertainty brought on by the Mexican Drug War. To conduct our analysis, we pair the Mexican Family Life Survey (MxFLS), a rich longitudinal dataset ideally suited for our purposes, with a dataset on homicide rates at the month and municipality-level. The homicide rates capture well the overall crime environment created by the drug war. The MxFLS elicits risk attitudes by asking respondents to choose between hypothetical gambles with different payoffs. Our strategy to identify a causal effect has two key components. First, we implement an individual fixed effects strategy which allows us to control for all time-invariant heterogeneity. The remaining time variant heterogeneity is unlikely to be correlated with changes in the local crime environment given the well-documented political origins of the Mexican Drug War. We also show supporting evidence in this regard. The second component of our identification strategy is to use an intent-to-treat approach to shield our estimates from endogenous migration. Our findings indicate that exposure to greater local-area violent crime results in increased risk aversion. This effect is not driven by changes in financial circumstances, but may be explained instead by heightened fear of victimization. Nonetheless, we find that having greater economic resources mitigate the impact. This may be due to individuals with greater economic resources being able to avoid crime by affording better transportation or security at work.

The third chapter, co-authored with Duncan Thomas, evaluates whether attitudes toward deferred gratification change after having children. For this study we also exploit the MxFLS, which elicits attitudes toward deferred gratification (commonly known as time discounting) by asking individuals to choose between hypothetical payments at different points in time. We implement a difference-in-difference estimator to control for all time-invariant heterogeneity and show that our results are robust to the inclusion of time varying characteristics likely correlated with child birth. We find that becoming a mother increases time discounting especially in the first two years after childbirth and in particular for those women without a spouse at home. Having additional children does not have an effect and the effect for men seems to go in the opposite direction. These heterogeneous effects suggest that child rearing may affect time discounting due to generated stress or not fully anticipated spending needs.

Advances in Dynamic Modeling and Computation for Count Data

A class of multi-process models is developed for collections of time indexed count data. Autocorrelation in counts is achieved with dynamic models for the natural parameter of the binomial distribution. In addition to modeling binomial time series, the framework includes dynamic models for multinomial and Poisson time series. Markov chain Monte Carlo (MCMC) and Po ́lya-Gamma data augmentation (Polson et al., 2013) are critical for fitting multi-process models of counts. To facilitate computation when the counts are high, a Gaussian approximation to the P ́olya- Gamma random variable is developed.

Three applied analyses are presented to explore the utility and versatility of the framework. The first analysis develops a model for complex dynamic behavior of themes in collections of text documents. Documents are modeled as a “bag of words”, and the multinomial distribution is used to characterize uncertainty in the vocabulary terms appearing in each document. State-space models for the natural parameters of the multinomial distribution induce autocorrelation in themes and their proportional representation in the corpus over time.

The second analysis develops a dynamic mixed membership model for Poisson counts. The model is applied to a collection of time series which record neuron level firing patterns in rhesus monkeys. The monkey is exposed to two sounds simultaneously, and Gaussian processes are used to smoothly model the time-varying rate at which the neuron’s firing pattern fluctuates between features associated with each sound in isolation.

The third analysis presents a switching dynamic generalized linear model for the time-varying home run totals of professional baseball players. The model endows each player with an age specific latent natural ability class and a performance enhancing drug (PED) use indicator. As players age, they randomly transition through a sequence of ability classes in a manner consistent with traditional aging patterns. When the performance of the player significantly deviates from the expected aging pattern, he is identified as a player whose performance is consistent with PED use.

All three models provide a mechanism for sharing information across related series locally in time. The models are fit with variations on the P ́olya-Gamma Gibbs sampler, MCMC convergence diagnostics are developed, and reproducible inference is emphasized throughout the dissertation.

Do the Messages Matter? An Investigation of Classroom Messages and College Students’ Personal Theories about Education

Students hold a number of personal theories about education that influence motivation and achievement in the classroom: theories about their own abilities, knowledge, and the learning process. Therefore, college instructors have a great interest in helping to develop adaptive personal theories in their students. The current studies investigated whether specific messages that instructors send in college classroom might serve as a mechanism of personal theory development. Across 2 studies, 17 college instructors and 401 students completed surveys assessing their personal theories about education at the beginning and end of college courses. Students and instructors reported hearing and sending many messages in the classroom, including instructor help messages, conciliatory messages, uncertainty in the field messages, differential ability messages and generalized positive and negative feedback. Between-class and within-class differences in message reports were associated with students’ personal theories at the end of their courses, controlling for initial personal theories. Students’ initial personal theories were also related to the messages students reported hearing. The findings demonstrate the utility of assessing non-content messages in college classrooms as potential mechanisms for changing students’ personal theories in college. Implications for research and practice are discussed.