Measuring the impact of scientific articles is important for evaluating the research output of individual scientists, academic institutions and journals. While citations are raw data for constructing impact measures, there exist biases and potential issues if factors affecting citation patterns are not properly accounted for. In this work, we address the problem of field variation and introduce an article level metric useful for evaluating individual articles’ visibility. This measure derives from joint probabilistic modeling of the content in the articles and the citations among them using latent Dirichlet allocation (LDA) and the mixed membership stochastic blockmodel (MMSB). Our proposed model provides a visibility metric for individual articles adjusted for field variation in citation rates, a structural understanding of citation behavior in different fields, and article recommendations which take into account article visibility and citation patterns. We develop an efficient algorithm for model fitting using variational methods. To scale up to large networks, we develop an online variant using stochastic gradient methods and case-control likelihood approximation. We apply our methods to the benchmark KDD Cup 2003 dataset with approximately 30,000 high energy physics papers.

Variable selection is gaining more attention because it plays an important role in deriving practical and reliable optimal treatment regimes for personalized medicine, especially when there are a large number of predictors. Most existing variable selection techniques focus on selecting variables that are important for prediction. With such methods, some variables that are poor in prediction but are critical for treatment decision making may be ignored. A qualitative interaction of a variable with treatment arises when the treatment effect changes direction as the value of the variable varies. Variables that have qualitative interactions with treatment are of clinical importance for treatment decision making. Gunter, Zhu and Murphy [Stat. Methodol. 8 (2011) 42–55] proposed the S-score method to characterize the magnitude of qualitative interaction of an individual variable with treatment. In this paper, we develop a sequential advantage selection method based on a modified S-score. Our method sequentially selects variables with a qualitative interaction and can be applied in multiple decision-point settings. To select the best candidate subset of variables for decision making, we also propose a BIC-type criterion that is based on the sequential advantage. The empirical performance of the proposed method is evaluated by simulation and an application to depression data from a clinical trial.

We hypothesize that if election results are manipulated or forged, then, due to the well-known human attraction to round numbers, the frequency of reported round percentages can be increased. To test this hypothesis, we analyzed raw data from seven federal elections held in the Russian Federation during the period from 2000 to 2012 and found that in all elections since 2004 the number of polling stations reporting turnout and/or leader’s result expressed by an integer percentage (as opposed to a fractional value) was much higher than expected by pure chance. Monte Carlo simulations confirmed high statistical significance of the observed phenomenon, thereby suggesting its man-made nature. Geographical analysis showed that these anomalies were concentrated in a specific subset of Russian regions which strongly suggests its orchestrated origin. Unlike previously proposed statistical indicators of alleged electoral falsifications, our observations can hardly be explained differently but by a widespread election fraud.

Heat waves merit careful study because they inflict severe economic and societal damage. We use an intuitive, informal working definition of a heat wave—a persistent event in the tail of the temperature distribution—to motivate an interpretable latent state extreme value model. A latent variable with dependence in time indicates membership in the heat wave state. The strength of the temporal dependence of the latent variable controls the frequency and persistence of heat waves. Within each heat wave, temperatures are modeled using extreme value distributions, with extremal dependence across time accomplished through an extreme value Markov model. One important virtue of interpretability is that model parameters directly translate into quantities of interest for risk management, so that questions like whether heat waves are becoming longer, more severe or more frequent are easily answered by querying an appropriate fitted model. We demonstrate the latent state model on two recent, calamitous, examples: the European heat wave of 2003 and the Russian heat wave of 2010.

In observational surveys, post-stratification is used to reduce bias resulting from differences between the survey population and the population under investigation. However, this can lead to inflated post-stratification weights and, therefore, appropriate methods are required to obtain less variable estimates. Proposed methods include collapsing post-strata, trimming post-stratification weights, generalized regression estimators (GREG) and weight smoothing models, the latter defined by random-effects models that induce shrinkage across post-stratum means. Here, we first describe the weight-smoothing model for prevalence estimation from binary survey outcomes in observational surveys. Second, we propose an extension of this method for trend estimation. And, third, a method is provided such that the GREG can be used for prevalence and trend estimation for observational surveys. Variance estimates of all methods are described. A simulation study is performed to compare the proposed methods with other established methods. The performance of the nonparametric GREG is consistent over all simulation conditions and therefore serves as a valuable solution for prevalence and trend estimation from observational surveys. The method is applied to the estimation of the prevalence and incidence trend of influenza-like illness using the 2010/2011 Great Influenza Survey in Flanders, Belgium.

Many panel studies collect refreshment samples—new, randomly sampled respondents who complete the questionnaire at the same time as a subsequent wave of the panel. With appropriate modeling, these samples can be leveraged to correct inferences for biases caused by nonignorable attrition. We present such a model when the panel includes many categorical survey variables. The model relies on a Bayesian latent pattern mixture model, in which an indicator for attrition and the survey variables are modeled jointly via a latent class model. We allow the multinomial probabilities within classes to depend on the attrition indicator, which offers additional flexibility over standard applications of latent class models. We present results of simulation studies that illustrate the benefits of this flexibility. We apply the model to correct attrition bias in an analysis of data from the 2007–2008 Associated Press/Yahoo News election panel study.

As a technique to investigate link-level loss rates of a computer network with low operational cost, loss tomography has received considerable attentions in recent years. A number of parameter estimation methods have been proposed for loss tomography of networks with a tree structure as well as a general topological structure. However, these methods suffer from either high computational cost or insufficient use of information in the data. In this paper, we provide both theoretical results and practical algorithms for parameter estimation in loss tomography. By introducing a group of novel statistics and alternative parameter systems, we find that the likelihood function of the observed data from loss tomography keeps exactly the same mathematical formulation for tree and general topologies, revealing that networks with different topologies share the same mathematical nature for loss tomography. More importantly, we discover that a reparametrization of the likelihood function belongs to the standard exponential family, which is convex and has a unique mode under regularity conditions. Based on these theoretical results, novel algorithms to find the MLE are developed. Compared to existing methods in the literature, the proposed methods enjoy great computational advantages.

We consider the problem of estimating a common period for a set of functions sampled at irregular intervals. The motivating problem arises in astronomy, where the functions represent a star’s observed brightness over time through different photometric filters. While current methods perform well when the brightness is sampled densely enough in at least one filter, they break down when no brightness function is densely sampled. In this paper we introduce two new methods for period estimation in this important latter case. The first, multiband generalized Lomb–Scargle (MGLS), extends the frequently used Lomb–Scargle method to naïvely combine information across filters. The second, penalized generalized Lomb–Scargle (PGLS), builds on MGLS by more intelligently borrowing strength across filters. Specifically, we incorporate constraints on the phases and amplitudes across the different functions using a nonconvex penalized likelihood function. We develop a fast algorithm to optimize the penalized likelihood that combines block coordinate descent with the majorization–minimization (MM) principle. We test and validate our methods on synthetic and real astronomy data. Both PGLS and MGLS improve period estimation accuracy over current methods based on using a single function; moreover, PGLS outperforms MGLS and other leading methods when the functions are sparsely sampled.

New representations of tree-structured data objects, using ideas from topological data analysis, enable improved statistical analyses of a population of brain artery trees. A number of representations of each data tree arise from persistence diagrams that quantify branching and looping of vessels at multiple scales. Novel approaches to the statistical analysis, through various summaries of the persistence diagrams, lead to heightened correlations with covariates such as age and sex, relative to earlier analyses of this data set. The correlation with age continues to be significant even after controlling for correlations from earlier significant summaries.

Event-related potentials (ERPs) are recordings of electrical activity along the scalp time-locked to perceptual, motor and cognitive events. Because ERP signals are often rare and weak, relative to the large between-subject variability, establishing significant associations between ERPs and behavioral (or experimental) variables of interest poses major challenges for statistical analysis.

Noting that ERP time dependence exhibits a block pattern suggesting strong local and long-range autocorrelation components, we propose a flexible factor modeling of dependence. An adaptive factor adjustment procedure is derived from a joint estimation of the signal and noise processes, given a prior knowledge of the noise-alone intervals. A simulation study is presented using known signals embedded in a real dependence structure extracted from authentic ERP measurements. The proposed procedure performs well compared with existing multiple testing procedures and is more powerful at discovering interesting ERP features.

The latent multinomial model (LMM) of Link et al. [Biometrics 66 (2010) 178–185] provides a framework for modelling mark-recapture data with potential identification errors. Key is a Markov chain Monte Carlo (MCMC) scheme for sampling configurations of the latent counts of the true capture histories that could have generated the observed data. Assuming a linear map between the observed and latent counts, the MCMC algorithm uses vectors from a basis of the kernel to move between configurations of the latent data. Schofield and Bonner [Biometrics 71 (2015) 1070–1080] shows that this is sufficient for some models within the framework but that a larger set called a Markov basis is required when errors are more complex. We address two further challenges: (1) that models with complex error mechanisms may not fit within the LMM framework and (2) that Markov bases can be difficult to compute for studies of even moderate size. We extend the framework to model the capture/demographic and error processes separately and develop a new MCMC algorithm using dynamic Markov bases. Our work is motivated by a study of queen snakes (Regina septemvittata) and we use simulation to compare estimates of survival rates when snakes are marked with PIT tags which have perfect identification versus brands which are prone to error.

Capture–recapture data are often collected when abundance estimation is of interest. In this manuscript we focus on abundance estimation of closed populations. In the presence of unobserved individual heterogeneity, specified on a continuous scale for the capture probabilities, the likelihood is not generally available in closed form, but expressible only as an analytically intractable integral. Model-fitting algorithms to estimate abundance most notably include a numerical approximation for the likelihood or use of a Bayesian data augmentation technique considering the complete data likelihood. We consider a Bayesian hybrid approach, defining a “semi-complete” data likelihood, composed of the product of a complete data likelihood component for individuals seen at least once within the study and a marginal data likelihood component for the individuals not seen within the study, approximated using numerical integration. This approach combines the advantages of the two different approaches, with the semi-complete likelihood component specified as a single integral (over the dimension of the individual heterogeneity component). In addition, the models can be fitted within BUGS/JAGS (commonly used for the Bayesian complete data likelihood approach) but with significantly improved computational efficiency compared to the commonly used superpopulation data augmentation approaches (between about 10 and 77 times more efficient in the two examples we consider). The semi-complete likelihood approach is flexible and applicable to a range of models, including spatially explicit capture–recapture models. The model-fitting approach is applied to two different data sets: the first relates to snowshoe hares where model $M_{h}$ is applied and the second to gibbons where a spatially explicit capture–recapture model is applied.

The availability of data from electronic health records facilitates the development and evaluation of risk-prediction models, but estimation of prediction accuracy could be limited by outcome misclassification, which can arise if events are not captured. We evaluate the robustness of prediction accuracy summaries, obtained from receiver operating characteristic curves and risk-reclassification methods, if events are not captured (i.e., “false negatives”). We derive estimators for sensitivity and specificity if misclassification is independent of marker values. In simulation studies, we quantify the potential for bias in prediction accuracy summaries if misclassification depends on marker values. We compare the accuracy of alternative prognostic models for 30-day all-cause hospital readmission among 4548 patients discharged from the University of Pennsylvania Health System with a primary diagnosis of heart failure. Simulation studies indicate that if misclassification depends on marker values, then the estimated accuracy improvement is also biased, but the direction of the bias depends on the direction of the association between markers and the probability of misclassification. In our application, 29% of the 1143 readmitted patients were readmitted to a hospital elsewhere in Pennsylvania, which reduced prediction accuracy. Outcome misclassification can result in erroneous conclusions regarding the accuracy of risk-prediction models.

While time series prediction is an important, actively studied problem, the predictive accuracy of time series models is complicated by nonstationarity. We develop a fast and effective approach to allow for nonstationarity in the parameters of a chosen time series model. In our power-weighted density (PWD) approach, observations in the distant past are down-weighted in the likelihood function relative to more recent observations, while still giving the practitioner control over the choice of data model. One of the most popular nonstationary techniques in the academic finance community, rolling window estimation, is a special case of our PWD approach. Our PWD framework is a simpler alternative compared to popular state–space methods that explicitly model the evolution of an underlying state vector. We demonstrate the benefits of our PWD approach in terms of predictive performance compared to both stationary models and alternative nonstationary methods. In a financial application to thirty industry portfolios, our PWD method has a significantly favorable predictive performance and draws a number of substantive conclusions about the evolution of the coefficients and the importance of market factors over time.

Most previous studies of the causal relationship between malaria and stunting have been studies where potential confounders are controlled via regression-based methods, but these studies may have been biased by unobserved confounders. Instrumental variables (IV) regression offers a way to control for unmeasured confounders where, in our case, the sickle cell trait can be used as an instrument. However, for the instrument to be valid, it may still be important to account for measured confounders. The most commonly used instrumental variable regression method, two-stage least squares, relies on parametric assumptions on the effects of measured confounders to account for them. Additionally, two-stage least squares lacks transparency with respect to covariate balance and weighing of subjects and does not blind the researcher to the outcome data. To address these drawbacks, we propose an alternative method for IV estimation based on full matching. We evaluate our new procedure on simulated data and real data concerning the causal effect of malaria on stunting among children. We estimate that the risk of stunting among children with the sickle cell trait decreases by 0.22 per every malaria episode prevented by the sickle cell trait, a substantial effect of malaria on stunting ($p$-value: 0.011, 95% CI: 0.044, 1).

Recent studies have examined racial disparities in stop-and-frisk, a widely employed but controversial policing tactic. The statistical evidence, however, has been limited and contradictory. We investigate by analyzing three million stops in New York City over five years, focusing on cases where officers suspected the stopped individual of criminal possession of a weapon (CPW). For each CPW stop, we estimate the ex ante probability that the detained suspect has a weapon. We find that in more than 40% of cases, the likelihood of finding a weapon (typically a knife) was less than 1%, raising concerns that the legal requirement of “reasonable suspicion” was often not met. We further find that blacks and Hispanics were disproportionately stopped in these low hit rate contexts, a phenomenon that we trace to two factors: (1) lower thresholds for stopping individuals—regardless of race—in high-crime, predominately minority areas, particularly public housing; and (2) lower thresholds for stopping minorities relative to similarly situated whites. Finally, we demonstrate that by conducting only the 6% of stops that are statistically most likely to result in weapons seizure, one can both recover the majority of weapons and mitigate racial disparities in who is stopped. We show that this statistically informed stopping strategy can be approximated by simple, easily implemented heuristics with little loss in efficiency.

Whole genome sequencing of pathogens from multiple hosts in an epidemic offers the potential to investigate who infected whom with unparalleled resolution, potentially yielding important insights into disease dynamics and the impact of control measures. We considered disease outbreaks in a setting with dense genomic sampling, and formulated stochastic epidemic models to investigate person-to-person transmission, based on observed genomic and epidemiological data. We constructed models in which the genetic distance between sampled genotypes depends on the epidemiological relationship between the hosts. A data-augmented Markov chain Monte Carlo algorithm was used to sample over the transmission trees, providing a posterior probability for any given transmission route. We investigated the predictive performance of our methodology using simulated data, demonstrating high sensitivity and specificity, particularly for rapidly mutating pathogens with low transmissibility. We then analyzed data collected during an outbreak of methicillin-resistant Staphylococcus aureus in a hospital, identifying probable transmission routes and estimating epidemiological parameters. Our approach overcomes limitations of previous methods, providing a framework with the flexibility to allow for unobserved infection times, multiple independent introductions of the pathogen and within-host genetic diversity, as well as allowing forward simulation.

We propose a Multiple Imputation Random Lasso (MIRL) method to select important variables and to predict the outcome for an epidemiological study of Eating and Activity in Teens. In this study $80\%$ of individuals have at least one variable missing. Therefore, using variable selection methods developed for complete data after listwise deletion substantially reduces prediction power. Recent work on prediction models in the presence of incomplete data cannot adequately account for large numbers of variables with arbitrary missing patterns. We propose MIRL to combine penalized regression techniques with multiple imputation and stability selection. Extensive simulation studies are conducted to compare MIRL with several alternatives. MIRL outperforms other methods in high-dimensional scenarios in terms of both reduced prediction error and improved variable selection performance, and it has greater advantage when the correlation among variables is high and missing proportion is high. MIRL is shown to have improved performance when comparing with other applicable methods when applied to the study of Eating and Activity in Teens for the boys and girls separately, and to a subgroup of low social economic status (SES) Asian boys who are at high risk of developing obesity.

Small area estimation techniques are now routinely used to generate local-level poverty estimates for aid allocation and poverty monitoring in developing countries. However, the widely implemented World Bank (WB) or Elbers, Lanjouw and Lanjouw [Econometrica 71 (2003) 355–364] (ELL) method can only be used when a survey and census are conducted at approximately the same time. The empirical best prediction (EBP) method of Molina and Rao [Canad. J. Statist. 38 (2010) 369–385] also requires a new census for updating. Hence, if small area estimation methods that use both survey and census unit record data are required, and the survey is rerun some years after the census, how to update small area estimates becomes an important issue. In this paper, we propose an intercensal updating method for local-level poverty estimates with estimated standard errors which we call Extended Structure PREserving Estimation (ESPREE). This method is a new extension of classical Structure PREserving Estimation (SPREE). We test our approach by applying it to inter-censal municipal-level poverty estimation and carrying out a validation exercise in the Philippines, comparing the estimates generated with an alternative ELL or EBP updating method due to Lanjouw and van der Wiede [Determining changes in welfare distributions at the micro-level: Updating poverty maps. (2006) Powerpoint presentation at the NSCB Workshop for the NSCB/World Bank Intercensal Updating Project] which uses time-invariant variables. The results show that the ESPREE estimates are preferable, generally being unbiased and concurring well with local experts’ opinion on poverty levels at the time of the updated survey.

Future climate change impacts depend on temperatures not only through changes in their means but also through changes in their variability. General circulation models (GCMs) predict changes in both means and variability; however, GCM output should not be used directly as simulations for impacts assessments because GCMs do not fully reproduce present-day temperature distributions. This paper addresses an ensuing need for simulations of future temperatures that combine both the observational record and GCM projections of changes in means and temporal covariances. Our perspective is that such simulations should be based on transforming observations to account for GCM projected changes, in contrast to methods that transform GCM output to account for discrepancies with observations. Our methodology is designed for simulating transient (nonstationary) climates, which are evolving in response to changes in CO$_{2}$ concentrations (as is the Earth at present). This work builds on previously described methods for simulating equilibrium (stationary) climates. Since the proposed simulation relies on GCM projected changes in covariance, we describe a statistical model for the evolution of temporal covariances in a GCM under future forcing scenarios, and apply this model to an ensemble of runs from one GCM, CCSM3. We find that, at least in CCSM3, changes in the local covariance structure can be explained as a function of the regional mean change in temperature and the rate of change of warming. This feature means that the statistical model can be used to emulate the evolving covariance structure of GCM temperatures under scenarios for which the GCM has not been run. When combined with an emulator for mean temperature, our methodology can simulate evolving temperatures under such scenarios, in a way that accounts for projections of changes while still retaining fidelity with the observational record. The emulator for variability changes is also of interest on its own as a summary of GCM projections of variability changes.

Histone modification is a vital epigenetic mechanism for transcriptional control in eukaryotes. High-throughput techniques have enabled whole-genome analysis of histone modifications in recent years. However, most studies assume one combination of histone modification invariantly translates to one transcriptional output regardless of local chromatin environment. In this study we hypothesize that the genome is organized into local domains that manifest a similar enrichment pattern of histone modification, which leads to orchestrated regulation of expression of genes with relevant biological functions. We propose a multivariate Bayesian Change Point (BCP) model to segment the Drosophila melanogaster genome into consecutive blocks on the basis of combinatorial patterns of histone marks. By modeling the sparse distribution of histone marks with a zero-inflated Gaussian mixture, our partitions capture local BLOCKs that manifest a relatively homogeneous enrichment pattern of histone marks. We further characterized BLOCKs by their transcription levels, distribution of genes, degree of co-regulation and GO enrichment. Our results demonstrate that these BLOCKs, although inferred merely from histone modifications, reveal a strong relevance with physical domains, which suggest their important roles in chromatin organization and coordinated gene regulation.

Longitudinal imaging studies have both spatial and temporal correlation among the multiple outcome measurements from a subject. Statistical methods of analysis must properly account for this autocorrelation. In this work we discuss how a linear model with a separable parametric correlation structure could be used to analyze data from such a study. The goal of this paper is to provide an easily understood description of how such a model works and discuss how it can be applied to real data. Model assumptions are discussed and the process of selecting a working correlation structure is thoroughly discussed. The steps necessitating collaboration between statistical and scientific investigators have been highlighted, as have considerations for missing data or uneven follow-up.

The results from a completed longitudinal cardiac imaging study were considered for illustration purposes. The data comes from a clinical trial for medical therapy for patients with mitral regurgitation, with repeated measurements taken at sixteen locations from the left ventricle to measure disease progression. The spatiotemporal correlation model was compared to previously used summary measures to demonstrate improved power as well as increased flexibility in the use of time- and space-varying predictors.