Variance estimation for systematic sampling from deliberately ordered populations

The systematic sampling (SYS) design (Madow and Madow, 1944) is widely used by statistical offices due to its simplicity and efficiency (e.g., Iachan, 1982). But it suffers from a serious defect, namely, that it is impossible to unbiasedly estimate the sampling variance (Iachan, 1982) and usual variance estimators (Yates and Grundy, 1953) are inadequate and can overestimate the variance significantly (Särndal et al., 1992). We propose a novel variance estimator which is less biased and that can be implemented with any given population order. We will justify this estimator theoretically and with a Monte Carlo simulation study.

Identifying adaptive genetic divergence among populations from genome scans

The identification of signatures of natural selection in genomic surveys has become an area of intense research, stimulated by the increasing ease with which genetic markers can be typed. Loci identified as subject to selection may be functionally important, and hence (weak) candidates for involvement in disease causation. They can also be useful in determining the adaptive differentiation of populations, and exploring hypotheses about speciation. Adaptive differentiation has traditionally been identified from differences in allele frequencies among different populations, summarised by an estimate of F-ST. Low outliers relative to an appropriate neutral population-genetics model indicate loci subject to balancing selection, whereas high outliers suggest adaptive (directional) selection. However, the problem of identifying statistically significant departures from neutrality is complicated by confounding effects on the distribution of F-ST estimates, and current methods have not yet been tested in large-scale simulation experiments. Here, we simulate data from a structured population at many unlinked, diallelic loci that are predominantly neutral but with some loci subject to adaptive or balancing selection. We develop a hierarchical-Bayesian method, implemented via Markov chain Monte Carlo (MCMC), and assess its performance in distinguishing the loci simulated under selection from the neutral loci. We also compare this performance with that of a frequentist method, based on moment-based estimates of F-ST. We find that both methods can identify loci subject to adaptive selection when the selection coefficient is at least five times the migration rate. Neither method could reliably distinguish loci under balancing selection in our simulations, even when the selection coefficient is twenty times the migration rate.

Conserving marginal populations of the foodplant (Impatiens noli-tangere) of an endangered moth (Eustroma reticulatum) in a changing climate

Impatiens noli-tangere is scarce in the UK and probably only native to the Lake District and Wales. It is the sole food plant for the endangered moth Eustroma reticulattum. Significant annual fluctuations in the size of I. noli-tangere populations endanger the continued presence of E. reticulatum in the UK. In this study, variation in population size was monitored across native populations of L noli-tangere in the English Lake District and Wales. In 1998, there was a crash in the population size of all metapopulations in the Lake District but not of those found in Wales. A molecular survey of the genetic affinities of samples in 1999 from both regions and a reference population from Switzerland was performed using AFLP and ISSR analyses. The consensus UPGMA dendrogram and a PCO scatter plot revealed clear differentiation between the populations of L noli-tangere in Wales and those in the Lake District. Most of the genetic variation in the UK (H-T= 0.064) was partitioned between (G(ST) = 0.455) rather than within (H-S = 0.034) regions, inferring little gene flow occurs between regions. There was similar bias towards differentiation between metapopulations in Wales, again consistent with low levels of interpopulation gene flow. This contrasts with far lower levels of differentiation in the Lake District which suggests modest rates of gene flow may occur between populations. It is concluded that in the event of local extinction of sites or populations, reintroductions should be restricted to samples collected from the same region. We then surveyed climatic variables to identify those most likely to cause local extinctions. Climatic correlates of population size were sought from two Lake District metapopulations situated close to a meteorological station. A combination of three climatic variables common to both sites explained 81-84% of the variation in plant number between 1990 and 2001. Projected trends for these climatic variables were used in a Monte Carlo simulation which suggested an increased risk of I. noli-tangere population crashes by 2050 at Coniston Water. but not at Derwentwater. Implications of these findings for practical conservation strategies are explored. (C) 2003 Elsevier Ltd. All rights reserved.

Linkage disequilibrium assessment via log-linear modeling of SNP haplotype frequencies

Analyses of high-density single-nucleotide polymorphism (SNP) data, such as genetic mapping and linkage disequilibrium (LD) studies, require phase-known haplotypes to allow for the correlation between tightly linked loci. However, current SNP genotyping technology cannot determine phase, which must be inferred statistically. In this paper, we present a new Bayesian Markov chain Monte Carlo (MCMC) algorithm for population haplotype frequency estimation, particulary in the context of LD assessment. The novel feature of the method is the incorporation of a log-linear prior model for population haplotype frequencies. We present simulations to suggest that 1) the log-linear prior model is more appropriate than the standard coalescent process in the presence of recombination (>0.02cM between adjacent loci), and 2) there is substantial inflation in measures of LD obtained by a "two-stage" approach to the analysis by treating the "best" haplotype configuration as correct, without regard to uncertainty in the recombination process. Genet Epidemiol 25:106-114, 2003. (C) 2003 Wiley-Liss, Inc.

Mixed initial-boundary value problem in particle modeling of microelectronic devices

The Boltzmann equation in presence of boundary and initial conditions, which describes the general case of carrier transport in microelectronic devices is analysed in terms of Monte Carlo theory. The classical Ensemble Monte Carlo algorithm which has been devised by merely phenomenological considerations of the initial and boundary carrier contributions is now derived in a formal way. The approach allows to suggest a set of event-biasing algorithms for statistical enhancement as an alternative of the population control technique, which is virtually the only algorithm currently used in particle simulators. The scheme of the self-consistent coupling of Boltzmann and Poisson equation is considered for the case of weighted particles. It is shown that particles survive the successive iteration steps.

Likelihood-free estimation of model evidence

Statistical methods of inference typically require the likelihood function to be computable in a reasonable amount of time. The class of “likelihood-free” methods termed Approximate Bayesian Computation (ABC) is able to eliminate this requirement, replacing the evaluation of the likelihood with simulation from it. Likelihood-free methods have gained in efficiency and popularity in the past few years, following their integration with Markov Chain Monte Carlo (MCMC) and Sequential Monte Carlo (SMC) in order to better explore the parameter space. They have been applied primarily to estimating the parameters of a given model, but can also be used to compare models. Here we present novel likelihood-free approaches to model comparison, based upon the independent estimation of the evidence of each model under study. Key advantages of these approaches over previous techniques are that they allow the exploitation of MCMC or SMC algorithms for exploring the parameter space, and that they do not require a sampler able to mix between models. We validate the proposed methods using a simple exponential family problem before providing a realistic problem from human population genetics: the comparison of different demographic models based upon genetic data from the Y chromosome.

Social organization of marine Tucuxi dolphins, Sotalia guianensis, in the Cananeia Estuary of southeastern Brazil

Social organization is an important component of the population biology of a species that influences gene flow, the spatial pattern and scale of movements, and the effects of predation or exploitation by humans. An important element of social structure in mammals is group fidelity, which can be quantified through association indices. To describe the social organization of marine tucuxi dolphins (Sotalia guianensis) found in the Cananeia estuary, southeastern Brazil, association indices were applied to photo-identification data to characterize the temporal stability of relationships among members of this population. Eighty-seven days of fieldwork were conducted from May 2000 to July 2003, resulting in direct observations of 374 distinct groups. A total of 138 dolphins were identified on 1-38 distinct field days. Lone dolphins were rarely seen, whereas groups were composed of up to 60 individuals (mean +/- 1 SD = 12.4 +/- 11.4 individuals per group). A total of 29,327 photographs were analyzed, of which 6,312 (21.5%) were considered useful for identifying individuals. Half-weight and simple ratio indices were used to investigate associations among S. guianensis as revealed by the entire data set, data from the core study site, and data from groups composed of <= 10 individuals. Monte Carlo methods indicated that only 3 (9.3%) of 32 association matrices differed significantly from expectations based on random association. Thus, our study suggests that stable associations are not characteristic of S. guianensis in the Cananeia estuary.

Numerical analysis of an incremented statistical sampling procedure in MCNP

MCNP has stood so far as one of the main Monte Carlo radiation transport codes. Its use, as any other Monte Carlo based code, has increased as computers perform calculations faster and become more affordable along time. However, the use of Monte Carlo method to tally events in volumes which represent a small fraction of the whole system may turn to be unfeasible, if a straight analogue transport procedure (no use of variance reduction techniques) is employed and precise results are demanded. Calculations of reaction rates in activation foils placed in critical systems turn to be one of the mentioned cases. The present work takes advantage of the fixed source representation from MCNP to perform the above mentioned task in a more effective sampling way (characterizing neutron population in the vicinity of the tallying region and using it in a geometric reduced coupled simulation). An extended analysis of source dependent parameters is studied in order to understand their influence on simulation performance and on validity of results. Although discrepant results have been observed for small enveloping regions, the procedure presents itself as very efficient, giving adequate and precise results in shorter times than the standard analogue procedure. (C) 2007 Elsevier Ltd. All rights reserved.

Stochastic lattice gas model describing the dynamics of the SIRS epidemic process

We study a stochastic process describing the onset of spreading dynamics of an epidemic in a population composed of individuals of three classes: susceptible (S), infected (I), and recovered (R). The stochastic process is defined by local rules and involves the following cyclic process: S -> I -> R -> S (SIRS). The open process S -> I -> R (SIR) is studied as a particular case of the SIRS process. The epidemic process is analyzed at different levels of description: by a stochastic lattice gas model and by a birth and death process. By means of Monte Carlo simulations and dynamical mean-field approximations we show that the SIRS stochastic lattice gas model exhibit a line of critical points separating the two phases: an absorbing phase where the lattice is completely full of S individuals and an active phase where S, I and R individuals coexist, which may or may not present population cycles. The critical line, that corresponds to the onset of epidemic spreading, is shown to belong in the directed percolation universality class. By considering the birth and death process we analyze the role of noise in stabilizing the oscillations. (C) 2009 Elsevier B.V. All rights reserved.

Risk of Giardia infection for drinking water and bathing in a peri-urban area in Sao Paulo, Brazil

A high incidence of waterborne diseases is observed worldwide and in order to address contamination problems prior to an outbreak, quantitative microbial risk assessment is a useful tool for estimating the risk of infection. The objective of this paper was to assess the probability of Giardia infection from consuming water from shallow wells in a peri-urban area. Giardia has been described as an important waterborne pathogen and reported in several water sources, including ground waters. Sixteen water samples were collected and examined according to the US EPA (1623, 2005). A Monte Carlo method was used to address the potential risk as described by the exponential dose response model. Giardia cysts occurred in 62.5% of the samples (0.1-36.1 cysts/l). A median risk of 10-1 for the population was estimated and the adult ingestion was the highest risk driver. This study illustrates the vulnerability of shallow well water supply systems in peri-urban areas.

Colorectal cancer screening in Australia: an economic evaluation of a potential biennial screening program using faecal occult blood tests

Objective: To evaluate whether the introduction of a national, co-ordinated screening program using the faecal occult blood test represents 'value-for-money' from the perspective of the Australian Government as third-party funder.  Methods: The annual equivalent costs and consequences of a   biennial screening program in 'steady-state' operation were estimated for the Australian population using 1996 as the reference year. Disability-adjusted life years (DALYs) and the years of life lost (YLLs) averted, and the health service costs were modelled, based on the epidemiology and the costs of colorectal cancer in Australia together with the mortality reduction achieved in randomised controlled trials. Uncertainty in the model was examined using Monte Carlo simulation methods. Results: We estimate a minimum or 'base program' of screening those aged 55 to 69 years could avert 250 deaths per annum (95% uncertainty interval 99–400), at a gross cost of $A55 million (95% UI $A46 million to $A96 million) and a gross incremental cost-effectiveness ratio of $A17,000/DALY (95% UI $A13,000/DALY to $A52,000/DALY). Extending the program to include 70 to 74-year-olds is a more effective option (cheaper and higher health gain) than including the 50 to 54-year-olds. Conclusions: The findings of this study support the case for a national program directed at the 55 to 69-year-old age group with extension to 70 to 74-year-olds if there are sufficient resources. The pilot tests recently announced in Australia provide an important opportunity to consider the age range for screening and the sources of uncertainty, identified in the modelled evaluation, to assist decisions on implementing a full national program.

Dopamine transporter (DAT1) VNTR polymorphism and alcoholism in two culturally different populations of South India

It is well established that the central dopaminergic reward pathway is likely involved in alcohol intake and the progression of alcohol dependence. Dopamine transporter (DAT1) mediates the active re-uptake of DA from the synapse and is a principal regulator of dopaminergic neurotransmission. The gene for the human DAT1 displays several polymorphisms, including a 40-bp variable number of tandem repeats (VNTR) ranging from 3 to 16 copies in the 3′-untranslated region (UTR) of the gene. To assess the role of this gene in alcoholism, we genotyped the VNTR of DAT1 gene in a sample of 206 subjects from the Kota population (111 alcohol dependence cases and 95 controls) and 142 subjects from Badaga population (81 alcohol dependence cases and 61 controls). Both populations inhabit a similar environmental zone, but have different ethnic histories. Phenotype was defined based on the DSM-IV criteria. Genotyping was performed using PCR and electrophoresis. The association of DAT1 with alcoholism was tested by using the Clump v1.9 program which uses the Monte Carlo method. In both Kota and Badaga populations, the allele A10 was the most frequent allele followed by allele A9. The genotypic distribution is in Hardy–Weinberg equilibrium in both cases and control groups of Kota and Badaga populations. The DAT1 VNTR was significantly associated with alcoholism in Badaga population but not in Kota population. Our results suggest that the A9 allele of the DAT gene is involved in vulnerability to alcoholism, but that these associations are population specific.

Inference in differences-in-differences with few treated groups and heteroskedasticity

Differences-in-Differences (DID) is one of the most widely used identification strategies in applied economics. However, how to draw inferences in DID models when there are few treated groups remains an open question. We show that the usual inference methods used in DID models might not perform well when there are few treated groups and errors are heteroskedastic. In particular, we show that when there is variation in the number of observations per group, inference methods designed to work when there are few treated groups tend to (under-) over-reject the null hypothesis when the treated groups are (large) small relative to the control groups. This happens because larger groups tend to have lower variance, generating heteroskedasticity in the group x time aggregate DID model. We provide evidence from Monte Carlo simulations and from placebo DID regressions with the American Community Survey (ACS) and the Current Population Survey (CPS) datasets to show that this problem is relevant even in datasets with large numbers of observations per group. We then derive an alternative inference method that provides accurate hypothesis testing in situations where there are few treated groups (or even just one) and many control groups in the presence of heteroskedasticity. Our method assumes that we can model the heteroskedasticity of a linear combination of the errors. We show that this assumption can be satisfied without imposing strong assumptions on the errors in common DID applications. With many pre-treatment periods, we show that this assumption can be relaxed. Instead, we provide an alternative inference method that relies on strict stationarity and ergodicity of the time series. Finally, we consider two recent alternatives to DID when there are many pre-treatment periods. We extend our inference methods to linear factor models when there are few treated groups. We also derive conditions under which a permutation test for the synthetic control estimator proposed by Abadie et al. (2010) is robust to heteroskedasticity and propose a modification on the test statistic that provided a better heteroskedasticity correction in our simulations.

Inference in differences-in-differences with few treated groups and heteroskedasticity

Differences-in-Differences (DID) is one of the most widely used identification strategies in applied economics. However, how to draw inferences in DID models when there are few treated groups remains an open question. We show that the usual inference methods used in DID models might not perform well when there are few treated groups and errors are heteroskedastic. In particular, we show that when there is variation in the number of observations per group, inference methods designed to work when there are few treated groups tend to (under-) over-reject the null hypothesis when the treated groups are (large) small relative to the control groups. This happens because larger groups tend to have lower variance, generating heteroskedasticity in the group x time aggregate DID model. We provide evidence from Monte Carlo simulations and from placebo DID regressions with the American Community Survey (ACS) and the Current Population Survey (CPS) datasets to show that this problem is relevant even in datasets with large numbers of observations per group. We then derive an alternative inference method that provides accurate hypothesis testing in situations where there are few treated groups (or even just one) and many control groups in the presence of heteroskedasticity. Our method assumes that we know how the heteroskedasticity is generated, which is the case when it is generated by variation in the number of observations per group. With many pre-treatment periods, we show that this assumption can be relaxed. Instead, we provide an alternative application of our method that relies on assumptions about stationarity and convergence of the moments of the time series. Finally, we consider two recent alternatives to DID when there are many pre-treatment groups. We extend our inference method to linear factor models when there are few treated groups. We also propose a permutation test for the synthetic control estimator that provided a better heteroskedasticity correction in our simulations than the test suggested by Abadie et al. (2010).

Cherry picking with synthetic controls

The synthetic control (SC) method has been recently proposed as an alternative method to estimate treatment e ects in comparative case studies. Abadie et al. [2010] and Abadie et al. [2015] argue that one of the advantages of the SC method is that it imposes a data-driven process to select the comparison units, providing more transparency and less discretionary power to the researcher. However, an important limitation of the SC method is that it does not provide clear guidance on the choice of predictor variables used to estimate the SC weights. We show that such lack of speci c guidances provides signi cant opportunities for the researcher to search for speci cations with statistically signi cant results, undermining one of the main advantages of the method. Considering six alternative speci cations commonly used in SC applications, we calculate in Monte Carlo simulations the probability of nding a statistically signi cant result at 5% in at least one speci cation. We nd that this probability can be as high as 13% (23% for a 10% signi cance test) when there are 12 pre-intervention periods and decay slowly with the number of pre-intervention periods. With 230 pre-intervention periods, this probability is still around 10% (18% for a 10% signi cance test). We show that the speci cation that uses the average pre-treatment outcome values to estimate the weights performed particularly bad in our simulations. However, the speci cation-searching problem remains relevant even when we do not consider this speci cation. We also show that this speci cation-searching problem is relevant in simulations with real datasets looking at placebo interventions in the Current Population Survey (CPS). In order to mitigate this problem, we propose a criterion to select among SC di erent speci cations based on the prediction error of each speci cations in placebo estimations