GEOSTATISTICAL INFERENCE UNDER PREFERENTIAL SAMPLING

Geostatistics involves the fitting of spatially continuous models to spatially discrete data (Chil`es and Delfiner, 1999). Preferential sampling arises when the process that determines the data-locations and the process being modelled are stochastically dependent. Conventional geostatistical methods assume, if only implicitly, that sampling is non-preferential. However, these methods are often used in situations where sampling is likely to be preferential. For example, in mineral exploration samples may be concentrated in areas thought likely to yield high-grade ore. We give a general expression for the likelihood function of preferentially sampled geostatistical data and describe how this can be evaluated approximately using Monte Carlo methods. We present a model for preferential sampling, and demonstrate through simulated examples that ignoring preferential sampling can lead to seriously misleading inferences. We describe an application of the model to a set of bio-monitoring data from Galicia, northern Spain, in which making allowance for preferential sampling materially changes the inferences.

Causal Inference in Observational Studies with Outcome-Dependent Sampling

In this paper, we consider estimation of the causal effect of a treatment on an outcome from observational data collected in two phases. In the first phase, a simple random sample of individuals are drawn from a population. On these individuals, information is obtained on treatment, outcome, and a few low-dimensional confounders. These individuals are then stratified according to these factors. In the second phase, a random sub-sample of individuals are drawn from each stratum, with known, stratum-specific selection probabilities. On these individuals, a rich set of confounding factors are collected. In this setting, we introduce four estimators: (1) simple inverse weighted, (2) locally efficient, (3) doubly robust and (4)enriched inverse weighted. We evaluate the finite-sample performance of these estimators in a simulation study. We also use our methodology to estimate the causal effect of trauma care on in-hospital mortality using data from the National Study of Cost and Outcomes of Trauma.

A common ancestral glycoprotein (GP) 9 1828A>G (Asn45Ser) gene mutation occurring in European families from Australia and Northern Europe with Bernard-Soulier Syndrome (BSS)

Bernard-Soulier syndrome (BSS) is an extremely rare hereditary bleeding disorder, caused by mutations occurring in the Glycoprotein (GP) Ibalpha, GPIbbeta and GP9 genes that encode for the corresponding subunits of platelet GPIb-V-IX adhesion receptor complex. BSS has been reported in many populations, mostly behaving in an autosomal-recessive manner.While the great majority of BSS mutations are unique to a single individual or family, the GP9 1828A>G Asn45Ser mutation, which we have identified in an undocumented Australian Caucasian, has already been reported in multiple unrelated Caucasian families from various Northern and Central European countries. Haplotype analysis of 19 BSS patients from 15 unrelated Northern European families (including 2 compound heterozygote siblings from a British family previously published, and 17 1828A>G Asn45Ser homozygotes), showed that 14 of these BSS patients from 11 of the 1828A>G Asn45Ser homozygote families share a common haplotype at the chromosomal region 3' to the GP9 gene. Hence, the results suggest that the GP9 1828A>GAsn45Ser mutation in these families is ancient, and its frequent emergence in the European population is the result of a founder effect rather than recurrent mutational events. Association of the 1828A>G Asn45Ser mutation with variant haplotypes in 4 other Northern European BSS families raised the possibility of a second founder event, or rare recombinations in these families. Additional members from these 'atypical' lineages would need to be screened to resolve this question.

Colonization history of the Swiss Rhine basin by the bullhead (Cottus gobio): inference under a Bayesian spatially explicit framework

The present distribution of freshwater fish in the Alpine region has been strongly affected by colonization events occurring after the last glacial maximum (LGM), some 20,000 years ago. We use here a spatially explicit simulation framework to model and better understand their colonization dynamics in the Swiss Rhine basin. This approach is applied to the European bullhead (Cottus gobio), which is an ideal model organism to study fish past demographic processes since it has not been managed by humans. The molecular diversity of eight sampled populations is simulated and compared to observed data at six microsatellite loci under an approximate Bayesian computation framework to estimate the parameters of the colonization process. Our demographic estimates fit well with current knowledge about the biology of this species, but they suggest that the Swiss Rhine basin was colonized very recently, after the Younger Dryas some 6600 years ago. We discuss the implication of this result, as well as the strengths and limits of the spatially explicit approach coupled to the approximate Bayesian computation framework.