The test-negative design: Opportunities, limitations and biases.

The purpose of this paper is to investigate how to interpret the outcome of a test-negative design study. We does so by systematically reviewing the properties of the design in relation to its potential applications. We first argue that the application of the design does not depend on certain assumptions (as is sometimes articulated in the literature), which may open new opportunities for using the design. Then, we relate a number of limitations of the design. The design cannot be used for studying the mortality effects of vaccines and is problematic for studies into the effect on hospitalization. The vaccine's effectiveness on the transmission of viruses is also potentially problematic, depending very much on the characteristics of the tests. The implication of our findings is that the test-negative designs can, at best, be seen as an indication of effectiveness in highly idealized situations that are often far away from reality.

Effects of mobile phone electromagnetic fields on brain waves in healthy volunteers.

The interaction between biological tissue and electromagnetic fields (EMF) is a topic of increasing interest due to the rising prevalence of background EMF in the past decades. Previous studies have attempted to measure the effects of EMF on brainwaves using EEG recordings, but are typically hampered by experimental and environmental factors. In this study, we present a framework for measuring the impact of EMF on EEG while controlling for these factors. A Bayesian statistical approach is employed to provide robust statistical evidence of the observed EMF effects. This study included 32 healthy participants in a double-blinded crossover counterbalanced design. EEG recordings were taken from 63 electrodes across 6 brain regions. Participants underwent a measurement protocol comprising two 18-min sessions with alternating blocks of eyes open (EO) and eyes closed (EC) conditions. Group 1 (n = 16) had EMF during the first session and sham during the second session; group 2 (n = 16) had the opposite. Power spectral density plots were generated for all sessions and brain regions. The Bayesian analysis provided statistical evidence for the presence of an EMF effect in the alpha band power density in the EO condition. This measurement protocol holds potential for future research on the impact of novel transmission protocols.

The Limits of Bayesian Thinking in Court.

We comment on the contributions of Dahlman and of Fenton et al., who both suggested a Bayesian approach to analyze the Simonshaven case. We argue that analyzing a full case with a Bayesian approach is not feasible, and that there are serious problems with assigning actual numbers to probabilities and priors. We also discuss the nature of Bayesian thinking in court, and the nature and interpretation of the likelihood ratio. In particular, we discuss what it could mean that a likelihood ratio is in some sense uncertain.

Reply to Lucas & Henneberg: Are human faces unique?

This paper offers a response to the 2015 article of Lucas and Henneberg "Are human faces unique?" (Forensic Sci. Int. 257, 514.e1-514.e6), as well as to the subsequent popularization thereof. In the first part, we assess the probabilistic claims made by the authors and find crucial parts to be unsupported. In particular, the authors offer no probabilistic model on which to base their conclusions. Even if we disregard the errors identified in the first part, we find a troubling discrepancy between the published findings and the popular summary.

Evaluating evidence in linked crimes with multiple offenders.

In de Zoete et al. (2015) a framework for the evaluation of evidence when an individual is a suspect of two separate offenses (based on Evett et al., 2006) is implemented using a Bayesian network. Here, we extend this to situations with multiple offenders. When we have multiple offenders, new questions arise: (1) Can we distinguish between the offenders, even if we do not know their identity? (2) Do we know that certain pieces of evidence originate from the same person? (3) Do we know the number of offenders? With the aid of a mock case example, we show that such subtle differences between situations can lead to substantially different conclusions in terms of posterior probabilities of a certain suspect being one of the offenders in a particular crime. We reach our conclusions by constructing appropriate Bayesian networks for each situation. Although we find it undesirable that Bayesian networks are demonstrated in court, they can be very helpful in guiding expert and legal reasoning, identifying pitfalls and assist in preventing them. Bayesian networks can be used as a tool to understand how the different pieces of evidence influence each others evidential value, and the probabilities of the hypotheses of interest.

Stochastic SIR epidemics in a population with households and schools.

We study the spread of stochastic SIR (Susceptible [Formula: see text] Infectious [Formula: see text] Recovered) epidemics in two types of structured populations, both consisting of schools and households. In each of the types, every individual is part of one school and one household. In the independent partition model, the partitions of the population into schools and households are independent of each other. This model corresponds to the well-studied household-workplace model. In the hierarchical model which we introduce here, members of the same household are also members of the same school. We introduce computable branching process approximations for both types of populations and use these to compare the probabilities of a large outbreak. The branching process approximation in the hierarchical model is novel and of independent interest. We prove by a coupling argument that if all households and schools have the same size, an epidemic spreads easier (in the sense that the number of individuals infected is stochastically larger) in the independent partition model. We also show by example that this result does not necessarily hold if households and/or schools do not all have the same size.

p-values should not be used for evaluating the strength of DNA evidence.

Recently, p-values have been suggested to explain the strength of a likelihood ratio that evaluates DNA evidence. It has been argued that likelihood ratios would be difficult to explain in court and that p-values would offer an alternative that is easily explained. In this article, we argue that p-values should not be used in this context. p-Values do not directly relate to the strength of the evidence. The likelihood ratio measures the strength of the evidence, while the p-value measures how rare it is to find evidence that is equally strong or stronger, which is something fundamentally different. In addition, a p-value is not always unambiguous. To illustrate our arguments, we present several examples from forensic genetics.

Optimal strategies for familial searching.

Familial searching is the process of finding potential relatives of the donor of a crime scene profile in a DNA database. Several authors have proposed strategies for generating candidate lists of potential relatives. This paper reviews four strategies and investigates theoretical properties as well as empirical behavior, using a comprehensive simulation study on mock databases. The effectiveness of a familial search is shown to highly depend on the case profile as well as on the tuning parameters. We give recommendations for proceeding in an optimal way and on how to choose tuning parameters both in general and on a case-by-case basis. Additionally we treat searching heterogeneous databases (not all profiles comprise the same loci) and composite searching for multiple types of kinship. An R-package for reproducing results in a particular case is released to help decision-making in familial searching.

The combined evidential value of autosomal and Y-chromosomal DNA profiles obtained from the same sample.

When a Y-chromosomal and a (partial) autosomal DNA profile are obtained from one crime sample, and both profiles match the suspect's profiles, we would like to know the combined evidential value. To calculate the likelihood ratio of observing the autosomal and Y-chromosomal DNA profiles combined, we need to know the conditional random match probability of the observed autosomal DNA profile, given the Y-chromosomal match. We examine this conditional probability in two ways: (1) with a database containing data of 2,085 men and (2) using a simulation model. We conclude that if the Y-chromosomal DNA profiles match, we can still regard the autosomal DNA profile as independent from the Y-chromosomal DNA profile if the matching person is not a descendant of the father of the donor of the (crime) sample. The evidential value can, in that case, be computed by multiplying the random match probabilities of the individual profiles.

Selection effects and database screening in forensic science.

We argue that it is, in principle, not difficult to deal with selection effects in forensic science. If a suspect is selected through a process that is related to the forensic evidence, then the strength of the evidence will be compensated by very small prior odds. No further correction is necessary. The same is true for so-called data-dependent hypotheses. These are allowed, since if the hypothesis is really "tailored around" the evidence, the evidential value will be high but the prior odds will compensate for that. The assessment of the prior odds is outside the scope of the forensic scientist, but he should make lawmakers, judges and juries aware of the phenomenon. This discussion applies to many situations--we discuss four concrete examples.

Simulation of biological evolution and the NFL theorems.

William Dembski (No free lunch: why specified complexity cannot be purchased without intelligence, 2002) claimed that the NFL theorems from optimization theory render darwinian biological evolution impossible. Häggström (Biology and Philosophy 22:217-230, 2007) argued that the NFL theorems are not relevant for biological evolution at all, since the assumptions of the NFL theorems are not met. Although I agree with Häggström (Biology and Philosophy 22:217-230, 2007), in this article I argue that the NFL theorems should be interpreted as dealing with an extreme case within a much broader context. This broader context is in fact relevant for scientific research of certain evolutionary processes; not in the sense that the theorems can be used to draw conclusions about any intelligent design inference, but in the sense that it helps us to interpret computer simulations of evolutionary processes. As a result of this discussion, I will argue that from simulations, we do not learn much about how complexity arises in the universe. This position is in contrast with certain claims in the literature that I will discuss.

The local threshold for geographical spread of infectious diseases between farms.

We investigated the influence of the spatial pattern of farms on the geographical spread of infectious livestock diseases, such as classical swine fever, foot-and-mouth disease and avian influenza in a combined analytical-numerical approach. Our purpose of this paper is to develop a method to identify the areas in which an infection has the potential to spread in an outbreak. In our model, each infected farm can infect neighbouring farms and the probability of transmission is a function of the inter-farm distance (spatial kernel). Therefore, the density of farms in an area is a good indicator for the probability of a major outbreak. In the epidemiological nomenclature, such density corresponds to a local reproduction ratio and we studied the critical behaviour of both the local density and the local reproduction ratio. We found that a threshold can be defined above which major outbreaks can occur, and the threshold value depends on the spatial kernel. Our expression for the threshold value is derived based on scaling arguments and contains two parameters in the exponents of the equation. We estimated these parameters from numerical results for the spatial spread using one particular mathematical function for the form of the spatial kernel. Subsequently, we show that our expression for the threshold using these estimated parameters agrees very well with numerical results for a number of different other functional forms of the spatial kernel (thus suggesting that we are dealing with universal parameters). As an illustration of the practical relevance of the presented method, we calculated the threshold value for avian influenza in the Netherlands and use it to produce a risk map for this disease.

A branching model for the spread of infectious animal diseases in varying environments.

This paper is concerned with a stochastic model, describing outbreaks of infectious diseases that have potentially great animal or human health consequences, and which can result in such severe economic losses that immediate sets of measures need to be taken to curb the spread. During an outbreak of such a disease, the environment that the infectious agent experiences is therefore changing due to the subsequent control measures taken. In our model, we introduce a general branching process in a changing (but not random) environment. With this branching process, we estimate the probability of extinction and the expected number of infected individuals for different control measures. We also use this branching process to calculate the generating function of the number of infected individuals at any given moment. The model and methods are designed using important infections of farmed animals, such as classical swine fever, foot-and-mouth disease and avian influenza as motivating examples, but have a wider application, for example to emerging human infections that lead to strict quarantine of cases and suspected cases (e.g. SARS) and contact and movement restrictions.

The evidential value in the DNA database search controversy and the two-stain problem.

Does the evidential strength of a DNA match depend on whether the suspect was identified through database search or through other evidence ("probable cause")? In Balding and Donnelly (1995, Journal of the Royal Statistical Society, Series A 158, 21-53) and elsewhere, it has been argued that the evidential strength is slightly larger in a database search case than in a probable cause case, while Stockmarr (1999, Biometrics 55, 671-677) reached the opposite conclusion. Both these approaches use likelihood ratios. By making an excursion to a similar problem, the two-stain problem, we argue in this article that there are certain fundamental difficulties with the use of a likelihood ratio, which can be avoided by concentrating on the posterior odds. This approach helps resolving the above-mentioned conflict.

Modeling and real-time prediction of classical swine fever epidemics.

We propose a new method to analyze outbreak data of an infectious disease such as classical swine fever. The underlying model is a two-type branching process. It is used to deduce information concerning the epidemic from detected cases. In particular, the method leads to prediction of the future course of the epidemic and hence can be used as a basis for control policy decisions. We test the model with data from the large 1997-1998 classical swine fever epidemic in The Netherlands. It turns out that our results are in good agreement with the data.