Problem Of Sampling Bias Research Articles

Large-scale forecasting of Heracleum sosnowskyi habitat suitability under the climate change on publicly available data

This research aims to establish the possible habitat suitability of Heracleum sosnowskyi (HS), one of the most aggressive invasive plants, in current and future climate conditions across the territory of the European part of Russia. We utilised a species distribution modelling framework using publicly available data of plant occurrence collected in citizen science projects (CSP). Climatic variables and soil characteristics were considered to follow possible dependencies with environmental factors. We applied Random Forest to classify the study area. We addressed the problem of sampling bias in CSP data by optimising the sampling size and implementing a spatial cross-validation scheme. According to the Random Forest model built on the finally selected data shape, more than half of the studied territory in the current climate corresponds to a suitability prediction score higher than 0.25. The forecast of habitat suitability in future climate was highly similar for all climate models. Almost the whole studied territory showed the possibility for spread with an average suitability score of 0.4. The mean temperature of the wettest quarter and precipitation of wettest month demonstrated the highest influence on the HS distribution. Thus, currently, the whole study area, excluding the north, may be considered as s territory with a high risk of HS spreading, while in the future suitable locations for the HS habitat will include high latitudes. We showed that chosen geodata pre-processing, and cross-validation based on geospatial blocks reduced significantly the sampling bias. Obtained predictions could help to assess the risks accompanying the studied plant invasion capturing the patterns of the spread, and can be used for the conservation actions planning.

Solving sampling bias problems in presence–absence or presence‐only species data using zero‐inflated models

AbstractAimLarge databases of species records such as those generated through citizen science projects, archives or museum collections are being used with increasing frequency in species distribution modelling (SDM) for conservation and land management. Despite the broad spatial and temporal coverage of the data, its application is often limited by the issue of sampling bias and consequently, zero inflation; there are more zeros (which are potentially ‘false absences’) in the data than expected. Here, we demonstrate how pooling species presence data into a ‘pseudo‐abundance’ count can allow identification and removal of sampling bias through the use of zero‐inflated (ZI) models, and thus solve a common SDM problem.LocationAll locationsTaxonAll taxaMethodsWe present the results of a series of simulations based on hypothetical ecological scenarios of data collection using random and non‐random sampling strategies. Our simulations assume that the locations of occurrence records are known at a high spatial resolution, but that the absence of occurrence records may reflect under‐sampling. To simulate pooling of presence–absence or presence‐only data, we count occurrence records at intermediate and coarse spatial resolutions, and use ZI models to predict the counts (species abundance per grid cell) from environmental layers.ResultsZI models can successfully identify predictors of bias in species data and produce abundance prediction maps that are free from that bias. This phenomenon holds across multiple spatial scales, thereby presenting an advantage over presence‐only SDM methods such as binomial GLMs or MaxEnt, where information about species density is lost, and model performance declines at coarser scales.Main ConclusionsOur results highlight the value of converting presence–absence or presence‐only species data to ‘pseudo‐abundance’ and using ZI models to address the problem of sampling bias. This method has huge potential for ecological researchers when using large species datasets for research and conservation.

Sampling bias correction in species distribution models by quasi-linear Poisson point process

Species distribution modeling has an essential role in ecology to investigate habitat suitability based on the relationship between species occurrences and environmental conditions. The presence-only data for organisms is usually assumed to be obtained randomly from the region of interest; however, it is often the case that it is biased toward the areas easily accessed and adversely affects prediction accuracy. To address this sampling bias problem relevant to the prediction accuracy and model fitting of habitat distributions, we propose a new Poisson point process (PPP) model named as a quasi PPP, where environmental effect and sampling bias are explicitly modeled in separate clusters in a framework of quasi-linear modeling. The quasi-linear modeling is designed for capturing homogeneity within clusters and heterogeneity between clusters to improve the estimation accuracy of species distribution. The proposed model includes conventional models such as thinned and superposed PPPs as special cases. We have found that the quasi PPP outperforms the other existing methods in terms of goodness of model fitting. A statistical index based on the quasi-linear modeling is proposed to measure how the presence-only data used for the estimation of the species habitat distribution is affected by the sampling bias. The utility of the quasi PPP has been illustrated using simulation studies as well as the comprehensive vascular plant data in Japan. Our proposed model flexibly incorporates the effect of sampling biases to improve the estimation accuracy of species distributions. The results of data analysis are easily reproducible and applications to other data sets are also easily implementable by a package of qPPP of a statistical software R.

Finite sample performance of a long run variance estimator based on exactly (almost) unbiased autocovariance estimators

This paper proposes a bias reduced long run variance (LRV) estimator of a univariate time series with unknown mean that addresses well known finite sample bias problems. The LRV estimator is based on the (almost) exactly unbiased autocovariance estimator proposed by Vogelsang and Yang (2016). Whereas using fixed-b critical values is known to correct downward bias in LRV estimates generated by demeaning the data, the approach we take also corrects the classic Parzen bias that is not captured by the fixed-b approach. When applied to the tests of the null hypothesis of the mean in a simple location model, a simulation study shows that the proposed LRV estimator leads to tests with less over-rejections while maintaining power at least as high and often higher as the standard robust t test based on fixed-b critical values. These simulations suggest further theoretical analysis of the bias reduction approach is warranted.

Active learning for classifying data streams with unknown number of classes

The classification of data streams is an interesting but also a challenging problem. A data stream may grow infinitely making it impractical for storage prior to processing and classification. Due to its dynamic nature, the underlying distribution of the data stream may change over time resulting in the so-called concept drift or the possible emergence and fading of classes, known as concept evolution. In addition, acquiring labels of data samples in a stream is admittedly expensive if not infeasible at all. In this paper, we propose a novel stream-based active learning algorithm (SAL) which is capable of coping with both concept drift and concept evolution by adapting the classification model to the dynamic changes in the stream. SAL is the first AL algorithm in the literature to explicitly take account of these concepts. Moreover, using SAL, only labels of samples that are expected to reduce the expected future error are queried. This process is done while tackling the problem of sampling bias so that samples that induce the change (i.e., drifting samples or samples coming from new classes) are queried. To efficiently implement SAL, the paper proposes the application of non-parametric Bayesian models allowing to cope with the lack of prior knowledge about the data stream. In particular, Dirichlet mixture models and the stick breaking process are adopted and adapted to meet the requirements of online learning. The empirical results obtained on real-world benchmarks demonstrate the superiority of SAL in terms of classification performance over the state-of-the-art methods using average and average class accuracy.

Nonparametric maximum likelihood estimation for the multisample Wicksell corpuscle problem.

We study nonparametric maximum likelihood estimation for the distribution of spherical radii using samples containing a mixture of one-dimensional, two-dimensional biased and three-dimensional unbiased observations. Since direct maximization of the likelihood function is intractable, we propose an expectation-maximization algorithm for implementing the estimator, which handles an indirect measurement problem and a sampling bias problem separately in the E- and M-steps, and circumvents the need to solve an Abel-type integral equation, which creates numerical instability in the one-sample problem. Extensions to ellipsoids are studied and connections to multiplicative censoring are discussed.

A Primary Way of Solving Sampling Bias Problem in Complex Internet Topology

Measuring complex Internet topology is primary research activities in Internet-related research fields and has become a focus recently. And due to the complexity of Internet topology, further processing is required and is thoroughly studied in this paper. First, Internet topology measurement is performed by CAIDA measurement approach with more than twenty monitors. Then the problem of sampling bias is quantitatively studied. Mathematical models are introduced into both the node bias problem and link bias problem, and final option is gained. Final results are that more than 156 CAIDA monitors are necessary in order to have a full Internet topology measurement result.

Slow explorers take less risk: a problem of sampling bias in ecological studies

Sampling bias is a key issue to consider when designing studies to address biological questions and its importance has been widely discussed in the literature. However, some forms of bias remain underestimated. We investigated the roosting decisions of free-living great tits utilizing nest-boxes in response to the installation of a novel object (a miniature video camera) inside their nest-boxes. We show that birds that score highly on a widely used exploration test (i.e., fast explorers) are more likely to accept and approach novel objects used in a seemingly unobtrusive sampling technique; thus, the sample collected overrepresents fast explorers. This form of behavior-related bias, sensitivity to novel objects, has largely been overlooked in sampling design. We demonstrate potential pitfalls of neglecting this behavior-related sampling bias in biological studies.

Properties of the CUE estimator and a modification with moments

In this paper, we analyze properties of the Continuous Updating Estimator (CUE) proposed by Hansen et al. (1996), which has been suggested as a solution to the finite sample bias problems of the two-step GMM estimator. We show that the estimator should be expected to perform poorly in finite samples under weak identification, in particular, the estimator is not guaranteed to have finite moments of any order. We propose the Regularized CUE (RCUE) as a solution to this problem. The RCUE solves a modification of the first-order conditions for the CUE estimator and is shown to be asymptotically equivalent to CUE under many weak moment asymptotics. Our theoretical findings are confirmed by extensive Monte Carlo studies.

Measuring and Processing Complex Internet Topology

Measuring complex Internet topology is primary research activities in Internet-related research fields and has become a focus recently. And due to the complexity of Internet topology, further processing is required and is thoroughly studied in this paper. First, Internet topology measurement is performed by CAIDA measurement approach with more than twenty monitors. Then the problem of sampling bias is quantitatively studied. Mathematical models are introduced into both the node bias problem and link bias problem, and final option is gained. Final results are that more than 156 CAIDA monitors are necessary in order to have a full Internet topology measurement result.

Factors that prevent physicians reporting adverse events

The purpose of this paper is to explore the reasons why Greek doctors are reluctant to report adverse events. The paper is an exploratory study using the adverse events questionnaire, administered to 209 doctors in 14 major Athens hospitals, comprising university as well as tertiary. The questionnaire showed a high degree of internal consistency (Cronbach's alpha 0.84), which revealed a four-factor solution, explaining 67.4 per cent of the variance. Three dominant reasons for not reporting, with which almost or more than 50 per cent of doctors strongly or slightly agreed, were identified as no tradition for bringing up adverse events and a belief that bringing up adverse events will not lead to any improvement and workload. Factors that dissuade doctors from bringing up adverse events may be projected not only by cultural aspects such as professional, national and organisational cultures but also by healthcare structural issues such as safety systems, rules and procedures, and relevant acts and regulations. The study has several noteworthy limitations. First, doctors' response was poor, which might conceal sample bias problems. Second, the present study identified four factors but the fourth factor was not well defined. The study provides an understanding why physicians fail to report adverse events so that systems can be introduced and cultures developed, which make this easier.

Spatial Structure of Forest Lepidopteran Communities in Oak Hickory Forests of Indiana

The response of forest insect communities to disturbances such as timber harvest will likely depend on the underlying spatial structure of species assemblages before the disturbance occurs. Unfortunately, many studies of forest management implicitly assume homogeneity of community structure before harvest; postlogging communities are inferred to be a direct product of the imposed management. The goal of this study is to examine variation in the community structure of forest Lepidoptera using the pretimber harvest data on Lepidoptera from 20 forest sites within three watersheds at Morgan Monroe State Forest, IN. A total of 14,537 individuals representing 324 species of Lepidoptera were sampled from Morgan-Monroe State Forest in 2007. Sampling efficiency was not a function of management unit, and, surprisingly, we found little evidence that management units differed in overall community composition. Diversity partitioning suggested that >99% of Simpson diversity (species dominance) was determined at the local scale, and each site contained the same 10 dominant taxa in rank order. Variation in species richness seemed to be more a problem of sampling bias than underlying differences in habitat preference by moth feeding guilds. Finally, Mantel tests suggested that forest moth communities at Morgan-Monroe are not spatially autocorrelated. The results here are encouraging because they strongly suggest that shifts in lepidopteran community structure should reflect the community response to disturbance rather than inherent spatial heterogeneity of species composition.

Spatial Structure of Forest Lepidopteran Communities in Oak Hickory Forests of Indiana

The response of forest insect communities to disturbances such as timber harvest will likely depend on the underlying spatial structure of species assemblages before the disturbance occurs. Unfortunately, many studies of forest management implicitly assume homogeneity of community structure before harvest; postlogging communities are inferred to be a direct product of the imposed management. The goal of this study is to examine variation in the community structure of forest Lepidoptera using the pretimber harvest data on Lepidoptera from 20 forest sites within three watersheds at Morgan Monroe State Forest, IN. A total of 14,537 individuals representing 324 species of Lepidoptera were sampled from Morgan-Monroe State Forest in 2007. Sampling efficiency was not a function of management unit, and, surprisingly, we found little evidence that management units differed in overall community composition. Diversity partitioning suggested that > 99% of Simpson diversity (species dominance) was determined at the local scale, and each site contained the same 10 dominant taxa in rank order. Variation in species richness seemed to be more a problem of sampling bias than underlying differences in habitat preference by moth feeding guilds. Finally, Mantel tests suggested that forest moth communities at Morgan-Monroe are not spatially autocorrelated. The results here are encouraging because they strongly suggest that shifts in lepidopteran community structure should reflect the community response to disturbance rather than inherent spatial heterogeneity of species composition.

Correcting for the Sampling Bias Problem in Spike Train Information Measures

Information Theory enables the quantification of how much information a neuronal response carries about external stimuli and is hence a natural analytic framework for studying neural coding. The main difficulty in its practical application to spike train analysis is that estimates of neuronal information from experimental data are prone to a systematic error (called "bias"). This bias is an inevitable consequence of the limited number of stimulus-response samples that it is possible to record in a real experiment. In this paper, we first explain the origin and the implications of the bias problem in spike train analysis. We then review and evaluate some recent general-purpose methods to correct for sampling bias: the Panzeri-Treves, Quadratic Extrapolation, Best Universal Bound, Nemenman-Shafee-Bialek procedures, and a recently proposed shuffling bias reduction procedure. Finally, we make practical recommendations for the accurate computation of information from spike trains. Our main recommendation is to estimate information using the shuffling bias reduction procedure in combination with one of the other four general purpose bias reduction procedures mentioned in the preceding text. This provides information estimates with acceptable variance and which are unbiased even when the number of trials per stimulus is as small as the number of possible discrete neuronal responses.

The greatest lower bound to the reliability of a test and the hypothesis of unidimensionality

To assess the reliability of congeneric tests, specifically designed reliability measures have been proposed. This paper emphasizes that such measures rely on a unidimensionality hypothesis, which can neither be confirmed nor rejected when there are only three test parts, and will invariably be rejected when there are more than three test parts. Jackson and Agunwamba's (1977) greatest lower bound to reliability is proposed instead. Although this bound has a reputation for overestimating the population value when the sample size is small, this is no reason to prefer the unidimensionality-based reliability. Firstly, the sampling bias problem of the glb does not play a role when the number of test parts is small, as is often the case with congeneric measures. Secondly, glb and unidimensionality based reliability are often equal when there are three test parts, and when there are more test parts, their numerical values are still very similar. To the extent that the bias problem of the greatest lower bound does play a role, unidimensionality-based reliability is equally affected. Although unidimensionality and reliability are often thought of as unrelated, this paper shows that, from at least two perspectives, they act as antagonistic concepts. A measure, based on the same framework that led to the greatest lower bound, is discussed for assessing how close is a set of variables to unidimensionality. It is the percentage of common variance that can be explained by a single factor. An empirical example is given to demonstrate the main points of the paper.

A Unified Explanation of Territorial Conflict: Testing the Impact of Sampling Bias, 1919-1992

This article develops a new unified territorial explanation of conflict that accounts for the possibility of certain factors affecting the rise of a militarized dispute, as well as the probability that a dispute will escalate to war. In the past, research linking territorial disputes to a relatively high probability of war outbreak has been criticized for underestimating the potential problem of sampling bias in the militarized interstate dispute (MID) data. This study utilizes newly available data on territorial claims going back to 1919 to determine, using a two-stage estimation procedure, whether the presence of territorial claims in the dispute onset phase affects the relationship between territorial militarized disputes and war in the second stage. It is found that territorial claims increase the probability of a militarized dispute occurring and that territorial MIDs increase the probability of war, even while controlling for the effect of territorial claims on dispute onset. The effect of territory across the two stages is consistent with the new territorial explanation of conflict and war and shows no sampling bias with regard to territory in the MID data.

What Is the Current Truth About Polygraph Lie Detection?

With reference to the continued debate about polygraph lie detection, this article applies standard criteria of scientific validity to the most common and influential polygraph procedure, the Control Question Test (CQT). Apart from the CQT's coding objectivity, which is satisfactory, the remaining validity criteria are neither met nor even tested seriously. Published evidence does not allow for item analyses and therefore does not speak to internal consistency and the assumed existence and invariance of the measured construct. The underlying decision model remains unspecified and untested, reflecting lack of construct validity. Optimistic estimates of the CQT's predictive validity suffer from a serious sampling-bias problem. And the notion of incremental validity is hardly considered at all. Due to ethical and professional reasons, it is argued that a fair and responsible appraisal of polygraph testing should clearly reveal this lack of validity, rather than seeking a 2-sided compromise. This article is based on a review that led a German Supreme Court to abandon the CQT from penal procedures.

Validity of the control question polygraph test: The problem of sampling bias.

Sampling bias is a potential problem in polygraph validity studies in which posttest confessions are used to establish ground truth because this criterion is not independent of the polygraph test. In the present study, criterion evidence was sought from polygraph office records and from independent police files for all 402 control question tests (CQTs) conducted during a 5-year period by federal police examiners in a major Canadian city. Based on blind scoring of the charts, the hit rate for criterion innocent subjects (65% of whom were verified by independent sources) was 55%; for guilty subjects (of whom only 2% were verified independently), the hit rate was 98%. Although the estimate for innocent subjects is tenable given the characteristics of the sample on which it is based, the estimate for the guilty subsample is not. Some alternatives to confession studies for evaluating the accuracy of the CQT with guilty subjects are discussed. The polygraph examination procedure used most frequently in criminal investigations is the control question test (CQT; cf. Barland & Raskin, 1973; Reid & Inbau, 1977), a procedure in which physiological responses to specific crime-relevant questions are compared with responses to control questions dealing with the subject's prior history of wrongdoing. The CQT relies on the assumption that guilty subjects should be more concerrted with and hence more reactive to the relevant questions, whereas innocent subjects should be more disturbed by the more broadly incriminatory control questions. This assumption is without a sound theoretical basis (Ben-Shakhar, Bar-Hillel, & Lieblich, 1986; Kleinmuntz & Szucko, 1982; Lykken, 1978,1981), however, and therefore carefully controlled empirical research is needed to establish the validity of the CQT. Ideally, an estimate of CQT validity would be derived from a representative sample of real-life examinations in which an unambiguous criterion of ground truth was available. Laboratory studies, although advantageous from the standpoint o fcriterion certainty, have generally failed to recreate the motivational and

Validity of the control question polygraph test: The problem of sampling bias.

Validity of the control question polygraph test: The problem of sampling bias.