Form Of Measurement Error Research Articles

Comparing measurements of violent crime in local communities: a case study in Islington, London

ABSTRACT Police-recorded crime data are prone to measurement error, affecting our understanding of the nature of crime. Research has responded to this problem using data from surveys and emergency services. These data sources are not error-free, and data from different sources are not always easily comparable. This study compares violent crime data recorded by police, ambulance services, two surveys and computer simulations in Islington, London. Different data sources show remarkably different results. However, crime estimates become more similar, but still show different distributions, when crime rates are calculated using workday population as the denominator and log-transformed. Normalising crime rates by workday population controls for the fact that some data sources reflect offences’ location while others refer to victims’ residence, and log-transforming rates mitigates the biasing effect associated with some multiplicative forms of measurement error. Comparing multiple data sources allows for more accurate descriptions of the prevalence and distribution of crime.

Nonparametric statistical learning based on modal regression

In this article, we propose a novel nonparametric statistical learning tool based on modal regression, which can complement the standard mean and quantile regression and has broad applicability in various fields. We first propose a local polynomial modal regression which focuses on the most likely conditional value (conditional mode) of a dependent variable Y given covariates x, and has several superiorities over conditional mean or quantiles, such as resistance to outliers and some forms of measurement error and having shorter prediction intervals when data are skewed. We employ the idea of local polynomial regression to estimate the modal regression nonparametrically. To broaden the applicability of the new technique to multivariate data or functional/longitudinal data, we further develop a varying coefficient modal regression. A Monte Carlo simulation study and an analysis of health care expenditure data demonstrate that the new proposed nonparametric modal regression is a promising complementary nonparametric data analysis tool to conventional nonparametric mean or quantile regression.

Social animal models for quantifying plasticity, assortment, and selection on interacting phenotypes.

Both assortment and plasticity can facilitate social evolution, as each may generate heritable associations between the phenotypes and fitness of individuals and their social partners. However, it currently remains difficult to empirically disentangle these distinct mechanisms in the wild, particularly for complex and environmentally responsive phenotypes subject to measurement error. To address this challenge, we extend the widely used animal model to facilitate unbiased estimation of plasticity, assortment and selection on social traits, for both phenotypic and quantitative genetic (QG) analysis. Our social animal models (SAMs) estimate key evolutionary parameters for the latent reaction norms underlying repeatable patterns of phenotypic interaction across social environments. As a consequence of this approach, SAMs avoid inferential biases caused by various forms of measurement error in the raw phenotypic associations between social partners. We conducted a simulation study to demonstrate the application of SAMs and investigate their performance for both phenotypic and QG analyses. With sufficient repeated measurements, we found desirably high power, low bias and low uncertainty across model parameters using modest sample and effect sizes, leading to robust predictions of selection and adaptation. Our results suggest that SAMs will readily enhance social evolutionary research on a variety of phenotypes in the wild. We provide detailed coding tutorials and worked examples for implementing SAMs in the Stan statistical programming language.

Cash-Management in Times of Covid-19

Abstract The incidence of COVID-19 has systematically decreased households’ use of cash as means of payment as well as the average size and frequency of cash withdrawals. We argue that the structure of Baumol–Tobin type inventory theoretical models and their extensions can be used to separate the confounding factors, such as the desired level of consumption and the choice of the fraction of consumption paid in cash, from the cash management behavior, i.e. the size and frequency of cash withdrawals. Using this insight we argue that the observed cash management is consistent with COVID-19 increasing the fixed cost of withdrawing cash. We use detailed data on ATM cash disbursements in Argentina, Chile, and the US to estimate how much the pandemic has changed the transaction cost of using cash. This estimation shows that if the intensity of the virus doubles in a county, cash transaction cost increases by approximately 2%. The results from Argentina, Chile, and the US are remarkably similar and robust to several forms of measurement error and endogeneity.

Cash-Management in Times of COVID-19

The incidence of COVID-19 has systematically decreased households’ use of cash as means of payment as well as the average size and frequency of cash withdrawals. We argue that the structure of Baumol-Tobin type inventory theoretical models and their extensions can be used to separate the confounding factors, such as the desired level of consumption and the choice of the fraction of consumption paid in cash, from the cash management behavior, i.e the size and frequency of cash withdrawals. Using this insight we argue that the observed cash management is consistent with COVID-19 increasing the fixed cost of withdrawing cash. We use detailed data on ATM cash disbursements in Argentina, Chile, and the US to estimate how much the pandemic has changed the transaction cost of using cash. This estimation shows that if the intensity of the virus doubles in a county, cash transaction cost increases by approximately 2%. The results from Argentina, Chile, and the US are remarkably similar and robust to several forms of measurement error and endogeneity.

Methods to Account for Uncertainty in Latent Class Assignments When Using Latent Classes as Predictors in Regression Models, with Application to Acculturation Strategy Measures.

Latent class models have become a popular means of summarizing survey questionnaires and other large sets of categorical variables. Often these classes are of primary interest to better understand complex patterns in data. Increasingly, these latent classes are reified into predictors of other outcomes of interests, treating the most likely class as the true class to which an individual belongs even though there is uncertainty in class membership. This uncertainty can be viewed as a form of measurement error in predictors, leading to bias in the estimates of the regression parameters associated with the latent classes. Despite this fact, there is very limited literature treating latent class predictors as measurement error models. Most applications ignore this issue and fit a two-stage model that treats the modal class prediction as truth. Here, we develop two approaches-one likelihood-based, the other Bayesian-to implement a joint model for latent class analysis and outcome prediction. We apply these methods to an analysis of how acculturation behaviors predict depression in South Asian immigrants to the United States. A simulation study gives guidance for when a two-stage model can be safely implemented and when the joint model may be required.

Are Estimates of Non-Standard Employment Wage Penalties Robust to Different Wage Measures? The Case of Zero Hours Contracts in the UK

A range of evidence suggests that non-standard jobs, including fixed-term and other temporary jobs such as casual jobs, pay lower wages than more standard, permanent jobs, even after controlling for differences in worker and job characteristics. A recent literature suggests this is also the case for zero hours contracts (ZHCs), a growing form of non-standard employment in several developed countries, including the UK. These studies typically rely on derived wage variables – derived from survey responses to questions on earnings and hours data – which are prone to various forms of measurement error, some of which may be correlated with employment contract. Many relevant surveys, however, also include stated-rate hourly wage questions which, although also likely measured with error, are not subject to the same measurement issues. This suggests potential for sensitivity in non-standard employment wage penalty estimates depending on the wage measure used. Using the example of ZHCs in the UK, we first use derived wages to replicate the ballpark conditional ZHC wage penalty typical of existing studies. We then show that there is no conditional ZHC wage penalty, on average, when using the stated-rate hourly wage measure. This also holds for other non-standard employment types, including casual and fixed-term employment. Further, whereas the derived wage measure suggests, in line with existing literature, that the ZHC wage penalty is largest at the bottom of the wage distribution, we show the opposite to be the case when using the stated-rate wage measure. We discuss implications for policy, our understanding of labour market behaviour, and also for the wider literature on non-standard work wage penalties.

Partial least squares path modeling: Quo vadis?

Structural equation modeling (SEM) is a family of statistical techniques that has become very popular in marketing. Its ability to model latent variables, to take various forms of measurement error into account, and to test entire theories makes it useful for a plethora of research questions. It does not come as a surprise that some of the most cited scholarly articles in the marketing domain are about SEM (e.g., Bagozzi and Yi 1988; Fornell and Larcker 1981), and that SEM is covered by two contributions within this volume. The need for two contributions arises from the SEM family tree having two major branches (Reinartz et al. 2009): covariance-based SEM (which is presented in Chap. 11) and variance-based SEM, which is presented in this chapter.

Modeling heaped duration data: An application to neonatal mortality

In 2005, the Indian Government launched a conditional cash-incentive program to encourage institutional delivery. This paper studies the effects of the program on neonatal mortality using district-level household survey data. We model mortality using survival analysis, paying special attention to substantial heaping, a form of measurement error, present in the data. The main objective of this paper is to provide a set of sufficient conditions for identification and consistent estimation of the (discretized) baseline hazard accounting for heaping and unobserved heterogeneity. Our identification strategy requires neither administrative data nor multiple measurements, but a correctly reported duration point and the presence of some flat segment(s) in the baseline hazard. We establish the asymptotic properties of the maximum likelihood estimator and derive a set of specification tests that allow, among other things, to test for the presence of heaping and to compare different heaping mechanisms. Our empirical findings do not suggest a significant reduction of mortality in treated districts. However, they do indicate that accounting for heaping matters for the estimation of the hazard parameters.

Correcting encoder interpolation error on the Green Bank Telescope using an iterative model based identification algorithm

Various forms of measurement errors limit telescope tracking performance in practice. A new method for identifying the correcting coefficients for encoder interpolation error is developed. The algorithm corrects the encoder measurement by identifying a harmonic model of the system and using that model to compute the necessary correction parameters. The approach improves upon others by explicitly modeling the unknown dynamics of the structure and controller and by not requiring a separate system identification to be performed. Experience gained from pin-pointing the source of encoder error on the Green Bank Radio Telescope (GBT) is presented. Several tell-tale indicators of encoder error are discussed. Experimental data from the telescope, tested with two different encoders, are presented. Demonstration of the identification methodology on the GBT as well as details of its implementation are discussed. A root mean square tracking error reduction from 0.68 arc seconds to 0.21 arc sec was achieved by changing encoders and was further reduced to 0.10 arc sec with the calibration algorithm. In particular, the ubiquity of this error source is shown and how, by careful correction, it is possible to go beyond the advertised accuracy of an encoder.

Modelling Heaped Duration Data: An Application to Neonatal Mortality

In 2005, the Indian Government launched a conditional cash-incentive program to encourage institutional delivery. This paper studies the effects of the program on neonatal mortality using district-level household survey data. We model mortality using survival analysis, paying special attention to substantial heaping, a form of measurement error, present in the data. The main objective of this paper is to provide a set of sufficient conditions for identification and consistent estimation of the (discretized) baseline hazard accounting for heaping and unobserved heterogeneity. Our identification strategy requires neither administrative data nor multiple measurements, but a correctly reported duration point and the presence of some flat segment(s) in the baseline hazard. We establish the asymptotic properties of the maximum likelihood estimator and derive a set of specification tests that allow, among other things, to test for the presence of heaping and to compare different heaping mechanisms. Our empirical findings do not suggest a significant reduction of mortality in treated districts. However, they do indicate that accounting for heaping matters for the estimation of the hazard parameters.

The potential and perils of preprocessing: Building new foundations

Preprocessing forms an oft-neglected foundation for a wide range of statistical and scientific analyses. However, it is rife with subtleties and pitfalls. Decisions made in preprocessing constrain all later analyses and are typically irreversible. Hence, data analysis becomes a collaborative endeavor by all parties involved in data collection, preprocessing and curation, and downstream inference. Even if each party has done its best given the information and resources available to them, the final result may still fall short of the best possible in the traditional single-phase inference framework. This is particularly relevant as we enter the era of “big data”. The technologies driving this data explosion are subject to complex new forms of measurement error. Simultaneously, we are accumulating increasingly massive databases of scientific analyses. As a result, preprocessing has become more vital (and potentially more dangerous) than ever before. We propose a theoretical framework for the analysis of preprocessing under the banner of multiphase inference. We provide some initial theoretical foundations for this area, including distributed preprocessing, building upon previous work in multiple imputation. We motivate this foundation with two problems from biology and astrophysics, illustrating multiphase pitfalls and potential solutions. These examples also emphasize the motivations behind multiphase analyses—both practical and theoretical. We demonstrate that multiphase inferences can, in some cases, even surpass standard single-phase estimators in efficiency and robustness. Our work suggests several rich paths for further research into the statistical principles underlying preprocessing. To tackle our increasingly complex and massive data, we must ensure that our inferences are built upon solid inputs and sound principles. Principled investigation of preprocessing is thus a vital direction for statistical research.

A time-varying measurement error model for age of onset of a psychiatric diagnosis: applied to first depressive episode diagnosed by the Composite International Diagnostic Interview (CIDI)

A substantial degree of uncertainty exists surrounding the reconstruction of events based on memory recall. This form of measurement error affects the performance of structured interviews such as the Composite International Diagnostic Interview (CIDI), an important tool to assess mental health in the community. Measurement error probably explains the discrepancy in estimates between longitudinal studies with repeated assessments (the gold-standard), yielding approximately constant rates of depression, versus cross-sectional studies which often find increasing rates closer in time to the interview. Repeated assessments of current status (or recent history) are more reliable than reconstruction of a person's psychiatric history based on a single interview. In this paper, we demonstrate a method of estimating a time-varying measurement error distribution in the age of onset of an initial depressive episode, as diagnosed by the CIDI, based on an assumption regarding age-specific incidence rates. High-dimensional non-parametric estimation is achieved by the EM-algorithm with smoothing. The method is applied to data from a Norwegian mental health survey in 2000. The measurement error distribution changes dramatically from 1980 to 2000, with increasing variance and greater bias further away in time from the interview. Some influence of the measurement error on already published results is found.

Results on Differential and Dependent Measurement Error of the Exposure and the Outcome Using Signed Directed Acyclic Graphs

Measurement error in both the exposure and the outcome is a common problem in epidemiologic studies. Measurement errors in the exposure and the outcome are said to be independent of each other if the measured exposure and the measured outcome are statistically independent conditional on the true exposure and true outcome (and dependent otherwise). Measurement error is said to be nondifferential if measurement of the exposure does not depend on the true outcome conditional on the true exposure and vice versa; otherwise it is said to be differential. Few results on differential and dependent measurement error are available in the literature. Here the authors use formal rules governing associations on signed directed acyclic graphs (DAGs) to draw conclusions about the presence and sign of causal effects under differential and dependent measurement error. The authors apply these rules to 4 forms of measurement error: independent nondifferential, dependent nondifferential, independent differential, and dependent differential. For a binary exposure and outcome, the authors generalize Weinberg et al.'s (Am J Epidemiol. 1994;140(6):565-571) result for nondifferential measurement error on preserving the direction of a trend to settings which also allow measurement error in the outcome and to cases involving dependent and/or differential error.

Measurement error in network data: A re-classification

Research on measurement error in network data has typically focused on missing data. We embed missing data, which we term false negative nodes and edges, in a broader classification of error scenarios. This includes false positive nodes and edges and falsely aggregated and disaggregated nodes. We simulate these six measurement errors using an online social network and a publication citation network, reporting their effects on four node-level measures – degree centrality, clustering coefficient, network constraint, and eigenvector centrality. Our results suggest that in networks with more positively-skewed degree distributions and higher average clustering, these measures tend to be less resistant to most forms of measurement error. In addition, we argue that the sensitivity of a given measure to an error scenario depends on the idiosyncracies of the measure's calculation, thus revising the general claim from past research that the more ‘global’ a measure, the less resistant it is to measurement error. Finally, we anchor our discussion to commonly-used networks in past research that suffer from these different forms of measurement error and make recommendations for correction strategies.

Correcting the Bias in the Concentration Index When Income is Grouped

The problem introduced by grouping income data when measuring socioeconomic inequalities in health (and health care) has been highlighted in a recent study in this journal. We re(examine this issue and show there is a tendency to underestimate the concentration index at an increasing rate when lowering the number of income categories. This bias results from a form of measurement error and we propose two correction methods. Firstly, the use of instrumental variables (IV) can reduce the error within income categories. Secondly, through a simple formula for correction that is based only on the number of groups. We compare the performance of these methods using data from 15 European countries and the United States. We find that the simple correction formula reduces the impact of grouping and always outperforms the IV approach. Use of this correction can substantially improve comparisons of the concentration index both across countries and across time.

‘Do you still like to play with him?'

There is a wealth of research on the relation of sociometric status and other psychological constructs and behavioural properties. Surprisingly, few studies focus on the stability of sociometric status. Existing studies focus on long-term stability, on short-term variability as a form of measurement error, and are limited to ratings that children receive from others. The present article argues that stability and variability are inextricably bound together. We investigate repeated short-term stability and variability in 6- to 8-year-old children’s given and received ratings. In addition to stability, children show a characteristic pattern of fluctuation and variability. In their patterns of giving ratings to others, children have a certain relatively constant ‘style’. The focus of the present article lies on the presentation of new methodological approaches to the study of variability and stability of sociometric ratings in children. (Netherlands Journal of Psychology 63, 86-101.)

A critical appraisal of epidemiological studies comes from basic knowledge: a reader's guide to assess potential for biases.

Scientific literature may be biased because of the internal validity of studies being compromised by different forms of measurement error, and/or because of the selective reporting of positive and 'statistically significant' results. While the first source of bias might be prevented, and in some cases corrected to a degree, the second represents a pervasive problem afflicting the medical literature; a situation that can only be 'corrected' by a change in the mindset of authors, reviewers, and editors. This review focuses on the concepts of confounding, selection bias and information bias, utilising explanatory examples and simple rules to recognise and, when possible, to correct for them. Confounding is a mixing of effects resulting from an imbalance of some of the causes of disease across the compared groups. It can be prevented by randomization and restriction, and controlled by stratification, standardization or by using multivariable techniques. Selection bias stems from an absence of comparability among the groups being studied, while information bias arises from distorted information collection techniques. Publication bias of medical research results can invalidate evidence-based medicine, when a researcher attempting to collect all the published studies on a specific topic actually gathers only a proportion of them, usually the ones reporting 'positive' results. The selective publication of 'statistically significant' results represents a problem that researchers and readers have to be aware of in order to face the entire body of published medical evidence with a degree of scepticism.

The design and statistical analysis of animal experiments.

Research scientists are strongly urged to read all of the papers in this issue if they want to improve the quality and quantity of their research and ensure that their papers are accepted in high-quality journals. There is always room for improvement; some experiments are performed quite badly (Festing 1994, 2000; Festing et al. 2002; Roberts et al. 2002). The main problem seems to be poor training in statistics and a general horror among some biologists of anything to do with mathematics. However, experience shows that most scientists have a keen interest in experimental design and are not nearly as frightened of statistical methods that are planned at the same time as their experiments. For this reason, I believe that courses for research workers should emphasize design aspects. After all, if an experiment is well designed, it is relatively easy to get help with the statistical analysis; but if it is incorrectly designed, it may be impossible to extract any useful information from it. Animal experiments are usually done to discover something about the biology of, for example, the species, strain, or sex being studied and indirectly to infer something about humans or other target species. Thus, laboratory animals are usually used as “models” of some other species. The use of models involves the following three distinct steps: (1) choice of a suitable model, based on our current knowledge about disease processes in the target species and in potential model organisms; (2) one or more experiments to indicate how the model responds to any applied treatments; and (3) consideration of the relevance of the results for the target species. This issue of ILAR Journal is concerned with the second step, the design and analysis of experiments using animals that have already been chosen as likely to be informative indirectly about the target species. Anyone planning an animal experiment should first consider the important ethical issues involved. A good framework for this consideration is to use Russell and Burch’s “3Rs” (Russell and Burch 1959), namely, whether the animal model can be replaced by a less or nonsentient alternative such as an insect, nematode, cell culture, or computer simulation. If not, the possibility of refining the experiment in an effort to minimize pain and distress for each individual should be considered. Animals should be housed in good conditions, free of pathogens. Surgical techniques should use appropriate anesthesia and analgesia, and humane endpoints should be used. Finally, the number of animals should be reduced to the minimum required to achieve the scientific objectives of the experiment. Reduction can be achieved by choosing the most appropriate animal model (Herman 2002), allowing for the fact that it is not always necessary for the models to mimic the human condition exactly (Elsea and Lucas 2002), and by good experimental design and statistical methods, as outlined in this issue. All animal models are subject to biological variation as a result of genetic and nongenetic variation and the interaction between them. Even genetically identical littermates will vary to some extent as a result of chance developmental effects, social hierarchy, and unequal exposure to environmental influences. Good experimental design aims to control this variation so that it does not obscure any treatment effect, with the statistical analysis being designed to extract all useful information and take into account any remaining variation. A first step is to review likely sources of variation (Howard 2002). Species, strains, sexes, and individuals differ as a result of genetics; however, there are also many environmental influences that can either bias the results in some way or can contribute so much “noise” that it is impossible to identify the effects of any experimental treatments. Some of this variation is even introduced by the experimenter in the form of measurement error. One of the most serious causes of interindividual variation is clinical or subclinical disease, but fortunately (at least in the case of rats and mice), so-called “specific pathogen-free” animals have been available for many years. This availability does not mean that the problem has disappeared. Barriers break down due to new and existing pathogens somehow gaining entry, particularly when investigators exchange strains of animals; however, at least this source of variation is well Michael F. W. Festing, M.Sc., Ph.D., D.Sc., CStat., BIBiol., is a Senior Research Scientist at the MRC Toxicology Unit, University of Leicester, UK.

Measurements of Canada Goose Morphology: Sources of Error and Effects on Classification of Subspecies

Subspecific classification of Canada geese (Branta canadensis) based on morphological measurements serves many management and research functions, such as determining harvest pressure on subspecies or estimating the population composition of wintering flocks. Despite this widespread use, the magnitude of error involved in such measurements, the effect of observer experience on measurement error, and the effect of measurement error on classification are not known. To investigate these issues, we carried out a study on Canada geese harvested in Wisconsin involving replicated measurements by observers of different experience levels. Measurement error for experienced observers was half as large as that for inexperienced observers (6-10% vs. 13-21% of all variability for all structures except the tarsus). Experienced observers measured the skull and culmen most precisely, the tarsus least precisely. Consistent differences among observers (observer bias) that could bias classification were smaller for experienced observers. We used reference data and distributional assumptions to estimate that without observer bias or other forms of measurement error, 8-9% of geese measured would be misclassified because of actual size overlap between subspecies. Without observer bias, remaining measurement error among experienced and inexperienced observers increased misclassification by 1% and 2%, respectively. Observer bias can increase misclassification substantially beyond these levels, depending on the magnitude and direction of observer bias and the prevalence of the subspecies. Misclassification of geese resulted in overestimating the prevalence of the less common subspecies in mixed populations, which may be important in developing management strategies. We recommend training observers and standardizing measurement procedures primarily to reduce observer bias that leads to biased classification of geese, and secondarily to reduce other components of measurement error.