Modern Statistical Software Research Articles

Causal mediation analysis: How to avoid fooling yourself that X causes Y.

The purpose of many preclinical studies is to determine whether an experimental intervention affects an outcome through a particular mechanism, but the analytical methods and inferential logic typically used cannot answer this question, leading to erroneous conclusions about causal relationships, which can be highly reproducible. A causal mediation analysis can directly test whether a hypothesised mechanism is partly or completely responsible for a treatment's effect on an outcome. Such an analysis can be easily implemented with modern statistical software. We show how a mediation analysis can distinguish between three different causal relationships that are indistinguishable when using a standard analysis.

МЕТОД ОЦЕНИВАНИЯ ЛОГИСТИЧЕСКИХ РЕГРЕССИЙ С ПОМОЩЬЮ АППАРАТА ЛИНЕЙНОГО ПРОГРАММИРОВАНИЯ

The article is devoted to the problem of estimating logistic regressions where the explanatory variable has only two values, 0 and 1. The predicted values of the explanatory variable of the estimated logistic regression are interpreted as the probabilities of the occurrence of some event. As a result, such models are widely used to solve classifi cation problems. In practice, the maximum likelihood estimation, which is implemented in many modern statistical packages, is mainly used to estimate logistic regressions. One of its disadvantages, for example, is that it does not provide unique estimates when grouping objects into two separate classes. The study proposes a new method for estimating logistic regressions. Conventionally, it can be divided into two stages. The fi rst stage consists of solving a specially formulated linear programming problem. As a result, the weighting coeffi cients of the linear combination of explanatory variables are found. In fact, classifi cation is already carried out at this stage. The second stage is to calibrate the probability scale. Computational experiments were carried out based on a real sample of volume 100. The new method has proven its effi ciency when objects are completely separable into two classes. In addition, in terms of the number of correctly predicted cases, the new method was never inferior to the maximum likelihood estimation, and even surpassed the latter in one of the experiments.

Training in modern statistical methodologies and software tools for the definition and analysis of (stochastic) quantitative microbial risk assessment models with a comparison between the Hungarian and Spanish food supply chains.

Human pathogenic Salmonella enterica strains have been infecting people since historical times. The original human pathogens, typhoid Salmonella strains (e.g. S. Typhi) played a huge role in the previous centuries but nowadays in the developed world the number of cases or outbreaks caused by these serotypes deceased due to the development of personal and public hygiene. Nowadays in these regions the animal-borne zoonotic serotypes (e.g. S. Enteritidis) became more important because of their high prevalence in intensive animal husbandry. But these bacteria can also appear in fruits and vegetables. The fellow joined the scientific work of the Polytechnic University of Cartagena, Spain about the safety of plant-based products, where he could gain experience in microbiological laboratory exercises and theoretical calculations of statistics and modelling. The activities in the laboratory were part of the research lines already established at the host institution, being based on the protocols they have already implemented. Nonetheless, the fellow had the opportunity to design his own experiment, do the experimental work required and analysed the data within the context of a qualitative microbiological risk assessment. The main focus was on the heat resistance of two strains of zoonotic Salmonella spp. at different temperatures. Experiments were done using a reference strain and an extremely resistant variant to evaluate this rare phenotype. The experiments were executed using a Mastia thermoresistometer, a device patented by the host institution that provides more control when studying thermal treatments than traditional methods. The data was analysed using the principles of predictive microbiology, using the D-value as an estimate of heat resistance that provides insight into the bacterial behaviour. For this, the fellow used the bioinactivation software, developed within the host group. Through the work and results the fellow learned the principles of quantitative microbiological risk assessment (QMRA) and predictive microbiology, which was the aim for the EU-FORA programme.

Statistical considerations for assessing precision of heterogeneous duplicate measurements: An application to pharmaceutical bioanalysis.

Duplicate analysis is a strategy commonly used to assess precision of bioanalytical methods. In some cases, duplicate analysis may rely on pooling data generated across organizations. Despite being generated under comparable conditions, organizations may produce duplicate measurements with different precision. Thus, these pooled data consist of a heterogeneous collection of duplicate measurements. Precision estimates are often expressed as relative difference indexes (RDI), such as relative percentage difference (RPD). Empirical evidence indicates that the frequency distribution of RDI values from heterogeneous data exhibits sharper peaks and heavier tails than normal distributions. Therefore, traditional normal-based models may yield faulty or unreliable estimates of precision from heterogeneous duplicate data. In this paper, we survey application of the mixture models that satisfactorily represent the distribution of RDI values from heterogeneous duplicate data. A simulation study was conducted to compare the performance of the different models in providing reliable estimates and inferences of percentile calculated from RDI values. These models are readily accessible to practitioners for study implementation through the use of modern statistical software. The utility of mixture models are explained in detail using a numerical example.

Modeling of Motor Insurance Extreme Claims through Appropriate Statistical Distributions

In the area of insurance, probability modeling has a wide variety of applications. In life insurance, the compensation sum is calculated in advance and may often be estimated using actuarial techniques, while in motor insurance, the claim amount is generally not known in advance. In the insurance business, the improvement of actuarial risk control strategies is an essential technique for controlling insurance risk. Although an insurance company’s risk assessment about its solvency is a complex and detailed problem, its solution begins with statistical modeling of individual claims’ amounts. This article emphasizes the possible ways of obtaining a suitable probability distribution model that accurately explains insurance risks and how to use such a model for risk management purposes. For this reason, we have applied modern programming techniques and statistical software implemented the methods provided based on data on premium amounts of third-party motor insurance claims.

Внутренняя структура способов выполнения учебных действий и стадий становления субъектности студентов

The topical problem of pedagogical psychology is to identify the conditions and factors of students’ personal agency development. One of the ways to solve the problem is to build adequate mathematical models. The purpose of the study is to justify the use of multimediator analysis in modelling a fractal psychological and pedagogical system so as to describe the ways and stages of students’ agency development in the course of the study. It is hypothesized that ways of performing learning actions are considered as elements of the stages of learners’ personal agency formation. And in this function they play a twofold role: they act both as predictors of the agency development stages and mediators that determine the interconnection between the ways and stages of the personal agency development. The test of the multimediator analysis was carried out in three groups of 2nd–4th-year university students, studying full-time (57 people; average age 20.6 years, Sd = 2.4 years; all females) and two groups of second-year students of college (40 people, 39 of them are boys, aged 17.5, Sd = 0.71). The practical application of the research results makes it possible to identify both the internal structure of the ways of performing learning activities and the stages of students’ personal agency development. The prospects of using multimediator analysis with the help of modern statistical packages are shown.

Looking Backwards, Looking Forwards: Evaluating the Roe Handaxe Methodology in the Twenty-first Century and the Introduction of a New ‘Roe-type’ Index

ABSTRACT The methodology developed by Derek Roe in the 1960s to describe the outline shape of handaxes has become an industry standard amongst scholars of the Acheulean. It was applied to the UK and in a modified form in East and Southern Africa. It is being used by a new generation of archaeologists to understand successive Acheulean occupations of the UK in the Middle Pleistocene. There has been a relatively little critique of the method. This paper explores some of the issues that the method raises and suggests and that while it may not fairly represent specific outlines shapes, in every case, it does fairly represent a basic bauplan. A new Roe-style index is suggested, the tip elongation index which helps structure the methodology when used with modern statistical and graphic software packages.

Neural Networks in Forecasting Disease Dynamics

Introduction.In recent years, computer technologies are more and more widely used in medicine. Thus, medical neuro‑ informatics solves diagnostic and forecasting tasks using neural networks.Materials and methods. Using the example of erysipelas, the possibility of forecasting the course and outcome of the dis‑ ease is demonstrated. A retrospective study of the medical histories of patients treated for erysipelas at the Ufa Clinical Hospital No.8 during 2006–2015 was carried out. Modern statistical packages and the MATLAB environment were used.Results and discussion.The conducted comparative analysis showed a 3-layer recurring network of direct distribution to be the most suitable neural network architecture. The optimal structure of the neural network was found to be: 27–6–1 (i.e. 27 neurons are used at the entrance; 6 — in a hidden layer; 1 — in the output layer). The best convergence of the network learning process is provided by the quasi-Newton and conjugated gradient algorithms. In order to assess the effectiveness of the proposed neural network in predicting the dynamics of inflammation, a comparative analysis was carried out using a number of conventional methods, such as exponential smoothing, moving average, least squares and group data handling.Conclusion.The proposed neural network based on approximation and extrapolation of variations in the patient’s medi‑ cal history over fixed time window segments (within the ‘sliding time window’) can be successfully used for forecasting the development and outcome of erysipelas.

Models for Counts and Particle Size Distributions of Subvisible Particle Data.

Traditional statistical analyses of subvisible particle data are usually based on either descriptive statistics, normal-based methods, or standard Poisson models. These methods often do not adequately describe the counts or particle size distribution. They usually ignore relevant information represented in the data, such as count correlation. Therefore, any meaningful analyses of subvisible particle data require a reasonable representation of counts and particle size distribution and the correlation in the data. Such comprehensive approaches are not widely available or used when analyzing subvisible particle data. In this article, we propose the use of generalized linear mixed models to analyze the counts and the particle size distribution of subvisible particle data. These models make optimal use of the information in the data and allow flexible approaches for the analyses of a wide range of data structures. They are readily accessible to practitioners through the use of modern statistical software. These models are demonstrated with two numerical examples using two different data structures.

Analysis of modern statistical software

Analysis of modern statistical software

Approaches to the Development of Drugs with the Use of Modern Statistical Software Concepts and Quality-by-Design

Introduction. The development of the composition of drugs is very difficult and long process. To implement it, it is currently recommended to apply a modern approach – quality through the development of Quality by design. One of the methods for introducing this approach is the use of multifactorial experiments.Aim. The aim of the work was to develop the composition and technology of effervescent tablets using the Minitab program.Materials and methods. For the study used the program Minitab (ver. 18.), Minitab LLC, USA. The program created an experiment plan and evaluated those obtained on the basis of its models using regression analysis.Results and discussion. The composition and technology of effervescent tablets was developed by direct compression using the principles of Quality by design. Algorithms of actions for using the Minitab software package for development are proposed. An experiment plan was constructed and models of the dependence of quality indicators on the composition and technological parameters were generated. The resulting sub-sections were analyzed using regression analysis tools.Conclusion. Based on the results of the work, the composition of effervescent tablets obtained by direct compression technology, as well as the algorithm of drug development in the Minitab program, was proposed.

A study on the role of trust factor in adopting personalised medicine

ABSTRACT Despite all the potential that Personalised Medicine (PM) has in reducing healthcare costs and treatment errors, as well as in bringing about significant improvement in the mode of healthcare, the adoption rate of Personalised Medicine is not yet up to the mark. Some research has been done in the area of Personalised Medicine and several new drugs have been already discovered. However, adequate research is yet to be performed to sort out various hurdles that are hindering implementation of this novel medicare system on a larger scale. One of the hurdles is the so-called trust factor. In the present paper, it has been our endeavour to investigate whether the trust issues have a moderating effect on the success of adoption of personalised medicine. A survey-based post facto methodology has been adopted for this purpose. Responses have been collected from a number of people, who are medical practitioners/medical researchers/pharmacists/regulatory board members/patients. The dataset obtained thereby has been analysed by using advanced statistical techniques and modern statistical software packages. The study indicates that the trust issues have a significant moderating effect on the successful adoption of personalised medicine.

Confidence intervals in dermatology

This study from the U.S.A. aimed to find out how often confidence intervals are used when reporting dermatology research. Confidence intervals relate to two values (‘intervals’) that the researchers, using data analysis, believe that the actual, true value will fall within. A 95% confidence interval indicates that if an experiment were repeated 100 times, and a 95% confidence interval was calculated each time, 95 of the intervals would contain the true value while five of the intervals would not contain the true value. A confidence interval might be used when you want to estimate something about a very large number of people, and can therefore only use a sample of those people. For example, if you want to know how many people in the UK have dry skin, you cannot ask every person, so you use a sample and then calculate a range for the mean number of people in your sample who have dry skin. As the sample is never going to be an exact representation of the whole population, you might say that you are 95% confident that the mean number of people in the UK with dry skin is, for example, between 20% and 30%. Confidence intervals help readers understand both the statistical significance and the clinical significance more easily. Most modern statistical software packages conveniently calculate confidence intervals and provide an intuitive way to understand study results. The authors collected dermatology articles from a variety of major dermatology journals as well as dermatology articles from the New England Journal of Medicine over the past decade. The authors found that reporting of this important metric is disappointingly low in dermatology journals, with only 1 in 5 articles containing confidence intervals, which was significantly lower than reporting in the New England Journal of Medicine. Articles published more recently were more likely to report confidence intervals. After controlling for factors like type of study and year of publication, the New England Journal of Medicine still reported confidence intervals more frequently. This is likely due to their extremely rigorous statistical peer review process. While such strict requirements cannot be adopted by all journals, increased awareness of this simple metric could significantly improve the statistical reporting of the dermatology literature. We urge dermatology researchers to include this important metric in their findings. Likewise, we recommend improved statistical training as part of dermatology residency training.

Fitting the damped trend method of exponential smoothing

The well-established forecasting methods of exponential smoothing rely on the “optimal” estimation of parameters if they are to perform well. A grid search procedure to minimise the MSE is often used in practice to fit exponential smoothing methods, especially in large inventory control applications. Grid searches are also found in some modern statistical software. We ask whether the ex ante forecast accuracy of the damped trend method of exponential smoothing can be improved by optimising parameters. Furthermore, we ask whether the method should be fitted according to a mean absolute error criterion rather than the mean squared error commonly used in practice. We found that model-fitting matters. Parameter optimization makes significant improvements in forecast accuracy regardless of the fit criterion. We also show that the mean absolute error criterion usually produces better results.

A guide to missing data for the pediatric nephrologist.

Missing data is an important and common source of bias in clinical research. Readers should be alert to and consider the impact of missing data when reading studies. Beyond preventing missing data in the first place, through good study design and conduct, there are different strategies available to handle data containing missing observations. Complete case analysis is often biased unless data are missing completely at random. Better methods of handling missing data include multiple imputation and models using likelihood-based estimation. With advancing computing power and modern statistical software, these methods are within the reach of clinician-researchers under guidance of a biostatistician. As clinicians reading papers, we need to continue to update our understanding of statistical methods, so that we understand the limitations of these techniques and can critically interpret literature.

The Value of Scatter Plots

The Value of Scatter Plots

Identifying the Structure of the Experimental Design

In many areas of scientific research, complex experimental designs are now routinely employed. The statistical analysis of data generated when using these designs may be carried out by a statistician; however, modern statistical software packages allow such analyses to be performed by non-statisticians. For the non-statistician, failing to correctly identify the structure of the experimental design can lead to incorrect model selection and misleading inferences. A procedure, which does not require expert statistical knowledge, is described that focuses the non-statistician's attention on the relationship between the experimental material and design, identifies the underlying structure of the selected design, and highlights any potential weaknesses it may have. These are important precursors to the randomization and subsequent statistical analysis and can be easily overlooked by a non-statistician. The process is illustrated using a generalization of the Hasse diagram and has been implemented in a program written in R.

A Note on the Relationship between the Pearson Product-Moment and the Spearman Rank-Based Coefficients of Correlation

This note derives the relationship between the Pearson product-moment coefficient of correlation and the Spearman rank-based coefficient of correlation for the bivariate normal distribution. This new derivation shows the relationship between the two correlation coefficients through an infinite cosine series. A computationally efficient algorithm is also provided to estimate the relationship between the Pearson product-moment coefficient of correlation and the Spearman rank-based coefficient of correlation. The algorithm can be implemented with relative ease using current modern mathematical or statistical software programming languages e.g. R, SAS, Mathematica, Fortran, et al. The algorithm is also available from the author of this article.

The Life of p: “Just Significant” Results are on the Rise

Null hypothesis significance testing uses the seemingly arbitrary probability of .05 as a means of objectively determining whether a tested effect is reliable. Within recent psychological articles, research has found an overrepresentation of p values around this cut-off. The present study examined whether this overrepresentation is a product of recent pressure to publish or whether it has existed throughout psychological research. Articles published in 1965 and 2005 from two prominent psychology journals were examined. Like previous research, the frequency of p values at and just below .05 was greater than expected compared to p frequencies in other ranges. While this overrepresentation was found for values published in both 1965 and 2005, it was much greater in 2005. Additionally, p values close to but over .05 were more likely to be rounded down to, or incorrectly reported as, significant in 2005 than in 1965. Modern statistical software and an increased pressure to publish may explain this pattern. The problem may be alleviated by reduced reliance on p values and increased reporting of confidence intervals and effect sizes.

Classification Techniques in Analyzing Surgical Outcomes Data

We live in a world that has been characterized by “big data.” Data that have been collected for years and even decades are now seen as important inputs to understanding current processes, predicting outcomes, and prescribing actions. This movement is so prominent that the term data mining has been supplanted by analytics to describe the broadly based and sophisticated methodologies that have been created and validated to allow these massive databases to be analyzed. Every major university has at least one course entitled “Analytics” and often degree programs to address the needs of business, engineering, psychology, and others. The NSQIP was initially introduced within the Veterans Affairs system in the mid-1990s as the first structured, risk-adjusted, prospective database of surgical outcomes. A private-sector version was developed and then implemented by the American College of Surgeons (ACS) and has rapidly evolved and matured into a national surgical outcomes database. This database now qualifies as “big data” and is available to investigators for analysis of specific surgical procedures and evaluation of surgical quality in hospitals and institutions. During the same years that the ACS NSQIP has “matured,” there has been considerable enhancement of the analytical techniques available to study large databases. These newer techniques were developed to address many of the limitations of the earlier methods and have been made possible because of the rapid expansion of the computing power of modern desktop computers and servers. The purpose of this review is to provide an overview of the analytical techniques currently available to enhance the ability of the reader, either clinician or investigator, to interpret and apply predictive modeling in surgical care. This review is not a treatise on statistics and is not intended to be an in-depth description of how to perform database queries or to use modern statistical software. The interested reader is referred to the focused references and