Importance Of Sample Size Research Articles

Using pseudo-absence models to test for environmental selection in marine movement ecology: the importance of sample size and selection strength

BackgroundUnderstanding the selection of environmental conditions by animals requires knowledge of where they are, but also of where they could have been. Presence data can be accurately estimated by direct sampling, sightings, or through electronic tag deployments. However, absence data are harder to determine because absences are challenging to measure in an uncontrolled setting. To address this problem, ecologists have developed different methods for generating pseudo-absence data relying on theoretical movement models. These null models represent the movement of environmentally naive individuals, creating a set of locations that animals could have been if they were not exhibiting environmental selection.MethodsHere, we use four different kinds of null animal movement models—Brownian motion, Lévy walks, Correlated random walks, and Joint correlated random walks to test the ability and power of each of these null movement models to serve as appropriate animal absence models. We use Kolmogorov-Smirnov tests to detect environmental selection using two data sets, one of simulated animal tracks biased towards warmer sea surface temperatures, and one of 57 observed blue shark tracks of unknown sea surface temperature selection.ResultsThe four different types of movement models showed minimal difference in the ability to serve as appropriate null models for environmental selection studies. Selection strength and sample size were more important in detecting true environmental selection. We show that this method can suffer from high false positive rates, especially in the case where animals are not selecting for specific environments. We provide estimates of test accuracy at different sample sizes and selection strengths to avoid false positives when using this method.ConclusionWe show how movement models can be used to generate pseudo-absences and test for habitat selection in marine organisms. While this approach efficiently detects environmental selection in marine organisms, it cannot detect the underlying mechanisms driving this selection.

Not just “big” data: Importance of sample size, measurement error, and uninformative predictors for developing prognostic models for digital interventions

There is strong interest in developing a more efficient mental health care system. Digital interventions and predictive models of treatment prognosis will likely play an important role in this endeavor. This article reviews the application of popular machine learning models to the prediction of treatment prognosis, with a particular focus on digital interventions. Assuming that the prediction of treatment prognosis will involve modeling a complex combination of interacting features with measurement error in both the predictors and outcomes, our simulations suggest that to optimize complex prediction models, sample sizes in the thousands will be required. Machine learning methods capable of discovering complex interactions and nonlinear effects (e.g., decision tree ensembles such as gradient boosted machines) perform particularly well in large samples when the predictors and outcomes have virtually no measurement error. However, in the presence of moderate measurement error, these methods provide little or no benefit over regularized linear regression, even with very large sample sizes (N = 100,000) and a non-linear ground truth. Given these sample size requirements, we argue that the scalability of digital interventions, especially when used in combination with optimal measurement practices, provides one of the most effective ways to study treatment prediction models. We conclude with suggestions about how to implement these algorithms into clinical practice.

The Three Glycotypes in the London Classification System of Sporadic Creutzfeldt-Jakob Disease Differ in Disease Duration.

Sporadic Creutzfeldt-Jakob disease (sCJD) is the most common form of CJD and is believed to be caused by the misfolding and aggregation of endogenous prion protein. Several classification systems have been developed to correlate the molecular characteristics of these misfolded prions (PrPSc) to the heterogeneous clinical presentations of sCJD. A central component of these systems is glycotyping, which involves the interpretation of the results of western immunoblotting of the protease-resistant fragment of the misfolded prion protein (PrPres). The two main classification systems differ in their recognition of a unique banding pattern on electrophoretic gels correlating to a putative clinical subtype. The perpetuation of both classification systems within scientific literature is, in part, due to a paucity of high-level evidence that conclusively addresses the merit of recognising each unique banding pattern. Here, 110 post-mortem confirmed cases of sCJD collected at the Australian Creutzfeldt-Jakob Disease Registry (ANCJDR) between 1993 and 2018 were analysed and classified as per the London classification system. The data presented here demonstrated that sCJD cases with 'type 1' and 'type 2' PrPSc as defined by the London classification system differ in their disease duration. No other differences in clinical phenotype or biological characteristics were found to be statistically significant. These findings highlight the importance of sample size and replicability in analyses of this rare disease process. Recognising these glycotypes as phenotypically distinct may represent 'best practice' in the collection and processing of sCJD samples within international registries for research purposes.

Biomarkers for Bladder Pain Syndrome/Interstitial Cystitis

In this manuscript, the literature published in the past 5 years about BPS/IC biomarkers was reviewed. Topics include BPS/IC biomarkers, the classical BPS/IC diagnostic, clinical outcomes as a new concept of BPS/IC phenotyping biomarkers, BPS/IC urinary biomarkers, BPS/IC bladder tissue biomarkers and BPS/IC biomarkers outside the bladder. The importance of patients’ phenotyping and sub-grouping emerged as a way of finding new BPS/IC biomarkers and treatment. Also, the use of non-invasive methodologies and of artificial intelligence approaches has been gaining momentum in BPS/IC biomarkers studies. Biomarker research points to the need of following an approach based on a systemic theory instead of a bladder-centred theory. Also, the importance of sample size and the choice of controls to validate BPS/IC biomarkers have been elevated.

The importance of sample size with regard to the robustness of postmortem reference values

Evaluating postmortem toxicological results is a challenging task due to multiple factors affecting blood concentrations after death. In order to improve the diagnostic accuracy in cases of suspected fatal intoxication different compilations of postmortem reference drug concentrations are often used. However, it is not clear what constitutes a reliable postmortem reference value.The current study presents reference concentrations for 13 substances from seven substance groups according to a standardized protocol. The reference concentrations were gathered from 3767 autopsy cases and subdivided into intoxications by one substance only (Group A, n=611), multi-substance intoxications (Group B, n=1355) and postmortem controls, in which incapacitation by drugs were excluded (Group C, n=1801). In particular, this study presents statistical information about the precision and conformity change with various sample sizes.Based on the present data >10 detections are usually needed, for the substances examined, to differentiate between intoxication cases and controls. Repeated samplings show that the median of small samples (N=≤5) has a high variation (normalized interquartile range 138–75%) and that a high number of detections (N=>20) in each group are needed to reduce the variation.

An N-pact factor for clinical psychological research.

Although an emphasis on adequate sample size and statistical power has a long history in clinical psychological science (Cohen, 1992), increased attention to the replicability of scientific findings has renewed focus on the importance of statistical power (Bakker, van Dijk, & Wicherts, 2012). These recent efforts have not yet circled back to modern clinical psychological research, despite the importance of sample size and power in producing a credible body of evidence. As one step in this process of scientific self-examination, the present study estimated an N-pact Factor (the statistical power of published empirical studies to detect typical effect sizes; Fraley & Vazire, 2014) in 2 leading clinical journals (the Journal of Abnormal Psychology [JAP] and the Journal of Consulting and Clinical Psychology [JCCP]) for the years 2000, 2005, 2010, and 2015. Study sample size, as one proxy for statistical power, is a useful focus because it allows comparisons with other subfields and may highlight some of the core methodological differences between clinical and other areas. We found that, across all years examined, the average median sample size in clinical research was 179 participants (175 for JAP and 182 for JCCP). The power to detect a small to medium effect size of .20 is just below 80% for both journals. Although the clinical N-pact factor was higher than that estimated for social psychology, the statistical power in clinical journals is still limited to detect many effects of interest to clinical psychologists, with little evidence of improvement in sample sizes over time. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

The importance of sample size in marine megafauna tagging studies.

Telemetry is a key, widely used tool to understand marine megafauna distribution, habitat use, behavior, and physiology; however, a critical question remains: "How many animals should be tracked to acquire meaningful data sets?" This question has wide-ranging implications including considerations of statistical power, animal ethics, logistics, and cost. While power analyses can inform sample sizes needed for statistical significance, they require some initial data inputs that are often unavailable. To inform the planning of telemetry and biologging studies of marine megafauna where few or no data are available or where resources are limited, we reviewed the types of information that have been obtained in previously published studies using different sample sizes. We considered sample sizes from one to >100 individuals and synthesized empirical findings, detailing the information that can be gathered with increasing sample sizes. We complement this review with simulations, using real data, to show the impact of sample size when trying to address various research questions in movement ecology of marine megafauna. We also highlight the value of collaborative, synthetic studies to enhance sample sizes and broaden the range, scale, and scope of questions that can be answered.

ML versus IV estimates of spatial SUR models: evidence from the case of Airbnb in Madrid urban area

In recent decades we have seen an increased interest in the use of seemingly unrelated regressions models (SUR) in a spatial context, with compelling case studies in different fields. This upsurge has favoured the development of new and more efficient inference techniques. At present, the user has a basic toolkit to deal with this kind of model, that is, however, in need of improvement. This paper focuses on the question of estimating, quickly and accurately, spatial SUR models. The most popular procedure is maximum likelihood (ML) which guarantees precision at the cost of a high computational burden. This is especially true in cases of large sample size and strong spatial structure. We explore simpler estimation algorithms such as instrumental variables (IV), which expedites calculation at the cost of a certain loss in quality of the estimates. We focus on the importance of sample size and the trade-off between accuracy and speed. To that end, we perform a comprehensive simulation experiment in which we compare ML and IV algorithms, looking for their strengths and weaknesses. The paper includes two applications to the case of Airbnb in the urban area of Madrid. First, we estimate a spatial SUR hedonic model of accommodation prices, using micro-data for three different cross sections a la Anselin, that is, considering temporal correlation plus spatial structure. Then, the data are aggregated by neighbourhoods. We specify a spatial SUR model with two equations (apartments and rooms) also using three cross sections. In both cases, the models are estimated by ML and IV using the spsur R package (Angulo et al. in spSUR: spatial seemingly unrelated regression models. R Package version 1.0.0.3, 2019), with the aim of illustrating its capabilities.

Acute rejection in kidney transplantation and the evaluation of associated polymorphisms (SNPs): the importance of sample size.

Background Acute rejection (AR) is one of the most frequent complications after kidney transplantation (KT). Scientific evidence reports that some single-nucleotide polymorphisms (SNPs) located in genes involved in the immune response and in the pharmacokinetics and pharmacodynamics of immunosuppressive drugs are associated with rejection in renal transplant patients. The aim of this study was to evaluate some SNPs located in six genes: interleukin-10 (IL-10), tumor necrosis factor (TNF), adenosine triphosphate-binding cassette sub-family B member 1 (ABCB1), uridine diphosphate glucuronosyltransferase family 1 member A9 (UGT1A9), inosine monophosphate dehydrogenase 1 (IMPDH1) and IMPDH2. Methods We enrolled cases with at least one AR after KT and two groups of controls: patients without any AR after KT and healthy blood donors. Genetic analysis on DNA was performed. The heterozygosity (HET) was determined and the Hardy-Weinberg equilibrium (HWE) test was performed for each SNP. The sample size was calculated using the QUANTO program and the genetic associations were calculated using the SAS program (SAS Institute Inc., Cary, NC, USA). Results In our previous preliminary study (sample size was not reached for cases), the results showed that patients with the C allele in the SNP rs1045642 and the A allele in the SNP rs2032582 of the ABCB1 gene had more frequent AR. In contrast, with the achievement of sample size, the trend of the previous data was not confirmed. Conclusions Our study highlights a fundamental aspect of scientific research that is generally presumed, i.e. the sample size of groups enrolled for a scientific study. We believe that our study will make a significant contribution to the scientific community in the discussion of the importance of the analysis and the achievement of sample size to evaluate the associations between SNPs and the studied event.

Random errors in research and the importance of sample size

Random errors in research and the importance of sample size

Teacher's Toolkit: An Inquiry-Based Activity to Show the Importance of Sample Size and Random Sampling

Teacher's Toolkit: An Inquiry-Based Activity to Show the Importance of Sample Size and Random Sampling

Invited Commentary: E Pluribus Unum for Epidemiology.

In this issue of the Journal, Amirian et al. (Am J Epidemiol. 2016;183(2):85-91) present a report from the Genetic Epidemiology of Glioma International Consortium (GLIOGENE Consortium), a new international consortium of glioma case-control studies. This report is noteworthy, because the GLIOGENE Consortium represents a new generation of epidemiologic consortia. GLIOGENE investigators have created an infrastructure that addresses important limitations of first-generation consortia efforts, which comprised a posteriori harmonization of exposure data and the inclusion of studies that did not include the same--or any--exposure data. As with these first-generation consortia efforts, the GLIOGENE Consortium embraces the primary importance of sample size, and to achieve that, the consortium tolerates different study designs that permit heterogeneity in case and control ascertainment. In contrast, however, the consortium's Glioma International Case-Control (GICC) Study incorporates systematic collection of exposure data from both cases and controls to facilitate downstream evaluation of exposure associations and gene-environment interactions. The described GICC Study thus serves as a model for future epidemiologic efforts that reflects a paradigm shift whereby studies are now being conducted with the expectation of downstream collaboration, thus demanding coordination and harmonization of apparently independent efforts at the time of study initiation rather than at study completion.

Importance of sample size, data type and prediction method for remote sensing-based estimations of aboveground forest biomass

Estimates of forest biomass are needed for various technical and scientific applications, ranging from carbon and bioenergy policies to sustainable forest management. As local measurements are costly, there is a great interest in obtaining reliable estimates over large areas from remote sensing data. Currently, such estimates are obtained with a variety of data sources, statistical methods and prediction standards, and there is no agreement on what are best practices for this task.To improve our understanding of how these different methods affect prediction quality, we first conducted a systematic review of the available literature to identify the most common sensor types and prediction methods. Based on the review, we identified sample size of the reference points on the ground, prediction method (stepwise linear regression, support vector machines, random forest, Gaussian processes and k-nearest neighbor), and sensor type as the main differences that could potentially affect predictive quality. We then compared those factors in two case study areas in Germany and Chile, for which airborne discrete return Light Detection And Ranging (LiDAR) and airborne hyperspectral as well as airborne discrete return LiDAR and spaceborne hyperspectral data were available. For each factor combination, we calculated Pearson's coefficient of correlation between observations and predictions (r2) and root mean squared error (RMSE) for bootstrapped estimates using k-fold cross-validation with a varying number of folds. Finally, Analysis of Variance (ANOVA) was used to quantify the influence of the factors on the predictive error of the biomass models.Our results confirm previous findings that predictor data (sensor) type is the most important factor for the accuracy of biomass estimates, with LiDAR being preferable to hyperspectral data. In contrast to some previous studies, complementing LiDAR with hyperspectral data did not improve predictive accuracy. Also the prediction method had a substantial effect on accuracy and was generally more important than the sample size. In most cases, random forest performed best and stepwise linear models worst, judging from r2 and RMSE under cross-validation. Additional results suggested that r2 may deliver unrealistically large values when the hold-out sample during the cross-validation is too small.In conclusion, our literature review revealed that different methods for biomass estimation are currently used, with no general agreement on best practices. In our case studies, we found substantial accuracy differences between those methods, with LiDAR data, in combination with a random forest algorithm and a large number of reference sample units on the ground yielding the lowest error for biomass predictions. The comparatively high importance of the statistical prediction method seems particularly relevant, as they suggest that choosing the appropriate statistical method may be more effective than obtaining additional field data for obtaining good biomass estimates. Considering the costs of improving accuracy of global and regional biomass estimates by ground measurements, it seems sensible to invest in further comparative studies, preferably with a wider range of sites and including also RADAR sensors, to establish robust best-practice recommendations for obtaining regional and global biomass estimates from remote-sensing data.

Summed radiocarbon calibrations as a population proxy: a critical evaluation using a realistic simulation approach

The logic of using summed radiocarbon (14C) calibrations (cumulative probability density functions for large numbers of calibrated 14C dates) as proxies for past populations rests on the presumption of a proportional relationship between population size and the production, and subsequent preservation, recovery, and analysis, of 14C-datable material. Critiques of this approach have generally focused on the various problems that may undermine the validity of this assumption.Here, instead, we presume a perfect correspondence between population size and the quantity of datable material produced at any given time, and explore the question of how well summed 14C calibrations can track demographic changes under such ideal circumstances. We introduce a method of generating a random sample of simulated 14C determinations, from a specified distribution, with variable data densities and measurement errors. In other words, we generate a random sample of 14C dates not from an ideal statistical distribution but rather using a defined population curve to determine the probability distribution from which the calendar dates of the simulated 14C samples are drawn. We generate simulated 14C ages for these samples, calibrate them, and sum those calibrations. We compare the resulting proxy population curve to the known population distribution from which it was generated, to see whether known population fluctuations are unambiguously visible on a proxy curve derived from 14C data sets that are realistic in terms of the number and precision of the 14C determinations included.Results highlight 1) the critical role played by the magnitude and duration of any population fluctuation, and 2) the importance of sample size, and the reality that the numbers of samples required to detect significant population changes are generally far higher than those available to researchers proposing demographic reconstructions on the basis of literature searches for radiocarbon dates. We conclude that even if archaeological 14C data sets could be corrected for taphonomic filters and research biases, demographic signals would be difficult to distinguish from statistical noise in summed probability distributions. We suggest that simulation studies should be integral components of any attempt to reconstruct prehistoric demography from 14C dates.

Design and execution of clinical trials in orthopaedic surgery.

High-quality randomised controlled trials (RCTs) evaluating surgical therapies are fundamental to the delivery of evidence-based orthopaedics. Orthopaedic clinical trials have unique challenges; however, when these challenges are overcome, evidence from trials can be definitive in its impact on surgical practice. In this review, we highlight several issues that pose potential challenges to orthopaedic investigators aiming to perform surgical randomised controlled trials. We begin with a discussion on trial design issues, including the ethics of sham surgery, the importance of sample size, the need for patient-important outcomes, and overcoming expertise bias. We then explore features surrounding the execution of surgical randomised trials, including ethics review boards, the importance of organisational frameworks, and obtaining adequate funding.Cite this article: Bone Joint Res 2014;3:161–8.

Importance of sample size in clinical trials

Adequate sample size is an important issue in clinical trials. This article aimed to assess the effect of various factors on sample size and the importance of adequate sample size in clinical trials. Recent data pertinent to study objective was searched and collected from Pub-med and other sources were analysed. It was found that factors determining adequate sample size have paramount importance in assessing the accurate results, while less or more than the required sample size has many disadvantages. A researcher needs to focus on these issues while determining sample size in a clinical trial. Emphasizing and appropriate handling of all the concerned parameters related to sample size before initiating a clinical trial will improve the validity of the study.

Review of Sample Size for Structural Equation Models in Second Language Testing and Learning Research: A Monte Carlo Approach

The importance of sample size, although widely discussed in the literature on structural equation modeling (SEM), has not been widely recognized among applied SEM researchers. To narrow this gap, we focus on second language testing and learning studies and examine the following: (a) Is the sample size sufficient in terms of precision and power of parameters in a model using Monte Carlo analysis? (b) How are the results from Monte Carlo sample size analysis comparable with those from the N ≥ 100 rule and from the N: q ≥ 10 (sample size–free parameter ratio) rule? Regarding (a), parameter bias, standard error bias, coverage, and power were overall satisfactory, suggesting that sample size for SEM models in second language testing and learning studies is generally appropriate. Regarding (b), both rules were often inconsistent with the Monte Carlo analysis, suggesting that they do not serve as guidelines for sample size. We encourage applied SEM researchers to perform Monte Carlo analyses to estimate the requisite sample size of a model.

Improving the accuracy of livestock distribution estimates through spatial interpolation

Animal distribution maps serve many purposes such as estimating transmission risk of zoonotic pathogens to both animals and humans. The reliability and usability of such maps is highly dependent on the quality of the input data. However, decisions on how to perform livestock surveys are often based on previous work without considering possible consequences. A better understanding of the impact of using different sample designs and processing steps on the accuracy of livestock distribution estimates was acquired through iterative experiments using detailed survey. The importance of sample size, sample design and aggregation is demonstrated and spatial interpolation is presented as a potential way to improve cattle number estimates. As expected, results show that an increasing sample size increased the precision of cattle number estimates but these improvements were mainly seen when the initial sample size was relatively low (e.g. a median relative error decrease of 0.04% per sampled parish for sample sizes below 500 parishes). For higher sample sizes, the added value of further increasing the number of samples declined rapidly (e.g. a median relative error decrease of 0.01% per sampled parish for sample sizes above 500 parishes. When a two-stage stratified sample design was applied to yield more evenly distributed samples, accuracy levels were higher for low sample densities and stabilised at lower sample sizes compared to one-stage stratified sampling. Aggregating the resulting cattle number estimates yielded significantly more accurate results because of averaging under- and over-estimates (e.g. when aggregating cattle number estimates from subcounty to district level, P <0.009 based on a sample of 2,077 parishes using one-stage stratified samples). During aggregation, area-weighted mean values were assigned to higher administrative unit levels. However, when this step is preceded by a spatial interpolation to fill in missing values in non-sampled areas, accuracy is improved remarkably. This counts especially for low sample sizes and spatially even distributed samples (e.g. P <0.001 for a sample of 170 parishes using one-stage stratified sampling and aggregation on district level). Whether the same observations apply on a lower spatial scale should be further investigated.

Increasing taxonomic resolution in pollen identification: Sample size, spatial sampling bias and implications for palaeoecology

Many studies proposing methods for differentiating fossil pollen of different species derive diagnostic identification criteria from small sample sizes and/or modern pollen collected from limited areas within species ranges. Here, we evaluate the importance of sample size and population representation in pollen identification methods, in an examination of morphological measurements on over 21,000 pollen from three alder species (Alnus viridis subsp. sinuata, Alnus incana subsp. tenuifolia and Alnus rubra) in western North America. We assess the sensitivity of morphometric analyses and multi-trait classification models of Alnus pollen to changes in sample size and spatial sampling breadth, by randomly reducing our large dataset in size and by splitting the dataset into regional subsets. Sensitivity analyses demonstrate that using small sample sizes to identify diagnostic morphological traits for pollen identification can result in misleading conclusions about interspecific morphological differences in pollen types and ultimately in misidentification of pollen. Pollen identification methods based on reference samples collected from limited regions are prone to bias, due to regional differences in modern pollen morphology. Our study reveals the importance of using large sample sizes from across species ranges when attempting to determine the diagnostic morphological features for fossil pollen identification. We recommend basing identification criteria on pollen from as many individual plants per species as possible, as opposed to numerous pollen grains from only a small number of individuals, because at least in Alnus, individual plants produce morphologically similar pollen.

W09-04 - Industry perspectives on the importance of sample size, statistical power and protocol fidelity

An inadequate sample size, low statistical power and poor reliability in a clinical trial exposes subjects to potentially harmful treatments without advancing knowledge. An unnecessary number of subjects may be exposed to a potentially harmful treatment, or are denied a potentially beneficial one. For such an important issue, there is a surprisingly small amount of published literature. One step in determining statistical power and fidelity requires educing an effect size of scientific interest. From an industry viewpoint, it is not up to a statistician to decide this; however, it is the statistician's responsibility to try to extract this information from the industry researchers involved in planning the study. A major problem being observed is that researchers don’t recognize it as a question that they are responsible for answering. This presentation will outline a model of protocol fidelity in clinical trials. The model addresses the complexity of clinical trial research, by expanding the aspects of the research process that require attention to fidelity from more simple models focusing on delivery and performance that incorporates study design and training. This will allow a more precise understanding of the elements responsible for a treatment's success or failure, improving replicability and translation to clinical or community settings. Learning Objectives From this workshop, clinicians and statisticians in industry and academia will learn: 1. The importance of taking protocol fidelity into account at the earliest stages of the research, beginning with procedures to ensure that the treatment components accurately reflect the theoretical underpinnings of the intervention, and that sites are trained in standardized fashion, to criterion, with emphasis on maintenance rater skills over time. 2. Mapping the treatment to theory, description of the core mechanisms of the treatment, and standardized training, monitoring, and maintenance of rater skills over time are essential to ensuring that the treatment is being operationalized as proposed.