No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Preventing human error: The impact of data entry methods on data accuracy and statistical results
Abstract
Human data entry can result in errors that ruin statistical results and conclusions. A single data entry error can make a moderate correlation turn to zero and a significant t-test non-significant. Therefore, researchers should design and use human computer interactions that minimize data entry errors. In this paper, 195 undergraduates were randomly assigned to three data entry methods: double entry, visual checking, and single entry. After training in their assigned method, participants entered 30 data sheets, each containing six types of data. Visual checking resulted in 2958% more errors than double entry, and was not significantly better than single entry. These data entry errors sometimes had terrible effects on coefficient alphas, correlations, and t-tests. For example, 66% of the visual checking participants produced incorrect values for coefficient alpha, which was sometimes wrong by more than .40. Moreover, these data entry errors would be hard to detect: Only 0.06% of the errors were blank or outside of the allowable range for the variables. Thus, researchers cannot rely upon histograms and frequency tables to detect data entry errors. Single entry and visual checking should be replaced with more effective data entry methods, such as double entry.
... High inter-rater reliability was established through various methods, including twoway mixed absolute agreement, intra-class correlations, and single measures. Any discrepancies were resolved following the procedures outlined by Barchard and Pace [29], minimizing measurement error. ...
... Two-way mixed absolute agreement, within-class correlations, and single measures identified high inter-rater reliability. Discrepancies were addressed and resolved as per Barchard and Pace [29] to add only minimal measurement error. ...
The recent surge of generative artificial intelligence (AI) in higher education presents a fascinating landscape of opportunities and challenges. AI has the potential to personalize education and create more engaging learning experiences. However, the effectiveness of AI interventions relies on well-considered implementation strategies. The impact of AI platforms in education is largely determined by the particular learning environment and the distinct needs of each student. Consequently, investigating the attitudes of future educators towards this technology is becoming a critical area of research. This study explores the impact of generative AI platforms on students’ learning performance, experience, and satisfaction within higher education. It specifically focuses on students’ experiences with varying levels of technological proficiency. A comparative study was conducted with two groups from different academic contexts undergoing the same experimental condition to design, develop, and implement instructional design projects using various AI platforms to produce multimedia content tailored to their respective subjects. Undergraduates from two disciplines—Early Childhood Education (n = 32) and Computer Science (n = 34)—participated in this study, which examined the integration of generative AI platforms into educational content implementation. Results indicate that both groups demonstrated similar learning performance in designing, developing, and implementing instructional design projects. Regarding user experience, the general outcomes were similar across both groups; however, Early Childhood Education students rated the usefulness of AI multimedia platforms significantly higher. Conversely, Computer Science students reported a slightly higher comfort level with these tools. In terms of overall satisfaction, Early Childhood Education students expressed greater satisfaction with AI software than their counterparts, acknowledging its importance for their future careers. This study contributes to the understanding of how AI platforms affect students from diverse backgrounds, bridging a gap in the knowledge of user experience and learning outcomes. Furthermore, by exploring best practices for integrating AI into educational contexts, it provides valuable insights for educators and scholars seeking to optimize the potential of AI to enhance educational outcomes.
... Compliance of a center to the required regulations will make the center's data more reliable. Non-compliance events may lead to errors in patient inclusion criteria, operating procedures and to various types of data entry errors [2,3]. Additionally, data tampering or fraud may occur in a single center [4]. ...
... For continuous outcomes, Fig. 2c and d show the simulations of linear model as a comparator to the non-parametric approach for balanced and unbalanced experimental design respectively. As anticipated, the non-parametric method shows increased type I error for small sample sizes [3,5,10]. A linear model is known to control the familywise type I error rate, the purpose of this comparison is to show the ability of the non-parametric method to control the type I error similarly to the linear model, specifically for extreme settings. ...
Monitoring of clinical trials is a fundamental process required by regulatory agencies. It assures the compliance of a center to the required regulations and the trial protocol. Traditionally, monitoring teams relied on extensive on-site visits and source data verification. However, this is costly, and the outcome is limited. Thus, central statistical monitoring (CSM) is an additional approach recently embraced by the International Council for Harmonisation (ICH) to detect problematic or erroneous data by using visualizations and statistical control measures. Existing implementations have been primarily focused on detecting inlier and outlier data. Other approaches include principal component analysis and distribution of the data. Here we focus on the utilization of comparisons of centers to the Grand mean for different model types and assumptions for common data types, such as binomial, ordinal, and continuous response variables. We implement the usage of multiple comparisons of single centers to the Grand mean of all centers. This approach is also available for various non-normal data types that are abundant in clinical trials. Further, using confidence intervals, an assessment of equivalence to the Grand mean can be applied. In a Monte Carlo simulation study, the applied statistical approaches have been investigated for their ability to control type I error and the assessment of their respective power for balanced and unbalanced designs which are common in registry data and clinical trials. Data from the German Multiple Sclerosis Registry (GMSR) including proportions of missing data, adverse events and disease severity scores were used to verify the results on Real-World-Data (RWD).
... This process could paradoxically be risky for disease transmission, as the assessor may have visited other farms and used the same materials on multiple farms without disinfection (Kim et al., 2017;Mee et al., 2012;Ssematimba et al., 2013). In addition, if the data on paper require transcription, this process may contribute to the entry of data with errors (Barchard and Pace, 2011). ...
The aim of this study was to identify which biosecurity assessment methods (BAMs) are currently used in practice in animal farms. To address this, a structured questionnaire was developed to gather information such as the animal species, main objectives, type of enforcement, output generated and feedback of the result. In the context of the BETTER Cost Action project, country representatives identified in each of their countries which BAMs were used and completed an online survey. The survey was prepared and translated in 23 languages. Besides a descriptive analysis, clusters of BAMs were determined using a multiple correspondence analysis. Responses, collected between December 2022 and July 2023, included 74 BAMs used in 28 countries. Most of them were used in a single country while three were used in multiple countries. This study provides a comprehensive picture of existing BAMs and insights into their diversity, such as variations in objectives, implementation, evaluators, respondents, feedback, or assessment outputs. Moreover, we identified four BAMs clusters differentiated by their objective, evaluator and type of feedback provided. This study might also represent the basis for future research on strengths and weaknesses of different BAMs.
... Moreover, the human error rate for manual data entry is approximately 1-5%, which can quickly become a risk for quality assurance, in particular in cases where data has to be copied multiple times between different systems. Even typical quality assurance measures like visual checks do not improve the error rate drastically [16]. ...
As with the previous revolutions, the goal of the fourth revolution is to make manufacturing, design, logistics, maintenance, and other related fields faster, more efficient, and more customer centric. This holds for classical industries, for civil engineering, and for NDE and goes along with new business opportunities and models.
Core components to enable those fourth revolutions are semantic interoperability, converting data into information; the Industrial Internet of Things (IIoT) offering the possibility for every device, asset, or thing to communicate with each other using standard open interfaces; and the digital twin converting all the available information into knowledge and closing the cyber-physical loop.
For NDE this concept can be used (1) to design, improve, and tailor the inspection system hardware and software, (2) to choose and adapt to best inspection solution for the customer, (3) to enhance the inspection performance, and (4) to enable remote NDE interfaces and instrumentation, thus enabling better quality, speed, and cost at the same time.
On a broader view, the integration of NDE into IIoT and digital twin is the chance for the NDE industry for the overdue change from a cost center to a value center. In most cases, data gathered by NDE is used for a quality assurance assessment resulting in a binary decision. But the information content of NDE goes way deeper and is of major interest for additional groups: engineering and management. Some of those groups might currently not be aware of the benefits of NDE data and the NDE industry makes the access unnecessarily difficult by proprietary interfaces and data formats. Both those challenges need to be taken on now by the NDE industry. The big IT players are not waiting and, if not available on the market, they will develop and offer additional data sources including ultrasonics, X-ray, or eddy current. This chapter is based on content from “The Core of the Fourth Revolutions: Industrial Internet of Things, Digital Twin, and Cyber-Physical Loops” (Vrana, J Nondestruct Eval 40:46, 2021).
... Hence, statistical software packages such as GRAPHPAD, MINITAB, SPSS, R, STATA, and MATLAB, among others, have been developed for the 21 st century to overcome the problems associated with manual data analysis(Kimberly & Larry, 2011;Abatan & Olayemi, 2014). ...
Adoption of the Competency-Based Curriculum was a shift from the historical Content-Based Curriculum, which necessitated the need to bring teachers up to speed with the requirements of the Competency-Based Curriculum in terms of teacher cognition and classroom practices in the delivery of the Competency-Based Curriculum that would improve students' academic performance and help them acquire critical competencies embedded in the curriculum.
Despite implementing the Competency-Based Curriculum over five years, high school graduates in Liberia continue to underperform academically and lack relevant competencies. National examination results and anecdotal reports highlight that students' performance remains substandard. The effectiveness and competence of teachers in implementing the Competency-Based curriculum have come under scrutiny and criticism from stakeholders. Predicated upon the alarming undesirable results and the blame game, this study proceeded to dig deep and investigate the problem by examining teachers' cognition and its dimensionality and related factors, teacher classroom practices, and impact on students' academic performance and competency acquisition.
The Theories of Planned Behavior and Situated Evaluation guided the study and applied a mixed-methods multistrand design. Through proportional stratified random sampling, 175 teachers and 455 students were selected from 35 schools for the quantitative data while purposive sampling selected 35 teachers for the qualitative data. The study analysis employed a series of statistical analyses using the Statistical Package of Social Science version 23.0 for the quantitative data and abductive thematic analysis using ATLAS. ti 7.5 to analyze the qualitative data.
The study revealed several important findings about teacher cognition and implementation of the Competency-Based Curriculum. Teachers generally demonstrated below-average understanding of the curriculum, with their cognition consisting of three main components: knowledge, thinking, and belief. Traditional assessment methods were more commonly used than authentic assessment methods. Both teacher-related and
IX
contextual factors significantly influenced teachers' understanding of the curriculum. Interestingly, while quantitative results showed a strong link between teacher cognition and classroom practices, qualitative results indicated a mismatch between teacher cognition of the curriculum and what they implemented in classrooms. Together, teacher cognition and classroom practices did not have a significant impact on student’s academic performance and acquisition of embedded competencies. The findings highlight critical policy and practical implications for the competency-based curriculum's implementation.
The study introduced two innovative frameworks: a curriculum implementation framework and a teacher capacity-building framework to address these gaps. These innovations will enhance teacher cognition about the Competency-Based Curriculum and classroom practices to achieve better learning outcomes. By adopting the proposed frameworks, education systems can improve teacher cognition and ensure teacher classroom practices align with the Competency-Based Curriculum, which would translate into improved learning outcomes in Liberia and other related contexts.
... The results of the RPN calculation are used as a reference in ranking each risk source. Then, the RPN critical value calculation is carried out to determine the priority of handling of each existing risk, where risks with RPN values above the critical value are included in (Barchard et al. 2011). The existence of risks in the data entry process can hinder the conversion of data into useful information (Plutoshift, 2018). ...
Background: The facultative administration of general reinsurance of PT. XYZ has experienced several operational risks, such as scattered and incomplete bargaining documents between companies and emails from insurance companies that were not promptly responded to. However, in practice, the facultative administration of general reinsurance of PT. XYZ has not implemented operational risk management.Purpose: The purpose of this study is to identify operational risks and their causes, analyze the priority level of handling, and recommend handling operational risks. Design/methodology/approach: This research was conducted from November 2023 to January 2024 and required 9 respondents. Furthermore, the information received based on the respondents' statements will be processed using Fishbone Diagram, FMEA (Failure Mode and Effect Analysis), and Pareto Diagram methods.Finding/Results: 11 risk sources were identified and grouped into process, human, system, and external risk categories, of which 3 were prioritized for handling, namely the risk that emails related to offers from insurance companies were not responded on time, an error occurred on the company's website database, and the information in the business offer slip did not explain in detail the coverage value of each risk to be reinsured. Conclusion: Some of the handling efforts proposed in this study include adding competent human resources, carrying out regular system maintenance, and providing reminders to insurance companies.Originality/value (State of the art): The facultative administration of general reinsurance of PT XYZ does not escape the existence of several operational risks that fall into the categories of process, human, system and external risks. Keywords: facultative administration, FMEA, operational risk, Pareto diagram, risk priority
... Uncertainties associated with data integration into numerical models arise from a variety of sources, such as parameter estimation, inaccurate remote sensing, geophysical data, or even incomplete groundwater monitoring time series. These can be due to human error [8] during data collection, translation, storage, or inaccurate recording such as defective measuring devices for long-term measurements. Incomplete data can further occur due to the loss of archives or even lack of financial resources [9,10]. ...
State-of-the-art hydrogeological investigations use transient calibrated numerical flow and transport models for multiple scenario analyses. However, the transient calibration of numerical flow and transport models still requires consistent long-term groundwater time series, which are often not available or contain data gaps, thus reducing the robustness and confidence of the numerical model. This study presents a data-driven approach for the reconstruction and prediction of gaps in a discontinuous groundwater level time series at a monitoring station in the Allertal (Saxony-Anhalt, Germany). Deep Learning and classical machine learning (ML) approaches (artificial neural networks (TensorFlow, PyTorch), the ensemble method (Random Forest), boosting method (eXtreme gradient boosting (XGBoost)), and Multiple Linear Regression) are used. Precipitation and groundwater level time series from two neighboring monitoring stations serve as input data for the prediction and reconstruction. A comparative analysis shows that the input data from one measuring station enable the reconstruction and prediction of the missing groundwater levels with good to satisfactory accuracy. Due to a higher correlation between this station and the station to be predicted, its input data lead to better adapted models than those of the second station. If the time series of the second station are used as model inputs, the results show slightly lower correlations for training, testing and, prediction. All machine learning models show a similar qualitative behavior with lower fluctuations during the hydrological summer months. The successfully reconstructed and predicted time series can be used for transient calibration of numerical flow and transport models in the Allertal (e.g., for the overlying rocks of the Morsleben Nuclear Waste Repository). This could lead to greater acceptance, reliability, and confidence in further numerical studies, potentially addressing the influence of the overburden acting as a barrier to radioactive substances.
... One of the major advantages of online surveys versus either paper surveys or interviews is that data entry is automatic, eliminating a time-consuming process that includes a high risk of error (Barchard & Pace, 2011). Of course, any dataset requires variables to be coded and combined into constructs for analysis. ...
This chapter presents practical examples and makes recommendations about studying emerging adulthood using crowd-sourced research approaches. Examples are drawn from three studies (one ongoing) to assess the Emerging Adulthood Measured at Multiple Institutions project. Spanning 20 years, it is one of the earliest crowd-sourcing research projects in psychology; these studies offer lessons based on findings and the trials and errors in managing them. In addition to reviewing each study and its implications, this chapter discusses the management of crowd-sourcing projects, including administration and project management. Ideally, readers will gain an understanding of the most realistic way to study emerging adulthood with a crowd.
... This approach considers the specific features of the dataset as well as the objectives of the research (De Waal, Goedegebuure, and Geradts 2011). Barchard and Pace (2011) identify data loss as a frequent challenge encountered in repetitive tasks, primarily resulting from manual data entry processes, equipment malfunctions, measurement inaccuracies, intentional omissions, and other causes. When there are only a limited number of observations for a specific variable, the sample size can significantly diminish. ...
A vast amount of data in many different formats is produced and stored daily, offering machine learning a valuable resource to enhance its predictive capabilities. However, the pervasiveness of inaccuracies in real‐world data presents a significant barrier that can seriously limit the effectiveness of learning algorithms. The ensemble models and hyper‐tuned multi‐layer perceptron (MLP) with need‐based hidden neuron layers are effective frameworks for data imputation. Addressing the issue of missing data is a complex and demanding task, and much remains to be explored in developing effective and precise methods for predicting and imputing missing values across different datasets. The study offers important perspectives on using algorithms in machine learning to predict and impute gaps in data in recently updated datasets. The findings indicate that finely tuned MLP classifiers notably improve prediction accuracy and dependability compared to models with a static or reduced number of neurons. Furthermore, the study highlights the promising potential of ensemble models within the error‐correcting output code (ECOC) framework as an effective approach for this task. It also suggests future research directions to refine further and strengthen machine learning‐based imputation methods regarding precision and stability. ECOC framework includes all kinds of MLP classifiers and regressors such as binary classifiers, multi‐class classifiers, or regression models. MLP models predict complex relationships in modern datasets. Hugging Face, COSMIC, SKlearn, and Kaggle have relevant and updated datasets. The weighted average recognition (96%) shows that the proposed MLP‐based stochastic learning strategies achieved better results.
... The average mean of 2.51 with a highly accepted verbal description for the system's security based on ISO/IEC 25010 criteria (see Table 2) validates the system's capability to protect confidential data from unauthorized access (Martin et al., 2017). Barchard et al. (2011) stated that errors are expected when people enter data. They also added that data input mistakes could have disastrous consequences. ...
Accreditation activities are critical for an institution pursuing excellent recognition. However, the segmented approach to records or document management employed for accreditation, which relies on paper-based records, has several flaws. Impediments such as misplacement, crumpling, termites, and other conditions that might result in damaged or lost data are examples of these failure points. Hence, this study determines the key informants' perception of a web-based technology platform built in the Philippines for digitized data for state university and college accreditation. A quantitative and qualitative method was utilized in the study. The ISO/IEC 25010 standard was utilized to determine how the system satisfies the implied criteria. Emergent themes were also used to develop a thematic map to validate results. The research findings emphasized the wide variety of advantages that digitizing data via a web-based technology platform may provide. The system highlights the maximum data security the system provides, where data encryption and password protection only allow authorized users to access the system. It also permits remote access backup of stored data, ensuring that the system will have a backup copy in a disaster. Furthermore, under widely accepted conventions and standards, the system ensures no violation of data privacy and plagiarism; whose policies, methods, and performance are transparent would significantly validate the significant impact of technology acceptance and adoption. According to the authors' knowledge, this is the first web-based technology platform for digitized data for state university and college accreditation established in the Philippines.
... Moreover, the human error rate for manual data entry is approximately 1-5%, which can quickly become a risk for quality assurance, in particular in cases where data has to be copied multiple times between different systems. Even typical quality assurance measures like visual checks do not improve the error rate drastically [16]. ...
As with the previous revolutions, the goal of the fourth revolution is to make manufacturing, design, logistics, maintenance, and other related fields faster, more efficient, and more customer centric. This holds for classical industries, for civil engineering, and for NDE and goes along with new business opportunities and models.
Core components to enable those fourth revolutions are semantic interoperability, converting data into information; the Industrial Internet of Things (IIoT) offering the possibility for every device, asset, or thing to communicate with each other using standard open interfaces; and the digital twin converting all the available information into knowledge and closing the cyber-physical loop.
For NDE this concept can be used (1) to design, improve, and tailor the inspection system hardware and software, (2) to choose and adapt to best inspection solution for the customer, (3) to enhance the inspection performance, and (4) to enable remote NDE interfaces and instrumentation, thus enabling better quality, speed, and cost at the same time.
On a broader view, the integration of NDE into IIoT and digital twin is the chance for the NDE industry for the overdue change from a cost center to a value center. In most cases, data gathered by NDE is used for a quality assurance assessment resulting in a binary decision. But the information content of NDE goes way deeper and is of major interest for additional groups: engineering and management. Some of those groups might currently not be aware of the benefits of NDE data and the NDE industry makes the access unnecessarily difficult by proprietary interfaces and data formats. Both those challenges need to be taken on now by the NDE industry. The big IT players are not waiting and, if not available on the market, they will develop and offer additional data sources including ultrasonics, X-ray, or eddy current. This chapter is based on content from “The Core of the Fourth Revolutions: Industrial Internet of Things, Digital Twin, and Cyber-Physical Loops” (Vrana, J Nondestruct Eval 40:46, 2021).
... Currently, the AIPs and CCET are produced through template spreadsheets and manual data input. However, human data entries are prone to errors and may compromise data quality (Barchard & Pace, 2011). Some LGUs may have adopted information systems to streamline their processes, but the quality and level of integration and interoperability of these systems across and within LGUs still vary. ...
Local government units (LGUs) in the Philippines are situated
at the forefront of climate change action. As they translate local plans
into investment programs, LGUs are required to implement climate
change expenditure tagging (CCET) to track budgeted items related to
climate adaptation and mitigation. However, numerous LGUs in the
Philippines have faced challenges in achieving sufficient compliance with
CCET. This study, therefore, aims to contribute to this area by assessing
the state of local CCET and subsequently identifying institutional and
policy recommendations to improve its implementation. Using the
evaluation criteria developed by the Organization for Economic Cooperation and Development (OECD), this study employs a qualitative
descriptive design involving document analysis, literature review and key
informant interviews (KII). In general, the findings highlight the need for
supporting legislation to ensure sustainability and propose an expansion
of the tagging mechanics. This expansion may involve indicating
financing sources, integrating adaptation and mitigation objectives within
programs, projects, and activities (PPA), applying degrees of relevance
through corresponding weights, accounting for negative expenditures, and
tracking PPAs’ alignment with the five comprehensive development plan
(CDP) sectors. To improve implementation effectiveness and efficiency,
integrating CCET across LGUs’ planning, budgeting, and legislative
functions is recommended, alongside institutionalizing administrative
reforms for sufficient institutional capacities for CCET implementation.
... High inter-rater reliability was confirmed through two-way mixed absolute agreement, within-class correlations, and single measures. Any discrepancies were addressed and resolved following the guidelines of Barchard and Pace (2011), resulting in only minimal measurement error. ...
Students’ perspectives on using generative artificial intelligence (AI) chatbots and machine learning are crucial in shaping the design, development, and implementation of their learning projects across various disciplines. Cognitive thinking, a key aspect of AI-related machine learning, aims to replicate human intelligence and behavior. However, the relation between cognitive thinking and knowledge acquisition is often overlooked. This cross-sectional study empirically examines the relationship between academic achievement and students’ attitudes toward machine learning, particularly through the use of generative AI chatbots. It specifically focuses on the role of higher-order thinking skills—such as problem-solving, critical thinking, and creativity—as both mediators and moderators in this relationship. A total of four hundred sixteen undergraduate students (n=416) from diverse academic backgrounds voluntarily took part in a project, in which they designed and developed generative AI chatbots in media and information literacy courses. The findings indicate that creativity mediated the relationship between academic achievements and attitudes toward machine learning, but its moderating impact was not significant. Problem-solving and critical thinking did not show significant mediating effects on attitudes toward machine learning, while they showed significant moderating effects in the connection between academic performance and attitudes toward machine learning. This study contributes by elucidating the interrelationships between students’ higher-order thinking skills, academic performance, and attitudes on the use of AI and machine learning technologies. By highlighting the mediating role of creativity and the moderating effects of problem-solving and critical thinking, this study offers a deeper understanding of how these skills shape students’ perceptions of AI.
... The extended Bland-Altman Limits of agreement plots showed that all raters made errors. Barchard and Pace investigated the impact of human errors in data entry on research [43]. The current study also revealed that interpretable cardiac and pulmonary function information predictors were sources of variability, for example, for variables such as dyspnea, chronic respiratory failure, and congestive heart disease. ...
Background
Multiple preoperative calculators are available online to predict preoperative mortality risk for noncardiac surgical patients. However, it is currently unknown how these risk calculators perform across different raters. The current study investigated the interrater reliability of three preoperative mortality risk calculators in an elective high-risk noncardiac surgical patient population to evaluate if these calculators can be safely used for identification of high-risk noncardiac surgical patients for a preoperative multidisciplinary team discussion.
Methods
Five anesthesiologists assessed the preoperative mortality risk of 34 high-risk patients using the preoperative score to calculate postoperative mortality risks (POSPOM), the American College of Surgeons surgical risk calculator (SRC), and the surgical outcome risk tool (SORT). In total, 170 calculations per calculator were gathered.
Results
Interrater reliability was poor for SORT (ICC (C.I. 95%) = 0.46 (0.30–0.63)) and moderate for SRC (ICC = 0.65 (0.51–0.78)) and POSPOM (ICC = 0.63 (0.49–0.77). The absolute range of calculated mortality risk was 0.2–72% for POSPOM, 0–36% for SRC, and 0.4–17% for SORT. The coefficient of variation increased in higher risk classes for POSPOM and SORT. The extended Bland–Altman limits of agreement suggested that all raters contributed to the variation in calculated risks.
Conclusion
The current results indicate that the preoperative risk calculators POSPOM, SRC, and SORT exhibit poor to moderate interrater reliability. These calculators are not sufficiently accurate for clinical identification and preoperative counseling of high-risk surgical patients. Clinicians should be trained in using mortality risk calculators. Also, clinicians should be cautious when using predicted mortality estimates from these calculators to identify high-risk noncardiac surgical patients for elective surgery.
... In addition, labelling raw data requires a great deal of human effort and often specialty training when dealing with potential intricacies and ramifications. Furthermore, human error is unavoidable and its propagation must always be accounted for, given it may undermine the validity of seemingly successful results [2]. ...
Lack of data on which to perform experimentation is a recurring issue in many areas of research, particularly in machine learning. The inability of most automated data mining techniques to be generalized to all types of data is inherently related with their dependency on those types which deems them ineffective against anything slightly different. Meta-heuristics are algorithms which attempt to optimize some solution independently of the type of data used, whilst classifiers or neural networks focus on feature extrapolation and dimensionality reduction to fit some model onto data arranged in a particular way. These two algorithmic fields encompass a group of characteristics which when combined are seemingly capable of achieving data mining regardless of how it is arranged. To this end, this work proposes a means of combining meta-heuristic methods with conventional classifiers and neural networks in order to perform automated data mining. Experiments on the MNIST dataset for handwritten digit recognition were performed and it was empirically observed that using a ground truth labeled dataset's validation accuracy is inadequate for correcting labels of other previously unseen data instances.
... Furthermore, data for this study was collected manually from a single participant. According to a study examining the error rate of manually entering data [10], participants made 12.03 mistakes out of 1260 entries meaning there's roughly a 1% error rate in manually inputting data. These mistakes wouldn't be large such as inputting 11 instead of 1 due to the noticeable effect on the results, however, it's possible that~1% of entries were placed in the wrong cell or have an incorrect value. ...
... However, the platform's built-in functionalities for workflow management and automation are somewhat restricted, often requiring manual intervention or complex workarounds. This limitation hinders the efficiency and scalability of data processing tasks, especially for repetitive processes or large-scale projects [1]. To address this challenge, the open-source Python library AlteryxConnector emerges as a valuable tool. ...
This paper introduces AlteryxConnector, a Python library designed to enhance workflow managementand automation through the Alteryx API. AlteryxConnector simplifies interactions with Alteryx, enablingadvanced workflow automation, dynamic configuration, and integration with other systems. By providingprogrammatic access to Alteryx workflows, the library significantly enhances the platform's data blending andanalytics capabilities, which are otherwise limited by the standard interface. This study combines detaileddescriptions of AlteryxConnector's functionalities, real-world applications, and potential for futureenhancements to illustrate its impact on data science and engineering practices.
... Double data entry and validation is one way of reducing the errors. [9] Although ESC reduces such problems, the feasibility of advocating EDC should be thought of by the researchers while planning for the research. [10] conclusIon Missing data pose a significant challenge in research, potentially compromising the integrity and validity of study findings. ...
This article addresses the pervasive issue of missing data in biomedical research and the potential solutions to prevent it. The focus is on proactive strategies for prevention at various stages of the research process. Advocating for a comprehensive study design, clear communication through information sheets during consent, and the use of well‑designed data collection tools are key to preventing missing data issues in health research. In addition, the training and supervision for data collectors, especially in the context of potential biases arising from sensitive questions, are critical. Electronic data collection has the advantages of validation and checks to prevent missing data. Hence, adequate planning
is a must to reduce missing values and improve the validity of the study outcome.
... 28,29 Outliers must be prevented by training the data entry staff and verification mechanisms during entry. 28 Furthermore, digit preference can also be reduced by training and quality control programs, automatic numbering systems, 30 and close supervision. ...
Sri Lanka is among the 16 countries with the highest leprosy burden. There have been no reviews of the globally reported data from Sri Lanka and data quality has not been ascertained. Methods We conducted a detailed analysis of cases in the leprosy databases available at the Anti-Leprosy Campaign (ALC). Details of all reported cases of leprosy from 2015 to 2019 were analysed by a panel of experts. In the event of discrepancies or missing data, additional information and new datasheets were obtained from ALC through personal communication. Results The total number of leprosy cases reported over the period (2015-2019) was 9294, and the average annual number of cases was within the range of 1749-2002. Regarding the accuracy and validity of the key indicators, the major indicators were within the range of other parallel data sources, with the exception of 2015. However, when analysing the datasets, a clear observation was that databases were created using different terminology, and there were no standard data entry guidelines available throughout this time period, which may have affected the completeness of the data, thus affecting data quality. Digit preferences, duplicates, and outliers, which were mainly confined to a few years, were observed. 1 2 R. Ferdinando et al. Conclusion and recommendations Although the case registry available at the ALC provides a good database for the comprehensive analysis of cases, it should be strengthened to improve the data quality. Because no guidelines currently exist for data coding and entry, developing guidelines and training data entry operators are recommended.
... Unfortunately, data entry errors can have deleterious effects on research results. Simple data entry errors -such as typing an incorrect number can ruin the results of a statistical analysis (70). Hence the data for these states is mentioned but not considered for the conclusion. ...
Background: Vaccine hesitancy was one of the problems India faced after the COVID-19 vaccine roll-out. Earlier studies carried out to understand the perception of people towards the vaccine were mainly based on online or offline community-based surveys. These studies do not help to understand the actions taken by people towards vaccination. Hence, this study has been designed to understand the exact behaviour of people toward the vaccine. Methods: The study population is divided into three age groups, 18+ years, 12-14 years, and 15-18 years. The data analysis has been done using cumulative coverage of vaccines in the said age groups. Results: The study shows a substantial population has missed second and booster doses of vaccine at the state, regional, and national levels in all three three-age groups. Even for the states that have shown the smallest number of people who missed their dose, their number is in the thousands. Conclusion: Further research is needed to know, in total population, how many people have not even taken a single dose of vaccine. Policy-level efforts are needed to cover the entire population of the country for at least a single dose and vulnerable population, not only for the primary series (1st and 2nd dose) but also for the booster dose of the COVID-19 vaccine.
... During the research, we performed a double entry on our data and visual checking to reduce the risk of human error, such as duplicated publications or incomplete data. Double entry produced far fewer errors than only conducting visual checking or single entry [75]. ...
In recent decades, the focus of health research has shifted to the impact of disease or impairment on how people proceed, behave, and experience quality of life. People’s lives are affected by oral diseases in various ways. Oral health-related quality of life (OHRQoL) is inextricably linked to general health and well-being, and it has far-reaching consequences for clinical practice and dentistry research. Particularly in Indonesia, increasing attention to OHRQoL is related to several concerning oral conditions, such as the extremely high number of cases of tooth decay and inflammation of dental supportive tissue that inexplicably lowers the population’s OHRQoL. To date, there has yet to be a bibliometric study of OHRQoL research in Indonesia. We intend to map the existing scientific literature on OHRQoL research in Indonesia during the last five years and investigate its research gaps. Scopus and the Sinta Database (a national database through Google Scholar) were used to retrieve Indonesian OHRQoL research publications from 2018 to 2023. Bibliographic data were analyzed using SPSS Statistics 25.0 and VOS Viewer 1.6.19. The data demonstrate that the number of OHRQoL-related publications in Indonesia and the number of local writers have increased over time. More of these publications were published in prestigious national journals than foreign ones. The study found that local researchers tended to conduct OHRQoL research on children and older populations, raising the issue of tooth decay or tooth loss. Exploring other subjects, such as dental anxiety, patient satisfaction, chewing performance, aesthetics, and appearance, and other populations (people with oral cancer and other systemic conditions) could broaden the environment of OHRQoL research in Indonesia.
... A two-way mixed absolute agreement, within-class correlations, and single measures in the higher range were used to evaluate inter-rater reliability. If there were any differences between the two assessments, the two raters addressed them, and all data discrepancies were examined, as Barchard and Pace (2011) recommended. Only a small measurement error was added to the coding procedure. ...
Integrating web-based platforms into student learning assessment has gained increasing attention among researchers and instructors. However, limited research addresses the practical challenges of involving undergraduates in the creation of web-based assessment platforms. This study investigates the effects of this process on students' learning outcomes and their psycho-emotional experiences, focusing on motivation and engagement. A comparative study involving 156 students revealed that those using Kahoot! platform achieved significantly better learning outcomes and displayed higher motivation and emotional engagement than those using Google Forms. It also highlights the role of interface design in improving web-based assessment platforms, with implications for user interface design.
... The poor data problem is further augmented at the data cleaning stage, where a large amount of data is eliminated, affecting the power of estimates obtained from prediction models and thereby blurring management decisions. Barchard and Pace (2011) statistically showed the devastating effects of data errors, prevalent in humanbased systems, on estimated results. A shortage of experts has also resulted in reduced inspection coverage leaving significant amounts of infrastructure neglected (Maeda et al., 2018). ...
Pavement management has traditionally relied on human-based decisions. In many countries, however, the pavement stock has recently increased, while the number of management experts has declined, posing the challenge of how to efficiently manage the larger stock with fewer resources. Compared to efficient computer-based techniques, human-based methods are more prone to errors that compromise analysis and decisions. This research built a robust probabilistic pavement management model with a safety metric output using inputs from image processing tested against the judgment of experts. The developed model optimized road pavement safety. The study explored image processing techniques considering the trade-off between processing cost and output accuracy, with annotation precision and intersection over union (IoU) set objectively. The empirical applicability of the model is shown for selected roads in Japan.
... We manually entered the VHI-K30 data obtained from the participants into an MS Excel sheet and cross-checked this data to minimize the data entry errors such as mismatch or out-of-range errors. 48 From the entered raw data, total scores were calculated for the 30 items of VHI-K30 and 10 items corresponding to VHI-K10. The raw data, and the total scores corresponding to VHI-30 and VHI-10, were transferred to Statistical Package for Social Sciences (SPSS) software (version 26) for further statistical analysis. ...
Objectives:
This study aimed to derive the 10-item voice handicap index in the Kannada language (VHI-K10) from the existing VHI-30 in Kannada (VHI-K30). We also aimed to examine several psychometric properties of the newly derived VHI-K10, such as internal consistency, reliability, concurrent validity, discriminant validity, and diagnostic accuracy.
Methods:
Initially, VHI-K10 was derived from the existing VHI-K30 through item reduction, consistent with the recommendations for item reduction of the voice handicap index. This newly derived VHI-K10 was administered to 273 individuals (199 individuals with dysphonia and 74 individuals with normal voice quality). We also obtained phonation and reading samples from the participants. The obtained data were subjected to appropriate statistical analysis to determine several psychometric properties.
Results:
The newly derived VHI-K10 was found to have a strong internal consistency (Cronbach’s α = 0.93). We also found strong test-retest reliability for VHI-K10, with an intraclass correlation coefficient of 0.933. There was a strong statistically significant correlation between the VHI-K10 and the existing VHI-K30 for both individuals with dysphonia (ρ = 0.924, P < 0.001) and individuals with normal voice quality (ρ = 0.798, P < 0.001). However, the correlation of VHI-K10 with the auditory-perceptual measure of GRBAS was fair (ρ = 0.353, P < 0.001) for individuals with dysphonia and was not statistically significant for individuals with normal voice quality. Further, the diagnostic accuracy of VHI-K10 was found to be excellent, with an area under the curve (AROC) value of 0.926 with a cut-off point of ≥6.5, which was slightly superior to that of VHI-K30 (AROC = 0.909, cut-off point ≥21.5).
Conclusions:
The shortened 10-item version of the voice handicap index in Kannada is consistent with versions of the VHI-10 in other languages. This version of the VHI-10 in Kannada is found to be a robust tool with strong psychometric properties.
... These lists are only partially manageable and usually stored separately for each stakeholder and not linked to each other leading to "decision making based on acts-of-faith" (Jupp and Awad, 2017). Apart from the costs incurred, this leads to unquantifiable transmission and transfer errors due to human misinterpretation and misinteraction (compare the extensive work of Sträter, 1997 and the field studies of Barchard and Pace, 2011). ...
The research gap consists of unspecified data and requirements at the beginning of a construction project that a client has for the facility management (FM) phase of an asset to operate it efficiently and effectively. This with the aim to enable further applications based on Building Information Modelling (BIM).
Building owners are dependent on accurate data for the FM. Yet, the data currently available due to digital methods are mostly not structured or do not focus on the FM but on the shorter construction phase. It is therefore only possible to provide additional services and applications to the user, FM and the public with an increased expenditure of resources.
The Design Science Research approach is applied. The result is a generic, model-based, reusable and extensible conceptual framework to incorporate FM data based within the three-dimensional model-based design and construction of an asset to enable smart applications, which are introduced. The approach is exemplified by a use case of the reservation of a meet-ing room. The conceptual framework is composed of empirical data from expert interviews, questionnaires and factual analysis from 13 projects of different sizes. The findings were assessed by an international panel of experts. The conceptual framework shows which phases need which data, who needs them, and which added value can be generated if intelligent data structuring is used at the beginning of the construction project and bridges the gap between requirement and practice.
... Second, as data transfer of HLA typing results is yet not fully automated, clerical errors on both sides (local and central) might have contributed to false entries as well. Overall, manual entry of such sensitive data seems outdated and no longer appropriate as manual data processing is generally associated with a high susceptibility to error (15). Fully digital interfaces could simplify the entry of HLA data in the future and reduce data errors. ...
Introduction
Antibody mediated rejection (ABMR) is the most common cause of long-term allograft loss in kidney transplantation (KT). Therefore, a low human leukocyte antigen (HLA) mismatch (MM) load is favorable for KT outcomes. Hitherto, serological or low-resolution molecular HLA typing have been adapted in parallel. Here, we aimed to identify previously missed HLA mismatches and corresponding antibodies by high resolution HLA genotyping in a living-donor KT cohort.
Methods
103 donor/recipient pairs transplanted at the University of Leipzig Medical Center between 1998 and 2018 were re-typed using next generation sequencing (NGS) of the HLA loci -A, -B, -C, -DRB1, -DRB345, -DQA1, -DQB1, -DPA1, and -DPB1. Based on these data, we compiled HLA MM counts for each pair and comparatively evaluated genomic HLA-typing with pre-transplant obtained serological/low-resolution HLA (=one-field) typing results. NGS HLA typing (=two-field) data was further used for reclassification of de novo HLA antibodies as “donor-specific”.
Results
By two-field HLA re-typing, we were able to identify additional MM in 64.1% (n=66) of cases for HLA loci -A, -B, -C, -DRB1 and -DQB1 that were not observed by one-field HLA typing. In patients with biopsy proven ABMR, two-field calculated MM count was significantly higher than by one-field HLA typing. For additional typed HLA loci -DRB345, -DQA1, -DPA1, and -DPB1 we observed 2, 26, 3, and 23 MM, respectively. In total, 37.3% (69/185) of de novo donor specific antibodies (DSA) formation was directed against these loci (DRB345 ➔ n=33, DQA1 ➔ n=33, DPA1 ➔ n=1, DPB1 ➔ n=10).
Conclusion
Our results indicate that two-field HLA typing is feasible and provides significantly more sensitive HLA MM recognition in living-donor KT. Furthermore, accurate HLA typing plays an important role in graft management as it can improve discrimination between donor and non-donor HLA directed cellular and humoral alloreactivity in the long range. The inclusion of additional HLA loci against which antibodies can be readily detected, HLA-DRB345, -DQA1, -DQB1, -DPA1, and -DPB1, will allow a more precise virtual crossmatch and better prediction of potential DSA. Furthermore, in living KT, two-field HLA typing could contribute to the selection of the immunologically most suitable donors.
Process mining facilitates the discovery, conformance, and enhancement of business processes using event logs. However, incomplete event logs and the complexities of concurrent activities present significant challenges in achieving accurate process models that fulfill the completeness condition required in process mining. This paper introduces a Timed Genetic-Inductive Process Mining (TGIPM) algorithm, a novel approach that integrates the strengths of Timed Genetic Process Mining (TGPM) and Inductive Mining (IM). TGPM extends traditional Genetic Process Mining (GPM) by incorporating time-based analysis, while the IM is widely recognized for producing sound and precise process models. For the first time, these two algorithms are combined into a unified framework to address both missing activity recovery and structural correctness in process discovery. This study evaluates two scenarios: a sequential approach, in which TGPM and IM are executed independently and sequentially, and the TGIPM approach, where both algorithms are integrated into a unified framework. Experimental results using real-world event logs from a health service in Indonesia demonstrate that TGIPM achieves higher fitness, precision, and generalization compared to the sequential approach, while slightly compromising simplicity. Moreover, the TGIPM algorithm exhibits lower computational cost and more effectively captures parallelism, making it particularly suitable for large and incomplete datasets. This research underscores the potential of TGIPM to enhance process mining outcomes, offering a robust framework for accurate and efficient process discovery while driving process innovation across industries.
Outcome registries in vascular surgery are increasingly used to drive quality improvement by vascular societies. The VASCUNET collaboration, within the European Society for Vascular Surgery (ESVS), and the International Consortium of Vascular Registries (ICVR) developed a set of variables for quality improvement registries on abdominal aortic aneurysm (AAA) repair as a registry standard.
Digitalized financial processes are redefining 21st-century business landscapes. One area that has undergone significant transformation is financial reporting, which involves the preparation and presentation of essential reports that enhance strategic decision making and regulatory compliance. Automation and data processing play a crucial role in this context by enhancing the efficiency, accuracy, and transparency of the reports generated. This chapter explores in detail how advances in these two areas of digitalization are revolutionizing financial reporting and then analyzes the relevant technologies and their benefits and challenges, as well as best practices for their implementation. Regulatory guidelines are also examined in terms of their implications and future trends in digital financial reporting.
The One-of-a-kind production is characterized by manual and specialized manufacturing processes. Human adaptability is essential here. As a result, production data acquisition is mainly manual. In many cases, the recorded data are inaccurate or even subject to errors. Consequently, the opportunities for process optimization are impeded. This article therefore analyzes data quality problems and their causes. Subsequently, methods for the evaluation of process data such as process mining and performance indicators are considered and applied to production data sets. The comparison enables the derivation of organizational measures. The result is a practice-oriented method for continuously checking the data quality of manual production data acquisition.
Data shift is a phenomenon present in many real-world applications, and while there are multiple methods attempting to detect shifts, the task of localizing and correcting the features originating such shifts has not been studied in depth. Feature shifts can occur in many datasets, including in multi-sensor data, where some sensors are malfunctioning, or in tabular and structured data, including biomedical, financial , and survey data, where faulty standardization and data processing pipelines can lead to erroneous features. In this work, we explore using the principles of adversarial learning, where the information from several discriminators trained to distinguish between two distributions is used to both detect the corrupted features and fix them in order to remove the distribution shift between datasets. We show that mainstream supervised classifiers, such as random forest or gradient boosting trees, combined with simple iterative heuristics, can localize and correct feature shifts, outperforming current statistical and neural network-based techniques. The code is available at https://github.com/AI-sandbox/DataFix.
In medical research, the accuracy of data from electronic medical records (EMRs) is critical, particularly when analyzing dense functional data, where anomalies can severely compromise research integrity. Anomalies in EMRs often arise from human errors in data measurement and entry, and increase in frequency with the volume of data. Despite the established methods in computer science, anomaly detection in medical applications remains underdeveloped. We address this deficiency by introducing a novel tool for identifying and correcting anomalies specifically in dense functional EMR data. Our approach utilizes studentized residuals from a mean‐shift model, and therefore assumes that the data adheres to a smooth functional trajectory. Additionally, our method is tailored to be conservative, focusing on anomalies that signify actual errors in the data collection process while controlling for false discovery rates and type II errors. To support widespread implementation, we provide a comprehensive R package, ensuring that our methods can be applied in diverse settings. Our methodology's efficacy has been validated through rigorous simulation studies and real‐world applications, confirming its ability to accurately identify and correct errors, thus enhancing the reliability and quality of medical data analysis.
Students' perspectives on using generative artificial intelligence (AI) chatbots and machine learning are crucial in shaping the design, development, and implementation of their learning projects across various disciplines. Cognitive thinking, a key aspect of AI-related machine learning, aims to replicate human intelligence and behavior. However, the relation between cognitive thinking and knowledge acquisition is not always clear. Therefore, it is essential for students to engage in higher-order thinking, which allows them to critically analyze diverse viewpoints, assess their relevance, and understand the complex relationship between cognitive thinking and knowledge acquisition. This empirical study investigates the role of higher-order thinking skills, such as problem-solving, critical thinking, and creativity, in the relationship between academic achievements and attitudes toward machine learning technologies using generative AI chatbots. Four hundred sixteen undergraduate students ( n = 416) from diverse academic backgrounds voluntarily took part in a project, in which they designed and developed generative AI chatbots in media and information literacy courses. The findings indicate that creativity mediated the relationship between academic achievements and attitudes toward machine learning, but its moderating impact was not significant. Problem-solving and critical thinking did not show significant mediating effects on attitudes toward machine learning, while they showed significant moderating effects in the connection between academic performance and attitudes toward machine learning. This study contributes by elucidating the interrelationships between students’ higher-order thinking skills, academic performance, and attitudes on the use of AI and machine learning technologies. By highlighting the mediating role of creativity and the moderating effects of problem-solving and critical thinking, this study offers a deeper understanding of how these skills shape students' perceptions of AI. The findings have significant implications for educational practices, suggesting that fostering higher-order thinking skills is crucial in preparing students to embrace AI and machine learning technologies.
Importance
Pediatric obesity is associated with impaired cognitive function; however, the mechanisms underlying this association demand assessment. Sleep may be a relevant moderator, as poor sleep predicts both increased adiposity and impaired cognitive function.
Objective
To determine the effects of adiposity and sleep on adolescent cognitive function.
Design, Setting, and Participants
This single-blind randomized crossover trial was conducted from September 2020 to October 2022. Parents or caregivers provided demographic information for adolescent participants. Body mass index percentile and bioelectrical impedance analysis assessed adiposity. Adolescents completed 2 actigraphy-confirmed sleep conditions, adequate and restricted, followed by in-person cognitive assessment. No additional follow-up was provided. Data collection for this population-based study took place in a behavioral medicine clinic in Birmingham, Alabama. A total of 323 participants were assessed for eligibility (ages 14-19 years and healthy). Of the 244 eligible adolescents, 157 declined participation. Eighty-seven were randomized and 26 dropped out postenrollment. The final sample included 61 adolescents, 31 with healthy weight and 30 with overweight or obesity. Data were analyzed from April to October 2023.
Interventions
Following a 2-day washout period of adequate sleep, adolescents completed 2 sleep conditions: adequate (mean [SD] duration, 8 hours, 54 minutes [58.0 minutes]) and restricted (mean [SD] duration, 4 hours, 12 minutes [50.7 minutes]).
Main Outcomes and Measures
The National Institutes of Health Cognitive Toolbox assessed global and fluid cognition, cognitive flexibility, working and episodic memory, attention, and processing speed. The Stroop Task assessed inhibition.
Results
The final sample included 61 adolescents (mean [SD] age, 16.3 [1.6] years; 35 [57.4%] female). Restricted sleep predicted poorer global cognition scores (restricted mean [SD], 98.0 [2.8]; adequate mean [SD], 103.2 [2.9]), fluid cognition scores (restricted mean [SD], 94.5 [3.2]; adequate mean [SD], 102.0 [3.6]), and cognitive flexibility scores (restricted mean [SD], 84.8 [3.0]; adequate mean [SD], 92.8 [3.0]) for adolescents with overweight or obesity. No differences emerged for adolescents with healthy weight. Adolescents with overweight or obesity also had poorer attention scores (mean [SD], 80.0 [2.3]) compared to adolescents with healthy weight (mean [SD], 88.4 [SD, 2.3]) following restricted sleep. No differences emerged following adequate sleep. Findings were similar for total body fat percentage (TBF%); however, for adolescents with TBF% above 42, restricted sleep also predicted poorer processing speed, and the association between sleep and attention did not vary based on TBF%.
Conclusions and Relevance
Adolescents with overweight or obesity may be more vulnerable to negative cognitive effects following sleep restriction. Improved sleep hygiene and duration in this group may positively impact their cognitive health.
Trial Registration
ClinicalTrials.gov Identifier: NCT04346433
Python for Information Professionals: How to Design Practical Applications to Capitalize on the Data Explosion is an introduction to the Python programming language for library and information professionals with little or no prior experience. As opposed to the many Python books available today that focus on the language only from a general sense, this book is designed specifically for information professionals who are seeking to advance their career prospects or challenge themselves in new ways by acquiring skills within the rapidly expanding field of data science.
Readers of Python for Information Professionals will learn to:
Develop Python applications for the retrieval, cleaning, and analysis of large datasets.
Design applications to support traditional library functions and create new opportunities to maximize library value.
Consider data security and privacy relevant to data analysis when using the Python language.
https://rowman.com/ISBN/9781538178249/Python-for-Information-Professionals-How-to-Design-Practical-Applications-to-Capitalize-on-the-Data-Explosion
The automatic processing of handwritten forms remains a challenging task, wherein detection and subsequent classification of handwritten characters are essential steps. We describe a novel approach, in which both steps - detection and classification - are executed in one task through a deep neural network. Therefore, training data is not annotated by hand, but manufactured artificially from the underlying forms and yet existing datasets. It can be demonstrated that this single-task approach is superior in comparison to the state-of-the-art two task approach. The current study focuses on hand-written Latin letters and employs the EMNIST data set. However, limitations were identified with this data set, necessitating further customization. Finally, an overall recognition rate of 88.28% was attained on real data obtained from a written exam.KeywordsIntelligent Character RecognitionHandwritten Character RecognitionAutomated Form Processing
Demolition projects today face challenges in adopting
smart digital tools such as Building Information Modeling
(BIM), Big Data, and Artificial Intelligence (AI) within
the construction sector. Despite the prevalence of these
technologies, their integration into daily construction
operations remains limited. This paper presents a case
study conducted with a German medium-sized
construction company, highlighting the challenges faced
in data collection and management practices. The
company prided itself on being digital and data-driven,
allowing for a comprehensive data collection process
across its databases and documents. The collected data
was utilized to apply various AI methods in predicting
durations for small earthwork and infrastructure projects.
However, the analysis of the results revealed that the
current data availability and quality were insufficient for
effective AI implementation in construction SMEs.
Consequently, the paper provides implications to enable
SMEs to harness the benefits of AI methods in the future.
Highway-Rail Grade Crossings (HRGCs) present a significant safety risk to motorists, pedestrians, and train passengers as they are intersections where roads and railways intersect. Every year, HRGCs in the US experience a high number of crashes leading to injuries and fatalities. Estimations of crash and severity models for HRGCs provide insights into safety and mitigation of the risk posed by such incidents. The accuracy of these models plays a vital role in predicting future crashes at these crossings, enabling necessary safety measures to be taken proactively.
In the United States, most of these models rely on the Federal Railroad Administration's (FRA) HRGCs inventory database, which serves as the primary source of information for these models. However, errors or incomplete information in this database can significantly impact the accuracy of the estimated crash model parameters and subsequent crash predictions.
This study examined the potential differences in expected number of crashes and severity obtained from the Federal Railroad Administration's (FRA) 2020 Accident Prediction and Severity (APS) model when using two different input datasets for 560 HRGCs in Nebraska. The first dataset was the unaltered, original FRA HRGCs inventory dataset, while the second was a field-validated inventory dataset, specifically for those 560 HRGCs. The results showed statistically significant differences in the expected number of crashes and severity predictions using the two different input datasets. Furthermore, to understand how data inaccuracy impacts model estimation for crash frequency and severity prediction, two new zero-inflated negative binomial models for crash prediction and two ordered probit models for crash severity, were estimated based on the two datasets. The analysis revealed significant differences in estimated parameters’ coefficients values of the base and comparison models, and different crash-risk rankings were obtained based on the two datasets.
The results emphasize the importance of obtaining accurate and complete inventory data when developing HRGCs crash and severity models to improve their precision and enhance their ability to predict and prevent crashes.
A common occurrence in structural equation models are “improper solutions.” Estimates of negative variances of measurement errors, negative variances of equation errors, or correlations between latent variables that are greater than one are instances of improper solutions. Recent work has begun to examine the causes and cures for these problems but the role of outliers in generating improper solutions has been overlooked. The purposes of this article are threefold: (1) to explain how outliers can lead to improper solutions, (2) to use a confirmatory factor analysis example to demonstrate this, and (3) to encourage researchers to check for this possibility.
Hundreds of articles in statistical journals have pointed out that standard analysis of variance, Pearson product- moment correlations, and least squares regression can be highly misleading and can have relatively low power even under very small departures from normality. In practical terms, psychology journals are littered with nonsignificant results that would have been significant if a more modern method had been used. Modern robust techniques, developed during the past 30 years, provide very effective methods for dealing with nonnormality, and they compete very well with conventional procedures when standard assumptions are met. In addition, modern methods provide accurate confidence intervals for a much broader range of situations, they provide more effective methods for detecting and studying outliers, and they can be used to get a deeper understanding of how variables are related. This article outlines and illustrates these results.
At a data analysis exposition sponsored by the Section on Statistical Graphics of the ASA in 1988, 15 groups of statisticians analyzed the same data about salaries of major league baseball players. By examining what they did, what worked, and what failed, we can begin to learn about the relative strengths and weaknesses of different approaches to analyzing data. The data are rich in difficulties. They require reexpression, contain errors and outliers, and exhibit nonlinear relationships. They thus pose a realistic challenge to the variety of data analysis techniques used. The analysis groups chose a wide range of model-fitting methods, including regression, principal components, factor analysis, time series, and CART. We thus have an effective framework for comparing these approaches so that we can learn more about them. Our examination shows that approaches commonly identified with Exploratory Data Analysis are substantially more effective at revealing the underlying patterns in the data and at building parsimonious, understandable models that fit the data well. We also find that common data displays, when applied carefully, are often sufficient for even complex analyses such as this.
Editor's Note: At the 1959 meetings of the American Statistical Association held in Washington D. C., Messrs. F. J. Anscombe and C. Daniel presented papers on the detection and rejection of 'outliers', that is, observations thought to be maverick or unusual. These papers and their discussion will appear in the next issue of Technometrics. The following comments of Dr. Kruskal are another indication of the present interest of statisticians in this important problem. The purpose of these remarks is to set down some non-technical thoughts on apparently wild or outlying observations. These thoughts are by no means novel, but do not seem to have been gathered in one convenient place. 1. Whatever use is or is not made of apparently wild observations in a statistical analysis, it is very important to say something about such observations in any but the most summary report. At least a statement of how many observa
It is noted that the usual estimators that are optimal under a Gaussian assumption are very vulnerable to the effects of outliers. A survey of robust alternatives to the mean, standard deviation, product moment correlation, t-test, and analysis of variance is offered. Robust methods of factor analysis, principal components analysis and multivariate analysis of variance are also surveyed, as are schemes for outlier detection.
The impact of outliers on Cronbach's coefficient α has not been documented in the psychometric or statistical literature. This is an important gap because coefficient α is the most widely used measurement statistic in all of the social, educational, and health sciences. The impact of outliers on coefficient α is investigated for varying values of population reliability and sample sizes for visual analogue scales. Results show that coefficient α is not affected by symmetric outlier contamination, whereas asymmetric outliers artificially inflate the estimates of coefficient α. Coefficient α estimates are upwardly biased and more variable sample to sample, with increasing asymmetry and proportion of outlier contamination in the population. However, these effects of outliers on the bias and sample variability of coefficient α estimates are reduced for increasing population reliability. The results are discussed in the context of providing guidance for computing or interpreting coefficient α for visual analogue scales.
Hundreds of articles in statistical journals have pointed out that standard analysis of variance, Pearson product-moment correlations, and least squares regression can be highly misleading and can have relatively low power even under very small departures from normality In practical terms, psychology journals are littered with nonsignificant results that would have been significant if a more modern method had been used. Modern robust techniques, developed during the past 30 years, provide very effective methods for dealing with nonnormality, and they compete very well with conventional procedures when standard assumptions are met. In addition, modem methods provide accurate confidence intervals for a much broader range of situations, they provide more effective methods for detecting and studying outliers, and they can be used to get a deeper understanding of how variables are related. This article outlines and illustrates these results.
In a recent Monte Carlo simulation study, Liu and Zumbo showed that outliers can severely inflate the estimates of Cronbach’s coefficient alpha for continuous item response data—visual analogue response format. Little, however, is known about the effect of outliers for ordinal item response data—also commonly referred to as Likert, Likert-type, ordered categorical, or ordinal/rating scale item responses. Building on the work of Liu and Zumbo, the authors investigated the effects of outlier contamination for binary and ordinal response scales. Their results showed that coefficient alpha estimates were severely inflated with the presence of outliers, and like the earlier findings, the effects of outliers were reduced with increasing theoretical reliability. The efficiency of coefficient alpha estimates (i.e., sample-to-sample variation) was inflated as well and affected by the number of scale points. It is worth noting that when there were no outliers, the alpha estimates were downward biased because of the ordinal scaling. However, the alpha estimates were, in general, inflated in the presence of outliers leading to positive bias.
Parallel to the development in regression diagnosis, this paper defines good and bad leverage observations in factor analysis. Outliers are observations that deviate from the factor model, not from the center of the data cloud. The effects of each kind of outlying observations on the normal distribution-based maximum likelihood estimator and the associated likelihood ratio statistic are studied through analysis. The distinction between outliers and leverage observations also clarifies the roles of three robust procedures based on different Mahalanobis distances. All the robust procedures are designed to minimize the effect of certain outlying observations. Only the robust procedure with a residual-based distance properly controls the effect of outliers. Empirical results illustrate the strength or weakness of each procedure and support those obtained in analysis. The relevance of the results to general structural equation models is discussed and formulas are provided.
This paper provides a survey of two classes of methods that can be used in determining and improving the quality of individual files or groups of files. The first are edit/imputation methods for maintaining business rules and for imputing for missing data. The second are methods of data cleaning for finding duplicates within files or across files.
Data entry errors can be disastrous. In many industries, data entry errors are minimised by paying highly skilled data entry personnel or by using expensive optical scanning technologies. However, researchers, small businesses, and non-profit organisations cannot afford highly skilled data entry personnel or the alternative technologies, and instead rely upon minimally trained staff and volunteers for manual data entry. Obtaining accurate data entry is a challenge in these circumstances. Some supervisors do all data entry themselves because they know of no procedures for obtaining accurate data entry from assistants. Others use visual data checking or after-the-fact diagnostic procedures to try to identify unlikely values. None of these approaches are guaranteed to prevent or correct all errors. This paper describes a data entry system that is essentially free, that will take no longer to use than visual checking, and that will virtually eliminate data entry errors.
In this article we extend and implement the forward search algorithm for identifying atypical subjects/observations in factor analysis models. The forward search has been mainly developed for detecting aberrant observations in regression models (Atkinson, 1994) and in multivariate methods such as cluster and discriminant analysis (Atkinson, Riani, & Cerioli, 2004).
Three data sets and a simulation study are used to illustrate the performance of the forward search algorithm in detecting atypical and influential cases in factor analysis models. The first data set has been discussed in the literature for the detection of outliers and influential cases and refers to the grades of students on 5 exams. The second data set is artificially constructed to include a cluster of contaminated observations. The third data set measures car's characteristics and is used to illustrate the performance of the forward search when the wrong model is specified. Finally, a simulation study is conducted to assess various aspects of the forward search algorithm.
WHO and Health Action International (HAI) have developed a standardised method for surveying medicine prices, availability, affordability, and price components in low-income and middle-income countries. Here, we present a secondary analysis of medicine availability in 45 national and subnational surveys done using the WHO/HAI methodology.
Data from 45 WHO/HAI surveys in 36 countries were adjusted for inflation or deflation and purchasing power parity. International reference prices from open international procurements for generic products were used as comparators. Results are presented for 15 medicines included in at least 80% of surveys and four individual medicines.
Average public sector availability of generic medicines ranged from 29.4% to 54.4% across WHO regions. Median government procurement prices for 15 generic medicines were 1.11 times corresponding international reference prices, although purchasing efficiency ranged from 0.09 to 5.37 times international reference prices. Low procurement prices did not always translate into low patient prices. Private sector patients paid 9-25 times international reference prices for lowest-priced generic products and over 20 times international reference prices for originator products across WHO regions. Treatments for acute and chronic illness were largely unaffordable in many countries. In the private sector, wholesale mark-ups ranged from 2% to 380%, whereas retail mark-ups ranged from 10% to 552%. In countries where value added tax was applied to medicines, the amount charged varied from 4% to 15%.
Overall, public and private sector prices for originator and generic medicines were substantially higher than would be expected if purchasing and distribution were efficient and mark-ups were reasonable. Policy options such as promoting generic medicines and alternative financing mechanisms are needed to increase availability, reduce prices, and improve affordability.
Structural equation modelling has been widely applied in behavioural, educational, medical, social, and psychological research. The classical maximum likelihood estimate is vulnerable to outliers and non-normal data. In this paper, a robust estimation method for the nonlinear structural equation model is proposed. This method gives more weight to data that are likely to occur based on the structure of the posited model, and effectively downweights the influence of outliers. An algorithm is proposed to obtain the robust estimator. Asymptotic properties of the proposed method are investigated, which include the asymptotic distribution of the estimator, and some statistics for hypothesis testing. Results from a simulation study and a real data example show that our procedure is effective.
Research electronic data capture (REDCap) is a novel workflow methodology and software solution designed for rapid development and deployment of electronic data capture tools to support clinical and translational research. We present: (1) a brief description of the REDCap metadata-driven software toolset; (2) detail concerning the capture and use of study-related metadata from scientific research teams; (3) measures of impact for REDCap; (4) details concerning a consortium network of domestic and international institutions collaborating on the project; and (5) strengths and limitations of the REDCap system. REDCap is currently supporting 286 translational research projects in a growing collaborative network including 27 active partner institutions.
An experiment was conducted to determine whether, using microcomputer-based data entry, double data entry (DE) significantly lowers data entry keying error rates when compared to single entry (SE). Clinical centers of the Cardiac Arrhythmia Suppression Trial (CAST) participated in a randomized crossover design experiment comparing SE and DE. A total of 42,278 data items (fields) were checked for consistency between the paper data form and the computer database. The overall error rate was 19 per 10,000 fields. Error rates were 22 and 15 per 10,000 fields for SE and DE, respectively; P = .09 by Poisson regression. DE took 37% longer than SE, costing each clinic approximately an extra 90 min per month.
A major responsibility of any Quality Assurance Unit (QUA) is to ensure data integrity. Errors made during data entry can lead to many problems in the study review process and decrease the quality, accuracy, and overall efficiency of data management. One technique that can reduce the number of data entry errors in computer data sets is the use of a dual entry data system. Currently available software allows creation of customized data entry screens that either closely resemble or duplicate the data collection forms used during studies. Two data entry operators enter data into two independent data sets. The use of an on-screen display that resembles the data collection form reduces the potential for keypunch errors. The two data sets can then be electronically compared. The comparison reports differences between the two data sets. When differences exist, the correct values can be determined by reference to the original data sheets and the two data files can then be corrected. Theoretically, the only key punch errors that will exist after making these corrections are when the two independent entry operators make the same exact data entry error. Typically, the time required for two people to enter data is minimal compared to the time required to manually identify and correct data entry discrepancies. With error-free data entry, we have found that electronic data quality, accuracy, and audit efficiency are improved at every subsequent step of data management, analysis, quality assurance auditing, and report generation.
Data entry and encoding errors can jeopardize the integrity of data sets generated in a variety of research settings. Despite researchers' pursuits of more accurate entry methods, data entry errors persist. Although techniques exist for identifying such errors, the PowerChecker program described here provides a more efficient method of data set validation. Rather than enter data twice and then manually search for the correct values when there are discrepancies between the two sets, with PowerChecker the user can correct entry errors as the data are entered the second time. In addition, the time-stamped record of changes to the original data set aids in meeting quality assurance requirements of the Good Clinical Practice standards.
Data entry and its verification are important steps in the process of data management in clinical studies. In Japan, a kind of visual comparison called the reading aloud (RA) method is often used as an alternative to or in addition to the double data entry (DDE) method. In a typical RA method, one operator reads previously keyed data aloud while looking at a printed sheet or computer screen, and another operator compares the voice with the corresponding data recorded on case report forms (CRFs) to confirm whether the data are the same. We compared the efficiency of the RA method with that of the DDE method in the data management system of the Japanese Registry of Renal Transplantation. Efficiency was evaluated in terms of error detection rate and expended time. Five hundred sixty CRFs were randomly allocated to two operators for single data entry. Two types of DDE and RA methods were performed. Single data entry errors were detected in 358 of 104,720 fields (per-field error rate=0.34%). Error detection rates were 88.3% for the DDE method performed by a different operator, 69.0% for the DDE method performed by the same operator, 59.5% for the RA method performed by a different operator, and 39.9% for the RA method performed by the same operator. The differences in these rates were significant (p<0.001) between the two verification methods as well as between the types of operator (same or different). The total expended times were 74.8 hours for the DDE method and 57.9 hours for the RA method. These results suggest that in detecting errors of single data entry, the RA method is inferior to the DDE method, while its time cost is lower.
A framework for comparing normal population means in the presence of heteroscedasticity and outliers is provided. A single number called the weighted effect size summarizes the differences in population means after weighting each according to the difficulty of estimating their respective means, whether the difficulty is due to unknown population variances, unequal sample sizes or the presence of outliers. For an ANOVA weighted for unequal variances, we find interval estimates for the weighted effect size. In addition, the weighted effect size is shown to be a monotone function of a suitably defined weighted coefficient of determination, which means that interval estimates of the former are readily transformed into interval estimates of the latter. Extensive simulations demonstrate the accuracy of the nominal 95% coverage of these intervals for a wide range of parameters.
This article compares several methods for performing robust principal component analysis, two of which have not been considered in previous articles. The criterion here, unlike that of extant articles aimed at comparing methods, is how well a method maximizes a robust version of the generalized variance of the projected data. This is in contrast to maximizing some measure of scatter associated with the marginal distributions of the projected scores, which does not take into account the overall structure of the projected data. Included are comparisons in which distributions are not elliptically symmetric. One of the new methods simply removes outliers using a projection-type multivariate outlier detection method that has been found to perform well relative to other outlier detection methods that have been proposed. The other new method belongs to the class of projection pursuit techniques and differs from other projection pursuit methods in terms of the function it tries to maximize. The comparisons include the method derived by Maronna (2005), the spherical method derived by Locantore et al. (1999), as well as a method proposed by Hubert, Rousseeuw, and Vanden Branden (2005). From the perspective used, the method by Hubert et al. (2005), the spherical method, and one of the new methods dominate the method derived by Maronna.
Single versus double data entry in CAST Controlled Clinical Trials Exploratory data analysis
- R A Reynolds-Haertle
- R Mcbride
- J W Tukey
Reynolds-Haertle, R. A., & McBride, R., (1992). Single versus double data entry in
CAST
Controlled
Clinical
Trials,
13,
487–494.
doi:10.1016/0197-
2456(92)90205-E
Tukey, J. W. (1977). Exploratory data analysis. Reading, MA: Addison-Wesley.
Data management: Recommended practices. Monographs of the Society for Research in Child Development
- M Burchinal
- E Neebe
Burchinal, M., & Neebe, E. (2006). Data management: Recommended practices.
Monographs of the Society for Research in Child Development, 71, 9-23.
doi:10.1111/j.1540-5834.2006.00354.x.
EpiData Entry (version v2.1). A comprehensive tool for validated entry and documentation of data
- J M Lauritsen
- M Bruus
Lauritsen, J. M., & Bruus, M., (2008). EpiData Entry (version v2.1). A comprehensive
tool for validated entry and documentation of data. Odense, Denmark: The
EpiData Association. <http://www.epidata.dk>. Accessed 29.11.10.
Poka-Yoke Data Entry System Version 1.18. Excel file that shows how to set up a double-entry data entry system for any number of measures and items
- K A Barchard
- L A Pace
Barchard, K. A., & Pace, L. A., (2010). Poka-Yoke Data Entry System Version 1.18.
Excel file that shows how to set up a double-entry data entry system for any
number of measures and items. <barchard@unlv.nevada.edu>.
- K A Barchard
- L A Pace
K.A. Barchard, L.A. Pace / Computers in Human Behavior 27 (2011) 1834–1839
Outliers and improper solutions Sociological Methods & Research
- Bollen
EpiData Entry (version v2.1). A comprehensive tool for validated entry and documentation of data: The EpiData Association. <
- J M Lauritsen Lauritsen
- M Bruus
The PowerChecker: A Visual Basic program for ensuring data integrity. Behavior, Research Methods, Innovation, & Computers. v31
- Beaty
Single versus double data entry in CAST
- R A Reynolds-Haertle
- R Mcbride
Reynolds-Haertle, R. A., & McBride, R., (1992). Single versus double data entry in
CAST
Controlled
Clinical
Trials,
13,
487-494.
doi:10.1016/0197-2456(92)90205-E
Tukey, J. W. (1977). Exploratory data analysis. Reading, MA: Addison-Wesley.