Search

Generative Artificial Intelligence (AI) systems, like ChatGPT, have the potential to perpetuate and amplify cultural biases embedded in their training data, which are predominantly produced by dominant cultural groups. This paper explores the philosophical and technical challenges of detecting and mitigating cultural bias in generative AI, drawing on Plato's Allegory of the Cave to frame the issue as a problem of limited and distorted representation. We propose a multifaceted approach combining technical interventions, such as data diversification and culturally aware model constraints, with a deeper engagement with the cultural and philosophical dimensions of the problem. Drawing on theories of extended cognition and situated knowledge, we argue that mitigating AI biases requires a reflexive interrogation of the cultural contexts of AI development and a commitment to empowering marginalized voices and perspectives. We claim that controlling cultural bias in generative AI is inseparable from the larger project of promoting equity, diversity, and inclusion in AI development and governance. By bridging philosophical reflection with technical innovation, this paper contributes to the growing discourse on responsible and inclusive AI, offering a roadmap for detecting and mitigating cultural biases while grappling with the profound cultural implications of these powerful technologies. [ABSTRACT FROM AUTHOR]

Machine learning (ML) has become increasingly prevalent in various domains. However, ML algorithms sometimes give unfair outcomes and discrimination against certain groups. Thereby, bias occurs when our results produce a decision that is systematically incorrect. At various phases of the ML pipeline, such as data collection, pre-processing, model selection, and evaluation, these biases appear. Bias reduction methods for ML have been suggested using a variety of techniques. By changing the data or the model itself, adding more fairness constraints, or both, these methods try to lessen bias. The best technique relies on the particular context and application because each technique has advantages and disadvantages. Therefore, in this paper, we present a comprehensive survey of bias mitigation techniques in machine learning (ML) with a focus on in-depth exploration of methods, including adversarial training. We examine the diverse types of bias that can afflict ML systems, elucidate current research trends, and address future challenges. Our discussion encompasses a detailed analysis of pre-processing, in-processing, and post-processing methods, including their respective pros and cons. Moreover, we go beyond qualitative assessments by quantifying the strategies for bias reduction and providing empirical evidence and performance metrics. This paper serves as an invaluable resource for researchers, practitioners, and policymakers seeking to navigate the intricate landscape of bias in ML, offering both a profound understanding of the issue and actionable insights for responsible and effective bias mitigation. [ABSTRACT FROM AUTHOR]

One can argue that one of the main roles of the subject of statistics is to characterize what the evidence in the collected data says about questions of scientific interest. There are two broad questions that we will refer to as the estimation question and the hypothesis assessment question. For estimation, the evidence in the data should determine a particular value of an object of interest together with a measure of the accuracy of the estimate, while for the hypothesis assessment, the evidence in the data should provide evidence in favor of or against some hypothesized value of the object of interest together with a measure of the strength of the evidence. This will be referred to as the evidential approach to statistical reasoning, which can be contrasted with the behavioristic or decision-theoretic approach where the notion of loss is introduced, and the goal is to minimize expected losses. While the two approaches often lead to similar outcomes, this is not always the case, and it is commonly argued that the evidential approach is more suited to scientific applications. This paper traces the history of the evidential approach and summarizes current developments. [ABSTRACT FROM AUTHOR]

The increasing use of artificial intelligence (AI) in education has raised questions about the implications of ChatGPT for teaching and learning. A systematic literature review was conducted to answer these questions, analyzing 112 scholarly articles to identify the potential benefits and challenges related to ChatGPT use in educational settings. The selection process was thorough to ensure a comprehensive analysis of the current academic discourse on AI tools in education. Our research sheds light on the significant impact of ChatGPT on improving student engagement and accessibility and the critical issues that need to be considered, including concerns about the quality and bias of generated responses, the risk of plagiarism, and the authenticity of educational content. The study aims to summarize the utilizations of ChatGPT in teaching and learning by addressing the identified benefits and challenges through targeted strategies. The authors outlined some recommendations that will ensure that the integration of ChatGPT into educational frameworks enhances learning outcomes while safeguarding academic standards. [ABSTRACT FROM AUTHOR]

The burgeoning landscape of scientific communication, marked by an explosive surge in published articles, journals, and specialized publishers, prompts a critical examination of prevailing assumptions. This article advocates a dispassionate and meticulous analysis to avoid policy decisions grounded in anecdotal evidence or superficial arguments. The discourse surrounding so-called predatory journals has been a focal point within the academic community, with concerns ranging from alleged lack of peer review rigor to exorbitant publication fees. While the consensus often leans towards avoiding such journals, this article challenges the prevailing narrative. It calls for a more nuanced understanding of what constitutes predatory practices and underscores the importance of skeptical inquiry within our daily academic activities. The authors aim to dispel misconceptions and foster a more informed dialogue by scrutinizing APCs, impact factors, and retractions. Furthermore, the authors delve into the evolving landscape of scientific publishing, addressing the generational shifts and emerging trends that challenge traditional notions of prestige and impact. In conclusion, this article serves as a call to action for the scientific community to engage in a comprehensive and nuanced debate on the complex issues surrounding scientific publishing. [ABSTRACT FROM AUTHOR]

News recommender systems (NRS) are crucial for helping users navigate the vast amount of content available online. However, traditional NRS often suffer from biases that lead to a narrow and unfair distribution of exposure across news items. In this paper, we propose a novel approach, the Contextual-Dual Bias Reduction Recommendation System (C-DBRRS), which leverages Long Short-Term Memory (LSTM) networks optimized with a multi-objective function to balance accuracy and diversity. We conducted experiments on two real-world news recommendation datasets and the results indicate that our approach outperforms the baseline methods, and achieves higher accuracy while promoting a fair and balanced distribution of recommendations. This work contributes to the development of a fair and responsible recommendation system. [ABSTRACT FROM AUTHOR]

Emotion decoding accuracy (EDA) plays a central role within the emotional intelligence (EI) ability model. The EI-ability perspective typically assumes personality antecedents and social outcomes of EI abilities, yet, traditionally, there has been very limited research to support this contention. The present paper argues that the way in which EDA has been conceptualized and operationalized in EI research has ignored developments in social perception theory and research. These developments point, on one hand, to the importance of embedding emotion expressions in a social context and, on the other, to reformulating the definitions of emotion decoding accuracy. The present paper outlines the importance of context in the framework of a truth and bias model of the social perception of emotions (Assessment of Contextualized Emotions, ACE) for EI abilities. [ABSTRACT FROM AUTHOR]

ChatGPT, a conversational AI interface that utilizes natural language processing and machine learning algorithms, is taking the world by storm and is the buzzword across many sectors today. Given the likely impact of this model on data science, through this perspective article, we seek to provide an overview of the potential opportunities and challenges associated with using ChatGPT in data science, provide readers with a snapshot of its advantages, and stimulate interest in its use for data science projects. The paper discusses how ChatGPT can assist data scientists in automating various aspects of their workflow, including data cleaning and preprocessing, model training, and result interpretation. It also highlights how ChatGPT has the potential to provide new insights and improve decision-making processes by analyzing unstructured data. We then examine the advantages of ChatGPT's architecture, including its ability to be fine-tuned for a wide range of language-related tasks and generate synthetic data. Limitations and issues are also addressed, particularly around concerns about bias and plagiarism when using ChatGPT. Overall, the paper concludes that the benefits outweigh the costs and ChatGPT has the potential to greatly enhance the productivity and accuracy of data science workflows and is likely to become an increasingly important tool for intelligence augmentation in the field of data science. ChatGPT can assist with a wide range of natural language processing tasks in data science, including language translation, sentiment analysis, and text classification. However, while ChatGPT can save time and resources compared to training a model from scratch, and can be fine-tuned for specific use cases, it may not perform well on certain tasks if it has not been specifically trained for them. Additionally, the output of ChatGPT may be difficult to interpret, which could pose challenges for decision-making in data science applications. [ABSTRACT FROM AUTHOR]

This paper proposes a generalized estimator of finite population variance using the auxiliary information under stratified random sampling. The expressions for bias and mean square error equations of the proposed estimator up to the first degree of approximation are derived. The theoretical efficiency conditions under which the proposed estimator is better than some existing estimators are obtained. The performances of the existing and proposed estimators were assessed using three real datasets based on the criteria of minimum mean square error and supreme percentage relative efficiency. Evidence from the study showed that the proposed estimator performed better and was more efficient than some existing estimators considered. [ABSTRACT FROM AUTHOR]

In this paper, we investigate optimal delegation mechanisms in the presence of countervailing conflicts of interest in the context of principal–agent problems. We introduce two dimensions of conflict of interest—pandering incentives related to the outside option and project biases. We compare three delegation mechanisms: full delegation, veto-based delegation, and communication (no delegation). Contrary to conventional one-dimensional conflict models, our findings reveal a non-monotonic relationship between pandering incentives and bias. These conflicts counterbalance each other, resulting in a principal's benefit from delegation not strictly decreasing with increased bias. Surprisingly, delegation to a biased agent can be advantageous in certain scenarios. This research sheds light on the complex dynamics of delegation mechanisms when confronted with countervailing conflicts of interest, offering insights into decision-making in principal–agent relationships. [ABSTRACT FROM AUTHOR]

ESA's Earth explorer mission CryoSat-2 has an ice-monitoring objective, but it has proven to also be a valuable source of observations for measuring impacts of climate change over oceans. In this paper, we report on our long-term ocean data analysis and validation and give our final words on CryoSat-2's Geophysical Ocean Products (GOP) Baseline-C. The validation is based on a cross comparison with concurrent altimetry and with in situ tide gauges. The highlights of our findings include GOP Baseline-C showing issues with the ionosphere and pole tide correction. The latter gives rise to an east–west pattern in range bias. Between Synthetic Aperture Radar (SAR) and Low-Resolution Mode (LRM), a 1.4 cm jump in range bias is explained by a 0.5 cm jump in sea state bias, which relates to a significant wave height SAR-LRM jump of 10.5 cm. The remaining 0.9 cm is due to a range bias between ascending and descending passes, exhibiting a clear north–south pattern and ascribed to a timing bias of +0.367 ms, affecting both time-tag and elevation. The overall range bias of GOP Baseline-C is established at −2.9 cm, referenced to all calibrated concurrent altimeter missions. The bias drift does not exceed 0.2 mm/yr, leading to the conclusion that GOP Baseline-C is substantially stable and measures up to the altimeter reference missions. This is confirmed by tide gauge comparison with a selected set of 309 PSMSL tide gauges over 2010–2022: we determined a correlation of R = 0.82, a mean standard deviation of σ = 5.7 cm (common reference and GIA corrected), and a drift of 0.17 mm/yr. In conclusion, the quality, continuity, and reference of GOP Baseline-C is exceptionally good and stable over time, and no proof of any deterioration or platform aging has been found. Any improvements for the next CryoSat-2 Baselines could come from sea state bias optimization, ionosphere and pole tide correction improvement, and applying a calibrated value for any timing biases. [ABSTRACT FROM AUTHOR]

Recent advances in facial recognition technology have achieved outstanding performance, but unconstrained face recognition remains an ongoing issue. Facial-image-quality-evaluation algorithms evaluate the quality of the input samples, providing crucial information about the accuracy of recognition decisions. By doing so, this can lead to improved results in challenging scenarios. In recent years, significant progress has been made in assessing the quality of facial images. The computation of quality scores has become highly precise and closely correlated with the model results. In this paper, we reviewed and analyzed the existing biases of cutting-edge quality-estimation techniques for face recognition. Our experimentation focused on the quality estimators developed by MagFace, FaceQNet, and SER-FIQ and were evaluated on the CelebA reference dataset. A study of bias in the face-recognition model was conducted by analyzing the quality scores presented in each article. This allowed for an examination of existing biases within both the quality estimators and the face-recognition models. [ABSTRACT FROM AUTHOR]

The Philippines is highly vulnerable to multiple climate-related hazards due to its geographical location and weak adaptation measures. Floods are the most catastrophic hazards that impact lives, livelihoods, and, consequently, the economy at large. Understanding the ability of the general circulation models to simulate the observed rainfall using the latest state-of-the-art model is essential for reliable forecasting. Based on this background, this paper objectively aims at assessing and ranking the capabilities of the recent Coupled Model Intercomparison Project Phase 6 (CMIP6) models in simulating the observed rainfall over the Philippines. The Global Precipitation Climatology Project (GPCP) v2.3 was used as a proxy to gauge the performance of 11 CMIP6 models in simulating the annual and rainy-season rainfall during 1980–2014. Several statistical metrics (mean, standard deviation, normalized root means square error, percentage bias, Pearson correlation coefficient, Mann–Kendall test, Theil–Sen slope estimator, and skill score) and geospatial measures were assessed. The results show that that CMIP6 historical simulations exhibit satisfactory effectiveness in simulating the annual cycle, though some models display wet/dry biases. The CMIP6 models generally underestimate rainfall on the land but overestimate it over the ocean. The trend analysis shows that rainfall over the country is insignificantly increasing both annually and during the rainy seasons. Notably, most of the models could correctly simulate the trend sign but over/underestimate the magnitude. The CMIP6 historical rainfall simulating models significantly agree on simulating the mean annual cycle but diverge in temporal ability simulation. The performance of the models remarkably differs from one metric to another and among different time scales. Nevertheless, the models may be ranked from the best to the least best at simulating the Philippines' rainfall in the order GFDL, NOR, ACCESS, ENS, MRI, CMCC, NESM, FIO, MIROC, CESM, TAI, and CAN. The findings of this study form a good basis for the selection of models to be used in robust future climate projection and impact studies regarding the Philippines. The climate model developers may use the documented shortcoming of these models and improve their physical parametrization for better performance in the future. [ABSTRACT FROM AUTHOR]

Children's access and opportunities to play sports are influenced by categories of difference, such as gender, race, and socioeconomic status. In order to provide an inclusive community and facilitate the recruitment and retention of diverse youth, athletic organizations should be aware of implicit bias and how this can affect the relationship between volunteers and the children they serve. This paper presents a formative process evaluation of a diversity, equity, and inclusion (DEI) training program for a non-profit athletic organization. Training was implemented in person in a group setting with multiple opportunities for group discussion. Mixed methods were used to monitor (1) the implementation of training, (2) its effects on attitudes throughout training, and (3) impact one month after training. Findings demonstrate that the program increased participants' ability to identify DEI initiatives and sense of belonging to the organization, but these effects had declined by the end of the first month after training. However, participants' self-beliefs regarding DEI and overall willingness to engage in DEI activities remained enhanced one month after training. Qualitative data were helpful in providing insight into how training impacted participants and their interactions within and outside of the organization. Through this mixed methods approach, we can conclude that DEI training did in fact have a positive impact on the organization, but further evaluation and training may be necessary to address the decline in some effects seen one month after training. [ABSTRACT FROM AUTHOR]

The purpose of this study is to show how machine learning can be leveraged as a tool to govern social impact and drive fair and equitable investments. Many organizations today are establishing financial inclusion goals to promote social impact and have been increasing their investments in this space. Financial inclusion is the opportunity for individuals and businesses to have access to affordable financial products including loans, credit, and insurance that they may otherwise not have access to with traditional financial institutions. Peer-to-peer (P2P) lending serves as a platform that can support and foster financial inclusion and influence social impact and is becoming more popular today as a resource to underserved communities. Loans issued through P2P lending can fund projects and initiatives focused on climate change, workforce diversity, women's rights, equity, labor practices, natural resource management, accounting standards, carbon emissions, and several other areas. With this in mind, AI can be a powerful governance tool to help manage risks and promote opportunities for an organization's financial inclusion goals. In this paper, we explore how AI, specifically machine learning, can help manage the P2P platform Kiva's investment risks and deliver impact, emphasizing the importance of prediction model retraining to account for regulatory and other changes across the P2P landscape to drive better decision-making. As part of this research, we also explore how changes in important model variables affect aggregate model predictions. [ABSTRACT FROM AUTHOR]

It is well known that ranked set sampling (RSS) is more efficient than simple random sampling (SRS). Furthermore, the presence of missing data vitiates the conventional results. Only a minuscule amount of work has been conducted under RSS with missing data. This paper makes a modest attempt to provide some efficient difference- and ratio-type imputation methods in the presence of missing values under RSS. The envisaged imputation methods are demonstrated to provide better results than the existing imputation methods. The theoretical results are enhanced by a computational analysis using real and hypothetically generated symmetric (Normal) and asymmetric (Gamma and Weibull) populations. The computational results show that the proposed imputation method outperforms the existing imputation methods in terms of its higher percent relative efficiency. Additionally, the impact of skewness and kurtosis on the efficiency of the suggested imputation methods has also been calculated. [ABSTRACT FROM AUTHOR]

The industrial environment has gone through the fourth revolution, also called "Industry 4.0", where the main aspect is digitalization. Each device employed in an industrial process is connected to a network called the industrial Internet of things (IIOT). With IIOT manufacturers being capable of tracking every device, it has become easier to prevent or quickly solve failures. Specifically, the large amount of available data has allowed the use of artificial intelligence (AI) algorithms to improve industrial applications in many ways (e.g., failure detection, process optimization, and abnormality detection). Although data are abundant, their access has raised problems due to privacy concerns of manufacturers. Censoring sensitive information is not a desired approach because it negatively impacts the AI performance. To increase trust, there is also the need to understand how AI algorithms make choices, i.e., to no longer regard them as black boxes. This paper focuses on recent advancements related to the challenges mentioned above, discusses the industrial impact of proposed solutions, and identifies challenges for future research. It also presents examples related to privacy-preserving and explainable AI solutions, and comments on the interaction between the identified challenges in the conclusions. [ABSTRACT FROM AUTHOR]

The development of algorithms, based on machine learning techniques, supporting (or even replacing) human judgment must take into account concepts such as data bias and fairness. Though scientific literature proposes numerous techniques to detect and evaluate these problems, less attention has been dedicated to methods generating intentionally biased datasets, which could be used by data scientists to develop and validate unbiased and fair decision-making algorithms. To this end, this paper presents a novel method to generate a synthetic dataset, where bias can be modeled by using a probabilistic network exploiting structural equation modeling. The proposed methodology has been validated on a simple dataset to highlight the impact of tuning parameters on bias and fairness, as well as on a more realistic example based on a loan approval status dataset. In particular, this methodology requires a limited number of parameters compared to other techniques for generating datasets with a controlled amount of bias and fairness. [ABSTRACT FROM AUTHOR]

Archaeologists often rely on passive airborne optical remote sensing to deliver some of the core data for (European) landscape archaeology projects. Despite the many technological and theoretical evolutions that have characterised this field of archaeology, the dominant aerial photographic surveys, but also less common approaches to archaeological airborne reconnaissance, still suffer from many inherent biases imposed by sub-par sampling strategies, cost, instrument availability and post-processing issues. This paper starts with the concept of landscape (archaeology) and uses it to frame archaeological airborne remote sensing. After introducing the need for bias reduction when sampling an already distorted archaeological population and expanding on the 'theory-neutral' claim of aerial survey, the paper presents eight key characteristics that all have the potential to increase or decrease the subjectivity and bias when collecting airborne optical imagery with passive sensors. Within this setting, the paper then offers some technological-methodological reflection on the various passive airborne optical imaging solutions that landscape archaeology has come to rely upon in the past decades. In doing so, it calls into question the effectiveness and suitability of these highly subjective approaches for landscape archaeology. Finally, the paper proposes a new, more objective approach to aerial optical image acquisition with passive sensors. In the discussion, the text argues that the suggested exhaustive (or total) airborne sampling of the preserved archaeological record might transcend particular theoretical paradigms, while the data generated could span various interpretational perspectives and oppositional analytical approaches in landscape archaeology. [ABSTRACT FROM AUTHOR]

Focus on fair lending has become more intensified recently as bank and non-bank lenders apply artificial-intelligence (AI)-based credit determination approaches. The data analytics technique behind AI and machine learning (ML) has proven to be powerful in many application areas. However, ML can be less transparent and explainable than traditional regression models, which may raise unique questions about its compliance with fair lending laws. ML may also reduce potential for discrimination, by reducing discretionary and judgmental decisions. As financial institutions continue to explore ML applications in loan underwriting and pricing, the fair lending assessments typically led by compliance and legal functions will likely continue to evolve. In this paper, the author discusses unique considerations around ML in the existing fair lending risk assessment practice for underwriting and pricing models and proposes consideration of additional evaluations to be added in the present practice. [ABSTRACT FROM AUTHOR]

This paper analyzes the statistic properties of the systematic error in terms of range and bearing during the transformation process. Furthermore, we rely on a weighted nonlinear least square method to calculate the biases based on the proposed models. The results show the high performance of the proposed approach for error modeling and bias estimation. [ABSTRACT FROM AUTHOR]

The rapidly evolving field of machine learning (ML), along with artificial intelligence in a broad sense, is revolutionising many areas of healthcare, including laboratory medicine. The amalgamation of the fields of ML and patient-based real-time quality control (PBRTQC) processes could improve the traditional PBRTQC and error detection algorithms in the laboratory. This narrative review discusses published studies on using ML for the detection of systematic errors, non-systematic errors, and combinations of different types of errors in clinical laboratories. The studies discussed used ML for detecting bias, the requirement for re-calibration, samples contaminated with intravenous fluid or EDTA, delayed sample analysis, wrong-blood-in-tube errors, interference or a combination of different types of errors, by comparing the performance of ML models with human validators or traditional PBRTQC algorithms. Advantages, limitations, the creation of standardised ML models, ethical and regulatory aspects and potential future developments have also been discussed in brief. [ABSTRACT FROM AUTHOR]

The World Trade Center (WTC) disaster exposed the responders to several hazards. Three cohorts i.e., the Fire Department of New York (FDNY), the General Responder Cohort (GRC), and the WTC Health Registry (WTCHR) surveyed the exposed responder population. We searched Pubmed and Web of Science for literature on a well-published association between the WTC exposures and asthma, focusing on new-onset self-reported physician-diagnosed asthma. The resulting five articles were qualitatively assessed for potential biases. These papers were independently reviewed by the co-authors, and conclusions were derived after discussions. While, the cohorts had well-defined eligibility criteria, they lacked information about the entire exposed population. We conclude that selection and surveillance biases may have occurred in the GRC and WTCHR cohorts, but were likely to have been minimal in the FDNY cohort. Health care benefits available to responders may have increased the reporting of both exposure and outcome in the former, and decreased outcome reporting in the FDNY cohort. Irrespective of the biases, the studies showed similar findings, confirming the association between WTC exposure and self-reported physician-diagnosed asthma among responders. This suggests that health data gathered under great duress and for purposes other than epidemiology can yield sound conclusions. Potential biases can, however, be minimized by having validated survey instruments and worker registries in place before events occur.

Previous research has shown that merely calling an indoor environment environmentally certified will make people favor that environment over a conventional alternative. In this paper we explore whether this effect depends on participants deliberately comparing the two environments, and whether different reasons behind the certification influence the magnitude of the effect. In Experiment 1, participants in a between-subjects design assigned higher comfort ratings to an indoor environment that had been labeled "environmentally certified" in comparison with the exact same indoor environment that was unlabeled, suggesting that the effect arises even when participants do not compare the two environments when making their estimates. The results from Experiment 2 indicate that climate change mitigation (as the reason for the certification) is a slightly better trigger of the effect compared to climate change adaptation. The results suggest that studies on psychological effects of "green" buildings should experimentally control for the influence from participants' judgmental biases. [ABSTRACT FROM AUTHOR]

Today, the significance of biometrics is more pronounced than ever in accurately allowing access to valuable resources, from personal devices to highly sensitive buildings, as well as classified information. Researchers are pushing forward toward devising robust biometric systems with higher accuracy, fewer false positives and false negatives, and better performance. On the other hand, machine learning (ML) has been shown to play a key role in improving such systems. By constantly learning and adapting to users' changing biometric patterns, ML algorithms can improve accuracy and performance over time. The integration of ML algorithms with biometrics, however, introduces vulnerabilities in such systems. This article investigates the new issues of concern that come about because of the adoption of ML methods in biometric systems. Specifically, techniques to breach biometric systems, namely, data poisoning, model inversion, bias injection, and deepfakes, are discussed. Here, the methodology consisted of conducting a detailed review of the literature in which ML techniques have been adopted in biometrics. In this study, we included all works that have successfully applied ML and reported favorable results after this adoption. These articles not only reported improved numerical results but also provided sound technical justification for this improvement. There were many isolated, unsupported, and unjustified works about the major advantages of ML techniques in improving security, which were excluded from this review. Though briefly mentioned, we did not touch upon encryption/decryption aspects, and, accordingly, cybersecurity was excluded from this study. At the end, recommendations are made to build stronger and more secure systems that benefit from ML adoption while closing the door to adversarial attacks. [ABSTRACT FROM AUTHOR]

The bias of the maximum likelihood estimator can cause a considerable estimation error if the sample size is small. To reduce the bias of the maximum likelihood estimator under the small sample situation, the maximum likelihood and parametric bootstrap bias-correction methods are proposed in this study to obtain more reliable maximum likelihood estimators of the unit exponential distribution parameters. The procedure to implement the bias-corrected maximum likelihood estimation method is derived analytically, and the steps to obtain the bias-corrected bootstrap estimators are presented. The simulation results show that the proposed maximum likelihood bootstrap bias-correction method can significantly reduce the bias and mean squared error of the maximum likelihood estimators for most of the parameter combinations in the simulation study. A soil moisture data set and a numerical example are used for illustration. [ABSTRACT FROM AUTHOR]

A decision maker (DM) must choose between two projects or decide on no project. The expected benefits of these projects are correlated. The DM seeks advice from an agent with private information about the projects' benefits. However, the agent's divergent preferences for projects and lack of consideration for the DM's implementation costs may introduce two types of biases: project bias, favoring the agent's project, or pandering bias, favoring the project preferred by the DM. Our findings reveal that project correlation leads to these biases countervailing each other, facilitating the transmission of information. The agent typically recommends a project based on private information to dissuade the DM from choosing no project, as this would be detrimental to the agent. Additionally, we explore optimal delegation within organizations. In contrast to the prevailing literature advocating for delegation to biased agents for enhanced information elicitation, our study suggests limited benefits in the context of project correlation. [ABSTRACT FROM AUTHOR]

Artificial intelligence models represented in machine learning algorithms are promising tools for risk assessment used to guide clinical and other health care decisions. Machine learning algorithms, however, may house biases that propagate stereotypes, inequities, and discrimination that contribute to socioeconomic health care disparities. The biases include those related to some sociodemographic characteristics such as race, ethnicity, gender, age, insurance, and socioeconomic status from the use of erroneous electronic health record data. Additionally, there is concern that training data and algorithmic biases in large language models pose potential drawbacks. These biases affect the lives and livelihoods of a significant percentage of the population in the United States and globally. The social and economic consequences of the associated backlash cannot be underestimated. Here, we outline some of the sociodemographic, training data, and algorithmic biases that undermine sound health care risk assessment and medical decision-making that should be addressed in the health care system. We present a perspective and overview of these biases by gender, race, ethnicity, age, historically marginalized communities, algorithmic bias, biased evaluations, implicit bias, selection/sampling bias, socioeconomic status biases, biased data distributions, cultural biases and insurance status bias, conformation bias, information bias and anchoring biases and make recommendations to improve large language model training data, including de-biasing techniques such as counterfactual role-reversed sentences during knowledge distillation, fine-tuning, prefix attachment at training time, the use of toxicity classifiers, retrieval augmented generation and algorithmic modification to mitigate the biases moving forward. [ABSTRACT FROM AUTHOR]

The question that the authors of this article are collectively concerned with is as follows: how is it possible to protect children without disempowering them? To this end, the authors work to change adultcentric scholarly and social norms that justify rationales that marginalize children. The article begins with a theoretical overview of childism, in its transformative sense, with special attention to how childism relates to intersectional analyses. In doing so, age is highlighted as an axis of marginalization with reference to adultcentrism. After that, the centrality of analyzing and problematizing adultism in educational research and practice is discussed. The discussion is followed by a presentation of the published results of 'The Adultcentrism Scale' research tool developed at the University of Bergamo and the University of LUMSA-Rome. The research tool is used to evaluate the presence of adultcentric bias in adults in relation to children and can be helpful to understanding the psychological dimensions of educational relationships. Finally, the conclusion offers suggestions for how the research tool might be a useful example to raise awareness of adultcentric bias, promoting reflections that can lead to age-inclusive transformations. Overall, then, the article initiates a pertinent dialogue for advancing children's participation rights in Nordic research and society. [ABSTRACT FROM AUTHOR]

Assessing the usefulness of reviews has been the aim of several research studies. However, results regarding the significance of usefulness determinants are often contradictory, thus decreasing the accuracy of reviews' helpfulness estimation. Also, bias in user reviews attributed to differences, e.g., in gender, nationality, etc., may result in misleading judgments, thus diminishing reviews' usefulness. Research is needed for sentiment analysis algorithms that incorporate bias embedded in reviews, thus improving their usefulness, readability, credibility, etc. This study utilizes fuzzy relations and fuzzy synthetic evaluation (FSE) in order to calculate reviews' usefulness by incorporating users' biases as expressed in terms of reviews' articulacy and sentiment polarity. It selected and analyzed 95,678 hotel user reviews from Tripadvisor, written by users from five specific nationalities. The findings indicate that there are differences among nationalities in terms of the articulacy and sentiment of their reviews. The British are most consistent in their judgments expressed in titles and the main body of reviews. For the British and the Greeks, review titles suffice to convey any negative sentiments. The Dutch use fewer words in their reviews than the other nationalities. This study suggests that fuzzy logic captures subjectivity which is often found in reviews, and it can be used to quantify users' behavioral differences, calculate reviews' usefulness, and provide the means for developing more accurate voting systems. [ABSTRACT FROM AUTHOR]

The significant advancements in applying artificial intelligence (AI) to healthcare decision-making, medical diagnosis, and other domains have simultaneously raised concerns about the fairness and bias of AI systems. This is particularly critical in areas like healthcare, employment, criminal justice, credit scoring, and increasingly, in generative AI models (GenAI) that produce synthetic media. Such systems can lead to unfair outcomes and perpetuate existing inequalities, including generative biases that affect the representation of individuals in synthetic data. This survey study offers a succinct, comprehensive overview of fairness and bias in AI, addressing their sources, impacts, and mitigation strategies. We review sources of bias, such as data, algorithm, and human decision biases—highlighting the emergent issue of generative AI bias, where models may reproduce and amplify societal stereotypes. We assess the societal impact of biased AI systems, focusing on perpetuating inequalities and reinforcing harmful stereotypes, especially as generative AI becomes more prevalent in creating content that influences public perception. We explore various proposed mitigation strategies, discuss the ethical considerations of their implementation, and emphasize the need for interdisciplinary collaboration to ensure effectiveness. Through a systematic literature review spanning multiple academic disciplines, we present definitions of AI bias and its different types, including a detailed look at generative AI bias. We discuss the negative impacts of AI bias on individuals and society and provide an overview of current approaches to mitigate AI bias, including data pre-processing, model selection, and post-processing. We emphasize the unique challenges presented by generative AI models and the importance of strategies specifically tailored to address these. Addressing bias in AI requires a holistic approach involving diverse and representative datasets, enhanced transparency and accountability in AI systems, and the exploration of alternative AI paradigms that prioritize fairness and ethical considerations. This survey contributes to the ongoing discussion on developing fair and unbiased AI systems by providing an overview of the sources, impacts, and mitigation strategies related to AI bias, with a particular focus on the emerging field of generative AI. [ABSTRACT FROM AUTHOR]

Given that some suspected perpetrators were wrongly convicted, a defective bitemark analysis is comparable to dentists' most crucial clinical decisions regarding assessment. Bias affects human bitemark analysis beyond the limitation of the evidence itself. The aim of this study was to explore the potential for different types of bias in bitemark analysis and the methods involved in that analysis by conducting a scoping review. Results showed that the 14 articles that explore the topic of bias in bitemark analysis were published from 2006-2022. Publications were from the USA mainly (n = 7), followed by the UK (n = 3), Australia (n = 2), New Zealand, (n = 1) and the Netherlands (n = 1). Of these publications, 36% addressed contextual bias, while 57% acknowledged cognitive bias. According to the findings, preventive measures consist of limiting the availability of unrelated data during research, employing several comparison samples for a more impartial assessment, and repeating the analysis while being blind to past findings. Nevertheless, the physical limitations of the evidence such as distortions are still strongly present. [ABSTRACT FROM AUTHOR]

In recent years, deep learning models have led to improved accuracy in industrial defect detection, often using variants of YOLO (You Only Look Once), due to its high performance at a low cost. However, the generalizability, fairness and bias of their outcomes have not been examined, which may lead to overconfident predictions. Additionally, the complexity added by co-occurring defects, single and multi-class defects, and the effect on training, is not taken into consideration. This study addresses these critical gaps by introducing new methodologies for analyzing dataset complexity and evaluating model fairness. It introduces the novel approach of co-occurrence impact analysis, examining how the co-occurrence of defects in sample images affects performance, and introducing new dimensions to dataset preparation and training. Its aim is to increase model robustness in the face of real-world scenarios where multiple defects often appear together. Our study also innovates in the evaluation of model fairness by adapting the disparate impact ratio (DIR) to consider the true positive rate (TPR) across different groups and modifying the predictive parity difference (PPD) metric to focus on biases present in industrial quality control. Experiments demonstrate by cross-validation that the model trained on combined datasets significantly outperforms others in accuracy without overfitting and results in increased fairness, as validated by our novel fairness metrics. Explainability also provides valuable insights on the effects of different training regimes, notably absent in prior works. This work not only advances the field of deep learning for defect detection but also provides a strategic framework for future advancements, emphasizing the need for balanced datasets and considerations of ethics, fairness, bias and generalizability in the deployment of artificial intelligence in industry. [ABSTRACT FROM AUTHOR]

The complexity of migration flows across the world has led to a redefinition of psychological and social services users. The access of migrants from different cultural backgrounds to clinical services or social health services has diversified the demand for concomitant help. Biases and misinterpretations have been created by unaccustomed professionals in this field, which could lead to serious consequences and invalidate diagnostic and treatment procedures. The purpose of this study is to summarize the evidence about errors or prejudices observed in clinical practices regarding the provision of social health services to people from different cultural backgrounds. Results show three main types of biases: racial stereotype activation, ethnocentrism and micro-aggressions. Some implications on the clinical setting were discussed, as being aware of these biases can help mental health professionals manage communication more consciously with users. [ABSTRACT FROM AUTHOR]

Understanding the public's diverse linguistic expressions about rainfall and flood provides a basis for flood disaster studies and enhances linguistic and cultural awareness. However, existing research tends to overlook linguistic complexity, potentially leading to bias. In this study, we introduce a novel algorithm capturing rainfall and flood-related expressions, considering the relationship between precipitation observations and linguistics expressions. Analyzing 210 million social media microblogs from 2017, we identified 594 keywords, 20 times more than usual manually created bag-of-words. Utilizing Large Language Model, we categorized these keywords into rainfall, flood, and other related terms. Semantic features of these keywords were analyzed from the viewpoint of popularity, credibility, time delay, and part-of-speech, finding rainfall-related terms most common-used, flood-related keywords often more time delayed than precipitation, and notable differences in part-of-speech across categories. We also assessed spatial characteristics from keyword and city-centric perspectives, revealing that 49.5% of the keywords have significant spatial correlation with differing median centers, reflecting regional variations. Large and disaster-impacted cities show the richest expression diversity for rainfall and flood-related terms. [ABSTRACT FROM AUTHOR]

To consistently produce high quality products, a quality management system, such as the ISO9001, 2000 or TS 16949 must be practically implemented. One core instrument of the TS16949 MSA (Measurement System Analysis) is to rank the capability of a measurement system and ensure the quality characteristics of the product would likely be transformed through the whole manufacturing process. It is important to reduce the risk of Type I errors (acceptable goods are misjudged as defective parts) and Type II errors (defective parts are misjudged as good parts). An ideal measuring system would have the statistical characteristic of zero error, but such a system could hardly exist. Hence, to maintain better control of the variance that might occur in the manufacturing process, MSA is necessary for better quality control. Ball screws, which are a key component in precision machines, have significant attributes with respect to positioning and transmitting. Failures of lead accuracy and axial-gap of a ball screw can cause negative and expensive effects in machine positioning accuracy. Consequently, a functional measurement system can incur great savings by detecting Type I and Type II errors. If the measurement system fails with respect to specification of the product, it will likely misjudge Type I and Type II errors. Inspectors normally follow the MSA regulations for accuracy measurement, but the choice of measuring system does not merely depend on some simple indices. In this paper, we examine the stability of a measuring system by using a Monte Carlo simulation to establish bias, linearity variance of the normal distribution, and the probability density function. Further, we forecast the possible area distribution in the real case. After the simulation, the measurement capability will be improved, which helps the user classify the measurement system and establish measurement regulations for better performance and monitoring of the precision of the ball screw. [ABSTRACT FROM AUTHOR]

In this paper, a novel temperature compensation method for a dual-mass MEMS gyroscope is proposed based on drive mode vibration characteristic compensation using a temperature variable resistor. Firstly, the drive and sense modes of the gyroscope re analyzed and investigated, and it is found that the scale factor is proportional to the drive mode amplitude controlling reference voltage. Then, the scale factor temperature compensation method is proposed, and a temperature variable resistor is utilized to compensate the drive amplitude working point and make it change with temperature. In addition, the temperature compensation circuit is designed and simulated. After that, the temperature bias drift is compensated in a modular output. The experimental results show that scale factor and bias variation during the temperature range from −40 °C to 60 °C decrease from 3.680% to 1.577% and 3.880% to 1.913%, respectively. In addition, the bias value improves from 103.395 °/s to 22.478 °/s (optimized 78.26%). The bias stability and angular rate walking parameter are also optimized to 45.97% and 16.08%, respectively, which verify the method proposed in this paper. [ABSTRACT FROM AUTHOR]

Are traditional tests of forecast evaluation well behaved when the competing (nested) model is biased? No, they are not. In this paper, we show analytically and via simulations that, under the null hypothesis of no encompassing, a bias in the nested model may severely distort the size properties of traditional out-of-sample tests in economic forecasting. Not surprisingly, these size distortions depend on the magnitude of the bias and the persistency of the additional predictors. We consider two different cases: (i) There is both in-sample and out-of-sample bias in the nested model. (ii) The bias is present exclusively out-of-sample. To address the former case, we propose a modified encompassing test (MENC-NEW) robust to a bias in the null model. Akin to the ENC-NEW statistic, the asymptotic distribution of our test is a functional of stochastic integrals of quadratic Brownian motions. While this distribution is not pivotal, we can easily estimate the nuisance parameters. To address the second case, we derive the new asymptotic distribution of the ENC-NEW, showing that critical values may differ remarkably. Our Monte Carlo simulations reveal that the MENC-NEW (and the ENC-NEW with adjusted critical values) is reasonably well-sized even when the ENC-NEW (with standard critical values) exhibits rejections rates three times higher than the nominal size. [ABSTRACT FROM AUTHOR]

This study investigated how professionals' decision-making is influenced by nudging and their resistance to such a form of conditioning. A total of 61 professionals performed a nudge task in which three different scenarios related to wellbeing and sustainable behaviours were presented to the participants under boosted and soft nudge conditions. After the presentation of each scenario, participants were required to decide between two options of choice: one choice was more nudge-induced, the other was not. Electrophysiological (EEG), autonomic, behavioural, and self-report data were collected to determine the correlates of resistance with nudge conditions. The findings showed that professionals' resistance to nudging is high and not influenced by boosted or soft nudges. Also, while the generalized increase in EEG delta, theta, and beta power localized and lateralized in the right temporoparietal regions can lay the foundation of "the neural architecture" of resistance to nudging, the significant increase in SCR for the boosted compared to soft condition highlighted the pivotal role of this marker as the only indicator that differentiates the two nudge conditions. Overall, evaluating the correlates of the resistance to nudge can be useful to render professionals aware of the explicit and implicit factors to be strengthened to resist to such form of conditioning. [ABSTRACT FROM AUTHOR]

Instances of racial disparities are well documented in the United States' criminal justice system. This study reviewed the literature and conducted quantitative analyses on the role of race in forensic decision making among practitioners and other stakeholders in the criminal justice system. We hypothesized that Black, Indigenous, and People of Color (BIPOC) individuals will be significantly more likely to be associated with adverse outcomes than White individuals. A search strategy was developed and registered before the study commenced. Quantitative data were extracted from eligible studies to estimate the pooled effect size (odds ratio) for the effects of race. A final sample of 11 data sources (published study or dataset) was identified. Decision making by all stakeholders in the criminal justice system, including forensic practitioners, case investigators, and juries were evaluated in these studies. Two datasets evaluated the decision-making process involving forensic psychology or psychiatry, three focused on forensic evidence, four on forensic pathology, one involved forensic anthropology cases, and one involved clinical forensic medicine cases. The pooled odds ratio was estimated to be 1.10 (95% confidence interval: 0.67–1.81), indicating a trivial or negligible effect of race (i.e., BIPOC individuals were no more likely to be associated with adverse outcomes given the current evidence). Importantly, the results of this study do not indicate that bias or disparity related to race does not exist in forensic decision making in the criminal justice system. More research into systemic bias in forensic decision making, especially in relation to race, is needed. Forensic anthropologists are uniquely positioned to study and address racial disparities in the criminal justice system involving forensic science because of its interdisciplinary and multidisciplinary nature. This study highlights the need for further research and advocates for forensic anthropologists to be more involved in the study of the science and the impacts of forensic science rather than focusing on methodological advancement. [ABSTRACT FROM AUTHOR]

This study rolls out a robust framework relevant for simulation studies through the Generalised Autoregressive Conditional Heteroscedasticity (GARCH) model using the rugarch package. The package is thoroughly investigated, and novel findings are identified for improved and effective simulations. The focus of the study is to provide necessary simulation steps to determine appropriate distributions of innovations relevant for estimating the persistence of volatility. The simulation steps involve "background (optional), defining the aim, research questions, method of implementation, and summarised conclusion". The method of implementation is a workflow that includes writing the code, setting the seed, setting the true parameters a priori, data generation process and performance assessment through meta-statistics. These novel, easy-to-understand steps are demonstrated on financial returns using illustrative Monte Carlo simulation with empirical verification. Among the findings, the study shows that regardless of the arrangement of the seed values, the efficiency and consistency of an estimator generally remain the same as the sample size increases. The study also derived a new and flexible true-parameter-recovery measure which can be used by researchers to determine the level of recovery of the true parameter by the MCS estimator. It is anticipated that the outcomes of this study will be broadly applicable in finance, with intuitive appeal in other areas, for volatility modelling. [ABSTRACT FROM AUTHOR]

The standardized precipitation index (SPI) is used for characterizing and predicting meteorological droughts on a range of time scales. However, in forecasting applications, when SPI is computed on the entire available dataset, prior to model-validation, significant biases are introduced, especially under changing climatic conditions. In this paper, we investigate the theoretical and numerical implications that arise when SPI is computed under stationary and non-stationary probability distributions. We demonstrate that both the stationary SPI and non-stationary SPI (NSPI) lead to increased information leakage to the training set with increased scales, which significantly affects the characterization of drought severity. The analysis is performed across about 36,500 basins in Sweden, and indicates that the stationary SPI is unable to capture the increased rainfall trend during the last decades and leads to systematic underestimation of wet events in the training set, affecting up to 22% of the drought events. NSPI captures the non-stationary characteristics of accumulated rainfall; however, it introduces biases to the training data affecting 19% of the drought events. The variability of NSPI bias has also been observed along the country's climatic gradient with regions in snow climates strongly being affected. The findings propose that drought assessments under changing climatic conditions can be significantly influenced by the potential misuse of both SPI and NSPI, inducing bias in the characterization of drought events in the training data. [ABSTRACT FROM AUTHOR]

To inform strategies to improve the dietary intakes of populations, robust evaluations of interventions are required. This paper is drawn from a workshop held at the International Society of Behavioral Nutrition and Physical Activity 2017 Annual Meeting, and highlights considerations and research priorities relevant to measuring dietary outcomes within intervention studies. Self-reported dietary data are typically relied upon in such studies, and it is recognized that these data are affected by random and systematic error. Additionally, differential error between intervention and comparison groups or pre- and post-intervention can be elicited by the intervention itself, for example, by creating greater awareness of eating or drinking occasions or the desire to appear compliant. Differential reporting can render the results of trials incorrect or inconclusive by leading to biased estimates and reduced statistical power. The development of strategies to address intervention-related biases requires developing a better understanding of the situations and population groups in which interventions are likely to elicit differential reporting and the extent of the bias. Also needed are efforts to expand the feasibility and applications of biomarkers to address intervention-related biases. In the meantime, researchers are encouraged to consider the potential for differential biases in dietary reporting in a given study, to choose tools carefully and take steps to minimize and/or measure factors such as social desirability biases that might contribute to differential reporting, and to consider the implications of differential reporting for study results. [ABSTRACT FROM AUTHOR]

The purpose of the paper is to explore the relative biases in the estimation of the Full BEKK model as compared with the Diagonal BEKK model, which is used as a theoretical and empirical benchmark. Chang and McAleer et al., 2017 show that univariate GARCH is not a special case of multivariate GARCH, specifically, the Full BEKK model, and demonstrate that Full BEKK, which, in practice, is estimated almost exclusively, has no underlying stochastic process, regularity conditions, or asymptotic properties. Diagonal BEKK (DBEKK) does not suffer from these limitations, and hence provides a suitable benchmark. We use simulated financial returns series to contrast estimates of the conditional variances and covariances from DBEKK and BEKK. The results of non-parametric tests suggest evidence of considerable bias in the Full BEKK estimates. The results of quantile regression analysis show there is a systematic relationship between the two sets of estimates as we move across the quantiles. Estimates of conditional variances from Full BEKK, relative to those from DBEKK are relatively lower in the left tail and higher in the right tail. The BEKK model is a commonly applied multivariate volatility model frequently used in modelling and forecasting volatilities in financial applications. Our results suggest that it is subject to considerable bias and this should be considered by potential users. [ABSTRACT FROM AUTHOR]

Atheism has stirred up controversy in the Arabic-speaking world since the 2011 uprisings, when atheists there began appearing in public. What role does Arabic mass media play in the modern politics of minorities such as atheists, given the heated debates that it hosts on atheism? This question is important because perceptions of media frames influence the behavior of politicians and the electorate—and, as a result, laws that affect minority groups such as atheists. This article focuses on Lebanon, where eight of the nine television channels are affiliated with and funded by religious–political parties. It explores the existence of bias against atheists on televised Lebanese talk shows and news reports (2010–2022). Our findings reveal significant bias (69% overall and over 85% in speaker prominence bias), with channels that promote communal religious practice exhibiting the highest levels. Ultimately, our findings demonstrate that television, as the most influential Lebanese and Arabic mass medium, likely affects the public's negative perceptions of Arabic-speaking atheists. Our findings reflect the decrease in objectivity in conflict-based media and such media's poor understanding or intentional disregard for media's crucial role in building a fair, democratic society. [ABSTRACT FROM AUTHOR]

Nanotechnology application in cancer treatment is promising and is likely to quickly spread worldwide in the near future. To date, most scientific studies on nanomaterial development have focused on deepening the attitudes of end users and experts, leaving clinical practice implications unexplored. Neuro-oncology might be a promising field for the application of nanotechnologies, especially for malignant brain tumors with a low-survival rate such as glioblastoma (GBM). As to improving patients' quality of life and life expectancy, innovative treatments are worth being explored. Indeed, it is important to explore clinicians' intention to use experimental technologies in clinical practice. In the present study, we conducted an exploratory review of the literature about healthcare workers' knowledge and personal opinions toward nanomedicine. Our search (i) gives evidence for disagreement between self-reported and factual knowledge about nanomedicine and (ii) suggests the internet and television as main sources of information about current trends in nanomedicine applications, over scientific journals and formal education. Current models of risk assessment suggest time-saving cognitive and affective shortcuts, i.e., heuristics support both laypeople and experts in the decision-making process under uncertainty, whereas they might be a source of error. Whether the knowledge is poor, heuristics are more likely to occur and thus clinicians' opinions and perspectives toward new technologies might be biased. [ABSTRACT FROM AUTHOR]

Passive bistatic radar research is essential for accurate 3D target tracking, especially in the presence of missing or low-quality bearing information. Traditional extended Kalman filter (EKF) methods often introduce bias in such scenarios. To overcome this limitation, we propose employing the unscented Kalman filter (UKF) for handling the nonlinearities in 3D tracking, utilizing range and range-rate measurements. Additionally, we incorporate the probabilistic data association (PDA) algorithm with the UKF to handle cluttered environments. Through extensive simulations, we demonstrate a successful implementation of the UKF-PDA framework, showing that the proposed method effectively reduces bias and significantly advances tracking capabilities in passive bistatic radars. [ABSTRACT FROM AUTHOR]

Though a wide array of definitions and conceptualisations of critical thinking have been offered in the past, further elaboration on some concepts is required, particularly with respect to various factors that may impede an individual's application of critical thinking, such as in the case of reflective judgment. These barriers include varying levels of epistemological engagement or understanding, issues pertaining to heuristic-based thinking and intuitive judgment, as well as emotional and biased thinking. The aim of this review is to discuss such barriers and evaluate their impact on critical thinking in light of perspectives from research in an effort to reinforce the 'completeness' of extant critical thinking frameworks and to enhance the potential benefits of implementation in real-world settings. Recommendations and implications for overcoming such barriers are also discussed and evaluated. [ABSTRACT FROM AUTHOR]

Although widely used in the judgment under uncertainty literature, the so-called Lawyer–Engineer problem does not have a Bayesian solution because the base rates typically oppose qualitative stereotypical information, which has an undefined diagnostic value. We propose an experimental paradigm that elicits participants' subjective estimates of the diagnosticity of stereotypical information and allows us to investigate the degree to which participants are able to integrate both sources of information (base rates and stereotypical descriptions) according to the Bayesian rule. This paradigm was used to test the hypothesis that the responses (probability estimates) to the Lawyer–Engineer problem from more rational individuals deviate from normative Bayesian solutions in a way that shows smaller but more systematic bias. The results further suggest that the estimates of less rational participants are noisier (less reliable) but may be more accurate when aggregated across several problems. [ABSTRACT FROM AUTHOR]