Prognostic significance of circulating lymphocyte subsets in patients with nasopharyngeal carcinoma

The full text of this preprint has been withdrawn by the authors while they make corrections to the work. Therefore, the authors do not wish this work to be cited as a reference. Questions should be directed to the corresponding author.Additionally, the authors have provided this withdrawal declaration: "In this preprint, the criteria for inclusion and exclusion of study designs were not rigorous enough. After data verification, we found that there were some missing or unknown people in the follow-up results of the original data. This part of data may cause information bias. we recognize that the results and conclusions obtained based on these data may be unreliable. After careful discussion, all authors have agreed that, based on the need of scientific accuracy and honesty, it is necessary to withdraw the preprint."

Continued Effectiveness of COVID-19 Vaccination among Urban Healthcare Workers during Delta Variant Predominance

Background: Data on COVID-19 vaccine effectiveness (VE) among healthcare workers (HCWs) during periods of delta variant predominance are limited. Methods: We followed a population of urban Massachusetts HCWs (45% non-White) subject to epidemiologic surveillance. We accounted for covariates such as demographics and community background infection incidence, as well as information bias regarding COVID-19 diagnosis and vaccination status. Results and Discussion: During the study period (December 16, 2020 to September 30, 2021), 4615 HCWs contributed to a total of 1,152,486 person-days at risk (excluding 309 HCWs with prior infection) and had a COVID-19 incidence rate of 5.2/10,000 (114 infections out of 219,842 person-days) for unvaccinated person-days and 0.6/10,000 (49 infections out of 830,084 person-days) for fully vaccinated person-days, resulting in an adjusted VE of 82.3% (95% CI: 75.1-87.4%). For the secondary analysis limited to the period of delta variant predominance in Massachusetts (i.e., July 1 to September 30, 2021), we observed an adjusted VE of 76.5% (95% CI: 40.9-90.6%). Independently, we found no re-infection among those with prior COVID-19, contributing to 74,557 re-infection-free person-days, adding to the evidence base for the robustness of naturally acquired immunity.

On the Relationship among User’s Reading Behavior, Algorithm Recommendation Mechanism and the Manufactured Filter Bubbles Effect

This paper analyses the causal logic of algorithm recommendation and it employs the Pielou index to measure the distribution of news contents to provide empirical evidence to indicate whether the algorithm recommendation mechanism may produce filter bubbles. Moreover, this research takes Headlines Today as the research object to better understand the realization of tailored news and how their reading behaviour affect the algorithm recommendation mechanism. Meanwhile, the conclusion reinforces that users should enhance their information literacy in the era of artificial intelligence and big data, make rational use of algorithm recommendation mechanism, and pay close attention to the diversity of information sources to avoid information bias. This paper also helps the information flow platform to reflect on the shortcomings of the algorithm mechanism and optimise its strengths while avoiding those manufactured negative effects and proposes that in the optimisation of algorithm recommendation mechanism, the positive guidance to users should also be emphasised. Indicators such as content influence and mainstream media recommendation can be added to generate a multi-index recommendation.

Competition and Distortion: A Theory of Information Bias on the Peer-to-Peer Lending Market

Despite the popular emergence of peer-to-peer (P2P) lending platforms, relevant research investigating the role of these platforms on P2P markets still lags. In this paper, we present a model to study the market incentives of P2P lending platforms' optimal information-reporting strategies when the following exist: (i) uncertainty on the return of loans and (ii) competition from entrants. We focus on the information bias of platforms driven by demand-side actors—investors’ optimism/pessimism about risk—while we keep the platforms being rational. We characterize platforms' equilibrium reporting strategies under different market conditions. Surprisingly, we find that when uncertainty is significant, and the threat of entry is strong but not detrimental, the platform has incentives to bias information toward investors' biased beliefs. This result demonstrates a case where competition and uncertainty may jointly lead to information bias. However, a properly designed uncertainty-resolution mechanism could reduce the incentive. Our findings contribute to the literature on the P2P lending market by analyzing platform decisions and offer policy implications for regulating P2P lending market.

A Perfect Storm

In an age where news information is created by millions and consumed by billions over social media ( SM ) every day, issues of information biases, fake news, and echo-chambers have dominated the corridors of technology firms, news corporations, policy makers, and society. While multiple disciplines have tried to tackle the issue using their disciplinary lenses, there has, hitherto, been no integrative model that surface the intricate, albeit “dark” explainable AI confluence of both technology and psychology. Investigating information bias anchoring as the overarching phenomenon, this research proposes a theoretical framework that brings together traditionally fragmented domains of AI technology, and human psychology. The proposed Information Bias Anchoring Model reveals how SM news information creates an information deluge leading to uncertainty, and how technological rationality and individual biases intersect to mitigate the uncertainty, often leading to news information biases. The research ends with a discussion of contributions and offering to reduce information bias anchoring.

Manipulações estatísticas e anomalias visuais: design de visualização de dados e reconhecimento de vieses estatísticos | Statistical manipulations and visual anomalies: data visualization design and statistical bias recognition

O mundo contemporâneo é caracterizado por um amplo volume de informações produzidas. Contudo, proceder a seleção e leitura dessas informações por meio de relatos de pesquisa ou de notícias ainda é um desafio. Entre os obstáculos presentes se destacam os vieses da informação, originados por tratamentos de jornalistas ou pesquisadores, ou mesmo provocados intencionalmente para subverter a representação da realidade a partir dos dados obtidos. Assim, o presente estudo visa discutir a interpretação de informações visuais em representações gráficas de cálculos estatísticos de modo a contextualizar alguns dos principais recursos visuais de enviesamento de pesquisa. Para tanto, aborda os principais modos de enviesamento em pesquisas a partir das representações da estatística e da visualização de dados e identifica alguns passos nos quais o enviesamento se traduz em informações visuais. A partir do levantamento realizado, sugere-se que a compreensão visual dos recursos de visualização de dados pode ao menos instigar a indagação do leitor acerca do possível viés.*****The contemporary world is characterized by a large volume of produced information. However, selecting and reading this information through research reports or news is still a challenge. Among the present obstacles stand out the information bias, originated by treatments of journalists or researchers, or even intentionally provoked to subvert the representation of reality from the obtained data. Thus, the present study aims to discuss the interpretation of visual information in graphical representations of statistical calculations in order to contextualize some of the main visual bias features of research. To this end, it addresses the main modes of search bias from statistical representations and data visualization and identifies some steps in which bias translates into visual information. From the study, it is suggested that the visual understanding of data visualization resources may at least instigate the reader's question about the possible bias.

DECISION MAKING BIASES AND ERRORS AMONG MUSLIM GROUPS AND SECTS IN BANGLADESH: A MANAGEMENT PERSPECTIVE

Different Muslim groups/sects in Bangladesh are very intolerant to other groups/sects where Islam teaches brotherhood and unity. This study tried to see the issue from a management perspective, especially in the area of decision making. The study tried to identify the decision-making biases and/or errors among Muslim groups/sects in Bangladesh, and their impacts on their decision-making process. The study adopted a model developed by Kieren Jamieson and Paul Hyland (2006). This study followed a qualitative approach. It interviewed 20 Islamic scholars and unity initiators who are working for establishing brotherhood and unity among Muslim groups/sects in Bangladesh. Guidelines have been used while conducting a face-to-face interview. Firstly, the study tried to find whether there are biases and/or errors in the decision-making process among different Muslim groups/sects in Bangladesh, and we found some serious biases and/or errors that can surely lead to biased/inappropriate decisions about other Muslim groups/sects. Secondly, the study tried to specifically find the nature and impacts of those biases and/or errors according to the research framework. It categorized those biases and/or errors in information bias, cognitive bias, risk bias, and uncertainty bias. Those biases and/or errors occurred in the information load, and in the decision-making process. Cognitive biases, the study found, have the most impacts on decision making. From the study, we developed a model to present the decision-making biases and/errors, and their impacts on decisions Muslim groups/sects in Bangladesh take.