Multilevel Explanation Research Articles

Stratified GNN Explanations through Sufficient Expansion

Explaining the decisions made by Graph Neural Networks (GNNs) is vital for establishing trust and ensuring fairness in critical applications such as medicine and science. The prevalence of hierarchical structure in real-world graphs/networks raises an important question on GNN interpretability: "On each level of the graph structure, which specific fraction imposes the highest influence over the prediction?" Currently, the prevailing two categories of methods are incapable of achieving multi-level GNN explanation due to their flat or motif-centric nature. In this work, we formulate the problem of learning multi-level explanations out of GNN models and introduce a stratified explainer module, namely STFExplainer, that utilizes the concept of sufficient expansion to generate explanations on each stratum. Specifically, we learn a higher-level subgraph generator by leveraging both hierarchical structure and GNN-encoded input features. Experiment results on both synthetic and real-world datasets demonstrate the superiority of our stratified explainer on standard interpretability tasks and metrics such as fidelity and explanation recall, with an average improvement of 11% and 8% over the best alternative on each data type. The case study on material domains also confirms the value of our approach through detected multi-level graph patterns accurately reconstructing the knowledge-based ground truth.

A Multi-Level Model of Institutional Innovation and Knowledge Creation Capabilities: Insights From the Development of Taiwan's Nation Health Insurance.

This study proposes a multi-level model of institutional innovation in the healthcare sector-in other words, field-level institutional change pressures that start as network-level institutional innovation by hospitals and government for their organizational performance, with an emphasis on the effect of organizational-level construct-knowledge creation capabilities. A case study using in-depth interviews and a historical inquiry approach has been used to qualitatively analyze our cases during the development of Taiwan's National Health Insurance (NHI). Our results propose a multi-level explanation of institutional innovation by showing how field-level institutional change pressures can stimulate the government's institutional innovation at the network level. Moreover, knowledge creation capabilities may positively influence the government hospitals' ongoing institutional change pressures induced institutional innovation activity for their performance at the organizational level in an institutional setting. This study contributes to health organization management researchers and administrators by developing explanations of institutional innovation and creating a much-needed multi-level insight into hospital behavior in the highly institutionalized healthcare sector.

Honor in the Wild

“Culture of honor” means that individuals deter others by signaling their commitment to violent retaliation. We develop a multilevel explanation of cross-level interdependence of honor and violence. According to our concept of system-level honor, a social system is loaded with deterrence signaling if culture of honor is highly prevalent in the system. In line with the Smith and Price (1973, in Nature,https://www.nature.com/articles/246015a0) model, we argue that high system-level honor discourages Prober-Retaliator behavior: some individuals might tend to challenge others they assume to be inferior to increase their own reputation. Both individual culture of honor and system-level honor contribute to an increase in violence (H1; H2). However, as system-level honor and deterrence become more prevalent, the impact of individual honor diminishes because engaging in violent behavior becomes increasingly expensive within such a system (H3). As a second contextual effect, inequality in culture of honor should therefore increase violent behavior because it encourages Prober-Retaliator behavior (H4). We analyze the effect of culture of honor on school violence among 15-year-old adolescents. Disentangling the micro- and context-level effects of culture of honor on violent behavior in a multilevel analysis framework allows the estimation of a cross-level interaction using a large data set from more than 25,000 adolescents in more than 1,300 schoolroom contexts. Results are in line with our H3, but not with H4. Model-based predictions show that the deterrent effect must be unrealistically high to generate an equilibrium of average violence.

A multilevel account of hippocampal function in spatial and concept learning: Bridging models of behavior and neural assemblies.

A complete neuroscience requires multilevel theories that address phenomena ranging from higher-level cognitive behaviors to activities within a cell. We propose an extension to the level of mechanism approach where a computational model of cognition sits in between behavior and brain: It explains the higher-level behavior and can be decomposed into lower-level component mechanisms to provide a richer understanding of the system than any level alone. Toward this end, we decomposed a cognitive model into neuron-like units using a neural flocking approach that parallels recurrent hippocampal activity. Neural flocking coordinates units that collectively form higher-level mental constructs. The decomposed model suggested how brain-scale neural populations coordinate to form assemblies encoding concept and spatial representations and why so many neurons are needed for robust performance at the cognitive level. This multilevel explanation provides a way to understand how cognition and symbol-like representations are supported by coordinated neural populations (assemblies) formed through learning.

Generating multi-level explanations for process outcome predictions

Process mining focuses on the analysis of event log data to build various process analytical capabilities. Predictive process analytics has emerged as one of such key capabilities and it uses machine learning techniques to construct process prediction models. In recent years, deep neural networks have gained increasing interest in process prediction since they can handle multi-dimensional sequential inputs with minimal information loss. However, they are considered black-box models and existing studies in explaining deep neural network-based process predictions rely on only event-level features for explanation. In this paper, we propose a new approach for generating explanations for process outcome predictions at multiple levels. The approach is underpinned by three different prediction models: a transparent model for generating global explanations based on case-level features, an attention-based deep neural network for generating local explanations based on event-level features, and a novel eXplainable Dual-learning Deep network (XD2-net) for generating local explanations based on case-level features. Using three publicly available datasets, we have tested the applicability of the approach and further examined the multi-level explanations generated by the approach through an elaborate case study. Unlike others, the design of our approach promotes the idea of leveraging the complementary capabilities of different models and utilizing their strengths, rather than focusing on model performance competition. This will contribute towards generating more comprehensive explanations that meet the needs of different end users and purposes in the future.

Green innovation in events: the role of institutional pressures, future orientation, and past experience

Green innovation is pivotal to events. However, little is known about how exhibitors manage the challenge of adopting green innovation in the fast-paced and short-oriented events. Drawing upon institutional theory and organisational learning perspective, this study investigates how institutional pressures are linked to exhibitors’ willingness to adopt (WTA) eco-exhibiting via two temporal variables: future orientation (mediator) and past experience (moderator). PLS-SEM is applied to analyse a survey from 685 event exhibitors. Results show that coercive and normative pressures are significantly associated with exhibitors’ future orientation, which then predicts WTA. Besides, past experience moderates the relationship between institutional pressures and WTA. Specifically, experienced exhibitors learn by doing and their WTA is more associated with normative pressure, while novice exhibitors learn vicariously and their WTA links stronger to mimetic pressures. This paper integrates macro-level institutional pressures and organisational-level learning heterogeneity, offering a multi-level explanation of green innovation adoption. This research also contributes to the green events and sustainable tourism literature with novel insights from the time perspective, verifying events as pro-environmental learning spaces for exhibitors. We highlight time perspective and environmental education for future research.

How do dynamic capabilities affect performance? A systematic review of mediators

Scholars constantly seek insights into the effects of dynamic capabilities (DC) on performance. As a result, researchers followed various paths, including microfoundations and multilevel explanations of DC, and propose different measures of and perspectives on performance, reflecting their distinct conceptual foundations. Consequently, debates about precisely how DC affect performance continue. In line with its theoretical origins, research on DC has recently begun to blossom around the indirect effects of DC. However, the field is highly fragmented. To advance this domain theoretically, a consolidated, systematic review is needed. To provide one, this article starts with a detailed analysis of the indirect effects of DC at individual, organizational, and environmental levels. Then, as a second step, this study provides a comprehensive consolidation of the underlying arguments and theoretical approaches used to predict that certain variables function as mediators. According to these findings, the authors identify research gaps and propose an instructive research agenda.

Four Decades of Challenges by Culture to Mainstream Psychology: Finding Ways Forward

Over 5 days at the Nag’s Head Conference Center, USA in 1987, social and cross-cultural psychologists discussed what would be required if research relating to culture were to gain greater attention from psychology in general, and in particular from what was perceived at the time as its mainstream. The criteria for gaining greater credibility laid down by three leading social psychologists proved daunting in relation to the cross-cultural work presented at the meeting but subsequently inspired cross-culturalists to “raise their game.” In this paper, we describe these crucial challenges and how they have been addressed more recently by cross-cultural psychologists. We assess the extent to which studies focused on cultural differences are now thoughtfully represented in social, personality, and organizational psychology by briefly surveying the content of a single year’s issues of the Journal of Personality and Social Psychology, Journal of International Business Studies, and the Journal of Personality in relation to the concurrent content of the Journal of Cross-Cultural Psychology. We identify the perils of assimilation to psychology in general by diluting the concept of culture and by tyrannizing research with over-specified criteria of statistical rectitude. We also identify studies published in top-rated journals that have nonetheless advanced our field. We reiterate the need for defensible measures of cultural difference and methods for identifying and examining them as a basis for multi-level explanations of cultural effects and cultural change. We conclude by proposing a gold standard for assaying cross-cultural studies of psychological processes and outcomes.

How science-based start-ups and their entrepreneurial ecosystems co-evolve: A process study

There is increasing recognition of the role that ecosystems play in supporting start-ups: particularly start-ups commercializing science-based innovations. In contrast to existing research that tends to focus on established ecosystems, this study uses evolutionary theory to provide a multi-level explanation for the emergence of nascent ecosystems. We employ a qualitative process-based methodology to investigate eight cases of science-based start-ups and their entrepreneurial ecosystems. Our empirical study allowed us to identify a range of retention mechanisms which enable the ecosystem to develop over time – but also, countervailing forces in the form of dissipative processes. Our theoretical contribution is a process model of entrepreneurial ecosystem emergence that makes explicit how retention and dissipation are simultaneous processes. Our model demonstrates the potential of evolutionary theory to understand the interdependence of firm-ecosystem development over time.

How We Tell Apart Fiction from Reality

Abstract The human ability to tell apart reality from fiction is intriguing. Through a range of media, such as novels and movies, we are able to readily engage in fictional worlds and experience alternative realities. Yet even when we are completely immersed and emotionally engaged within these worlds, we have little difficulty in leaving the fictional landscapes and getting back to the day-to-day of our own world. How are we able to do this? How do we acquire our understanding of our real world? How is this similar to and different from the development of our knowledge of fictional worlds? In exploring these questions, this article makes the case for a novel multilevel explanation (called BLINCS) of our implicit understanding of the reality–fiction distinction, namely that it is derived from the fact that the worlds of fiction, relative to reality, are bounded, inference-light, curated, and sparse.

ORGANISATIONAL BEHAVIOUR IN HUMAN RESOURCE DEVELOPMENT

Organizational behaviour is a discipline of social science that look for explanations for human behaviour in organizations. Organisational behaviour draws on core disciplines such as psychology, sociology, anthropology, economics, communication, and law to create and investigate multilevel explanations of why people take part in particular behaviour, and which behaviour under which situation lead to better outcomes in organizations. Created using an applied or pragmatic lens and tested with a wide range of both quantitative and qualitative methodologies, most Organisational behaviour theories and research have direct implications for managers and for other organizational participants. Not surprisingly, one focal area of Organisational behaviour research concerns safety in organizations, and a growing body of safety-oriented literature in Organisational behaviour is based on data collected during simulation training across a variety of organizations such as hospitals, airlines, nuclear power plants, and other high reliability organizations. Keywords: Organizational Behaviour, Human Resource Management, Human Behaviour

Adoption and continued use of mobile contact tracing technology: multilevel explanations from a three-wave panel survey and linked data

ObjectiveTo identify the key individual-level (demographics, attitudes, mobility) and contextual (COVID-19 case numbers, tiers of mobility restrictions, urban districts) determinants of adopting the NHS COVID-19 contact tracing app and continued...

Linking transparency and privacy to the impact of Enterprise Social Networks

Despite widespread diffusion of Enterprise Social Networks, research about such technologies’ impact on employee performance remains limited. While most of the research has focused on how ENS’s communication visibility increases knowledge about coworker’s knowledge leading to better performance, the research has highly overlooked the role of privacy at the workplace. Furthermore, most online privacy research has been performed in consumer settings. So, we don’t have a clear idea of how privacy will affect information sharing in the enterprise context. In this research, I answer how information privacy concerns from workers regarding ESN affect individual metaknowledge accumulation? I extend on Communication Visibility Theory by including elements of a behavioral theory of privacy and transparency. I explain how ESN’s implementation at the organization might lead to a higher awareness of being watched by others, triggering hidden behaviors based on privacy concerns. Consequently, I propose a multi-level explanation of how privacy affects the returns individuals can get for ESN usage. First, individual privacy concerns decrease ESN usage, decreasing their opportunities to take advantage of its potential. Second, as privacy concerns are serious and shared within a team, the accuracy of information provided on the ESN might decrease, decreasing individuals’ chances of getting accurate metaknowledge to improve their performance. This research expands current knowledge about the impact of ESN at the workplace, by bringing understanding to the role of privacy on the impact of Enterprise Social Networks. And it highlights the importance of clarifying how privacy and transparency jointly affect human behavior in the workplace.

Knowledge Sharing in Roadmapping: Toward a Multilevel Explanation

Roadmapping is a workshop-based tool that draws on diverse sets of knowledge from individuals to create shared knowledge in the form of a roadmap for navigating a strategic landscape. Theoretical explanations of knowledge sharing in roadmapping are underdeveloped, with only limited insights into the relationship between individuals’ knowledge (at the micro level) and organizational-level outcomes. This article uses a multilevel lens to examine knowledge sharing in two roadmapping projects empirically. The analysis identifies five mechanisms that enable or constrain knowledge sharing in roadmapping: 1) a neutral space for face-to-face interactions between individuals that fosters; 2) a positive disposition of individuals to share knowledge, together with (3) imperatives to trigger knowledge contributions. Individually shared knowledge is transformed into collective outcomes via 4) shared meanings and 5) shared commitment to act. The analysis also highlights the important role of the facilitator in relation to knowledge sharing. This article extends the theoretical understanding of individuals and their behavior in the roadmapping process. In doing so, it demonstrates how microlevel actions relate to macrolevel outcomes. Practitioners, including facilitators, can use these insights as they design and deliver roadmapping interventions to improve knowledge sharing and, therefore, better strategic outcomes.

Grand Challenges, Covid‐19 and the Future of Organizational Scholarship

Grand Challenges, Covid‐19 and the Future of Organizational Scholarship

Międzyosobowy poziom wyjaśniania umysłu i zdolności poznawczych

What is the influence of the social and (to a lesser extent) the cultural on mind and cognition? The question will concern: (i) the nature of the relation between what is social and cultural, on the one hand, and the mind, cognitive ability, and cognitive development on the other; (ii) aspects of the mind, cognitive ability, and cognitive development affected by what is social and cultural; (iii) processes in which what is social and cultural becomes a part of individual cognitive functioning. This paper attempts to provide a conceptual framework within which the above questions can be answered. In the literature concerning the explanation of mind and cognitive abilities, attention is drawn to the importance of distinguishing levels of explanation. One of such approaches to multi-level explanation of the mind and cognition is the distinction drawn by Daniel Dennett between the personal level of explanation (e.g., level of beliefs, desires) and the subpersonal level of explanation (e.g., neural or computational). The purpose of this article is to argue that — recognising that the social and the cultural are one of the constitutive conditions of mind and cognition — we should posit a third level: the interpersonal level of explanation (ILE). The reason for postulating ILE is that there are specific domains in which interpersonal interactions affect elementary cognitive abilities and processes, which may thus result in their transformation into higher cognitive abilities and processes. The entire group of such domains will be defined as ILE, and at the centre of its explanatory potential will be the different types and forms of the said transformation. One of the main research objectives pursued at the ILE is to identify and formulate a taxonomy of transformative features and processes. Two models of such processes will be presented: the Vygotsky model and the Tomasello model. In the final part of this paper, the basic methodological assumptions of the ILE will be presented: the relative autonomy of this level of explanation, multi-level analysis, and the mechanism-based explanation. These assumptions will also be used to formulate general theses of the ILE.

Legitimizing the Judiciary: a Multilevel Explanation of Factors Influencing Public Confidence in Asian Court Systems

The study analyzed large-scale data from several sources, covering more than 42,000 individuals in 28 Asian countries to comparatively assess the effects of country-level variables on Asians’ confidence in their court systems. Three HLM ordinal logistic regression models were fitted instantaneously to achieve the objective of the study. Results revealed that a country’s levels of democracy and peacefulness had a significant and positive relationship with citizens’ confidence in the courts. Moreover, citizens living in countries with absolute press freedom reported greater levels of confidence. However, individuals living in countries that were highly corrupt tended to have reduced confidence in the courts. The impact of terrorism on a country had no influence on public confidence.

Donative labor effect of the nonprofit pay: A multilevel explanation

AbstractStudies on the nonprofit pay differential find that nonprofit workers in the child day care industry earn more than comparable for‐profit workers, whereas nonprofit lawyers earn less than lawyers in for‐profit firms. Are nonprofit day care center workers less altruistic than for‐profit day care workers or nonprofit lawyers? The answer is yet to come. The study aims to test the donative labor hypothesis, which is derived from altruistic motivation. To estimate the donative labor effects on the individual level, I apply cross‐classified multilevel modeling to disentangle the confounding effects on the industry level and occupation level. Data are pooled from Census 2000 and American Community Survey 2005–2016 to provide individual, industry, and occupation information. Industries are cleaned based on the Statistics of Income data from the National Center for Charitable Statistics. My analysis finds that nonprofit workers earn 4.3% less than comparable for‐profit workers net of industry and occupation effects. The study contributes to elucidating the explanatory levels of different theories. The random‐effects modeling has established an exhaustive inventory of nonprofit pay differential across industry and occupation levels.

A Multi-level Model to Understand Cervical Cancer Disparities in Appalachia.

The Appalachian region experiences higher incidence and mortality due to cervical cancer compared with other regions of the United States. The goal of the Ohio State University Center for Population Health and Health Disparities (CPHHD), called the Community Awareness Resources and Education (CARE) project, was to understand reasons for this disparity. The first wave (2003-2008) of funding included three projects focusing on the known risk factors for cervical cancer, lack of screening, smoking, and infection with human papillomavirus (HPV). On the basis of the results of these projects, the second wave (2011-2017) included four projects, designed to address a multi-level model of factors contributing to cervical disparities in Appalachia. The results of these projects were then used to refine a multi-level model that explains cervical cancer disparities in Appalachia. Future funded projects will take these multi-level explanations for cervical disparities and focus on implementation science strategies to reduce the burden of cervical cancer morbidity and mortality in Appalachia.See all articles in this Special Collection Honoring Paul F. Engstrom, MD, Champion of Cancer Prevention.

Correctional rehabilitation and human functioning: An embodied, embedded, and enactive approach

Theories of crime are based in an underlying understanding of human functioning. In this paper we argue that current theories of correctional rehabilitation imply a model of functioning which is reward - oriented , multifactorial , norm - based , and non - agential . This approach is seen as affording limited explanatory value, leading to problems with treatment efficacy. We outline an alternative perspective of functioning as being embodied , embedded , and enactive (3e). 3e places an emphasis on the individual as an embodied whole, in an adaptive relationship with their physical and social environment. We argue that 3e prioritizes the experience and agency of the individual, with a commitment to viewing a person as a functional whole drawing on comprehensive multilevel explanations. 3e places a strong emphasis on the role of individual affectivity and emotion as a core component of functioning. Finally, implications of the 3e model for correctional practice are discussed.