We consider multiple comparison tests for comparing several treatments with a control in a randomized block design under simple ordered restrictions. Williams (1971) proposed a closed testing procedure based on similar to two-sample t-tests under assuming normality in multi-sample models. We propose closed testing procedures based on similar to Shiraishi and Matsuda (2015), the maximal component of the cumulative chi-squared statistic, and linear statistic. The powers of the proposed tests are investigated in Monte-Carlo simulation studies. As a result, closed testing procedures based on and the maximal component of the cumulative chi-squared statistic are superior to the similar to Williams (1971).

Multi-stage modeling is an essential method for analyzing consumer behavior. Purchasing decisions often proceed through multiple stages, such as category purchase, brand choice, store choice, and retail format choice. Understanding these stage-specific processes and systematically analyzing them using statistical models is crucial for uncovering consumer behavioral characteristics and decision-making factors. This paper organizes the theoretical background of multi-stage modeling and primarily reviews its applications in the marketing field. Specifically, it provides an overview of prominent methods used in multi-stage modeling, including discrete choice models such as logit and probit models, hierarchical Bayesian models, and mixed logit models. It also addresses practical data formats, such as scanner panel data, which are commonly utilized for data collection and analysis in real-world applications. Furthermore, this paper examines approaches to integrating multiple stages into unified models, with a focus on methods that consider interdependencies between categories and heterogeneity among consumers. These insights have potential applications not only in marketing strategies but also in other fields, such as transportation behavior analysis and travel planning. The objective of this paper is to reevaluate the role of multi-stage modeling and clarify its practical value across various domains.

Does the concept of ‘flow’ exist in baseball games? Here, ‘flow’ refers to the phenomenon where a single play influences the subsequent development of the game. Several previous studies suggest that ‘flow’ in baseball is merely a cognitive bias. However, few studies have clearly defined the factors that generate or alter ‘flow’ and incorporated them into their analyses. In this study, we re-evaluate a well-known baseball adage, “opportunities arise after overcoming adversity,” using data from Japanese high school baseball tournaments. Using the Leverage Index to quantify high-pressure situations, we propose an analytical approach that considers “being right after a tough half-inning or not” as a treatment variable. Employing on-base percentages and scores as outcome variables, we examine the treatment effect through model selection and regression analyses. We also examine our results with and without the application of propensity score matching.

The goal of fairness-aware machine learning is to analyze data while taking into account potential issues of fairness, discrimination, neutrality, and/or independence. This survey shows the causes of unfairness in the use of machine learning, formal criteria to define the fairness in machine learning, and the methods to discover unfairness and to mitigate the unfairness in prediction.

One major challenge in using machine learning is that models often work like “black boxes”, that is, it is hard for humans to understand how they make decisions. To tackle this issue, there has been growing interest in research focused on making machine learning models more explainable. In this paper, we review some of the main approaches to model explanation. Recently, however, studies have shown that these explanation methods may not always be reliable. We also reviews recent research on the reliability of these explanations.

DP (Differential Privacy) is known as a de facto standard for privacy metrics when performing statistical analysis while protecting individual privacy. It has been adopted in industry and government. This paper provides a brief explanation of the basics of DP, including its definition, the type of privacy it guarantees, and the models in DP.