Efficient and Robust Block Designs for Order-of-Addition Experiments

Chang-Yun Lin · 2025

Read Efficient and Robust Block Designs for Order-of-Addition Experiments by Chang-Yun Lin (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-07096.txt.

Introduction

Chang-Yun Lin designs blocked experiments where the order of adding parts changes the result, in this 2025 paper. Such order-of-addition studies matter in fields from cooking to chemistry. When test units fall into unlike groups, blocking is needed to keep group differences from polluting the results.

Blocked designs for this setting were little explored. Lin extends a bookkeeping framework to cover them and proposes a word length pattern as the score for choosing robust designs. Algorithms built on Latin squares cut the search cost sharply. Checks show the designs keep effects from tangling with each other. Simulations match the power of complete blocked designs at a fraction of the effort.

Catalog record

CREATORChang-Yun Lin
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE78 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-07096

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Efficient and Robust Block Designs for Order-of-Addition Experiments Chang-Yun Lin Abstract Designs for Order-of-Addition (OofA) experiments have received growing attention due to their impact on responses based on the sequence of component addition. In certain cases, these experiments involve heterogeneous groups of units, which necessitates the use of blocking to manage variation effects. Despite this, the exploration of block OofA designs remains limited in the literature. As experiments become increasingly complex, addressing this gap is essential to ensure that the designs accurately reflect the effects of the addition sequence and effectively handle the associated variability. Motivated by this, this paper seeks to address the gap by expanding the indicator function framework for block OofA designs. We propose the use of the word length pattern as a criterion for selecting robust block OofA designs. To improve search efficiency and reduce computational demands, we develop algorithms that employ orthogonal Latin squares for design construction and selection, minimizing the need for exhaustive searches. Our analysis, supported by correlation plots, reveals that the algorithms effectively manage confounding and aliasing between effects. Additionally, simulation studies indicate that designs based on our proposed criterion and algorithms achieve power and type I error rates comparable to those of full block OofA designs. This approach offers a practical and efficient method for constructing block OofA designs and may provide valuable insights for future research and applications. KEY WORDS : Component orthogonal array, indicator function, position-based, orthogonal Latin squares, word length pattern 1 Introduction Designs for Order-of-Addition (OofA) experiments have received increasing attention in recent years. In these designs, the response is influenced by the sequence in which components or materials are added. An early example of an OofA experiment is Fisher’s (1937) study, where a lady claimed she could discern by taste whether the milk or the tea was added first to the cup. Such experiments have since been applied in many areas, including medical research (Black et al., 2001), biochemistry (Preuss et al., 2009), measurement science (PerkinElmer, 2016), and task arrangements (Voutchkov et al., 2005; Wilson et al., 2018). Several models have been proposed in the literature for analyzing data from OofA experiments. The most widely used model is the pairwise ordering (PWO) model, first proposed by Van Nostrand (1995), in which the response is fully accounted for by the pairwise ordering of the components. Design constructions under the PWO model have been extensively discussed and developed, often utilizing the D criterion to search for optimal OofA designs. Notable contributions include the works of Voelkel (2019), Voelkel and Gallagher (2019), Lin and Peng (2019), Mee (2020), Chen et al. (2021), Tsai (2022), Wang and Mee (2022), Zhao et al. (2022), and Xiao et al. (2024). These studies have significantly advanced the methodology for crafting efficient OofA designs under the PWO model, improving the capacity to analyze the influence of component order on response variables. In contrast to the PWO model, which considers the relative positions of each pair of components, Yang et al. (2021) introduced the component-position (CP) model, which emphasizes the absolute positions of the components. They introduced the concept of component orthogonal arrays (COA), which exhibit high D-efficiency for estimating parameters of the CP model. Stokes and Xu (2022) further advanced OofA designs by presenting them with position matrices instead of component matrices. Building on these position matrices, they introduced the position-based (PB) approach and developed position models. This approach significantly reduces the number of parameters required in position models compared to the PWO and CP models. In many situations, OofA experiments may not be conducted with a homogeneous group of units. These experiments might involve diverse patient groups or be performed in different environments. As a result, variations arising from group-to-group or environment-to-environment differences can influence the analysis outcomes of the OofA experiment. For instance, in biochemistry, environmental factors such as humidity or equipment variability can influence experimental results. In measurement science, differences in calibration, maintenance, or performance between different instruments can introduce variations in measurement data. To mitigate the influence of these variations, blocking is often employed. This involves organizing units into blocks such that the variation within each block is notably smaller than the variation observed between blocks. By doing so, the effects of the blocks are controlled, allowing for a clearer comparison of the effects of the components. A well-recognized real-world application of block OofA experiments is the five-drug study in the pharmaceutical industry described in Mee (2020). This study utilized two blocks of 20-run drug sequences to mitigate batch-to-batch variations. Wang et al. (2020) also conducted a similar experiment using three batch blocks of 20 runs each. These studies highlight the essential role of block OofA designs in real-world applications. However, methods for constructing such block designs and the criteria for their evaluation remain largely unexplored in the literature. As experiments become increasingly complex, addressing this gap is essential to ensure that designs accurately capture the effects of the addition sequence and effectively handle associated variability. Motivated by this, this paper aims to bridge the gap by leveraging position models to introduce a suitable criterion for assessing block OofA designs and developing a method for constructing highly efficient and robust block OofA designs based on the proposed criterion. To achieve this, we utilize the indicator function introduced by Cheng and Ye (2004), originally developed for factorial designs. We find that position models align with the framework of the indicator function and thus adapt this concept to tailor the indicator function specifically for OofA designs. Additionally, we incorporate block factors into the indicator function and develop word length patterns to guide the selection of optimal block OofA designs. The remainder of this article is organized as follows. In Section 2 , we create an indicator function based on the framework of position models for OofA designs. In Section 3 , we extend the concept from Section 2 to develop an indicator function for block OofA designs and propose a criterion based on this indicator function to evaluate these designs. In Section 4 , we present algorithms for selecting optimal block OofA designs using the criterion proposed in Section 3 . Several examples are provided to illustrate the process of obtaining these optimal designs. In Section 5 , we conduct a case study to demonstrate the analysis of the block OofA experiment using the proposed methods and the constructed designs. In Section 6 , we evaluate the performance of the proposed block OofA designs using correlation plots and simulation studies. Section 7 provides the concluding remarks. 2 Position models and indicator function for OofA designs We employ the position models proposed by Stokes and Xu (2022) for OofA designs, as these models involve fewer parameters compared to the PWO and CP models. In this section, we first review OofA designs and position models, and then establish a framework for the indicator function for OofA designs, adapted from the framework proposed by Cheng and Ye (2004) for factorial designs. The modification of the indicator function for block OofA designs will be developed in the Section 3 . 2.1 The OofA designs Table 1 : Two 3-component OofA designs with 6 runs (a) D_{1} Run Z_{1} Z_{2} Z_{3} 1 1 2 3 2 1

Read the complete text