Temporal Check All That Apply

© 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 1 Preprint. Temporal Check All That Apply (TCATA) Glenn Birksø Hjorth Andersen †[0000−0002−7698−1146], Jonas Yde Junge †[0000−0002−6319−6894], and John C. Castura[0000−0002−1640−833X] † These authors contributed equally to the manuscript. Abstract Eating and drinking food or beverages usually extends over a time period of seconds to minutes. How sensations evolve over time is important in some product categories, but not well captured by static sensory methods. The temporal check-all-that-apply (TCATA) method captures how several product attributes simultaneously evolve over time. This chapter covers important considerations regarding the choice of assessor type, product set, and attributes. It describes how to design and conduct the TCATA study. The data analysis section addresses pre-processing steps, assessor performance evaluation, and visualization of the product’s temporal profile as well as multiple analytical approaches including count analysis, factorial design analysis, and penalty-lift methodology for linking temporal perceptions to consumer acceptance. A beer case study demonstrates the application of TCATA in real sensory research. Keywords: sensory temporal method, applicability, perception dynamics, sensory evaluation © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 2 1. Introduction When eating or drinking, a single intake of food or beverage extends over a time period of seconds to minutes, depending on factors including oral processing requirements, mouth coating, and aftertaste, including any retronasal aromas, trigeminal effects, or other perceptions [28]. How sensations evolve over the full intake experience is important in some product categories, but not well captured by traditional sensory methods that measure perceptions at one or more discrete time points or integrated over the full intake experience. Continuous temporal methods aim to capture dynamic changes in sensory perception uninterruptedly. One such method is temporal check all that apply (TCATA; [13]), which can be used to track the evolution of multiple sensory attributes simultaneously over a relatively short time course, typically lasting 30 s to 180 s. The goal of this chapter is to provide practical information describing why, when, and how the TCATA method can be applied. TCATA extends the static check-all-that-apply (CATA; [32], [4]) question that presents a list of attributes to assessors who check the attributes that describe the sample under evaluation. In the TCATA method, assessors have the task of continuously checking attributes such that at any moment the selected attributes describe the sample at that time point. TCATA data are then aggregated to obtain a temporal profile of each product encompassing all the applicable sensory attributes. Products from a wide range of product categories have been evaluated using the TCATA method. The TCATA method has been applied to assess the temporal profiles of food and beverage products, as well as non-food products. The TCATA method has been performed by trained assessors, as well as untrained consumers [9]. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 3 Associations between TCATA and other types of data can be investigated. For example, TCATA evaluations can be conducted alongside temporal hedonic assessments to provide insights into which sensory attributes influence liking positively or negatively at different moments during consumption. Combining temporal sensory and hedonic measurements enables a deeper understanding of how the dynamic sensory characteristics of a product shape the consumer’s hedonic response over time. Collecting static hedonic ratings or purchase intent responses allow for exploring temporal drivers of these responses. Abbreviations: CATA, check all that apply; PCA, principal component analysis TCATA, temporal check all that apply; TDS, temporal dominance of sensations 2. Aim Temporal check all that apply (TCATA) is a sensory method that aims to continuously capture dynamic perceptions of multiple sensory attributes over relatively short time periods (e.g., usually up to 3 minutes) during product consumption. 3. Materials This section outlines the main considerations for designing a TCATA experiment, including the selection of assessors (Section 3.1), products (Section 3.2), and attributes (Section 3.3). Data collection recommendations are also given (Section 3.4). 3.1. Assessors The simplicity of the TCATA task allows for a broad range of applications. TCATA can be conducted with either trained assessors or untrained consumers. Which type of assessor © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 4 to use depends on the research objectives. In the subsections that follow, we discuss how assessor choice affects the study. 3.1.1. Trained assessors Many TCATA studies are conducted with trained assessors. In such studies, trained assessors are typically screened for their sensory abilities, trained to describe samples using a shared lexicon, and practice using rating scales to quantify the sensory characteristics of products. A strength of using trained assessors is their consensus understanding of attributes, which is achieved through practice, discussion, and exposure to attribute references or standards. Their aligned use of descriptive terminology provides interpretability benefits. TCATA studies can be conducted with relatively few trained assessors. Trained assessors are often scheduled for multi-day studies, which they attend since they are motivated and have an ongoing relationship with the research lab. Typically, a trained panel is comprised of 9 to 16 trained assessors, where each assessor evaluates each product two or three times in as many sessions, but more or fewer assessors and sessions can be used. 3.1.2. Consumers Many TCATA studies are conducted with untrained consumers. The responses of untrained consumers provide insight into the diversity of consumer perceptions, with the drawback that consumers might lack a shared understanding of the attributes used to describe the samples. Since the aim is to understand perceptions within the target consumer population or its segments, participants are typically recruited according to predetermined criteria. Quotas are commonly used so the consumer sample matches the target consumer population in relevant characteristics, such as gender and age. Usually, consumers are tasked with evaluating products using a short list of simple, familiar attributes that consumers would understand spontaneously © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 5 (e.g. sweet, sour, lemon, chocolate, hard), without requiring detailed explanations or exposure to references. Testing with consumers allows for understanding the perceptions of the target user population, including the variability and heterogeneity of their perceptions, as well as the opportunity to explore the relationship between dynamic perceptions and specific outcome measures, such as hedonic responses and purchase intentions [9]. A drawback of testing with consumers is they may lack a common understanding of attribute identities (i.e. they do not agree on what the attributes mean and therefore interpret them differently) and may provide imprecise responses due to lack of practice with the methodology [9]. Consumers are usually scheduled to attend only a single session. If consumers are required to attend a multi-session study over multiple days, the researcher should consider how to avoid attrition problems arising from no-shows after the first session. TCATA consumer studies usually include approximately 50 to 100 consumers, but either fewer or more consumers are sometimes used. When conducting TCATA with consumers, each consumer usually evaluates each product at least once. 3.2.Products The researcher must consider how many products to include in the study. Usually, all products are from the same product category and share many attributes (Section 3.3). Since usually all assessors evaluate all products in each session and enough data must be acquired to obtain meaningful results, the number of products is often limited to the number of samples an assessor can reasonably evaluate in a single session. Cognitive fatigue, sensory fatigue, sensory adaptation, and inebriation are examples of factors that place a practical upper limit on the number of samples that can be reasonably evaluated per session. For this reason, what constitutes an appropriate number of products depends on the product category. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 6 Additionally, products might in fact be prototypes or formulations based on a factorial design [1, 26, 36]. In such studies, analysis of variance can investigate how the levels of different factors and their interactions affect citation proportions over time. Contrast effects [28] can influence how a particular sample is characterized. For example, a slightly bitter product will usually be described as bitter more often if other products are not bitter than if other products are intensely bitter. Subtle differences between typical products might be obscured if overshadowed by products with extreme sensory profiles. For this reason, it is recommended to select the product set carefully. In most cases, select relevant, comparable products within a reasonable sensory range that provide information with practical value to the researcher. 3.3. Attributes The TCATA method requires curating a list of relevant attributes that will describe and differentiate the products. Usually, the number of attributes is 15 or fewer [23]. The attributes selected must be appropriate for the type of assessor. In consumer studies, they should be easily understood by the average participant. Often, attributes are identified through a pilot study in which consumers provide free comments describing the products in the set. Using a consumer panel to generate attributes for a TCATA consumer study helps to ensure only consumerrelevant terminology is used. In trained-panel studies, more technical attributes can be supported with definitions and standard references or exemplars. Attributes used by a trained panel will often originate from panel discussion, the scientific literature, lexicons for the product category, or previous studies on similar products. Attributes from multiple sensory modalities are sometimes presented together in a single list. For example, consumers might evaluate samples using a list of attributes that includes tastes, flavours, textures, and other in-mouth sensations related to pain or temperature. Attributes might © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 7 also be separated into different lists; for example, a consumer panel might evaluate samples based on emotion terms in one list and sensory terms in another list. Emotion and sensory terms might also be evaluated in different sessions. A panel might be tasked with evaluating only one sensory modality at a time. In this case, the attributes presented will pertain only to the sensory modality under evaluation. For example, inmouth texture evaluations by a trained panel would include only relevant texture attributes and omit attributes related to taste and flavour. 3.4. Data collection Conducting a TCATA study requires computerized data collection. The system must be able to present the question with attributes in a manner that allows for the necessary data to be collected. These data include the time when the assessor starts and stops each evaluation and which attributes are selected in the interim. Vendors may offer differ features, such as attribute fading and pop-up instructions, that provide additional flexibility. TCATA evaluations are mediated by the device screen. For this reason, the researchers should enforce minimum requirements for the device and screen display. The screen must be large enough to display all aspects of the TCATA question clearly, since attributes not visible on the screen are unlikely to be checked as often as attributes that are visible. The device should allow assessors to respond in an unimpeded manner. A benefit of testing in a controlled central location is the opportunity to optimize the sensory ballot for display on the devices used. If assessors bring their own devices, then the ballot might in some cases render sub-optimally if devices are not checked or minimum device requirements are not enforced. In this case, the researcher could pre-test the ballot in horizontal (landscape) and vertical (portrait) smartphone orientations, then provide consumers with smartphone orientation instructions or recommendations. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 8 4. Task and procedures This section provides an overview of the TCATA task and procedural decisions involved when conducting a TCATA experiment. It details the assessor task (Section 4.1), the method for unchecking selected attributes (Section 4.2), provides guidance for presenting samples (Section 4.3), presenting attributes (Section 4.4), question layout (Section 4.6), task instructions and suggestions for orientation (Section 4.5), task design and interaction (Section 4.6), evaluation duration (Section 4.7), and specifies intake procedures for reliable data collection (Section 4.8). 4.1.TCATA task The researcher must ensure assessors understand and are ready to perform what is required to evaluate a single sample using the TCATA procedure. 1. Comprehend attributes - Assessors must be familiar with the attributes provided for the products. There should be an opportunity to clarify attribute meanings before the evaluation begins. The assessor must also be familiar with the position of the attributes on the screen since the task requires attributes to be checked promptly. This could be achieved by presenting 2. Comprehend task - Evaluation instructions must clearly communicate what is expected. For example, in a conventional TCATA variant, assessors are instructed to check attributes that describe the sample (or its effect on the assessor), then uncheck selected attributes that no longer describe the product (or its effect on the assessor). Especially with trained assessors, the task might be to check sensations that are present and uncheck selected attributes if that sensation becomes absent. We would expect low-intensity attributes to be checked more often if using presence-absence instructions than if assessors are asked to characterize or describe a product, since in the latter case assessors might tend to check attributes only if the attribute © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 9 intensity is above some baseline for the product category. What baseline is meaningful will depend on the product category. 3. Start evaluation - Assessors should know when to press the Start button and when to begin the sample evaluation. 4. Track changes - Assessors should understand the task is to track the attributes that describe the product over the time course of the evaluation. Attributes that describe the sample should be checked when they are noticed. Assessor should also know how attributes are unchecked, which will be discussed shortly (Section 4.2). 5. Indicate events - Assessors should know which events (e.g. swallow, expectorate) need to be indicated during evaluation and make these indications where appropriate. This feature is typically available in the TCATA software and is shown as pop-up boxes during evaluation. 6. Stop evaluation - Assessors should know to press the Stop button if they are allowed to end the evaluation when no attributes describe the product. Otherwise, the evaluation will stop when the evaluation time has elapsed. 4.1.1. Should a warm-up sample be used? Besides providing thorough instructions, familiarization can be achieved with a preevaluation trial with one or more warm-up samples. Usually, every assessor gets the same warm-up sample, which is a product similar to the other products under evaluation. If a reference product exists, then this reference product provides an obvious choice for the warmup sample. The warm-up sample should be the same for all assessors. Using a warm-up sample has advantages related to data quality. First, it helps assessors become familiar with positions of the attributes on the screen. Second, it gives assessors the opportunity to practice following the TCATA procedure. Third, it removes the first-sample- © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 16 assessors to chew and swallow naturally. When evaluating liquid foods, assessors might be instructed to swirl the sample in the mouth for a given time. When evaluating solid foods, it might feel unnatural for assessors if they are given detailed oral processing instructions. Assessors might be asked to swallow at a specific to capture what is in-mouth evaluation and what is aftertaste evaluation. Assessors can be asked to indicate when they swallowed, but compliance may vary. • Multiple intakes - If samples will be evaluated over a series of intakes (e.g. multiple bites, sips, spoonfuls, etc.), then a multiple-intake evaluation procedure should be used [5, 14, 43]. Often, the procedure starts when the first intake (bite, sip,etc.) of the sample is put into the mouth. In some cases, the evaluation proceeds with an instruction to swallow at a specific time, a delay, a new intake of the sample, evaluation, and instructions to swallow after a given time. This procedure can be repeated as required. In other cases, the TCATA duration elapses at the end of the first intake, followed by a delay and a new TCATA question corresponding to each subsequent intake. Multiple intake procedures can be extended to whole product evaluations [43] 4.9. Best practices and troubleshooting This subsection includes general Dos and Don’ts when running TCATA as well as other product-based sensory studies. Dos • Do perform a dry run: It is usually advisable to conduct a dry run of a TCATA study with a few people who are not part of the evaluation panel. A dry run might identify aspects of the TCATA study that require refinement. For example, a dry run might reveal the instructions are unclearly communicated or difficult to follow, that TCATA Fading times are too slow or too fast, or other problems that could be identified and corrected before the evaluation begins. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 17 • Do familiarize assessors with the evaluation procedure. Assessors should be familiarized with the testing protocol before starting the actual evaluation [23]. • Do present samples using monadic sequential presentation format. This presentation format, in which samples are presented, evaluated, then removed, followed by a delay before the next sample is presented evaluation, is recommended. The delay helps to reduce carryover effects and the presentation format allows each sample to be evaluated without the opportunity for direct comparison with previous samples. Don’ts • Don’t forget best practices. A researcher who is implementing the TCATA method for the first time might get caught up what differs from other sensory procedures, forgetting most aspects are the same. Following best practices for sensory evaluation is key. As always, be absolutely certain to track which products have been assigned which blinding codes. 5. Results and analyses Data from a single TCATA evaluation represent the responses over time from one assessor for one sample. The data can be organized into a two-way matrix with rows representing attributes and columns representing time points at 0.1-s or 1.0-s intervals. The matrix contains Bernoulli data, where 1 indicates the attribute was checked at that time point and 0 indicates it was not checked. 5.1.Pre-processing TCATA data 5.1.1. Trimming data Assessors are instructed when to press Start and when to end the evaluation by pressing Stop. In an evaluation, the assessor’s first attribute might be checked almost immediately after pressing Start or there might be a long delay. In some cases, these differences may arise from © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 18 individual variation in the time taken to perform the initial tasting instructions; for example, one assessor might check the first attribute 1 s after pressing Start, whereas another assessor might check the first attribute 5 s after pressing Start. Whether the second assessor has struggled to identify the first attribute or has failed to follow the directions is unclear. In other cases, assessors may be instructed to press Start but not begin their evaluation immediately; for example, in a wine evaluation, assessors were instructed to press Start when sipping the sample, hold it in the mouth, then expectorate and begin evaluating after 10 s [5]. Reviewing the raw data is strongly recommended; such reviews may reveal some assessors failed to follow the instructions provided [10]. Data trimming can be applied to TCATA data on a per-evaluation basis. Data can be “left-trimmed” by deleting time between Start and the first attribute was checked. Data can be “right-trimmed” by deleting the time between the deselection of the last selected attribute and the Stop time. Data are often kept untrimmed. If data are trimmed, then usually both leftand right-trimming are applied before any time standardization, which is described next. 5.1.2. Standardizing time Time standardization attempts to align temporal sensory data by converting the time duration in each evaluation to quantiles. Typically, time standardization finds percentiles on a scale from 0 to 1. Time standardization can be applied to either to data that has been trimmed or to raw data that is untrimmed. Some researchers apply time standardization routinely, but we strongly discourage its routine application. Following rationale given by [9], we advocate time standardization be applied judiciously and often not at all for the following reasons. First, the duration when sensations are perceived might be substantially longer in some products and shorter in other products. For example, [29] found wines with higher carbonation levels had longer durations of © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 19 attribute citation than wines with lower carbonation levels. Standardizing every evaluation to the same timescale will distort the test results by removing this carbonation effect. Second, the assumption that time standardization can align the perceptions of assessors who differ in oral processing efficiency is probably incorrect; for example, [19] showed the food bolus in slow chewers differs structurally from the food bolus in fast chewers at the time of swallowing, which leads to different sensory perceptions. In this case, time standardization would align swallowing events but obscure individual differences that can only be understood in unstandardized units of time. For this reason, the rationale used to justify routine time standardization is probably wrong and introduces a time-warping that obscures product differences without aligning the individuals on their oral processing or sensory trajectories. 5.1.3. Filling gaps In TCATA Fading (see Section 4.2), attributes that remain applicable must be rechecked before they fade to a deselected state. Gaps in the temporal profile of an attribute arise if assessors do not re-check a fading attribute quickly enough. If TCATA data are interpreted as a literal representation of what was perceived, then a very short gap must indicate the attribute was applicable before the gap, not applicable for the very short duration of the gap, then applicable again. A plausible explanation for a short gap is the attribute was applicable before, during, and after the short gap, where the gap arose from a fast-moving evaluation in which the assessor failed to re-check the attribute while it was fading [3]. For this reason, one might want to consider filling short gaps in the temporal data. Judgment will be required when considering which gaps are short enough to fill. Filling gaps [41] may restore the temporal profile to a more regular temporal curve. Filling gaps have been shown to increase the duration of significant differences between products [38]. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 20 5.2.Assessing performance In sensory studies, the goal is often to determine the size and direction of product differences. Proficiency is measured to determine whether the assessors agree in their TCATA evaluations and whether the assessors and panel are responding repeatably if the same product is evaluated more than once. These measurements can reveal why product differences might have been missed or, alternatively, can add confidence in conclusions made about the effects observed. 5.2.1. Assessing agreement and repeatability Assessor agreement is the extent to which an assessor’s responses align with the rest of the panel. Agreement is nearly always of interest in sensory evaluation regardless of the panel type. Assessor repeatability is the consistency of responses across replicate evaluations. Assessor repeatability is of interest when (usually trained) assessors provide repeated evaluations of one or more products. Repeatability and agreement will be quantified using methods proposed by [13]. These methods will be described assuming the data are from only one attribute. • Assessor repeatability: Data from each session are organized into a product-by-time matrix. The mean absolute difference between each pair of session matrices is obtained. Their mean is the assessor repeatability. • Assessor agreement: First, a product-by-time matrix is obtained for each assessor by averaging data over sessions. The mean absolute difference between an assessor and the rest of the panel is obtained. The agreement of an assessor with the rest of the panel is one less this difference. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 21 Averaging assessor repeatability across attributes gives an overall measure of repeatability for this assessor. Averaging assessor agreement across attributes gives an overall measure of agreement of this assessor with the other assessors. These measures can be adapted to investigate either a subset of products, attributes, assessors, or time durations. The degree of consensus is often hight when citation rates are low, as often occurs at the very start and at the end of TCATA evaluations when most attributes are unchecked so data are sparse. Data sparsity is, in fact, a type of consensus, as discussed previously by [34]. Other approaches, which will not be discussed here, include using alternative similarity indices combined with inference testing [21], Gwet’s AC1 coefficient [22] to evaluate chancecorrected repeatability [34] and Cronbach’s alpha coefficient [18] to evaluate the internal consistency of a panel’s TCATA responses [8]. Two approaches based on the intraclass correlation coefficient (ICC) have been proposed, where each one uses a different type of ICC. One approach treats TCATA responses for each product, attribute, and time point as yes-no data, then quantifies the reliability of these yes-no responses using the ICC [8]. Another approach treats TCATA responses for each attribute and time point as CATA data, then quantifies the reliability of the panel in discriminating the products using the ICC [39]. Interested readers are referred to the respective publications. 5.3.Analyzing citation frequencies and citation proportions The most straightforward way to analyze TCATA data is to examine the number of “checks” for each product, attribute, and time point across all evaluations. Each count—referred to as the attribute citation frequency—is then divided by the total number of observations to obtain an attribute citation proportion. The attribute citation proportions across time points are time-series data. These time points should not be considered as independent data since adjacent © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 22 time points in continuous data tend to be highly correlated. Also, small fluctuations or “blips” should not be overinterpreted since a certain amount of random variation is expected. For these reasons, the attribute citation proportions across time points are usually smoothed algorithmically. By drawing strength from neighbouring observations, smoothing removes momentarily fluctuations that may arise from random error, including the inherent imprecision of temporal sensory measurement. Visualizing smoothed citation rates emphasizes the main patterns in the data and show how the sensations evolve over time. 5.3.1. Citation proportion and perceived intensity Studies have shown that attribute citation frequency may be regarded as a proxy for attribute intensity [24]. The relationship tends to follow a sigmoidal response pattern, where the probability of checking a CATA attribute increases more slowly at the extreme ends of the intensity scale (1–2 and 6–7) but linearly in the mid-range. In other words, the probability of a CATA attribute being selected typically increases with perceived intensity according to a sigmoidal response pattern [25]. 5.4. Evaluating TCATA data per-time-point TCATA data have a complicated correlation structure that is unknown. For this reason, then analyzing TCATA data for significant difference between products, the analysis is typically performed at each time point for each attribute. Data from a particular attribute and time point are typically analyzed across products using nonparametric data analysis methods. These analyses rely either on a theoretical null distribution (see, e.g., [13], [35]), such as Cochran’s Q test [17] and McNemar’s test [30], which are widely available in statistical software, or on an empirical null distribution obtained from randomization or permutation tests (see, e.g., [34]), which may require manual coding. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 23 CATA data analysis commonly includes an overall test (e.g., Cochran’s Q test) to evaluate whether there are overall differences among products for a given attribute. If thereissufficient confidence that product citation rates differ, then paired comparison tests (e.g., McNemar’s test) are then conducted to check for differences in the product pairs. These analyses are usually conducted using raw data. It is useful to check the conclusions of the statistical tests here to determine whether they converge with interpretations based on smoothed results (Section5.3). Usually, these interpretations diverge only trivially, but strong divergence might suggest oversmoothing or the need for adjustments in the statistical testing approach. 5.5.Visualizing the temporal profile We describe three visualizations of TCATA data that are commonly considered. • TCATA curves - For each given product, attribute curves are plotted showing citation rate (y axis) vs time (x axis) [13]. These attribute curves are usually smoothed to show the temporal evolution of the product. Often, to understand whether an attribute citation rate is high or low, a reference curve is developed per attribute. In many cases, the most appropriate reference curve for an attribute contains attribute citation proportions for the mean product over time [34]. Statistical tests can determine at which times the product can be concluded to be above or below the reference line. Differences between products lasting for very short time intervals (e.g. less than a few seconds) might not represent meaningful differences so should be interpreted with caution [9, 10, 13]. We tend to avoid overinterpreting these “blips” that might be due to responses fluctuating due to random error. • Difference plot - A difference plot can be used to assess pairwise differences in the temporal profiles of two products of interest. A difference plot shows the differences in attribute citation proportions for two products (y axis) vs time (x axis). Sometimes, the difference © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 24 curves are shown only when the difference in citation proportions for the two products is statistically significant helps avoid overinterpreting differences that might have arisen by chance. • Trajectory plot - The temporal profiles of all products evaluated by TCATA can be summarized multivariately [11, 14]. Objects are all combinations of products and time points. Variables are attributes (columns). Data are the mean citation proportions at the intersection of each row and column. Variables are centred but not variance-standardized. These data are then submitted to principal component analysis (PCA). Scores from each product are usually smoothed for to emphasize systematic trends and de-emphasize random fluctuations. Results can be visualized in PCA biplots, which usually show two components at a time. Each product is shown as a smooth curved line showing how sensations evolve over time. Related component-based approaches that treat TCATA data multivariately can be found in [5–7, 15, 16] 5.6.Analyzing data from a factorial design When the evaluated products are developed using a factorial design, allowing for the investigation of main effects and interactions between experimental factors, analysis of variance (ANOVA) becomes a suitable method for analyzing TCATA data [26]. To apply this approach, the first step is to create dummy variables that represent the underlying factorial structure of the products. In contrast to count analysis at individual time points, factorial analysis using ANOVA requires aggregating the data over broader temporal segments [26]. Time points can be aggregated into larger time bins typically ranging from 5 s to 30 s (e.g., [1, 2, 20, 26]), depending on the desired level of temporal resolution and the expected temporal dynamics of product. It is possible to use either a moving window [10] or discrete time bins, where, for example, having © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 25 three time bins leads to an early-middle-late division. Alternatively, the TCATA timeline is divided into three perceptual periods: (1) the build-up of sensations, (2) the plateau and decay of sensations, and (3) the lingering or aftertaste period. Citations can be summed within each period for each attribute, then expressed as citation proportions, which then serve as dependent variables in the ANOVA. Factorial dummy variables (or alternatively, product identifiers) are used as independent variables. Where the ANOVA detects significant effects, multiple comparison tests can identify differences between individual factor levels. A potential bias is introduced since the researcher determines the number of bins or bin durations. For this reason, the researcher should check whether conclusions hold if a different number of bins or bin durations is used, especially if the number of bins is not predetermined or is decided after reviewing the data. 5.7.Applying penalty-lift analysis TCATA allows consumers to indicate how sensations evolve over time in the products under study. Products might differ significantly in how they are perceived in ways that might impact relevant outcomes, such as hedonic responses, purchase intentions, or other key outcomes. Penalty-lift analysis [31, 35] can be used to assess how temporal changes affect such outcomes. Penalty-lift analysis can be applied to consumer-relevant outcome measures to support decision-making. The following examples describe applications of penaltylift analysis to different types of data provided by the same consumers. • TCATA data + hedonic responses—TCATA data can be analyzed alongside temporal or static hedonic responses to identify which attributes drive product liking or disliking. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 32 Fig. 2. Procedure for the per-attribute aggregation of TCATA data for a particular product. The top panel shows raw data from two assessors who each evaluated the Sweet attribute in three replicates (in three sessions). Grey bars show periods of time when the attribute was checked. The middle panel shows the citation frequencies per-time-point after aggregating the data across all assessors and sessions. The smoothed citation rate curve (black line) is shown atop the aggregate bars. The bottom panel shows the full temporal profile, which is obtained by aggregating and smoothing citation frequencies from all attributes. Further details are given in Section 5.3. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 33 6.3.Results For practical reasons, only the first 55 s of the 90-s evaluation time were analyzed. The results reported below are based on this truncated period. 6.3.1. Assessor agreement and repeatability Table 1 presents the mean agreement and repeatability scores for the seven sensory attributes. Agreement scores ranged from 0.58 (Bitter) to 0.77 (Sour). The highest agreement was observed for Sour, Fruity/Hops, Malty, and Sweet (0.77, 0.75, 0.75, and 0.75, respectively). One reason the moderate agreement was not higher is it was measured across all assessors regardless of thermal-tasting status. Previously, thermal-tasting status was found to be a significant factor affecting the sensory perception of de-alocoholized beers [36]. Repeatability scores were consistently high across all attributes (Table 1), which indicated good within-assessor consistency. The highest repeatability was found for Carbonation and Sour (0.89), closely followed by Fruity/Hops (0.88) and Malty (0.87). 6.3.2. Product differences Fig. 3 shows TCATA curves for the beer served at 6 °C with sound. The reference curves, where shown, visualize the mean citation rates of all beers. Significant differences between this beer and the mean beer were observed for the attributes Carbonation and Astringent. The cold serving temperature and effervescence sound had the most pronounced effect on Carbonation, which showed a short interval of significantly higher citation from 3 s to 5 s, followed by a longer continuous interval from 8 s to 22 s. Astringent was cited significantly less over a nearly continuous interval from 30 s to 35 s. Overall, the long-lasting and continuous period of higher citation for Carbonation indicates a clear difference between this beer and the mean of all beers. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 34 Table 1. Mean agreement and repeatability scores by attribute across assessors. Attribute Agreement Repeatability Astringent 0.67 0.83 Bitter 0.58 0.80 Carbonation 0.71 0.89 Fruity/Hops 0.75 0.88 Malty 0.75 0.87 Sour 0.77 0.89 Sweet 0.75 0.86 Fig. 3. TCATA curves for the beer served at 6 °C with sound. Smoothed citation proportions are shown for each attribute over time (solid lines). Where this beer differs significantly from the mean beer, the TCATA curve is emphasized (thick line) and the citation rate for the mean beer is shown (dotted line). © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 35 In Fig. 4, these same TCATA curves are shown. When significant differences between this beer and the mean beer were detected for an attribute, the mean agreement for the assessors is shown. Often, when sensory discrimination is relatively high, the mean agreement is relatively low. The reason is at times when citation proportions are relatively high (e.g. 60%), there are still many (e.g. 40%) assessors who disagreed. Mean agreement tends to be high when the citation proportion is low since, as noted in Section 5.2.1, data sparsity is a type of agreement [34]. Fig. 4. TCATA curves and assessor agreement shown for the beer served at 6 °C with sound. Smoothed citation proportions for each attribute are shown (solid line). Where this beer differs significantly from the mean beer, the smoothed curve is emphasized (thick line) and the smoothed mean agreement across assessors is shown (dot-dash line). Fig. 5 shows a TCATA difference plot comparing beer served at 6 °C with sound to beer served at 21 °C with sound. The difference plot only shows time points where there was a significant difference between the two beer treatments. Significant differences were observed © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 36 for Carbonation during a discontinuous interval around 3 s followed by a continuous interval from 6 s to 24 s, where the 6 °C condition received more citations, with the peak difference near 15 s. Astringency showed a continuous interval from 37 s to 40 s where the 6 °C condition was cited less than the 21 °C condition. Fig. 5 A difference plot comparing citation proportions at time points where significant difference have been found of beer served at 6 °C with sound versus 21 °C with sound. To illustrate the raw data, the TCATA data in this plot were not smoothed before plotting. 6.3.3. Product trajectories Fig. 6 shows the product trajectories (see Section 5.5). The trajectories show how the sensory perception of the de-alcoholized beer treatments evolve over time. The first plane extracts 92.7% of the variation in the TCATA data, with 50.2% of the variation extracted in PC1 and 42.6% in PC2. The attributes Carbonation, Sour, Bitter, and © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 37 Astringency were prominent in this plane. The beer treatments had attribute citation rates of zero at the start and nearly zero at the end of the evaluation period. Citation rates were largest around the time of swallowing. The perception of Carbonation was highest in beers served cold (i.e. beers 3 and 4 served at 6 °C). As indicated by the wide trajectory of beer ⟨3⟩, the perception of carbonation in cold beer was further enhanced by the sound of effervescence. Fig. 6 Project trajectories of four beers show the evolution of flavours and textures in the directions indicated by arrows at two time points: start (0 s) and swallowing (10 s). Treatments are indicated by numbered boxes at the trajectory end (55 s): ⟨1⟩ beer served with sound at 21 °C, ⟨2⟩ beer served without sound at 21 °C, ⟨3⟩ beer served with sound at 6 °C, and ⟨4⟩ beer served without sound at 6 °C. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 38 6.4. Practical guidelines This case study demonstrated an application of the TCATA method with fading to investigate how serving temperature and auditory stimuli influence the temporal sensory characteristics of a de-alcoholized beer among thermal tasters and thermal non-tasters. The results showed serving temperature and sound of effervescence affected the sensory profile, particularly the perception of carbonation and astringency. Beer served cold elicited stronger and more persistent carbonation sensations. The product trajectories indicated that the sound of effervescence enhanced perceptions of carbonation, especially in colder samples. The temporal profiles obtained from the TCATA method provided a detailed description of how these sensory attributes evolved from before swallowing to the finish, offering insights that would not have been accessible through static sensory methods. This case study illustrates how temporal data can be processed and interpreted using software such as the R package tempR. Measuring agreement and repeatability showed how to evaluate consistency at the attribute level and even within particular time intervals. TCATA curves showed how a single product’s temporal profile was influenced by the experimental factors. A difference plot highlighted the time points where two specific products differed significantly. These visualizations showed how temporal profiles were affected by the experimental factors. Product trajectories illustrated the sensory evolution of the entire product set over time. These analyses show the potential of the TCATA method to investigate how experimental manipulations affect the dynamics of product perception. More broadly, this case study underscores the importance of aligning study objectives, product design, assessor training, and analytical tools in temporal sensory applications. When carefully designed and © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 39 implemented, the TCATA method provides a robust and informative approach for studying how sensory perception develops and changes throughout product evaluation. References [1] G.B.H. Andersen, C.L.D. Christensen, J.C. Castura, N. Alexi, D.V. Byrne, and U. Kidmose. Sugar replacement in chocolate-flavored milk: Differences in consumer segments’ liking of sweetener systems relate to temporal perception. Beverages, 10(3):54, 2024. [2] G.H. Andersen, N. Alexi, K. Sfyra, D.V. Byrne, and U. Kidmose. Temporal check-all-thatapply on the sensory profiling of sucrose-replaced sweetener blends of natural and synthetic origin. Journal of Sensory Studies, 38(4):e12838, 2023. [3] G. Ares, J.C. Castura, L. Antuñez, L. Vidal, A. Gimenez, B. Coste, A. Picallo, M.K. Beresford, S.L. Chheang, and S.R. Jaeger. Comparison of two TCATA variants for dynamic sensory characterization of food products. Food Quality and Preference, 54:160–172, 2016. [4] G. Ares and S.R. Jaeger. Check-all-that-apply (CATA) questions with consumers in practice: Experimental considerations and impact on outcome. In Rapid sensory profiling techniques, pages 257–280. Elsevier, 2023. [5] A.K. Baker, J.C. Castura, and C.F. Ross. Temporal check-all-that-apply characterization of Syrah wine. Journal of Food Science, 81(6):S1521–S1529, 2016. [6] D. Beaton and M. Meyners. Powerful visualization of product-attribute associations for temporal data. Food Quality and Preference, 79:103572, 2020. [7] I. Berget, J.C. Castura, G. Ares, T. Næs, and P. Varela. Exploring the common and unique variability in TDS and TCATA data – a comparison using canonical correlation and orthogonalization. Food Quality and Preference, 79:103790, 2020. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 40 [8] J. Bi and C. Kuesten. Product effect size estimation and performance accuracy (validity and reliability) assessment for CATA and TCATA data. Food Quality and Preference, 127:105441, 2025. [9] J.C. Castura. Dynamics of consumer perception. In Methods in Consumer Research, Volume 1, pages 211–240. Elsevier, 2018. [10] J.C. Castura. Investigating temporal sensory data via a graph theoretic approach. Food Quality and Preference, 79:103787, 2020. [11] J.C. Castura. Exploring Variability of Perceptions and Heterogeneity of Opinions. Norwegian University of Life Sciences, 2023. [12] J.C. Castura. tempR: Temporal sensory data analysis, 2024. Version 0.10.1.1. https://CRAN.R-project.org/package=tempR. [13] J.C. Castura, L. Antuñez, A. Gimenez, and G. Ares. Temporal check-all-that-´ apply (TCATA): A novel dynamic method for characterizing products. Food Quality and Preference, 47:79–90, 2016. [14] J.C. Castura, A.K. Baker, and C.F. Ross. Using contrails and animated sequences to visualize uncertainty in dynamic sensory profiles obtained from temporal check-all-thatapply (TCATA) data. Food Quality and Preference, 54:90–100, 2016. [15] J.C. Castura, D.N. Rutledge, C.F. Ross, and T. Næs. Discriminability and uncertainty in principal component analysis (PCA) of temporal check-all-that-apply (TCATA) data. Food Quality and Preference, 96:104370, 2022. [16] J.C. Castura, P. Varela, and T. Næs. Investigating only a subset of paired comparisons after principal component analysis. Food Quality and Preference, 110:104941, 2023. [17] W.G. Cochran. The comparison of percentages in matched samples. Biometrika, 37(3/4):256–266, 1950. © 2025. Preprint distributed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/). Citation: Andersen, G.B.H., Junge, J.Y., & Castura, J.C. (2025). Temporal Check All That Apply (TCATA). Unpublished chapter prepared for A.G. Cruz, E.A. Esmerino & T.C. Pimentel (eds.): Classic and Novel Sensory Analysis, Springer. (Forthcoming). https://doi.org/10.5281/zenodo.17601146 41 [18] L.J. Cronbach. Coefficient alpha and the internal structure of tests. Psychometrika, 16(3):297–334, 1951. [19] M. Devezeaux de Lavergne, J.A.M. Derks, E.C. Ketel, R.A. de Wijk, and M. Stieger. Eating behaviour explains differences between individuals in dynamic texture perception of sausages. Food Quality and Preference, 41:189– 200, 2015. [20] C. Dietz, D. Cook, Q. Yang, C. Wilson, and R. Ford. A TCATA by modality approach to study the multisensory temporal profile of hop bitter and flavour products applied in lager. Food Quality and Preference, 97:104470, 2022. [21] C. Frascolla, M. Visalli, H. Cardot, and P. Schlich. Indexes of individual repeatability and product discrimination in TDS and TCATA and their statistical inference. Food Quality and Preference, 110:104929, 2023. [22] K.L. Gwet. Computing inter-rater reliability and its variance in the presence of high agreement. British Journal of Mathematical and Statistical Psychology, 61(1):29–48, 2008. [23] S.R. Jaeger, M.K. Beresford, D.C. Hunter, F. Alcaire, J.C. Castura, and G. Ares. Does a familiarization step influence results from a TCATA task? Food Quality and Preference, 55:91–97, 2017. [24] S.R. Jaeger, S.L. Chheang, D. Jin, C.M. Roigard, and G. Ares. Check-all-that-apply (CATA) questions: Sensory term citation frequency reflects rated term intensity and applicability. Food Quality and Preference, 86:103986, 2020. [25] S.R. Jaeger, S.L. Chheang, D. Jin, G.S. Ryan, and G. Ares. How do CATA questions work? relationship between likelihood of selecting a term and perceived attribute intensity. Journal of Sensory Studies, 38(4):e12833, 2023.