1 Introduction
NCA is a methodological approach that uses a necessity causal perspective for identifying necessary conditions from data. Necessity causal logic in NCA implies that if a certain level of the condition is not present, a certain level of the outcome will not occur: If not \(X\), then not \(Y\). No other factor can compensate for the absence of the necessary condition (or a required level of it). The necessary condition allows the outcome to exist, but does not produce it.1 A necessary condition operates independently of the rest of the causal structure and serves as a ‘critical success factor’ or ‘must-have’. When absent, it creates a ‘bottleneck’ or ‘constraint’ that prevents the target outcome, which cannot be compensated by other factors. For example, a student must have a high school diploma for admission to a university. Without such a diploma, the student will not be admitted. Other factors like a student’s high motivation cannot compensate for the lack of a diploma.
Necessity causal logic and NCA differ from sufficiency causal logic and related methods. In sufficiency causal logic the condition produces the outcome: if \(X\), then \(Y\). A single factor may be sufficient for the outcome, for example when students graduate, they receive a diploma. Often, not a single factor but a combination of factors (a ‘configuration’) may be sufficient for the outcome: if [combination of \(X\)’s], then \(Y\). A university diploma alone is not sufficient for admission to a PhD program; however, combining it with strong motivation, good grades, and a recommendation letter will secure it. Such a configurational sufficiency logic is used in methodologies and frameworks like Rothman’s Sufficient-Component Cause model in medicine (SCC, Rothman, 1976), Wright’s Necessary Element of a Sufficient Set in law (NESS, Wright, 1985), and Ragin’s Qualitative Comparative Analysis in the social sciences (QCA, Ragin, 1987).
Necessity causal logic also differs from probabilistic causal sufficiency logic where the single factor is a ‘contributing cause’ that helps to produce the outcome and contributes to the outcome ‘on average’: ‘if \(X\), then probably \(Y\).’ Highly motivated people are more likely to be admitted to a PhD program than those with lower motivation, although there are highly motivated people who are not admitted and less motivated people who are admitted. Such a probabilistic causal sufficiency logic is applied in statistical methodologies for causal inference, such as in Pearl’s Directed Acyclic Graphs (DAGs) and do-operator approach (Pearl, 2009) and Angrist et al.’s Instrumental Variable approach (IV, Angrist et al., 1996).
The different causal lenses (single necessity causes, configurational sufficiency causes, and single probabilistic sufficiency causes) offer fundamentally distinct perspectives on causality. These lenses reflect different theoretical expectations (hypotheses) about how outcomes arise. This means that phenomena and data can be approached with different causal lenses (Chapter 2). The choice of lens may be guided by the specific nature of the phenomenon, the study question, or the analyst’s theoretical priorities2. For instance, if the question is whether a certain resource (e.g., access to capital) is required for a firm to enter a market, a necessity lens is appropriate. This lens helps identify bottleneck factors: conditions that must be in place for success. In contrast, if the aim is to identify combinations of national conditions that are sufficient for strong climate policy commitment (e.g., public support + high income, or climate risk + high fossil fuel dependence), a configurational sufficiency perspective, as used in QCA, may be appropriate. This approach assumes that different pathways can each lead to the same outcome. If the goal is to understand whether self-efficacy increases the likelihood of engaging in a behavior, a probabilistic sufficiency lens is fitting. This approach helps estimate how much more likely the outcome is when a cause is present.
No lens is inherently superior; each has unique strengths and limitations. Thus, there are no better or worse lenses; there are no more or less relevant lenses. These lenses are different and shed a different light on the same issues. What matters is that the chosen lens aligns with the goals of the study, the nature of the causal claim being made, and indeed the preference of the analyst. Once a causal lens is selected, it is essential to apply a methodological approach that matches the logic of that lens, ensuring theory–method fit, which is essential for the credibility and usefulness of the findings (Chapter 11).
This book’s emphasis on causality has both theoretical and practical reasons. Theoretically, scientific theories aim to explain why outcomes occur, not merely to describe patterns of co-occurrence (Chapters 3 and 7). Practically, changing or influencing an outcome requires intervention on causes of the outcome, not on co-occurring factors (Chapter 12). The assumption that \(X\) is a necessary cause of \(Y\) implies the assumption that \(X\) temporally precedes \(Y\) (Chapter 2).
NCA is an approach that is specifically designed to analyze necessity. It has two intertwined parts: a ‘soft’ part and a ‘hard’ part. The soft part is NCA’s soul: the causal perspective of necessity (NCA methodology). The hard part is NCA’s data analysis method (NCA method) that fits this perspective. Therefore, the term ‘NCA’ refers to a combination of the NCA methodology with its causal necessity perspective, and the corresponding NCA method for analyzing data.3 NCA’s necessity causal lens extends beyond common binary conditional necessity logic. Specifically, NCA adds causal reasoning to conditional logic, incorporates a continuous view on necessity, and permits exceptions (the ‘typicality’ perspective, see Chapter 2).
The book is organized as follows. Part I discusses the methodological principles of NCA, explaining its causality (e.g., compared to other causal perspectives), theory (e.g., main elements of a necessity theory, different types of necessity theories), mathematics (e.g., mathematical estimation of ceiling line, NCA parameters such as effect size, and measures of fit) and statistics (e.g., NCA’s null hypothesis tests, Monte Carlo simulations with NCA), and NCA’s credibility to identify necessity (e.g., using statistical and empirical criteria).
Part II of the book outlines how to conduct NCA. Specifically, this part discusses NCA’s hypothesis formulation, data collection, data analysis, reporting, how to apply NCA in a multimethod study, and how NCA can be used in practice. The book concludes with a summary and personal reflection on 10 years of NCA, and Part III provides additional materials.
With its innovative approach to causality and data analysis, NCA has rapidly entered the social, medical, and technical sciences.4 The publication in 2016 of NCA’s core paper (Dul, 2016b) marks the start of NCA, although proto-versions of NCA were introduced earlier (Dul et al., 2010; Dul & Hak, 2008).5 Subsequently, several extensions were developed, including a statistical significance test (Dul, Van der Laan, et al., 2020), and specialized software (Appendix B). Various publications, such as a textbook (Dul, 2020) and an online book (Dul, 2021) with practical guidelines on applying the method were developed to facilitate the understanding and dissemination of NCA. Furthermore, publications introducing and summarizing NCA have emerged across various fields of business and management, including human resource management (Hauff et al., 2021), marketing and sales (Conde, 2025; Dul, Hauff, et al., 2021), tourism management (Tóth et al., 2019), entrepreneurship (Linder et al., 2023), supply chain management (Bokrantz & Dul, 2023), and international business (Richter & Hauff, 2022). Also outside the business and management field, the method was introduced in fields such as public health (Greco et al., 2022), clinical psychology (Marchetti et al., 2026), education (Tynan et al., 2020), and creativity (Dul, Karwowski, et al., 2020).
Editors of academic journals have recognized the value of NCA by publishing editorial comments on the approach (e.g., Dul, 2025; Fainshmidt et al., 2020; Hernaus & Černe, 2022; S. Robinson et al., 2022; Sarstedt & Liu, 2023). Review articles (e.g., Aguinis et al., 2020; Boon et al., 2019; Dabić et al., 2021; Del Sordo & Zattoni, 2025; Frazier et al., 2017; Rozenkowska, 2023) and a plethora of journal articles suggest using NCA in future studies, for example, in the fields of education (e.g., Credé & Tynan, 2021), human resource management (e.g., Hauff, 2021; Parker & Knight, 2024), information systems (e.g., P. N. Sharma et al., 2024), international business/management (e.g., Richter et al., 2022; Zahoor et al., 2023), marketing (e.g., Guenther et al., 2023; Jovanovic & Morschett, 2022), operations and supply chain management (e.g., Acquah, Quaicoe, et al., 2023), strategy (e.g., Bouncken, Fredrich, Ritala, et al., 2020; Deist et al., 2023; Klimas et al., 2023; Ogundipe et al., 2022), and tourism and hospitality (e.g., Becker et al., 2023). At the time of writing this book, hundreds of articles that apply NCA have been published already. The number of publications in ‘Web of Science’ ranked journals6 that apply NCA has increased steadily. Appendix C shows these articles that have appeared through the end of 2025.
Published articles use NCA as the sole or primary method (Appendix C; Table C.1), or combine NCA in a multimethod study (Chapter 11), often with Qualitative Comparative Analysis (QCA, Table C.5) or with regression-based methods. Examples of regression-based methods include multiple linear regression (MLR, Table C.2) and structural equation modeling (SEM, Tables C.3 and C.4). NCA is also combined with a variety of other methods (Tables C.6 and C.7). Although the number of articles with NCA has grown rapidly, NCA is not always applied according to minimum standards (Chapter 10, Dul et al., 2023), and several misconceptions have been published.7 Therefore, the main goal of this book is to promote proper understanding and high-quality application of NCA.
Many articles that apply NCA conclude that necessity was identified, although the evidence is not always convincing. Other articles that apply NCA concluded that necessity was not found (e.g., Batey et al., 2021; Golini et al., 2016; Gu et al., 2022; L. Luo et al., 2022; B. Peng et al., 2022). Not finding a necessary condition might be a valuable result, because such a result shows that a supposed essential factor for an outcome does not need to be present and its absence can be compensated (e.g., Arenius et al., 2017). On the other hand, finding a large number of potential necessary conditions, for example in an exploration study (e.g., Gantert et al., 2022; Klimas et al., 2022; Stek & Schiele, 2021) does not mean that all are informative because several conditions may not have a theoretical meaning or may be trivial, which could have been easily identified before the analysis was done (Chapter 7).8
Conducting NCA consists of four stages:
- Formulate the formal necessity hypothesis.
- Collect data.
- Analyze data.
- Report results.
In the first stage (Chapter 7), necessity causal logic is employed to formulate a theoretical expectation about conditions that are necessary for the outcome9.
The second stage (Chapter 8) consists of collecting new or existing data for testing the hypothesis, which includes the consideration of study design, sampling or case selection, and measurement. NCA does not have a ‘measurement model’, as SEM and QCA have (in QCA it is called ‘calibration’). This means that any type of data can be used as input to NCA as long as the scores of \(X\) and \(Y\) are meaningful, valid, and reliable. NCA poses no new requirements on the data, although some aspects of data collection (case selection in qualitative studies; power analysis in quantitative studies, see, for example, Dul (2024b)) may differ. NCA can also be used with archival data to give a new perspective on previous findings (Dul et al., 2024).
In the third stage (Chapters 9, 11), NCA’s data analysis is conducted, which differs entirely from common types of data analysis.
In the fourth stage (Chapter 10), the results of NCA are reported.
Currently, most applications of NCA are in academic research, where researchers use the NCA approach to develop and test theories. NCA is also entering practical settings where NCA is applied by practitioners such as consultants and data analysts as an innovative approach for addressing challenges and exploring new opportunities. This is discussed in Chapter 12.
After discussing the fundamental backgrounds of the NCA methodology in Part I, the four stages of the NCA method are discussed in Part II.
References
This book uses specific terminology, such as condition and outcome, that is explained in the glossary of Appendix A.↩︎
In this book the general term for a user of NCA or other methodological approaches is analyst. The analyst can be an academic scholar or a practitioner. The general term for an academic research activity or a project in practice is study.↩︎
In previous publications I have inconsistently used the terms ‘approach’, ‘methodology’ and ‘method’ in the context of NCA. In this book, I follow the common distinction between methodology, referring to the philosophical and theoretical framework, and method referring to the specific techniques, tools, or procedures used to collect and analyze data. The combination of the two parts is called approach or methodological approach.↩︎
The NCA methodology and method can be used in any type of investigation that aims to describe causal relationships. Although usually quantitative data (numbers) are used in NCA, NCA can also be used with qualitative data (words, symbols, see Chapter 8) as long as scores (values, levels) for the condition and the outcome are available.↩︎
The dissemination of NCA’s core paper (Dul, 2016b) has been welcomed by Bergh et al. (2022): “The significant expansion of an unknown method along with its usability (including an illustration, user-friendly recommendations, and a software tool) makes the article a gold standard for a contribution via transfer” (p. 1840).↩︎
Journals ranked in Clarivate’s Journal Citation Reports.↩︎
Most misconceptions relate to a lack of awareness about the principles of NCA. Example misconceptions are: “NCA should specify the full data generation process (DGP)” (given its purpose, NCA specifies only a part of the DGP), “NCA is just a statistical method” (NCA is a broad methodological approach that combines theory, mathematics (geometry) and uses statistical tools), “NCA’s statistical test is a test of H1” (NCA’s statistical tests are null hypothesis tests, testing H0), “NCA is the necessity analysis of QCA” (the necessity analysis of NCA and that of fuzzy-set QCA are fundamentally different), “Necessary conditions according to NCA are the same as INUS conditions” (INUS conditions are local necessary parts of a sufficient configuration that produces the outcome; NCA considers overall necessary conditions for the outcome), “NCA cannot test sufficiency” (NCA can test the absence of \(X\) being necessary for absence of \(Y\), which is equivalent to \(X\) is sufficient for \(Y\)), etc. Several misconceptions are addressed in specific publications (Dul et al., 2019; Dul, Vis, et al., 2021; Dul, 2022).↩︎
A positive or negative conclusion about necessity depends on the analyst’s judgement of many elements, summarized in NCA’s SCoRe checklist for conducting and reporting an NCA study (Chapter 10).↩︎
In this book, it is assumed that hypothesis formulation is done before data collection. The reasons are discussed in Chapter 7.↩︎