Abstract
While many basic research fields of psychology are mostly concerned with the search for general laws, personality psychology is also—and perhaps even primarily—tasked with studying individuality. Describing, explaining, predicting, or changing the experiential-behavioral reality of the individual person requires research methodologies that support valid person-level inferences, grounded in precise assessments and suitable statistical analyses. Traditional population-based methodologies often fall short because population-to-individual generalizability cannot be readily assumed, and one-size-fits-all approaches do not do justice to the individual. We propose that, in personality psychology and beyond, research processes must be personalized to effectively capture and address the complex nuances of individual personalities. Personalization may include developing person-specific psychometric tools, study designs, analytical models, and interventions. Such tailored approaches could not only enhance personality research but also aid in uncovering general psychological principles that manifest differently, perhaps even uniquely, across individuals. Moreover, the broader trend towards personalized solutions in fields like psychotherapy, health psychology, and educational psychology creates an opportunity for personality psychology to demonstrate and broaden its practical relevance.
Plain language summary
While many areas of psychology focus on finding general rules that apply to everyone, personality psychology also focuses on studying single persons. To truly understand and influence a person’s behavior and experiences, researchers need the right methods for studying that person. Traditional research methods often fail because they assume what works for the average person works for everyone, which is rarely true. We suggest that in personality psychology, and other areas, research should be tailored to individuals. This could mean creating tailored tools and methods for each person. Personalized approaches could improve our understanding of personality and help find general psychological principles that apply differently to different people. Additionally, personality psychology could be practically useful for personalized approaches in fields like therapy, health psychology, and education.
Keywords
Enabled by technological advancements, personalization has emerged as a transformative force, fundamentally altering traditional practices across various domains, ranging from consumer-oriented sectors such as marketing (for overviews, see Anshari et al., 2019; Chandra et al., 2022) to fields like medicine (Erikainen & Chan, 2019; Goetz & Schork, 2018; Pokorska-Bocci et al., 2014), education (Bernacki et al., 2021; Tetzlaff et al., 2021; Zhang et al., 2020), and psychotherapy (Cohen et al., 2021; Deisenhofer et al., 2023; Nye et al., 2023). Despite its widespread use, the term “personalization” lacks a clear and generally agreed-upon definition due to its broad application encompassing a wide and multifaceted array of approaches (Bernacki et al., 2021; Chandra et al., 2022; Erikainen & Chan, 2019; Pokorska-Bocci et al., 2014; Shemshack & Spector, 2020; Vesanen, 2007; Zhang et al., 2020). We define personalization very broadly as the deliberate process of tailoring something to the individual person or a specified group of people with the ultimate goal of optimizing a desired outcome. Thus, personalization stands in contrast to the concept of one-size-fits-all, which implies a uniform approach that disregards individual or group differences and intraindividual variation. In psychology, personalization is arguably most commonly associated with interventions that are selected or adapted to better suit target individuals and improve intervention outcomes (e.g., Cohen et al., 2021; Fisher, 2015; Nye et al., 2023; Tetzlaff et al., 2021). However, changing behavior and experiences is only one applied goal of psychology, and personalized approaches might also prove invaluable in describing, predicting, or explaining psychological phenomena at the level of the individual.
In this article, we highlight what (personality) psychology can gain from a shift towards personalization in research and applied settings. This article is not a systematic review, but summarizes what we consider to be the major conceptual arguments in favor of personalization, and we provide only selected examples of available methodological tools. First, we very briefly elaborate on how personalization, as defined above, relates to other concepts and terms in psychology and neighboring disciplines. We then make the case that within personality psychology, embracing personalization is not only justified but essential for understanding and addressing each person’s unique experiential-behavioral reality. Next, we highlight the merits of adopting a person-specific perspective and using personalization in the pursuit of the main goals of the field—
Personalization and related concepts
Across fields, different terms have been applied interchangeably with personalization or to distinguish between specific approaches and concepts, including individualization, customization, tailoring, targeting, matching, stratification, segmentation, bespoke, adaptive, precision, and numerous others. 1 The term “precision” is especially popular in the biomedical literature (i.e., precision medicine; Erikainen & Chan, 2019; Jørgensen, 2019), but it is also sometimes used in psychology (e.g., to describe data-driven personalization approaches; Deisenhofer et al., 2023). We avoid this term because it has well-established alternate meanings in the context of assessment and machine learning, referring, for example, to the reliability, non-randomness, consistency, or reproducibility of a measure or to the positive predictive value of a classification model, which might easily cause misunderstanding. “Stratification” (e.g., stratified medicine, see Erikainen & Chan, 2019; Schleidgen et al., 2013) and “segmentation” (e.g., customer segmentation in marketing, see Chandra et al., 2022) are also common terms that refer to group-specific rather than person-specific approaches. According to our definition, personalization does not necessarily imply that every aspect or detail is tailored to the unique characteristics of the target individual, but tailoring can also be informed by and directed at those characteristics that a person shares with others in a sufficiently homogenous subpopulation or group. That is, personalization encompasses both person-specific and group-specific approaches. The broadest possible form of personalization characterizes each target person as belonging to one of two groups (i.e., binary classification), with more and more nuanced target characterizations possible as the number of groups increases. In a truly person-specific approach, a unique solution is obtained for each target person. We will focus predominantly on person-specific approaches in this article as the assumption of heterogeneity between individuals is inherent to personality psychology, and a key reason the field should embrace personalization.
In psychology, and personality psychology in particular, research approaches focussing on individual persons are commonly and historically referred to as idiographic. We explicitly avoid this term for two main reasons. First and most importantly, the idiographic–nomothetic distinction is ambiguous, having been (re-)interpreted in psychological literature in numerous ways (for a detailed discussion, see Phan et al., 2024). As a result, distinct concepts such as within-person, dynamic, qualitative, or person-centered have often been conflated under the idiographic label (see also Krauss, 2008). Even Allport who popularized the idiographic–nomothetic distinction in personality psychology (Allport, 1937, 1961) later abandoned this terminology “to insure more rapid progress” (Allport, 1962, p. 409). We follow his example, aiming to maintain conceptual clarity. At the same time, we recognize that many—albeit not all—interpretations of the idiographic approach correspond to or overlap with our understanding of personalization. Second, the terms “idiographic” and “nomothetic” are largely unfamiliar outside the field of psychology and a few other social sciences. Additionally, they are not self-explanatory or easily understood, which can hinder effective communication. The specialized nature of this terminology often creates barriers to interdisciplinary collaboration and public engagement. As noted earlier, the concept of personalization is not confined to psychological science but has become increasingly relevant across various fields. Psychological research stands to both contribute to and benefit from the growing focus on personalization in these areas. Adopting a shared, clear, and accessible language could help foster broader understanding and cooperation across disciplines.
Another common distinction in psychological research is between person-centered and variable-centered approaches. Variable-centered most commonly refers to between-person analyses (i.e., Galtonian methodology, Robinson, 2011), but may also denote examinations of variable patterns across occasions (see Kuper et al., 2024). Person-centered approaches, on the other hand, are concerned with within-person patterns. However, the term has been used to refer to a range of methodologies with varying inferential goals, such as (a) to characterize and directly compare individuals in terms of their multivariate intraindividual configurations (i.e., psychography and comparative research; Stern, 1911); (b) to examine between-person variation in within-person patterns at the level of a group or population 2 (e.g., Asendorpf, 2015), or (c) to identify subgroups of individuals with similar within-person patterns (e.g., Howard & Hoffman, 2018). 3 Personalization typically requires a person-centered approach in that individuals must be represented by many data points and within-person relations. Meanwhile, most psychological research that is considered to be person-centered does not leverage personalization as the same variables are typically assessed in the same way for each participant, and data is analyzed in a single (population) model.
We advocate for using “personalization” as a comprehensive umbrella term to subsume a wide array of approaches in psychology and beyond. This linguistic choice places the person at the very heart of the concept, underscoring that the personalization process is performed with the person as the focal point. We further use this term for its widespread comprehension and with the aim of promoting a holistic understanding of the general concept, fostering interdisciplinary collaboration and exchange of ideas. However, we recognize that many approaches that fall under our definition may not be commonly referred to as personalization, and narrower definitions or alternative terms may be appropriate, depending on the
The limitations of population-level (personality–)psychological research
Mainstream psychological science typically examines population-level patterns that—depending on the research question and the methodology—can be interpreted (a) as characterizing the particular population (psychodemography; Lamiell, 2019), 4 (b) as revealing the between-person patterning of variables (in the population; variable-centered or variable-oriented perspective; Stern, 1911), or (c) as establishing normative within-person patterns (i.e., characterizing the average person from the population; see below). The knowledge that is produced concerns the population as a whole (population-specific generalization; Phan et al., 2024) rather than the individual persons it contains. Of course, knowledge about populations can be both theoretically interesting and practically relevant. As population-level research makes up the majority of empirical psychology, its numerous benefits have been extensively discussed throughout the literature, and they cannot be adequately synthesized within the scope of this article (but see Lundh, 2023, 2024 for some general thoughts and selected examples). Arguably, however, most psychological research is ultimately not concerned with population-level patterns but with phenomena of the mind and behavior (e.g., phenomena of perception, cognition, affect, motivation, or learning), which are conceptually located within and manifested at the level of the individual person. 5 Thus, a disparity often exists between the level at which an object of psychological investigation is conceptualized and the level at which it is analyzed (e.g., Moeller 2022; Richters, 2021).
Population-level, between-person patterns only apply at the person level under the strict and rarely met conditions of ergodicity (Fisher et al., 2018; Hamaker, 2012; Molenaar, 2004; Molenaar & Campbell, 2009; but see also Adolf & Fried, 2019). An ergodic psychological phenomenon occurs in the same way across individuals (homogeneity) and over time (stationarity), which would constitute a natural law in a strictly normativist sense. If interindividual variation is observed (and cannot be attributed to error), what applies to people in the aggregate may not be true for each, most, or even (in extreme cases) for any individual person (Hamaker, 2012; Lamiell, 1986; Molenaar, 2004; Phan et al., 2024). Vice versa, phenomena occurring in some but not all individuals might not emerge at the population level, and can be obscured and overlooked when population aggregates are examined. For example, though multiple studies suggest that, on average, adversity does not foster positive personality change (Jayawickreme & Blackie, 2014), this cannot be interpreted as evidence that the phenomenon of post-traumatic growth does not exist at the individual level. The average person merely represents the central tendency of the population (Lamiell, 2019). It is a statistical abstraction, without body or psyche and incapable of “psychological doings” (Lamiell, 2019) such as behavior, thought, or emotion. Concepts developed, or principles established for the average person might be of little or no relevance to individuals. Also, interventions that have been developed on the basis of population-level aggregates may fail to adequately address the target person’s unique functioning (see also Johnston & Johnston, 2013). Arguably, this means that the average person should not be of particular interest to psychologists. Differential psychology (Revelle et al., 2011) alerts us to the shortcomings of the average person. However, between-person differential approaches—just like research approaches focused on the average person—produce knowledge about populations rather than about individuals. In recent years, longstanding population-based traditions have once again been criticized, and there has been a push to place the individual back at the center of (personality) psychological research (e.g., Beck & Jackson, 2020a, 2020b; Moeller, 2022; Molenaar, 2004; Phan et al., 2024; Renner et al., 2020), which requires both a conceptual and methodological reorientation.
Personalization makes room for individuality in personality research
Personality is, by definition, a feature of the individual.
6
Exact definitions of personality vary but most refer to a person’s relatively stable characteristic patterns of behavior, cognition, affect, and motivation (e.g., Baumert et al., 2017; Funder, 2004; Revelle & Condon, 2015) within environments (ABCD’Es, Renner et al., 2020) and to the within-person organization underlying and determining these patterns (e.g., Allport, 1961). Therefore, the study of personality necessarily entails the examination of individual-level within-person consistencies (Phan et al., 2024) that cannot be inferred from population-level patterns, unless ergodicity has been established (see earlier). Yet, personality psychology has focused on interindividual differences throughout much of the 20th century (Carlson, 1971; Lamiell, 2019; Molenaar, 2004; Phan et al., 2024), with countless studies examining between-person covariation among person-level variables (
The experiential-behavioral reality of the individual person, that is, their
Examples for individual-level research goals in personality psychology that can be addressed with personalized approaches.
Description: Characterizing individuals and the psychological phenomena they display
Descriptive research characterizes the phenomena being studied and forms the empirical groundwork of any field. Most empirical personality research explores between-person or within-person, contemporaneous or lagged associations between variables of interest in observed data and should therefore be considered largely descriptive (Mõttus et al., 2020). Yet, causal inference and predictive validity are often implied and desired (Grosz et al., 2020; Mõttus et al., 2020). Historically, personality description has largely been treated as a between-person problem. Numerous personality constructs have been developed to summarize phenomena that co-vary at the between-person level, and an even larger number of measures has been constructed to capture and quantify these constructs. Most notably, the Big Five model is routinely applied in personality psychology and beyond to broadly describe interindividual variation in how people tend to act, think, and feel. The theoretical and practical value of the Big Five and its measures is empirically supported, especially by evidence of cross-population generalizability (e.g., cross-cultural; Church, 2016; Jolijn Hendriks et al., 2003), lifespan rank-order stability (i.e., temporal stability of interindividual differences; e.g., Bleidorn et al., 2022; Damian et al., 2019; Roberts & DelVecchio, 2000), and substantial covariation with other interindividual differences variables (e.g., important life outcomes; Beck & Jackson, 2022a; Ozer & Benet-Martínez, 2006; Soto, 2021). However, as mentioned earlier, it is only possible to draw inferences about individuals from population-level findings under rare conditions (Adolf & Fried, 2019; Fisher et al., 2018; Hamaker, 2012; Molenaar, 2004; Molenaar & Campbell, 2009). Therefore, there is little reason to assume that the Big Five traits, as conceptualized and measured at the between-person level, capture personality as it operates within individuals in their everyday lives. The within-person structure of personality—relating either to fluctuations in trait scores across time (e.g., Grosz, 2024) or to individual-level co-variation of state manifestations (e.g., Hamaker et al., 2005)—is a current focus in personality research. Addressing this issue in depth requires focused conceptual reasoning and empirical exploration beyond the scope of the present work. Instead, we briefly dwell on the broader and fundamental problem of generalizing population-level insights to individual personality descriptions, focussing on interindividual differences constructs and measures which are typically at the core of (descriptive) personality research. We argue that a comprehensive psychological characterization of an individual’s personality cannot be reduced to their quantitative standing on a standard set of variables in comparison to others (e.g., trait profiles), and we introduce alternative methodological approaches.
Interindividual differences constructs and measures may be ill-suited for describing an individual’s personality
Interindividual differences constructs such as the Big Five provide useful summaries of between-person co-variation patterns, but these may not represent reasonable abstractions when describing individuals. To give a simple example: Though sociability and assertiveness are correlated at the between-person level and are therefore conceptualized as facets of extraversion (e.g., Soto & John, 2017), an individual person might be generally very sociable but not at all assertive or vice versa. Consequently, relevant details of a person’s characteristic patterns are obscured when the individual is described in terms of their extraversion level. What is more, the individual might become substantially less sociable and more assertive over time, while their extraversion level remains unchanged, falsely suggesting that no trait change has occurred. Of course, more nuanced descriptions can be achieved by simply including a larger number of narrower (i.e., less aggregated) interindividual differences variables (Mõttus et al., 2020). Arguably, however, any construct that is developed (through generalization, abstraction, and categorization; Uher, 2021) with the premeditation of emphasizing interindividual differences and imposing some degree of parsimony cannot adequately reflect the intricacies that constitute a person’s individuality. A personality trait that is highly relevant and consequential at the individual level may not be adequately represented in even the most comprehensive of interindividual differences taxonomies. Every trait is inherently unique to the individual in its specific contents, development, and nuanced manifestations. At the conceptual level, shared aspects, or common trait constructs, on which individuals in a population may be roughly compared, should be regarded as mere approximations, gists, or coarse abstractions of actual (personal) traits (Allport, 1937, 1961). At the measurement level, the constructs and processes that are most relevant for the individual might evade traditional standardized assessment methods entirely.
There has always been an area of tension in psychological assessment between the desire to develop measures and compute scores that are meaningful and relevant for specific individuals but that also allow for comparisons and aggregation across individuals (e.g., Nesselroade et al., 2007). Dominated by the interindividual differences paradigm, personality assessment tends to prioritize the suitability of a measure for between-person operations rather than person-level inferences. Construction and optimization of personality measures is typically guided by between-person criteria such as internal consistency, retest reliability, and validity correlations. However, psychometric indices that concern the between-person patterning of scores provided by a measure do not denote the degree to which individuals consistently manifest a trait and therefore cannot indicate whether a trait is relevant to the description of their personalities (Lamiell, 1981). 8 What is more, between-person standardization of assessment procedures does not guarantee that a shared target construct or process is validly and reliably captured in each individual. Uniform implementation, scoring, and interpretation of measures across different people (i.e., etic assessment; see also Kuper et al., 2024; Phan et al., 2024) ensures that observed systematic variation cannot be attributed to external forces arising from procedural heterogeneity, but a psychometric measure is likely also influenced by internal forces other than a person’s level of the target construct. This calls into question not only the person-level applicability of scores provided by interindividual differences measures but also their between-person comparability (Nesselroade & Molenaar, 2022). 9
Personality and personality-relevant psychological phenomena can only be assessed and described in individuals, where they are conceptually located. A between-person model representing (a) the structure of an interindividual differences construct (i.e., construct content and hierarchy), (b) the nomological relations between the construct and other interindividual differences other variables (i.e., concomitants, antecedents, or outcomes), or (c) the relationship between the construct and measurable indicators (i.e., the measurement model) may not hold at the individual level. Personality psychologists who aim to produce knowledge about individuals and their characteristic experiential-behavioral pattern rather than about populations should consider the approximation provided by interindividual differences constructs and measures deficient at best.
Personality descriptions require personalized approaches
According to Lamiell (1981), “the problem of describing an individual’s personality is essentially one of identifying those qualities or attributes that he or she manifests with some degree of regularity over time and across situations” (p. 279). This assertion is in line with common definitions of personality referring to a consistent patterning within the individual (see above). As consistency within individuals cannot be inferred from or empirically supported by patterns of between-person associations, personality description needs to be approached from a person-specific, within-person perspective. Personality traits are only revealed and can only be inferred when consistent experiential-behavioral patterns can be empirically demonstrated in the life of the individual (Allport, 1937, 1961; Lamiell, 1981). This requires longitudinal data and individual-level, or
While the majority of statistical tools in personality psychologists’ methodological toolbox have been developed for population-level analyses and especially for describing interindividual differences, a growing number of statistical approaches for
Prediction: Improving within-person predictions through personalization
Predictive accuracy is especially important when personality information is used to solve real-world problems where the outcome of interest is unknown but needs to be estimated from available data. A prediction model is optimized for out-of-sample performance (Mõttus et al., 2020). The model is constructed (or trained) using existing data in which the outcome is known to accurately estimate the outcome from a set of predictors in new (i.e., previously not accessed) data. Following this understanding, prediction models can be used for estimating future outcomes (i.e., forecasting or prognosis), concurrent outcomes (i.e., detection or diagnosis), or even outcome variables that lie in the past (i.e., statistical retrodiagnosis or backcasting). For example, a prediction model could be optimized to estimate college students’ future academic performance (e.g., grades on an upcoming exam), current performance (e.g., grades on a present exam), or past performance (e.g., current grades reflecting past accomplishments or high school grade point averages). On the surface, prediction models can look just like descriptive models. For example, both can be multiple regression models. Importantly however, predictions—unlike descriptions—do not typically require parsimony and interpretability, and they may actually benefit from complexity. Machine learning methods are particularly useful in this context, as they (a) implement cross-validation to achieve robust predictions, (b) allow for the incorporation of a multitude of predictors, and (c) can capture intricate patterns in the data (including nonlinear relationships and complex interactions) that are often overlooked by traditional approaches (Yarkoni & Westfall, 2017). Advancements in machine learning and technology-based data collection have led to a surge in predictive personality research in recent years (Bleidorn & Hopwood, 2019; Rauthmann, 2020; Stachl et al., 2020; Tay et al., 2022).
In predictive personality research, as in descriptive research, a distinction can be made between between-person and within-person level approaches: Outcome variables can be interindividual differences (e.g., Big Few traits or work, health, and other life outcomes) or variables that vary within individuals, such as states or symptoms. Between-person level prediction models can be especially useful when personalization is applied at scale and when the desired outcome is defined at the population level. For example, personality computing, that is, automated prediction of personality trait levels, usually from digital data (e.g., Phan & Rauthmann, 2021; Stachl et al., 2020) can be employed to estimate people’s trait profiles and then provide them with tailored content (see later and Matz et al., 2023). The desired outcome may be, for example, to increase the rate of people who click on an ad, buy a product, participate in a program, or change certain behaviors. This means that the success of these types of personalization approaches is not evaluated at the level of the individual person but at the population level. Similarly, machine learning models can also be used as a method for grouping people into relevant categories (see Dwyer et al., 2018, for an overview of machine learning approaches for diagnosis, prognosis, and treatment in clinical settings).
At the within-person level, prediction models can be constructed, for example, to detect momentary or forecast future behavior (e.g., Beck & Jackson, 2022b), affect (e.g., Bogomolov et al., 2013; LiKamWa et al., 2013; Meegahapola et al., 2023; Taylor et al., 2020), stress (e.g., Carneiro et al., 2019; Taylor et al., 2020), and other psychological states from a variety of data sources. In the traditional one-size-fits all approach, a prediction model is developed using cross-validation across sub-samples of individuals, meaning that the model is trained and then tested on disjoint sets of people (Meegahapola et al., 2023). As a result, predictive accuracy across different kinds of samples is maximized, on average. However, many of the issues raised in the previous section apply similarly to within-person prediction problems. As we have outlined above, different constructs are relevant for different people, and within-person patterns may vary widely from person to person. Therefore, an omnibus population-level model that has been optimized for the average person may not provide the most accurate outcome estimates for a given individual.
Personalized approaches to prediction could maximize the prediction accuracy for a desired outcome variable for the specific individual. For example, Beck and Jackson (2022b) used different machine learning techniques to predict future studying and procrastinating as well as experienced loneliness at the level of the individual from past states, situation variables, and time parameters. Personalized prediction models achieved impressive accuracies, and the most predictive features varied widely across individuals, indicating the potential advantage of a personalized approach. More direct evidence is provided by research suggesting that person-specific prediction models mostly outperform population models (e.g., LiKamWa et al., 2013). While
Personalized or partially personalized within-person prediction models can be utilized in numerous ways, offering a flexible and dynamic approach to understanding and influencing individual behavior and experiences. For example, (partially) personalized prediction models can be employed to detect or forecast mental health problems and shifts in mood, well-being, or stress, enabling timely and suitable interventions (see also
Explanation: Looking for causal relations at the individual level
Next, personality researchers may seek to explain phenomena within their domain by identifying the causal factors and mechanisms underlying their emergence. However, the abstract, decontextualized, and relativistic nature of many personality constructs might render it impossible to pinpoint their causes or causal impacts (Mõttus et al., 2020). Also, explanations involving the sort of generalized abstractions we often use to describe interindividual differences are especially prone to end in circularity. For example, traits are often conceptualized and operationalized as summary descriptions of (interindividual differences in) state patterns while, at the same time, serving as the causal entities that underlie these patterns (i.e., tautological fallacy; e.g., Cervone, 2021; Uher, 2013, 2021). That is, the fact that individuals behave, think, and feel in a certain manner cannot be the explanation for
The specific causes of personality phenomena and the mechanisms by which they are produced can only be revealed by examining them where they occur: at the level of the individual person and contextualized. Further, causal relationships may only be detectable when the causes are represented by units at a level of abstraction/concreteness that is appropriate for the phenomenon to be explained (Mõttus et al., 2020). Empirical findings of considerable interindividual variability in within-person patterns (see earlier) point to the conclusion that, in many respects, not all individuals function equally or even similarly. While some causal relations and mechanisms may be common to all or common to most, others might be person-specific or, at least, specific to certain subpopulations. Establishing causality usually entails the experimental manipulation of the conditions and/or causal processes to provoke change in the target outcome. For example, to identify dietary intolerances, patients are often advised to restrict certain foods for a certain amount of time and to meticulously document changes in symptoms. Done systematically and sequentially, this enables patients and doctors to rule out the foods that do not contribute to food intolerance symptoms and to delimit the foods that might cause them. Similar approaches can be used to empirically establish the person-specific causes of personality-relevant experiential-behavioral states.
Single-subject experimental designs are typically employed to evaluate the effectiveness of behavioral interventions but can also be used to test theories at the individual level and in real world settings (McDonald, Quinn et al., 2017). In
Developing and implementing personalized
Change: Personality and personalized interventions
Beyond the description, explanation, and prediction of phenomena, psychology also serves a fourth goal: to effectively change human behaviors and experiences. This goal is, of course, especially central for applied psychological disciplines but less emphasized in personality psychology and other basic research fields. However, personality research also pertains to change (as a major goal of psychology 13 ) in two main ways. First, personality (or personality-relevant variables) can be the target of change interventions. A growing subfield of personality psychology has been dedicated to examining the feasibility of effecting targeted change (aspect of) people’s personalities though intervention (e.g., Bleidorn et al., 2019; Hudson, 2021; Hudson et al., 2019; Hudson & Fraley, 2015; Jackson et al., 2021; Olaru et al., 2023; Roberts et al., 2017; Stieger et al., 2023). Personality change interventions are not inherently personalized as a one-size-fits-all protocol can be implemented across participants, but intervention designs often include at least some tailored elements (e.g., self-selected change goals; Hudson et al., 2019; Olaru et al., 2023; Stieger et al., 2021). Second, personality science may theoretically or empirically inform intervention research and development in other domains, from marketing to clinical psychology (for an overview, see Matz et al., 2023). Below, we focus on the latter, that is, the overall applied relevance of personality psychology which clearly includes but also goes far beyond personality change interventions.
Personality-based personalization of psychological interventions
The need for personalized psychological interventions has long been recognized, especially in the clinical field where over half a century ago Paul (1967) famously wondered “what treatment, by whom, is most effective for this individual with that specific problem, and under which set of circumstances?” (p. 111). This question can be generalized as follows: What works best for whom under which conditions, delivered how, when, and where (see also Taliç et al., 2023, for a similar formulation) The characteristic experiential-behavioral patterning of the individual and its underlying causes constitutes an obvious starting point for selecting or developing effective intervention. Here, we consider three broad approaches to personality-based interventions: (a) content matching based on interindividual differences, (b) adaptive personalization based on dynamic within-person data, and (c) person-specific tailoring. This categorization corresponds to different perspectives in personality psychology mentioned earlier (see
First, interventions can be matched to (groups of) targets based on interindividual difference variables (differential variable-centered perspective). Big Five traits and other interindividual differences have been identified as likely moderators of intervention effects such as the effects of health behavior change interventions (e.g., O’Connor, 2020), positive psychology interventions (e.g., Ng, 2015; Senf & Liau, 2013; Taliç et al., 2023), or school-based interventions (e.g., Mertens et al., 2022; Stoltz et al., 2013). The development of different interventions may be guided by a theoretical understanding of interindividual difference variables or by empirical findings regarding their associations with relevant outcomes. Individuals’ scores on these variables are then used to decide who receives which intervention option. For an overview of applied interindividual differences-based psychological targeting, we refer readers to Matz et al. (2023) who illustrated the potential of this approach through numerous examples from health, clinical, political, marketing, educational, and occupational psychology. 14 When the interindividual differences paradigm is used successfully to guide personalization decisions, on-average improvements of intervention outcomes (e.g., success rate increases in the population) can be expected. That is, between-person knowledge can be used to effectively personalize interventions at scale across the target population. However, at the individual level, the predictive value of between-person associations of a magnitude typically seen in psychology may be minimal to none (see also Mõttus, 2022). For instance, a significant correlation between a trait variable and a positive intervention outcome does not necessarily mean that an individual scoring high on that trait will probably respond well to the intervention. Therefore, personalized interventions based on an understanding of interindividual differences can be effective at the population level and at the same time ineffective for most individuals. To illustrate this point, we consider the following hypothetical scenario: A social media platform uses personality computing algorithms to accurately infer users’ trait profiles from their data and allows advertisers to opt for personality-targeting. A political campaign works with personality psychologists to tailor messages so that they may appeal to individuals with different trait configurations. Personality-tailored online ads are seen by 10% of the electorate, and another 10% receive a non-personalized message. Personalization is effective in increasing turnout among potential voters from 40% (in the non-personalized condition) to 41%—a gain that could amount to thousands of votes which may decide a close election. At the same time, this also means that only one in 60 non-voters has been swayed by the personalized message compared with the standard message (i.e., the share of non-voters dropped from 60% to 59%), meaning that personalization was ineffective for the vast majority of targeted non-voters (i.e., 1–1/60 = 98.3%). For interventions that target individuals (e.g., in clinical, educational, or work settings) rather than being implemented in larger groups to influence mass phenomena, interindividual difference information can provide indications as for which intervention might have the highest likelihood of success. Yet, depending on the strength of the between-person relationship (e.g., the moderating effect of the trait variable on the intervention effect), the expected effect at the individual level might be minimal. Further, many descriptive observations of trait-by-intervention interactions come from studies that have been conducted with a few broad trait variables, relatively small samples, and without cross-validation. Therefore, their out-of-sample predictive value is unknown (and likely limited), and in many cases, additional research is required before personality trait variables should be used to effectively match (groups of) individuals with particular interventions.
Second, interventions can be selected or adjusted based on interindividual differences in within-person patterns or based on intraindividual changes (person-centered perspective). The former requires a baseline phase during which ambulatory data is collected for each person, which is then analyzed to identify (lagged) within-person relations between relevant outcomes and candidate determinants. These patterns inform the selection of an intervention for the individual that targets their personal determinants.
Lastly, a truly person-specific intervention can be developed for the individual target, by tailoring general principles to their specific manifestations (i.e., surface personalization) or based on a person’s idiosyncratic experiential-behavioral patterns and processes (i.e., deep personalization; Wright & Woods, 2020). The former is, in all likelihood, commonplace in most applied settings where practitioners interact directly with a target, and adjust (elements of) standardized intervention protocols to better suit the particular individual based on their own judgment and experience. For example, this is often the case in clinical practice (Wright & Woods, 2020; see also Cohen et al.’s (2021) and Deisenhofer et al.’s (2023) overview articles on personalization in psychotherapy). Deep personalization, on the other hand, requires a functional understanding of a person’s individuality.
The examples provided above for different personalization strategies have mostly been focused on tailoring the content of the intervention (i.e., what) to the target person (i.e., for whom) and their momentary condition. However, similar strategies can also guide other intervention decisions such as the modality (i.e., how; e.g., intervention type, techniques, components, intensity, delivery format, etc.), the timing and frequency (i.e., when; e.g., temporal or event-contingent offset, duration, etc.), and the broader context (i.e., where; e.g., environment in which the target is most receptive to the intervention). Particular opportunities for personalization vary across cases of intervention application, and they depend on the practitioner’s capability to collect and utilize high quality data about the individual target. Developing and implementing personalized interventions is usually labor-intensive and time-consuming, and their incremental practical value over generic one-size-fits-all solutions should be carefully weighed against the higher costs (Matz et al., 2023). As technological advancements continue to give rise to new ways of leveraging large amounts of data and automating expensive processes, highly personalized interventions will likely become ever more feasible.
Solidifying the applied relevance of personality research
Personalization could have the potential to vastly increase the effectiveness of psychological interventions at the level of individuals. However, in most areas and especially in high-stakes applied settings (e.g., psychotherapy) a lot more basic research is needed to establish the efficacy, safety, and fairness of personalized approaches before they can become routine practice, and ethical considerations must be addressed (see Matz et al., 2023, for guidelines for ethically-informed personality-based targeting). We believe that there are grounds for cautious optimism. The personalization movement will continue to produce insights and innovations in areas of applied psychological research such as clinical psychology (e.g., Cohen et al., 2021; Deisenhofer et al., 2023; Wright & Woods, 2020), health psychology (e.g., Fisher & Soyster, 2019; McDonald, Quinn et al., 2017), educational psychology (e.g., Bernacki et al., 2021; Tetzlaff et al., 2021; Zhang et al., 2020), or marketing psychology (e.g., Anshari et al., 2019; Chandra et al., 2022)—with or without personality psychological input and involvement.
Arguably however, personality psychology can and should play a crucial role as a basic research discipline informing personalization strategies. Real-world personalization problems put the applied relevance of personality psychological theory and empirical knowledge to the test. With the (re-)emergence of the person-specific perspective, personality researchers bring to the table not only their grasp of interindividual differences, but they might also contribute their understanding of dynamic experiential-behavioral patterns within the individual (Jayawickreme et al., 2021; Kuper et al., 2021). In turn, basic personality psychology research might also benefit from personalization in applied research, both directly by gaining insights about individuals and indirectly by taking inspiration for further research. For example, insights from intervention personalization processes may reveal person-specific change mechanisms (Taliç et al., 2023). Further, empirical demonstrations of individual-level intervention effects can support causal explanations of personality-relevant psychological phenomena (see
Concluding remarks
All people are simultaneously identical in many respects and completely unique. A historical tension exists in psychology between the recognition of individual uniqueness and the desire to find general laws that apply to everyone (or at least to most or certain people). McAdams (1997) summarized this central problem in the form of a question: “If science seeks lawfulness across persons (nomothetics), then how can it make sense of, even leave room for, the uniqueness of the individual (idiographics)?” (p. 8). Personality psychologists have managed to mostly avoid having to provide a clear answer by treating the study of the individual and the study of people in general as complementary but distinct approaches to the subject matter of the field (i.e., idiographic vs. nomothetic research; see Phan et al., 2024 for a critical analysis of this dichotomy). Dominated by the interindividual differences paradigm and population-based research methodologies, the field has been getting away with largely ignoring individuality. However, population-based methodologies not only fail to acknowledge uniqueness, but they are also not suitable for establishing laws that apply across people (Hamaker, 2012; Lamiell, 2003; Phan et al., 2024). Therefore, research outputs often fail to accomplish the field’s major goals of describing, predicting, explaining, or changing experiences and behaviors at the individual level.
The goal of this article was to highlight the benefits of shifting (personality) psychology’s focus towards adopting a more person-specific perspective and personalized approaches. We have pointed to numerous and varied examples of tools and techniques that can be used to address individuality in personality research and beyond. Many of these are fairly novel as they require large amounts of data on individuals which has only recently become feasibly obtainable through digital technologies (Fahrenberg et al., 2007; Harari & Gosling, 2023; Miller, 2012; Renner et al., 2020; Schoedel & Mehl, 2023). Methodological innovation in the field has also been fueled by the re-emergence of the person-specific perspective and a growing recognition of the limitations of conventional approaches (Beck & Jackson, 2020; Hamaker, 2012; Lamiell, 2018; Molenaar, 2004; Phan et al., 2024). Importantly, the present work is not a comprehensive methods review, and our examples paint only a rough picture of the current methodological landscape, demonstrating that numerous viable alternatives to population-based approaches already exist. In descriptive research, personalized assessments and
While recent methodological advancements are promising, it is important to recognize that conducting and integrating person-specific research remains challenging, with much of the work still in a proof-of-concept phase. A key concern is that research may become overly driven by the available methods rather than by the fundamental questions that need to be addressed. When methods dictate the agenda, there is a risk of prioritizing the application of novel sophisticated techniques over the pursuit of genuine scientific insight, leading researchers to fit their questions to the tools rather than vice versa. Arguably, the widespread accessibility and continuous development of between-person methods has had a determining influence on the personality-psychological research agenda. This may serve as a cautionary tale for those who advocate for the adoption of personalized approaches. While these new approaches hold significant potential, they should serve the research objectives rather than define them. 17
Moreover, empirically-based
Personality, whether defined as an individual’s characteristic experiential-behavioral patterns (e.g., Baumert et al., 2017; Funder, 2004; Revelle & Condon, 2015) or as the dynamic system within the individual that produces these patterns (e.g., Allport, 1961) must be examined in individuals. Personalization ultimately aims to maximize the validity of research outputs at the individual level which makes it an obvious strategy for studying personality and related phenomena. Person-specific insights are essential for personality research, but they are also of primary concern in applied setting where the focal entity is a particular individual (e.g., a patient or a client).
18
In principle, the (personality) psychologist can study the individual without referring or generalizing to other people or to an entire population (see also Phan et al., 2024). Generally however, researchers aim for inferences that can be integrated at the between-person level. One-size-fits-all approaches produce data that—on the surface—appears directly comparable across individuals, but we have questioned the notion that uniformity implies comparability throughout the present work. Personalization, on the other hand, introduces heterogeneity that complicates statistical between-person operations, but this may actually be inevitable to provide information that shares the same
In conclusion, personalized research approaches may not only prove invaluable in advancing our scientific understanding of the experiences and behaviors of individuals—arguably the main objective of personality psychology as a field—but they may also provide a path toward a better grasp of general principles that operate within the complexity of real-life individual differences, offering a richer perspective on human experiences and behavior that encompasses both commonalities and interindividual variations. By understanding and discerning what is common to all, common to most, common to some, and what is truly unique to each individual, researchers may ultimately be in a position to integrate general, differential, and personality psychological perspectives and insights.
Footnotes
Authors’ note
This article is one of three articles that resulted from an expert meeting on “Dynamics of Personality: Integrating Nomothetic and Idiographic Approaches in one Conceptual Framework” sponsored by the European Association of Personality Psychology (EAPP).
Author contributions
Conceptualization: NM, LVP, SMD, and KHR; original draft: NM; and review and editing: NM, LVP, SMD, KHR, and AGCW.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This publication was made possible through the support of a European Association of Personality Psychology (EAPP) Expert Meeting Grant. The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the EAPP. Karl-Heinz Renner is supported by dtec.bw – Digitalization and Technology Research Center of the Bundeswehr. dtec.bw is funded by the European Union – NextGenerationEU.
