Resilience research in learning disabilities: Guiding principles from developmental psychopathology

Abstract

Background:

The learning disabilities literature has had a long-standing focus on identifying risk factors that can inform diagnostic assessment and guide intervention efforts. While many existing conceptual models of learning disabilities have acknowledged the theoretical possibility of resilience mechanisms (Bishop & Snowling, 2004; Catts & Petscher, 2022; McGrath, Peterson, & Pennington, 2020; Pennington, 2006), few studies have empirically investigated such processes. However, more attention has been recently directed towards resilience mechanisms that may operate in the context of learning disabilities. Initial studies have found that cognitive-linguistic abilities and individual differences in task behaviors may play a role in supporting academic skills (specifically reading) despite the presence of risk factors (Eklund, Torppa, & Lyytinen, 2013; van Viersen, de Bree, & de Jong, 2019). However, many empirical results that signal potential for resilience processes are difficult to synthesize because methods and analyses are not consistent across studies. For example, studies may operationalize and analytically investigate resilience mechanisms in differing ways, leading to challenges in understanding the replicability and generalization of findings.

Conclusions:

In this paper, we make recommendations for standardized practices for studying resilience mechanisms, drawing from the well-validated guidelines for resilience science that have been established within the developmental psychopathology (DP) literature (Masten, Narayan, & Wright, 2023).This theoretical paper outlines the DP perspective on resilience, demonstrates the application of specific DP principles to the learning disabilities field, and provides a set of methodological recommendations for the inclusion of resilience-based methods alongside the prevailing risk-based methodology in learning disability research.

1. Introduction

The field of learning disabilities (LD) has had a long-standing focus on identifying risk factors that can inform diagnostic assessment and guide intervention efforts. While most of the widely accepted conceptual models of LDs have acknowledged the theoretical possibility of resilience mechanisms (e.g., Bishop & Snowling, 2004; Catts & Petscher, 2022; Margalit, 2003; McGrath et al., 2020; Pennington, 2006), historically there has been little empirical work that explicitly investigates resilience processes (for examples see, Catts & Petscher, 2022; Haft, Myers, & Hoeft, 2016; Kiuru et al., 2013; Margalit, 2003; Slomowitz et al., 2021; van Viersen, de Bree, Kroesbergen, Slot, & de Jong, 2015; van Viersen et al., 2019; Yu, Zuk, & Gaab, 2018). Furthermore, differing methodologies that do assess resilience processes have made it difficult to synthesize results across empirical studies. As such, the LD field could benefit from standardized practices for studying resilience mechanisms to inform a more cohesive and comprehensive, resilience-informed corpus of research.

Well-validated guidelines for studying resilience have been established within the developmental psychopathology (DP) literature (Masten, 2001, 2006; Masten et al., 2023), which can be adapted for the learning disabilities research context. The goal of this paper is to provide guidance regarding the application of DP theories and methods to the learning disability field, particularly related to identifying and distinguishing promotive and protective factors. This theoretical paper outlines the DP perspective on resilience, demonstrates the application of specific DP principles to the learning disabilities field, and provides a set of methodological recommendations for the inclusion of resilience-based methods alongside the prevailing risk-based methodology in learning disability research.

In this paper, we are using the term learning disability (LD) to encompass the specific learning disorders described in the Diagnostic and Statistical Manual, Fifth Edition (DSM-5) and characterized by significant difficulties with learning and applying academic skills that are persistent (despite interventions to address the area of academic weakness), cause significant interference in school/work or daily life, and are not better explained by other factors (i.e., intellectual, physical, environmental, neurological, psychiatric factors; American Psychiatric Association, 2013). While this definition of a learning disability includes difficulties with one or more academic skills in the reading, writing, and/or math domains, we primarily draw on examples from the reading disabilities literature, because the existing work has largely focused on reading skills, most commonly difficulties with single-word decoding (also known as dyslexia).

1.1. Developmental psychopathology perspective

The DP perspective provides an expansive theoretical lens for conceptualizing mental health disorders and was initially established to better understand the significant variability in developmental pathways related to psychopathology (see Cicchetti & Toth, 2009; Masten, 2006; Masten, 2007; Masten & Reed, 2009; Masten et al., 2023, Sroufe & Rutter, 1984). The spirit of the DP perspective is to take a developmental approach to understanding why psychopathology arises and how developmental trajectories move both towards and away from psychopathology across the lifespan (Masten, 2006; Sroufe, Rutter, 1984). The DP perspective accomplishes these goals by considering both risk and resilience processes (Masten et al., 2023; Rutter, 2012). This perspective shaped the field of resilience research and the study of those individuals who seem to “beat the odds” or function well despite, amidst, or following adversity (Masten, 2001; Masten et al., 2008; Masten et al., 2023). The DP perspective has been crucial to uncovering a more nuanced and developmentally sensitive understanding of various mental health and neurodevelopmental conditions across the life span (Rutter, 2013). While this perspective has been incorporated into previous learning disability work to varying degrees (e.g., Catts & Petscher, 2022; Haft et al., 2016; McGrath et al., 2020; Yu et al., 2018), it is timely to now provide guidance and illustrative examples for how the DP perspective can be more deeply integrated into learning disability research. To this end, we specifically focus on methods to identify promotive and protective factors as an initial example of how to apply the DP framework to study resilience within learning disability research.

1.1.1. Definitions of resilience

Resilience is a familiar but often imprecisely defined term. According to prominent DP researchers, one of the most commonly used definitions of resilience, in its essence, is the process of obtaining better-than-expected outcomes or developmentally appropriate functioning due to successful adaptation in the context of risk or adversity (Masten, 2001). It is important to note that this is not the only proposed definition of resilience, and that definitions continue to develop as research on this concept evolves (Masten, Lucke, Nelson, & Stallworthy, 2021). We chose to use this definition because it is one of the most commonly used, and it generalizes well to learning disability research. This definition also highlights two important aspects of the concept of resilience that 1) risk is a prerequisite condition for resilience, and 2) resilience is a dynamic process and not a trait or finite outcome that can be measured at one point in time (Masten et al., 2023).

The first point emphasizes that the context of risk or adversity is a prerequisite for resilience (Masten et al., 2021). Of course, individuals in low-risk contexts can also demonstrate positive developmental trajectories. However, DP theory distinguishes these two scenarios such that if there is no adversity, then the mechanisms are referred to as “competence,” rather than resilience (Masten et al., 2023). Both competence and resilience mechanisms are important developmental processes, but only the latter is the focus here. For the purposes of this paper, we are interested in resilience processes that may impact academic outcomes specifically, defined as the process of achieving better-than-expected and/or developmentally appropriate academic outcomes in the context of risk for LDs or the presence of an LD diagnosis. Appels, van Viersen, van Erp, Hornstra, and de Bree (2024) use the term “academic resilience” to describe this process. While beyond the scope of this paper, other learning disability researchers may be interested in examining resilience in other non-academic aspects of functioning in individuals diagnosed with LDs, such as social-emotional functioning, educational or vocational attainment, general life satisfaction, etc. (Haft et al., 2016; Kiuru et al., 2012; Margalit, 2003; Murray, 2003; Shany, Wiener, & Assido, 2013).

The second point regarding the definition of resilience underscores that resilience is a process not an outcome (Masten, 2001; Rutter, 2012). While it is difficult to measure a developmental process that is unfolding, particularly in cross-sectional studies, it is still possible to investigate the factors that may contribute to resilience processes—namely, the presence of promotive and protective factors. Just as we assume that risk factors contribute to the presence of academic weaknesses, we assume that promotive and protective factors contribute to better-than-expected academic outcomes in the context of risk.

1.1.2. Critiques of resilience theory

It is important to highlight some critiques of resilience as a concept. In previous applications of resilience in the DP literature, this term was often conceptualized as a set of individual difference factors, while risk factors were typically considered to be external/environmental factors (Kalisch et al., 2019). This internalization of resilience as stemming from exclusively within-person individual difference factors may place the burden of this process on the individual; it also could be misinterpreted to cast implicit blame on those who did not experience resilience. What was missing in the earliest definitions of resilience were the ways that systems and environmental contexts could operate to support resilience (Masten et al., 2023). The current paper discusses resilience factors that are individual-level differences because these are the most commonly studied factors in the academic resilience research to date, and there has been less focus on systemic factors contributing to resilience mechanisms. Nevertheless, we encourage the consideration of risk and resilience from both systems and individual differences perspectives (Appels et al., 2024; Hunsu, Oje, Tanner-Smith, & Adesope, 2023; Masten et al., 2023). A secondary problem that arose out of early resilience conceptualizations is that individuals who were identified as “resilient” were sometimes falsely considered “invulnerable” to the impacts of risk across all domains of functioning (Kalisch et al., 2019; Rudd, Meissel, & Meyer, 2021; Rutter, 2023). However, while resilience processes may contribute to stronger-than-expected functioning in a particular domain, the presence of this process does not guarantee complete protection from risk over time and in other domains of functioning (Rutter, 2023).

Finally, we wanted to call attention to our use of the term “risk factor” to denote factors that are associated with lower academic scores and/or higher likelihood of an LD identification. Recent work has highlighted that terminology arising from traditional medical models, including the term “risk factor,” has contributed to harmful discourse regarding neurodiverse populations by being stigmatizing and negatively value-laden, specifically when used in reference to the neurodevelopmental disability itself (particularly within the autistic community; Dwyer et al., 2022). As such, we carefully considered alternative terms, such as “elevated likelihood” or “predictive factor” which are more value-neutral and lack the same connotations as “risk” (Dwyer et al., 2022). For conceptual clarity, we ultimately chose to proceed with the term “risk factor” to clearly illustrate the distinctions between risk, promotive, and protective factors, given the overall goal of this paper to introduce this new framework to the LD field. However, we acknowledge the limitations and connotations of this term in its alignment with the medical model. We recommend that future academic resilience work continue to critically evaluate terminology, methodology, and conceptualizations to promote equity and inclusivity.

1.2. Mechanisms of resilience: promotive and protective effects

As mentioned previously, resilience mechanisms can be quantified via two types of resilience factors, termed promotive and protective factors. While promotive and protective factors both contribute to mechanisms of resilience, they do so in distinct ways and are identified via separate analytic methods. While most have likely heard of the term “protective factors” in the context of resilience work (e.g., Catts & Petscher, 2022), promotive factors are rarely discussed in the learning disability literature (for an exception see, Slomowitz et al., 2021). The lack of distinction between these factors tends to result in resilience-related effects being broadly labeled as protective factors, although this is not always accurate according to the broader DP resilience framework (e.g., Masten et al., 2023). Instead, some of the resilience factors identified might more accurately be described as promotive factors, rather than protective factors (see explanation below for analytic distinctions). This is more than a semantics issue, as the misspecification of these terms results in difficulties with accurately identifying the types of resilience processes that may best support children at risk for LDs or with LD diagnoses. In particular, the distinction between promotive and protective factors matters because of their association with beneficial but “gap-maintaining” (i.e., promotive factors) versus “gap-closing” (i.e, protective factors) effects between high and low risk groups. For example, identifying intervention targets which contribute to “gap-closing” effects (i.e., protective factors) on academic outcomes for children with and without LD could be important for intervention development. Such interventions might encourage “catch-up growth” (Fielding, Kerr, & Rosier, 2007) for children with LD relative to children with typical academic development. Distinguishing between promotive and protective factors is a step towards understanding resilience mechanisms in LD more broadly.

While there is a plethora of important guidelines from the DP perspective that learning disability research can adopt to standardize investigations of resilience, we strategically focused this paper on adapting DP methods for identifying promotive and protective factors within the LD literature. This topic is important because it provides fundamental information on operationalizing the core components that give rise to resilience mechanisms. In what follows, we provide definitional information about promotive and protective factors, highlight the importance of distinguishing between them, give examples of how to empirically investigate them, and discuss future directions for resilience research within the learning disability literature.

1.2.1. Promotive factors

Promotive factors are associated with positive outcomes equally regardless of whether they occur in low- or high-risk contexts (Masten et al., 2023; Masten & Reed, 2009). Thus, promotive factors result in comparable improvement for both higher and lower risk groups leading to beneficial but “gap-maintaining” effects between risk groups (Slomowitz et al., 2021). While promotive factors benefit all individuals regardless of risk status, they are considered associated with positive development when identified in those with lower levels of risk and associated with resilience when identified in those with higher levels of risk. Analytically, promotive factors are typically identified via main effects in linear models (Fig. 1B).

Fig. 1.

A–D. Seesaw metaphor for risk/resilience mechanisms extended from Catts & Petscher’s cumulative risk/resilience seesaw model (Catts & Petscher, 2022).

Within the DP literature, examples of promotive factors for positive psychosocial outcomes include constructs such as positive parenting, stronger cognitive abilities, stronger executive functions, and higher socioeconomic status, as these factors are generally beneficial to most individuals regardless of their level of risk(Masten, 2006; Masten et al., 2008). Within the learning disability literature, some cognitive promotive factors for reading ability include stronger skills in vocabulary, rapid automatized naming, verbal working memory, and processing speed (Eklund et al., 2013; Slomowitz et al., 2021; van Viersen et al., 2019).

1.2.2. Protective factors

Protective factors partially or completely attenuate the effects of risk factors on outcomes, specifically as the level of risk rises. These factors buffer against negative outcomes by diminishing the impact that a risk factor has on an outcome. Thus, when comparing the effect of a protective factor between high and low risk groups, the higher risk group shows more benefit from the protective factor compared to the lower risk group (even if the lower-risk group’s scores/performance remains higher overall). In other words, protective factors contribute to a “gap-closing” effect between outcomes of the higher and lower risk groups (Masten & Reed, 2009; Masten et al., 2008). Protective factors are analytically represented as interaction terms in linear models (Risk x Protective Factor; Fig. 1C).

While some factors exist that only serve protective, but not promotive functions, the DP literature shows that many factors are both promotive and protective (Masten et al., 2023). In other words, typical resilience factors identified in previous literature are frequently associated with improved functioning in all individuals regardless of their risk status (i.e., promotive factors, main effects), and they also confer additional benefit particularly for those at higher levels of risk (i.e., protective factors, interactions). Some examples of protective factors that are not promotive are airbags/seatbelts, vaccines, and automated external defibrillators (AEDs). These tools become critical for health and/or survival in contexts of risk (e.g., in traffic collisions, when in contact with a virus, or when experiencing cardiac arrest), but are generally not otherwise helpful when risk is not present (Masten et al., 2023; Narayan, Merrick, Lane, & Larson, 2023). We give these examples for the purposes of teaching about the concept of protective factors but, of course, they are not directly relevant to LD research.

To provide an example of an academic-specific protective factor, we highlight the results from Torppa (2025), a study from this special issue. This study focused on identifying resilience in mathematic performance where risk for weaker mathematic performance by the child was based on parental mathematic skills and potential promotive/protective factors were aspects of the home numeracy environment (HNE). Torppa (2025) identified that certain types of HNE activities (i.e., explicit, instructional math-related activities, such as counting objects, printing numbers, etc.) served as protective factors for math performance in 1st-2nd grade. In other words, certain HNE activities only benefitted those at higher risk, thus contributing to a gap-closing effect on math outcomes between the higher and lower risk groups. Notably, the HNE types that contributed to protective effects were not also promotive factors (i.e., the main effects were not significant; the performance of lower risk individuals did not significantly benefit from these HNE activities). These protective effects were specific to only certain types of HNE activities, and, in some cases, the protective effects were developmentally sensitive (e.g., factors were protective at one time point, but non-protective at a later timepoint; see the Torppa (2025) study within this special issue for more details).

1.3. Visual depiction of promotive, protective, and risk factors: A seesaw analogy

Here, we provide a visual analogy (Fig. 1) to illustrate the ways in which promotive, protective, and risk factor effects operate in context with one another. This seesaw visual analogy was adapted from Catts and Petscher’s (2022) risk/resilience seesaw model. While Catts and Petscher (2022) used the seesaw analogy to explain cumulative risk/resilience mechanisms, we only consider single factors in our figures for ease of interpretation, and we add a visual illustration of the distinction between risk, protective, and promotive factors.

Seesaw/Scale.

The seesaw in all of the figures represents the net impact of all risk, promotive, and protective factors on development. Since these models are seesaws, a factor’s effect size is demonstrated through its size/weight. A downward tilt on the risk side of the equation (right hand side) indicates lower performance on a given academic skill. As such, factors that counterbalance this right-side downward tilt are describing resilience processes, promotive and/or protective effects.

Blocks.

The blocks represent a promotive factor if they have a “+” sign on them, or a risk factor if they have a “-” sign on them. The bigger the box, the “heavier” the weight. The weight of the block and how far it tips the scale represents the strength of a main effect. Importantly, resilience occurs when a risk factor’s weight is counterbalanced in any way. In other words, if any resilience factors contribute to a risk factor being “less heavy” to any degree (even if the effect of that risk factor is not totally negated), then this would be considered a resilience process (i.e., all but Fig. 1A represent resilience processes). Here we depict single blocks to illustrate single risk or promotive factors for simplicity, but these processes are cumulative, and we would expect multiple factors with differing effect sizes for both risk and promotive processes, consistent with the multiple factors model of learning disabilities (McGrath et al., 2020; Pennington, 2006).

Triangle & Lasso.

The triangle represents a protective factor. The lasso represents a protective interactive effect (i.e., Protective × Risk interaction), where the protective factor reduces the weight that the risk factor contributes to the seesaw. An important idea to keep in mind is that the protective factor itself is not the main ingredient for a protective effect—it is the lasso. In other words, what matters most is the presence of a protective interaction effect between a risk factor and an additional factor; there is nothing inherently protective about that additional factor unless it shares an interactive effect with risk. Again, we illustrate a single protective factor for simplicity, but we would expect there to be multiple protective factors that could mitigate risk in cumulative models.

Explanations of Risk, Promotive & Protective Factors in Fig. 1A depicts a risk effect.

It is analogous to the main effect typically expected of a risk factor without the presence (or consideration) of any resilience factors. Fig. 1B depicts the effects of a promotive factor in the context of a risk factor; the promotive factor is counterbalancing the weight of a risk factor and contributing to resilience. The promotive factor may partially or completely counterbalance the risk factor. Even if the promotive factor’s main effect was not as strong (“heavy”) as the risk factor’s main effect, the promotive factor would still contribute to resilience because performance would be better than expected based only on the presence of risk alone. Fig. 1C: This figure depicts how a protective factor operates and contributes to resilience in the context of a risk factor. Specifically, the protective interaction mitigates the weight of the risk factor, making the risk factor less “heavy” and resulting in the risk factor contributing less weight on the scale. This is analogous to how a protective interaction operates analytically: it contributes to resilience by making the risk factor less impactful on an outcome. Fig. 1D demonstrates how a promotive, protective, and risk factor could be expected to operate together. The main effect of the promotive factor negates some of the weight of the risk factor. In addition, the protective factor contributes to resilience by further reducing the impact or “weight” that the risk factor contributes to the scale.

Caveats to this Illustration.

While these graphs provide visual illustrations of risk and resilience effects, these are merely metaphors for the purposes of clearly distinguishing promotive from protective effects. For example, there are likely various patterns and mechanisms for how risk, promotive, and protective factors operate that cannot be captured by our simplified analogy. For the sake of simplicity, there are also other possible interaction patterns that were not discussed in these models (e.g., diathesis-stress interactions, skill-enhancing interactions (Slomowitz et al., 2021). Finally, this figure does not illustrate the cumulative effects of multiple factors or developmental effects that are likely at play for risk and resilience mechanisms. Nevertheless, we hope that these visual illustrations help communicate some of the central ideas regarding distinguishing promotive and protective effects in resilience research.

1.4. Brief comment on academic resilience literature

For a detailed and systematic scoping review of the extant literature on academic (specifically reading) resilience, readers should turn to Appels et al. (2024) within this special issue. Here, we briefly review previous work that examines resilience factors for reading and math outcomes to give a sampling of potential resilience factors for LD researchers to consider in future work. In the brief summary below, we label “undetermined resilience factors” as those where further research is needed to distinguish whether the factor is promotive and/or protective. This resilience work is just emerging in the learning disabilities literature and so no clear consensus yet exists. As such we offer these findings for hypothesis generation.

Within the reading resilience literature, resilience factors have been investigated across various levels of analysis. Specifically, for reading-related studies in which risk for weaker reading outcomes is defined in a variety of ways (e.g., familial risk, low early literacy skills, low phonological awareness skills, reading disability diagnostic status) past work has suggested the following resilience factors: cognitive factors (e.g., promotive factors: rapid automatized naming [RAN], processing speed, verbal working memory, vocabulary; undetermined resilience factors: verbal reasoning, grammar, morphology, executive functioning skills; Berninger & Abbott, 2013, Catts & Petscher, 2018; Cavalli et al., 2016; Eklund et al., 2013; Haft et al., 2016; Law, Wouters, & Ghesquière, 2015; Thompson et al., 2015; van Viersen, Kroesbergen, Slot, & de Bree, 2016, 2019; 2021), emotion/motivation factors (e.g., undetermined resilience factors: coping skills, growth mindset, task-focused behavior, self-understanding, hopeful thinking; Eklund et al., 2013; Goldberg, Higgins, Raskind, & Herman, 2003; Haft et al., 2016; Idan & Margalit, 2014; Ofiesh & Mather, 2023; Petscher, al Otaiba, & Wanzek, 2021), social/environmental factors (e.g., undetermined resilience factors: positive teacher support, peer acceptance, family cohesion; Catts & Petscher, 2018; Kiuru et al., 2013; Idan & Margalit, 2014; Ofiesh & Mather, 2023), and neural factors (e.g., undetermined resilience factor: increased right hemisphere structural development or activation during reading tasks; Yu et al., 2018; Zuk et al., 2021).

The math resilience literature is significantly smaller than that of the reading literature, but recent investigations in this area have also begun to uncover resilience factors. Specifically, in the context of risk for math difficulties defined by various factors (i.e., weaker parental math abilities, developmental language disorder status, low early arithmetic abilities), past work has suggested the following resilience factors: cognitive skills (promotive factors: rapid automatized naming [RAN], nonverbal reasoning, counting, spatial relations; protective factors: number concepts, oral language, nonverbal reasoning skills; undetermined resilience factor: verbal reasoning skills; Psyridou et al., 2023; Kleemans, Vissers, & Segers, 2025), motivation factors (undetermined resilience factors: self-concept, task value, homework persistence; Psyridou et al., 2023), and societal/environmental factors (promotive: socioeconomic status; protective: home numeracy environment; Torppa (2025), Psyridou et al., 2023). Despite the limited number of math resilience studies, a majority of these more recent studies have used designs that can distinguish promotive and protective effects.

To help inform future investigations of academic resilience, we next outline the methodological requirements and analytic approach needed to delineate promotive from protective effects. These methods are applicable to investigations of resilience regardless of the academic outcomes of interest, but we expect that some risk, promotive, and protective factors identified by this work will be specific to academic domain while some risk, promotive, and protective factors may be general across academic domains.

1.5. Distinguishing promotive and protective factors: applications to academic resilience research

To build an analysis that allows a researcher to identify and distinguish promotive and protective effects, the analytic plan must have: 1.) a statistical test of main and interaction effects and 2.) the full 2 × 2 conceptual matrix of presence/absence of a suspected resilience factor and high/low risk groups to determine gap-maintaining and gap-closing effects (Masten, 2001).

First, the model must test for promotive effects via a main effect and protective effects via an interaction. As described above, it is possible for a resilience factor to be both promotive and protective since main effects and interaction effects are not mutually exclusive and in fact commonly co-occur. Further, if a significant interaction is identified, this does not necessarily mean that the interaction effect is protective. The form of the interaction must show that the factor in question is more beneficial for the higher risk group than the lower risk group. In other words, there are interaction effects that would not be consistent with protective factors (e.g., skill-enhancing interactions identified in Slomowitz et al., 2021; Torppa, 2025) and so a careful investigation and plotting of the interactive effects are crucial.

Second, in order to definitively understand whether a factor is promotive and/or protective, the study design must align with a conceptual 2 × 2 matrix of the presence/absence of the resilience factor across the high/low risk groups (Masten, 2001; Masten et al., 2021). This 2 × 2 matrix results in high risk (HR) and low risk (LR) groups divided into those with and without the resilience factor (+ versus -) (i.e., HR+, HR-, LR+, LR-; see Table 1). These four groups are needed to determine whether a factor contributes to a gap-maintaining effect (i.e., promotive factors; the gap between + versus – groups is the same for HR and LR groups; Fig. 2A) and/or a gap-closing effect (i.e., protective factors; the gap between the + versus – groups is bigger for the HR groups than the LR groups; Fig. 2B) or other interactive form (i.e., skill enhancing; the gap between + versus – group is bigger for the LR groups than the HR groups).While the implementation of this conceptual 2×2 matrix design could guide data collection, it can also be applied to pre-existing datasets. However, when applying this matrix to previous datasets, researchers must ensure that they have a sufficient number of individuals within the lower-risk or non-LD group that both do and do not have the resilience factor of interest. For example, previous academic resilience work often divides the high-risk group into those with and without the resilience factor (i.e., HR+ and HR-groups), without similarly dividing the low-risk group into those with and without the resilience factor (LR; i.e., no separation of LR+ and LR-; see Table 1). This design limitation contributes to an inability to identify a protective effect because it does not allow for an interaction to be tested, thus it cannot be determined if the resilience factor is gap-maintaining or gap-closing between higher and lower risk groups (Masten et al., 2021, Fig. 2). Note that while we use categorical examples in this paper to simplify the discussion, it is also possible to think of the 2 × 2 matrix as dimensional where risk status and the resilience factor are measured continuously and entered into linear models as continuous main effects and continuous interaction terms.

Table 1.

2×2 conceptual resilience factor matrix.

	Resilience Factor Present	Resilience Factor Absent
Higher Risk for LD	HR+	HR−
Lower Risk for LD	LR+	LR−

LD = Learning Disability.

Fig. 2.

Hypothetical Promotive and Protective Factor Graphs.

To illustrate why it is important to include the full 2×2 conceptual matrix to definitively identify promotive and protective effects, we compare the results and conclusions of our recent paper (Slomowitz et al., 2021) to one of the common academic resilience study designs described above (i.e., investigating the presence/absence of a resilience factor in the HR group, but not the LR group; HR+, HR-, vs. LR). In Slomowitz et al. (2021), part of the study’s goal was to determine whether stronger vocabulary skills served as a promotive and/or protective factor for word-reading skills in children who were considered at higher-risk for reading disability due to weaker phonological awareness skills. This hypothesis was derived from the theory of semantic bootstrapping where vocabulary is viewed as one indicator of broader language abilities (Snowling & Melby-Lervåg, 2016; Muter & Snowling, 2009). Stronger vocabulary skills were tested as a promotive factor through a main effect on word-reading skills and were tested as a protective factor by a vocabulary x phonological awareness interaction. If the interaction showed that the association between weaker phonological awareness skills and weaker word-reading skills was mitigated by stronger vocabulary skills, it would be supportive of stronger vocabulary serving as a protective factor. In this case, rather than using categories (i.e., HR+, HR-, LR+, LR- see Table 2), we used dimensional measurement of the risk and resilience factors to establish main effects and interactions and a categorical follow-up of the interaction effect for interpretation (see Fig. 3).

Table 2.

Slomowitz et al. (2021) Matrix.

	Stronger Vocabulary Skills	Weaker Vocabulary Skills
Higher Risk (Weaker PA Skills)	HR+	HR−
Lower Risk (Stronger PA Skills)	LR+	LR−

Fig. 3.

Abstracted plot of results from Slomowitz et al. (2021) skill-enhancement interaction.

Results confirmed a main effect for vocabulary skills on word-reading outcomes, where stronger vocabulary skills were associated with stronger word-reading abilities. In addition, a significant interaction between vocabulary x phonological awareness was identified. However, upon further inspection of the form of the interaction, this study found that stronger vocabulary skills did not mitigate the effect of weaker phonological awareness (PA) on weaker reading outcomes, contrary to expectations for a protective effect. Instead, stronger vocabulary abilities enhanced word-reading abilities, specifically for children with higher phonological awareness skills (i.e., lower risk group) (Fig. 3). Since this is not an example of a protective effect, we labeled the interaction a “skill-enhancing effect” where vocabulary was especially beneficial for reading outcomes in children with low risk for LD (as defined by strong PA skills). As such, these results suggest that vocabulary is a promotive but not a protective factor. It is particularly important to emphasize that while this skill-enhancing interaction of vocabulary x phonological awareness on word reading performance was significant, its effect size was much smaller than that of the larger promotive effect of vocabulary on reading performance. As such, strong vocabulary skills (along with other cognitive factors including verbal working memory, processing speed, and rapid automatized naming tested in this study) served as important promotive factors for word reading skills.

Without the full 2 × 2 matrix in Slomowitz et al., 2021, we would not have been able to distinguish promotive from protective effects. If we had used the more common study design that is typical for the LD literature (i.e., HR+/HR-vs LR), we would have identified that stronger vocabulary skills were associated with stronger word reading abilities in the higher risk group, but we would not have been able to identify if stronger vocabulary was similarly associated with stronger word reading in the lower risk group and if the effect of stronger vocabulary was different in the risk groups. This latter part of the analysis is what is needed to identify an interactive effect that could be protective but could also follow a different interaction form as it did in Slomowitz et al., 2021 where there was a skill-enhancement interaction. We illustrate the application of the 2 × 2 matrix for the purpose of encouraging other researchers to consider this design which aligns with the DP framework and provides greater clarity on resilience mechanisms by distinguishing promotive and protective effects.

1.6. Future directions for academic resilience research

In addition to encouraging the explicit investigation of promotive and protective factors in future LD work, we provide additional suggestions for future work inspired by other tenets of the DP perspective. Namely, we discuss ideas for investigating risk/resilience mechanisms in longitudinal and cumulative factors designs.

1.6.1. Longitudinal investigations of risk/resilience mechanisms

Within broader risk/resilience studies, the use of longitudinal methods tends to be preferred because resilience is a process and the pathways characterizing it are dynamic over time (Masten et al., 2021, 2023). As such, longitudinal work can provide additional clarity regarding how/when resilience mechanisms emerge, their developmental course, and their stability over time (Masten, 2006). Importantly, the recommendation to use longitudinal methods does not negate the ongoing utility of cross-sectional work. Cross-sectional studies (e.g., Appels et al., 2024; Slomowitz et al., 2021) will continue to remain an important first step in identifying promising resilience mechanisms warranting further study in longitudinal designs.

One way to accomplish this goal of expanding developmental work on resilience factors is for existing, longitudinal LD datasets to incorporate additional measures of potential resilience factors (for examples see Eklund et al., 2013; Liew, Cao, Hughes, & Deutz, 2018; also see Torppa, 2025). The inclusion of explicit tests for promotive and protective factors (i.e., both main and interactive effects) in future longitudinal work on LDs would better establish what factors contribute to gap-closing, gap-maintaining, or gap-widening effects in academic outcomes for children across the risk gradient, thereby facilitating more precise early identification, prevention, and intervention efforts.

1.6.2. Cumulative investigations of risk/resilience mechanisms

A second avenue for future resilience research is to investigate cumulative risk/resilience designs given the prevailing multifactorial models for learning disabilities (e.g., Bishop & Snowling, 2004; Catts & Petscher, 2022; Margalit, 2003; McGrath et al., 2020; Pennington, 2006). A common cumulative factor study design is to utilize count variables of adverse and positive childhood experiences to determine cumulative risk or resilience mechanisms (Catts & Petscher, 2022). However, as Narayan et al. (2023) points out, researchers have begun to more critically consider how these cumulative models are implemented, as count variables may not capture the variability in the strength of the associations between various risk/resilience factors and outcomes (Narayan et al., 2023; although see Hayiou-Thomas, Smith-Woolley, & Dale, 2021 for a contrasting perspective). To provide more nuanced examinations of cumulative and individualized patterns of risk and resilience factors, it has been suggested that researchers could use person-centered statistical methods that allow for the identification of different clusters or combinations of risk, promotive, and protective factors for reading outcomes (Catts & Petscher, 2022; Hayiou-Thomas et al., 2021; Appels et al., 2024).

Building off of this idea, in this special issue, Appels et al. (2024) suggested variable-centered methods could be used alongside person-centered methods to investigate various subgroups of risk/resilience factors and to clarify their analytic mechanisms. Appels et al. (2024) suggested that person-centered methods could be used to initially derive academic resilience subgroups, such as a latent profile analysis (LPA) that might identify one or more resilient profiles where risk and resilience factors combined to result in better than expected reading outcomes. Potential resilience factors identified in the LPA could then be these entered into variable-centered analyses (e.g., regressions or ANOVAs with main and interactive effects) to determine potential promotive and/or protective effects. This is a promising combination of analytic methods, and we encourage readers to see Appels et al. (2024) for more detailed implementation recommendations.

Newer analytic methods, such as network models, might also be relevant to resilience science (Masten et al., 2021). For example, Kalisch et al. (2019) offered methods for examining cumulative risk/resilience mechanisms using network modeling. One application of network modeling is to investigate the way in which psychopathology arises from the interconnection of symptoms and risk factors or “nodes” (Robinaugh et al., 2020; Kalisch et al., 2019). Based on this idea, Kalisch et al. (2019) suggested that resilience related to mental health symptoms may be the result of separate resilience factor nodes (promotive and/or protective factors) being activated and resulting in dynamic, developmental processes that reduce the strength of association between symptom/risk nodes. The activation of promotive/protective factors in these networks is hypothesized to destabilize the network of psychopathology more broadly, presumably resulting in better functional outcomes. Kalisch et al. (2019) termed this network model a hybrid symptoms-and-resilience [HSR] factor network. This proposed network modeling method checks a number of theoretical and methodological boxes discussed above: it allows for multiple risk and resilience factors to be examined simultaneously, it can test for both main and interactive effects, and it allows for longitudinal modeling. While careful consideration would be needed to apply this network approach to learning disabilities research, there are some helpful parallels. In place of mental health symptom nodes, the LD field could substitute measures of academic skills while integrating potential cognitive, socio-emotional, and environmental risk and resilience factors for LD. Applying a network modeling approach integrating risk and resilience factors would be innovative for LD research designs.

1.7. Clinical implications of academic resilience research

In past LD research, there has been a tendency to focus on risk factors associated with LDs which has led resilience to be understudied. However, the field is undergoing an expansion and widening its purview by including increased theoretical and empirical considerations of resilience alongside risk. For example, McGrath et al. (2020) recommended that the Multiple Deficit Model [MDM] be renamed to the Multiple Factors Model [MFM] to be inclusive of risk and resilience mechanisms. Other conceptual models have also highlighted the importance of resilience mechanisms, including the Cumulative Risk and Protection Model (Catts & Petscher, 2022) and other authors have highlighted resilience mechanisms as well (e.g., Appels et al., 2024; Haft et al., 2016; Kiuru et al., 2013; Margalit, 2003; Slomowitz et al., 2021; van Viersen et al., 2015; van Viersen et al., 2019; Yu et al., 2018). Widening the scope to include studies of resilience mechanisms in learning disability research could yield important findings for assessment and intervention.

Further research is needed to continue to improve assessment practices in learning disabilities. One challenge in assessment is the heterogeneity in cognitive-linguistic profiles of children with LDs. For example, previous reading research has consistently determined that weaknesses in phonological awareness are one of the more robust risk factors for word reading difficulties (Snowling & Hulme, 2021), yet there is a sizable subset of children who have this PA risk factor but do not have a reading disability and children who do not have a PA risk factor but do have a reading disability (Pennington et al., 2012). This might be because risk factors are probabilistic, but it might also be attributable to unaccounted for resilience mechanisms. For example, if there is a protective factor that interacts with PA to mitigate the effect of PA on reading outcomes, it would explain why some children show PA difficulties but do not have a reading disability. Thus, a more comprehensive focus on resilience mechanisms in LD research could improve the field’s ability to accurately and comprehensively assess the factors influencing learning disabilities (Catts & Petscher, 2022).

Regarding how the identification of resilience mechanisms could benefit intervention approaches, consider that most intervention methods are designed to address underlying weaknesses associated with academic challenges (Catts & Petscher, 2022; van Viersen et al., 2019). Few interventions for LDs aim to enhance promotive or protective factors (Catts & Petscher, 2022; van Viersen et al., 2015), partly because such factors are just beginning to be identified. In addition to continuing interventions that attempt to remediate core weaknesses in LDs, it could be beneficial to develop complementary intervention approaches that also strengthen promotive and protective factors (Catts & Petscher, 2022; Masten, 2006).

2. Conclusions

This paper defined resilience terminology from the DP perspective, explained the importance of distinguishing promotive from protective effects, illustrated the application of DP methods to the LD field, and discussed future directions for academic resilience research. We encourage researchers to apply the DP perspective and adapt methodology and analytic strategy to expand resilience research in LD. Our field’s primary focus on risks for LDs resulted in important advances in assessment and intervention for LD, but an incomplete picture of the full range of risk and resilience factors influencing developmental trajectories. A comprehensive understanding of both risk and resilience factors is necessary to uncover the complex mechanisms that contribute to developmental outcomes in individuals with LDs.

Data availability

No data was used for the research described in the article.