Abstract
In an attempt to investigate the factors contributing to knowledge retention, this study examines the impact of sense of belonging on learning endurance. A Sense of Belonging (SB) scale was developed to measure the extent to which students perceive personal relevance in their learning material. The study involved 80 Grade 9 students, divided into an experimental and a comparison group. The experimental group was taught science using a contextualized approach aimed at fostering a sense of belonging, whereas the comparison group underwent conventional instruction. Learning endurance was evaluated through immediate and delayed tests, with the delayed test conducted one year later. Statistical analysis revealed that the experimental group achieved significantly higher SB scores and demonstrated superior knowledge retention compared to the comparison group, with differences in both immediate and delayed test scores reaching statistical significance (p < 0.05). These results suggest that instructional methodologies associated with higher sense of belonging relate to improved long-term retention, emphasizing that the power of personal connection to the material significantly influences learning endurance.
Keywords
Introduction
In 1964, B.F. Skinner stated, “Education is what survives when what has been learned has been forgotten.” This quote underscores Skinner's focus on the enduring value of education beyond rote memorization; it highlights the importance of applying acquired knowledge and skills in real-world contexts, rather than merely retaining facts. Education, he suggests, helps us discern between knowledge worth retaining and using versus knowledge that likely to fade. While the ultimate test of successful learning is its application, the learning process itself involves critical stages: (1) acquisition and (2) endurance of learning. Some students successfully acquire and embed knowledge in their long-term memory, whereas others tend to easily forget what they have learned, despite reinforcement and review. Does this suggest that our current educational approaches have limited impact on knowledge retention over time? If so, what conditions are essential to foster learning endurance and permanence?
In this article, we aimed to investigate and understand the endurance of learning by exploring the question: Why can some students retain knowledge over an extended period, while others cannot? Beyond individual intelligence levels, certain exceptional students can retain and effectively apply their knowledge long-term, regardless of instructional methods. Additionally, based on nurturing principles, strategies such as remediation, review, enrichment, intervention, and reinforcement are widely regarded as effective in promoting lasting learning (Murre & Dros, 2015; Radvansky et al., 2022). Whether learning endurance arises from innate ability or environmental support, we propose one shared factor among students with greater reported learning endurance: a sense of belonging.
We define sense of belonging as a content-embedded form of relatedness and identity alignment within academic contexts. Traditionally, belonging reflects the fundamental human need to form meaningful connections and feel accepted within social groups (Baumeister & Leary, 1995; Walton & Cohen, 2007). In educational settings, belonging has been consistently linked to enhanced motivation, engagement, and persistence, emphasizing social inclusion and support from peers, teachers, and the broader learning environment (Gopalan & Brady, 2020; Murphy et al., 2020; OECD, 2017). Building on this foundation, we conceptualize sense of belonging as a process that begins with social relatedness, whereby students experience acceptance, recognition, and connection in the classroom. These relational experiences facilitate internalization, as students adopt learning goals and see themselves as capable participants within the academic community (Vansteenkiste et al., 2004). Internalization, in turn, enhances the perceived relevance, familiarity, and applicability of content, linking instructional material to students’ prior knowledge, personal experiences, and cultural context. This perceived relevance fosters deeper cognitive processing, integration with existing schemas, and sustained engagement, ultimately supporting long-term learning retention and academic persistence (National Research Council, 2000; Hidi & Renninger, 2006; Renninger & Hidi, 2016). Unlike constructs such as task value or situational interest, which primarily focus on the perceived importance of a task or momentary engagement, content-based belonging emphasizes identity-related connectedness to the material, bridging social, cognitive, and affective dimensions of learning. By framing sense of belonging in this manner, we provide a clear theoretical bridge from canonical social belonging to academic outcomes, indicating that students’ relational experiences in the classroom are associated with more enduring perceptions of learning.
In our empirical study and informal interviews with random Grade 10 science students, a sense of belonging emerged as a significant predictor of learning endurance. To assess retention, we administered a competency-based test using Grade 7 science competencies that students had aimed to master four years earlier. Despite some initial forgetfulness, several science competencies were successfully retained and recalled. In interviews about this retention, students used descriptors like “Nakakatulong” (helpful), “Magagamit” (usable), “Kaugnay” (relevant), and “Kailangan” (necessary), and similar terms linked to belonging. Many participants expressed astonishment at their ability to retrieve the information effortlessly, despite the extended period without deliberate recall.
We briefly review both recent and earlier studies on sense of belonging and learning, along relevant theories, to provide context for the present study. We exclude other factors from consideration, such as the effects of mental stability on learning endurance, the relationship between age and knowledge retention, the impact of giftedness and learning delays on long-term retention, and the influence of initial learning acquisition on endurance. Additionally, we omit literature discussing the detailed documentation of task repetition as a reinforcement strategy for long-term learning.
Memory Retention and the Role of Belonging
Education relies on cumulative learning, where foundational knowledge from lower grade levels is essential for understanding advanced topics. This cumulative structure, based on the mastery of prerequisite concepts, aims to minimize learning gaps that could hinder knowledge retention. We reviewed Piaget's Theory of Cognitive Development (Piaget, 1952; Piaget, 1972), and while it provides valuable insights into learning acquisition, it overlooks factors that influence long-term retention. Recent research cites that factors like mnemonics, exercise, nutrition, testing, and sleep support retention (Albajes-Eizagirre et al., 2014), yet there is limited focus on how prior knowledge and a sense of belonging contribute to learning endurance.
Flooding the literature are memory studies that highlight review and practice as key strategies for durable learning (Duman, 2015; Cihangir-Çankaya, 2011; Sünbül, 2010; Elmas et al., 2016). While Mazza et al. (2016) concluded that sleep interleaved with practice enhances retention, Ebbinghaus’s (1885) “forgetting curve” demonstrates rapid knowledge loss in the absence of consistent practice, a pattern corroborated by Murre and Dros (2015). Additionally, Radvansky et al. (2022) indicate that spaced learning intervals and repeated practice sustain memory over time, implying that learning endurance benefits from continuous reinforcement.
We also identified that much of the forgetting literature has focused on rote memorization. Research on the stages of memory processing—encoding, storage, and retrieval (Sridhar et al., 2023)—proposes that integrating new information with prior knowledge enhances retention (van Kesteren et al., 2018). From a belonging perspective, such integration is not merely cognitive but relational: students who perceive new material as connected to their existing schemas also tend to report higher levels of content-based relatedness (Allen et al., 2021). This perceived relatedness satisfies the basic psychological need for connection, which in turn facilitates the internalization of external inputs into one's self-system (Van den Broeck et al., 2016). Internalized content is more likely to be appraised as personally meaningful or relevant (Eccles & Wigfield, 2002), thereby increasing the likelihood of elaborative encoding and deeper processing (Craik & Lockhart, 1972). This implies a link between a sense of belonging to new knowledge and slowed forgetting, although research directly examining this connection remains limited.
We also find that applying previously learned knowledge to new material strengthens retention (Kosar & Bedir, 2018). Our analysis shows that Sobrinho and Souza (2023) found prior knowledge improves memory for congruent objects, while Bransford and Johnson (1972) demonstrated that advance cues enhance prose recall. Kole and Healy (2007) showed that knowledge retention was stronger for facts about familiar people, and we observe that reminders enhance retention when applied to related yet distinct problems (Gick & Holyoak, 1980). Together, our findings suggest that robust prior knowledge supports new learning and underscore that gaps in foundational knowledge may limit our ability to achieve enduring learning. However, beyond structural familiarity, belonging theory suggests that familiarity becomes motivationally potent when learners interpret content as aligned with their social identity and prior experiences (Walton & Cohen, 2007; Baumeister & Leary, 1995). In this way, social relatedness extends into identity-relevant internalization, which strengthens perceived relevance and promotes sustained cognitive investment in the material (Vansteenkiste et al., 2017).
Theoretical Underpinnings of Memory Retention
Our examination of theories on memory retention offers valuable insights into how a sense of belonging can enhance long-term learning. For instance, constructivist theory manifests that learning is shaped by a person's thinking and social interactions, with social constructivism emphasizing the role of cultural experiences and radical constructivism focusing on the individual's personal way of understanding the world (Malik & Marwaha, 2023). Belonging provides the motivational mechanism through which socially mediated knowledge becomes personally endorsed: social connectedness supports the internalization of shared meanings, transforming externally presented content into self-relevant understanding (Vansteenkiste et al., 2004). Students’ sense of belonging is associated with how they integrate new knowledge with prior knowledge, which corresponds to more enduring learning outcomes, while a sense of belonging is an academic stance toward learning new information based on what is already familiar to students. On the other hand, reinforcement theory (Skinner, 1957) states that behaviors are influenced by the rewards or consequences they produce. In an educational context, a sense of belonging can act as a positive reinforcement, encouraging students to retain and engage with new information. Yet, unlike purely behavioral reinforcement, belonging operates through internalization rather than external reward, shifting motivation from compliance to autonomous engagement (Ryan et al., 2008). This creates a supportive environment that fosters long-term memory and learning.
Additionally, associative learning theory (Pavlov, 1927) and spreading activation theory (Collins & Loftus, 1975) explain how new information is remembered better when linked to existing knowledge. Associative learning conveys that connecting new information to what we already know enhances retention, while spreading activation proposes that activating one memory can trigger related memories, making recall easier. Belonging strengthens these connections, allowing students to integrate new knowledge more deeply. When learners perceive instructional content as relevant to their lived experiences, activation spreads not only across semantic networks but across identity-linked schemas, increasing elaborative rehearsal and durable encoding (Craik & Lockhart, 1972; National Research Council, 2000).
We also consulted self-determination theory (SDT) (Deci & Ryan, 1985), which highlights the importance of intrinsic motivation arising from the fulfillment of basic psychological needs, including connection, competence, and autonomy. By satisfying the need for connection, a sense of belonging enhances students’ motivation and engagement with learning. Within SDT, social relatedness facilitates internalization, which enhances perceived value and relevance of tasks, leading to deeper processing and persistence (Van den Broeck et al., 2016). When students feel they belong, they are more likely to link new knowledge with what they already know, leading to better retention and recall.
Together, these theories connote that while repetition is important for memory (Duman, 2015; Cihangir-Çankaya, 2011), a sense of belonging is equally vital for sustaining learning over time. Engaged learning methods and real-world connections, as emphasized by Sünbül (2010) and Elmas et al. (2016), all contribute to better memory and knowledge retention. By theoretically bridging belonging research with cognitive processing models, this study positions belonging as the mechanism through which social relatedness fosters internalization, internalization enhances perceived relevance, and perceived relevance drives deeper cognitive processing—ultimately supporting learning endurance. Figure 1 shows the triangulation of theories to demonstrate the sense of belonging in learning endurance.
Theoretical model of content-embedded sense of belonging and learning endurance.
The Role of Sense of Belonging in Learning Endurance
Grounded in the theoretical framework of this study, we sought to identify works and research methodologies that examine learning from the perspective of its relation to students’ prior knowledge. Our review led us to the studies of Ocfemia (2019), Bornilla (2019), and Besmonte (2018), who employed teaching strategies focused on contextualization. While their research emphasizes knowledge acquisition through context-based instructional methods, they overlook the implications of contextualization for long-term retention.
Over a century before our research on sense of belonging, we found that most literature focused on testing acquired knowledge, rather than learning endurance. Although there have been efforts to measure learning endurance and retention, these did not address the role of sense of belonging. Previous studies have established that students’ prior knowledge is important for long-term retention, but no research has linked instruction based on prior knowledge to learning endurance.
Upon expanding our review, we found a significant amount of literature supporting spaced instruction for long-term memory retention, though it did not connect to sense of belonging (Kelly & Whatson, 2013). A 1932 book, “Remembering”, discussed how reading instruction based on schema could improve retention with appropriate intervals before recall (Bartlett, 1932). This led us to hypothesize that the combination of schema-based learning and spaced intervals can synergistically enhance learning endurance. The integration of prior knowledge with new information, reinforced through spaced instruction, creates a more durable connection to the material, leading to better retention. However, an unpublished study found that even schema-based instruction could not guarantee long-term retention if factors like attention span, personality, and lesson meaning were not considered in the lesson design (Geneva, 2018).
The lack of scientific clarity in the existing research on belonging and learning limits our understanding and application of effective educational practices. Based on the reviewed literature and the arguments presented, we have defined the following specific research goals:
To examine how a sense of belonging influences learning endurance. To analyze the relationship between self-reported sense of belonging and the learning retention.
Research Methodology
The methods described in this section documented the specific procedures we undertook to meet the defined study goals, beginning with the development and validation of the sense of belonging scale and advancing to the implementation, data collection, and analysis strategy.
Development of Sense of Belonging (SB) Scale
To measure content-embedded sense of belonging, a concise, stand-alone instrument was developed to meet three established specific criteria. First, (a) the indicators had to capture students’ immediate perceptions of how lesson content relates to their prior knowledge, interests, and learning goals, fostering familiarity and relevance. Second, (b) the measure needed to demonstrate face validity across a diverse student population. Third, (c) the scale was designed to reflect sense of belonging as an academic stance, operationalized through students’ perceived relevance, familiarity, and emotional resonance with the topic; factors that support sustained engagement and integration within the learning process.
We constructed the SB scale based on the interviews conducted with a cross-sectional sample of 150 Grade 10 students from three secondary schools in Cagraray Island, Philippines. The location was purposefully selected because its strong cultural foundation provides a meaningful context to explore content-embedded sense of belonging. By situating the study in a culturally coherent environment, we could assess whether students experienced recognition, connection, and relevance of classroom material to their prior knowledge, personal experiences, and local context. Table 1 shows the number of participants purposefully balanced by gender, aged 15 to 21, and residents of the schools’ catchment areas, ensuring representativeness within the target population.
Randomly Selected Student-Respondents on the Development of SB Scale.
Our process began with administering a 50-item test assessing competencies aligned with Grade 7 science standards. Following the test, we conducted semi-structured interviews focused on specific competencies the students understood best, determined by their performance in discrete content areas. Teachers verified that participants did not exhibit exceptional capabilities in retaining trivial information, minimizing potential biases. Our interviews produced a pool of 16 coded items, originally expressed in Filipino, reflecting students’ felt connection to the instructional content. The coding process emphasized recurring themes across participants, including familiarity, relevance, usefulness, and alignment with prior knowledge—core dimensions aligned with the conceptualization of sense of belonging. The English translations of the original Filipino SB scale items were performed following recommended forward-back translation and expert review procedures to ensure that the original contextual and conceptual meaning of the items was retained.
We converted the 16 coded items into a 4-point Likert scale and piloted it with the same students, which allowed us to look at how the items grouped together. To better understand the patterns in students’ responses, we performed initial Exploratory Factor Analysis (EFA), a method that identifies the main dimensions or “themes” underlying the items. We first checked whether our data were suitable for this analysis: the Kaiser-Meyer-Olkin (KMO) measure was 0.78, and Bartlett's test of sphericity was significant (χ2 = 132.54, p < 0.001), confirming that the data were appropriate for factor analysis. Using principal axis factoring with oblimin rotation, we identified the underlying dimensions in the scale. The factor loadings ranged from 0.34 to 0.62, and the communalities ranged from 0.34 to 0.48, showing that the identified dimensions captured the patterns in students’ responses across all items. Then, we decided to combine items with correlation coefficients greater than 0.56 because the matrix of factor loadings showed lower coefficients, ranging from 0.21 to 0.34, with the next highest coefficient being 0.56, which is an acceptable threshold. This process resulted in eight final items in Filipino, which were subjected to further qualitative validation by five experts to ensure the new items captured the same ideas as the original ones.
The final 8-item SB scale was subsequently subjected to further validation. We administered the scale to 150 Grade 9 students in a nearby secondary school (School X) that shares a similar cultural and instructional context as the target implementation school site, ensuring that students’ learning experiences were comparable. This allowed us to examine whether the scale reliably and accurately measures students’ sense of belonging, reflects the intended factor structure, relates appropriately to similar constructs, and produces meaningful, interpretable scores. We conducted statistical analyses as follows:
a. Secondary Exploratory Factor Analysis (sEFA)
We conducted another sEFA to see how the 8 items of the final SB scale naturally grouped together, revealing the main patterns in students’ sense of belonging. All items contributed meaningfully, with factor loadings ranging from 0.45 to 0.60 and communalities from 0.35 to 0.50, and no items were identified as candidates for removal. The Kaiser-Meyer-Olkin measure (0.80) and Bartlett's test of sphericity (χ2 = 140.25, p < 0.001) confirmed that the data were suitable for this analysis. These results indicate that the 8 items collectively reflect a coherent dimension of familiarity, relevance, and emotional connection.
b. Confirmatory Factor Analysis (CFA)
We then conducted a CFA to validate the factor structure suggested by the EFA. This step tests whether the items reliably measure a single underlying concept as expected. Model fit indices indicated strong agreement between the data and the intended structure: CFI = 0.94, TLI = 0.92, RMSEA = 0.056, and SRMR = 0.048. Factor loadings ranged from 0.56 to 0.63, demonstrating that each item meaningfully contributes to measuring students’ sense of belonging. Overall, the CFA confirmed that the scale is structurally sound and represents a coherent, unidimensional construct.
c. Internal Consistency
To ensure the scale produces stable and consistent measurements, we assessed internal consistency. Cronbach's alpha (α = 0.81) and McDonald's omega (ω = 0.82) showed that the items reliably measure the same concept. Additionally, item-total correlations (0.40–0.45) confirmed that each item meaningfully contributes to the overall score. These results indicate that our scale consistently captures students’ sense of belonging across different items.
d. Convergent and Discriminant Validity
We also examined how the SB scale relates to other relevant constructs. Convergent validity was supported by an average variance extracted (AVE = 0.38), meaning the items share a reasonable amount of common variance in measuring sense of belonging. Discriminant validity was demonstrated through correlations with related constructs — task value, interest, and competence beliefs — which ranged from moderate to strong (r = 0.41–0.52). This indicates that while the SB scale is appropriately related to similar ideas, it measures a distinct concept that is not captured by these other constructs.
e. Scoring and Interpretation
The SB scale consists of 8 items, including both positively and reverse-coded statements. Total scores are calculated by summing all items after reverse-coding where necessary, resulting in a possible range of 8 to 40. We examined the score distribution for skewness, kurtosis, and floor or ceiling effects, confirming that the scores are well-balanced and interpretable. This ensures that the scale can meaningfully differentiate students based on their sense of belonging.
Table 2 summarizes the validity testing of the final SB scale, reporting the psychometric properties for each item and model fit indicator.
Secondary Exploratory and Confirmatory Factor Analysis of the 8-Item Sense of Belonging Scale (sEFA N = 150; CFA Split-Half N = 75).
Model Fit and Reliability Statistics:
a. KMO = 0.78; Bartlett's χ2 = 132.54, p ≤ .001
b. EFA Internal Consistency: α = 0.81; ω = 0.82
c. CFA Fit Indices: CFI = 0.94, TLI = 0.92, RMSEA = 0.056, SRMR = 0.048
Note: EFA factor loadings ranged from 0.45 to 0.60 and met the ≥ 0.40 retention criterion. Communalities ranged from 0.35 to 0.50. Reverse-coded items were properly recoded prior to computing total scores. CFA loadings are standardized, and fit indices indicate acceptable model fit based on conventional thresholds.
The SB Scale Validity
We developed the SB scale to measure content-embedded sense of belonging, capturing students’ experiences of connectedness, recognition, and alignment with instructional content. Sense of belonging in this study encompasses social, cognitive, and affective dimensions, beginning with social relatedness in the classroom and facilitating internalization of learning goals and sustained engagement (Baumeister & Leary, 1995; Deci & Ryan, 2000; Ryan & Deci, 2017). Unlike task value or situational interest, which primarily reflect transient engagement, the SB scale targets identity-related connectedness with academic content, bridging relational, cognitive, and emotional domains.
Empirical evidence demonstrates that the SB scale measures and taps belonging. Our exploratory factor analysis revealed a coherent factor structure, with loadings ranging from 0.45 to 0.60, communalities between 0.35 and 0.50, and item–total correlations from 0.40 to 0.45. Confirmatory factor analysis indicated good model fit (CFI = 0.94, TLI = 0.92, RMSEA = 0.056, SRMR = 0.048), and internal consistency was satisfactory (α = 0.81; ω = 0.82). Although the Goodenow (1993) Social Belonging scale was not administered/adopted in this study, it serves as a conceptual benchmark. The SB scale aligns with the same underlying construct (social, cognitive, and affective belonging) but is content-specific and classroom-centered. Convergent validity is supported through correlations with theoretically related constructs, including task value, interest, and competence beliefs (r = 0.41–0.52, p < 0.001), while discriminant validity is confirmed through weaker associations with unrelated constructs (r = 0.19–0.25, p < 0.05). These results provide robust empirical evidence that the SB scale measures the intended construct of belonging in a content-embedded context. In addition, the scale also meets our design criteria: items assess students’ immediate perceptions of how lesson content relates to prior knowledge, interests, and learning goals; items were reviewed for clarity and cultural appropriateness; and the scale operationalizes belonging as an academic stance, capturing relevance, familiarity, and emotional resonance that support sustained engagement and internalization of learning goals.
Based on the foregoing, we did not adopt existing belonging scales, as they do not operationalize the content-specific, classroom-centered nature of the construct. Our review indicated that widely used instruments (Goodenow, 1993; Fuchs et al., 2021; Anderson-Butcher & Conroy, 2002; Madson et al., 2024) predominantly assess generalized peer or institutional connectedness, whereas other measures (De Putter-Smits et al., 2013; Teshager et al., 2021) emphasize discrete topics or activities rather than the relational and cognitive alignment with instructional content. Employing these instruments would have compromised the conceptual fidelity, cultural relevance, and interpretive validity of our study within the classroom context.
Teaching Approach and the Lesson
We explored teaching strategies associated with fostering belonging by employing search terms such as prior knowledge, relevance, mental construct, cognitive architecture, schema-based, and indigenization, which led us to a context-based instructional approach. The study by Boda and Brown (2020) supports this approach by demonstrating how context-specific Virtual Reality (VR) learning experiences can enhance student engagement and perceptions of science through local relevance. This aligns with Sevian et al. (2018), who emphasize that context-based learning connects academic content to real-life situations to boost engagement, and Srinivasa et al. (2022), who highlight that adaptive, tech-integrated strategies foster meaningful learning for digital-native students. This further aligns with Bibon (2022) perspective on context-based instruction, where teaching is grounded in students’ existing knowledge. Building on this, the approach was implemented during the 1st quarter of the AY 2017-2018 in Biology lessons covering topics like the Respiratory and Circulatory Systems, Mendelian and Non-Mendelian Genetics, and Biodiversity.
Twelve (12) weekly lessons were developed and subsequently validated by three Science Master Teachers, assessing them on (1) relevance, (2) alignment, (3) contextual content, (4) technical rigor, and (5) assessment criteria, using a point Likert scale. The validation results indicated that the lessons were highly valid (scoring between 3.47 and 3.66), with only minor adjustments needed for time allocation and alignment. Likewise, the same Master Teachers conducted observations of the actual lesson implementations to determine the fidelity of instruction.
The Participants
During the implementation of the study, a random sampling procedure was employed to select 80 Grade 9 students from a secondary school in Cagraray Island, Philippines during the AY 2017-2018. To further enhance the validity of the findings, we systematically controlled for confounding variables, including extreme cognitive abilities (no records of exceptionally high or low intelligence), advanced subject knowledge (not repeating the subject), and age (restricted to 14–16 years). Hence, students younger than 13, older than 16, those repeating the grade level, or consistently ranking at the top or bottom of their class were excluded to prevent contamination of the participant pool.
It is important to note that the participants in the experimentation (Grade 9, n = 80) were distinct from those involved in instrument validation. We validated the 8-item SB scale using a separate cohort of 150 Grade 9 students from a nearby secondary school (School X). Likewise, we validated the 60-item science competency test with the same group of students at School X, ensuring that our instruments were tested in a context comparable to the target implementation site.
In addition, we established eligibility criteria requiring participants to (1) be actively enrolled in Grade 9 during the specified academic year, (2) have been continuously enrolled in the same school since Grade 7, and (3) be residents of the community from which the lesson context was derived. We employed a random sampling method to ensure an unbiased distribution of demographic and academic characteristics, thereby enhancing the study's internal validity and generalizability.
All 80 students (40 per group) completed the one-year delayed test (ODT) and the follow-up sense of belonging survey (SB2), with no attrition occurring during this period. This ensured that comparisons between the experimental and comparison groups were not influenced by differential participation.
Data Gathering Method, Instrumentation, and Analysis Plan
For the purposes of this study, we operationally defined learning endurance as the sustained retention of learned science competencies over an extended period following initial instruction. We measured this construct through performance on a one-time, unannounced delayed test administered after a one-year retention interval. In this design, the retention interval refers to the period between the immediate post-instruction assessment and the delayed measurement of retained knowledge. Accordingly, higher delayed test scores were interpreted as evidence of greater learning endurance, reflecting the durability of encoded knowledge associated with students’ sense of belonging to the instructional content.
This quantitative study employed a quasi-experimental design, systematically comparing outcomes between an experimental group that received the intervention and a control group that did not (Shadish et al., 2002). To achieve our research objectives, we used two primary instruments: (a) the validated SB scale, and (b) a validated, researcher-made 60-item test assessing Grade 9 science competencies. We conducted the validation process of the 60-item science competency test with 150 Grade 9 students from a nearby secondary school (School X)—the same students who participated in the validation of the 8-item SB scale—administering both instruments on the same day. This was followed by item analysis to refine the selected items. The revised test was subsequently re-administered, yielding a KR-21 reliability coefficient of 0.78, deemed adequate for classroom assessment. Furthermore, we conducted an exploratory factor analysis (EFA) using principal component extraction and varimax rotation, which confirmed the unidimensional structure of the test, with factor loadings exceeding 0.50 for all items. Complementarily, an item response theory (IRT) analysis employing a two-parameter logistic (2PL) model indicated acceptable item discrimination parameters (a = 0.65–1.92) and difficulty indices (b = –1.15–1.35). Model fit statistics (RMSEA = 0.04, CFI = 0.96, TLI = 0.95) demonstrated good fit, providing robust evidence of the test's construct validity and measurement precision. We used the same validated test for both assessments (pre- and post-instruction) to ensure consistency in measuring the targeted competencies and comparability of scores over time. Several methodological features minimized potential test-specific memory effects: the delayed test was unannounced, preventing students from preparing or reviewing the items; the one-year interval between administrations was sufficiently long to reduce recall of specific questions; and the test emphasized conceptual understanding rather than rote memorization, further decreasing the likelihood of students remembering item-specific answers.
The study involved 80 Grade 9 students, who were randomly assigned to experimental and control groups, with 40 students in each. In the experimental group, we implemented context-based lessons, whereas the control group received standard curriculum-prescribed lesson designs. The study followed a single-blind design, with students unaware of the study's purpose. To reduce expectancy effects and ensure instructional fidelity, all lessons were fully scripted, and adherence was monitored using observation checklists. The same teacher delivered instruction to both groups to maintain consistency. We also ensured that all research procedures involving participants were conducted in strict accordance with ethical standards for research with minors. Ethical clearance was obtained from the Institutional Review Board (IRB) of the school venue, ensuring compliance with institutional and national guidelines. Written informed consent was secured from parents or legal guardians of all participants, and assent was obtained from the students prior to participation. To protect confidentiality, all participant information was anonymized and securely stored, with access restricted to the research team.
Immediately following instruction, we administered the SB scale (SB1) to both groups, alongside the validated 60-item science competency test as an immediate test (IT) to assess initial learning acquisition. Over a year later, at the beginning of the second quarter of AY 2018–2019, students completed the SB scale again (SB2) based on their retrospective experience. They also completed the same 60-item validated test as a one-time, unannounced delayed test (ODT) to evaluate retention, specifically targeting enduring knowledge retention after a prolonged retention interval period. To minimize extraneous influences during the one-year retention interval, we implemented exposure controls: school curriculum guides, teacher logs, and student records were reviewed to ensure that neither group received formal instruction, remediation, or enrichment activities directly overlapping with the targeted Grade 9 science competencies. In addition, participants were continuously enrolled in the same school, allowing consistent monitoring of instructional exposure and reducing the likelihood of uncontrolled tutoring or curriculum repetition. These procedures ensured that differences observed in the delayed test (ODT) primarily reflected sustained learning associated with the initial instructional condition rather than subsequent rehearsal, review, or external academic support.
After tabulating the SB scale and 60-item test results, we conducted data analysis utilizing both descriptive and inferential statistical techniques. We first computed descriptive statistics, including means, standard deviations, and ranges, to summarize the distribution and central tendency of students’ scores on both the SB scale and the validated science competency test for the experimental and comparison groups. These descriptive measures allowed us to observe overall performance patterns and initial differences between groups before conducting inferential analyses. We then employed Pearson's correlation coefficient (r) to examine relationships among SB scale items, while analysis of covariance (ANCOVA) was used to compare the delayed test (ODT) scores between groups while statistically controlling for immediate test (IT) scores, thereby adjusting for initial differences in learning. To quantify the magnitude of our intervention effects, we calculated Cohen's d and Hedge's g along with 95% confidence intervals. All analyses were interpreted in the context of fidelity checks and exposure controls, which ensured consistent lesson implementation and minimized extraneous learning during the one-year retention interval, thereby supporting the conclusion that differences in ODT scores and SB measures reflect the intervention's effect on enduring learning rather than external influences. Notably, t-tests were not used to establish baseline equivalency, as our primary interest was in enduring retention and the role of sense of belonging rather than immediate performance on standard evaluation scales. Table 3 presents a detailed rationale for each component of the study's methodological design.
Methods of the Study and its Corresponding Rationale.
Results and Discussion
The following discussions were the significant findings we obtained from our experimental results, and information from the pool of data.
Instructional Fidelity of Contextualized Lesson
First, to ensure that the experimental condition was implemented as intended, we conducted a fidelity check
These observations and reflections confirm that contextualized instruction was both faithfully implemented
Findings
The Influence of Sense of Belonging on Learning Endurance
Our investigation into the sense of belonging among both the experimental and comparison groups revealed significant insights into its influence on learning retention. Although some self-report responses in the indicators showed a slight decrease over time, these changes were statistically significant (p < 0.05), indicating measurable shifts in students’ sense of belonging, with the experimental group consistently maintaining higher scores than the comparison group. Table 4 summarizes the self-report responses from the SB scale for both the experimental and comparison groups.
Comparison of Sense of Belonging Indicators Between Experimental and Comparison Groups.
*Reverse-scored items.
Table 4 indicates that the experimental group consistently obtained higher mean scores across all indicators of sense of belonging than the comparison group. Although the item “I have started with a little knowledge of the topic” showed a slight decrease in the experimental group from SB1 = 3.42 to SB2 = 3.23, the post-test value remained substantially higher than that of the comparison group (SB1 = 2.54; SB2 = 2.51). A similar pattern was observed for “Applicable to our routines,” where the experimental group sustained a higher post-test mean (SB2 = 2.94) relative to the comparison group (SB2 = 2.46).
From a statistical standpoint, we interpret the consistently higher mean scores together with the larger effect sizes in the experimental group (Cohen's d = 0.73–1.32; mean d = 0.87) as indicating a strong association between contextualized instruction and higher levels of perceived sense of belonging. The magnitude of these effects, falling within the moderate-to-large range, suggests that higher sense-of-belonging scores tend to co-occur with the contextualized instructional condition. In contrast, the comparison group's smaller effect sizes (Cohen's d = 0.63–0.88; mean d = 0.74) reflect comparatively weaker associations between standard instruction and changes in the same construct. We infer that the convergence of higher mean scores and larger effect sizes in the experimental group points to a consistent statistical association between contextualized instruction and enhanced sense of belonging across measurement points. The slight reductions in certain post-test indicators do not alter this interpretation, as the overall magnitude of differences remains within a statistically meaningful range, indicating a stable pattern of association rather than a deterministic causal effect.
Table 5 is presented to provide a psychometrically rigorous evaluation of students’ sense of belonging (SB) over time and across groups. While Table 4 demonstrates the magnitude of group differences and effect sizes, Table 5 establishes construct validity, temporal stability, and measurement equivalence, ensuring that observed differences are not attributable to measurement bias or random fluctuations.
Mean Scores, Standard Deviations, 95% Confidence Intervals, and Measurement Invariance of Sense of Belonging Indicators.
*Reverse-scored items.
Our analysis indicates that the experimental group consistently reported higher SB scores than the comparison group across all measured indicators. The associated standard deviations and 95% confidence intervals demonstrate that these differences are statistically reliable, reflecting systematic patterns rather than random fluctuations. We also observe that the experimental group exhibited narrower confidence intervals, indicating that students’ perceptions of belonging were more consistent and homogeneous. In contrast, the slightly wider intervals in the comparison group suggest greater variability, implying that without contextualized instruction, students’ sense of belonging was comparatively less stable.
We further conducted measurement invariance testing to confirm that the observed differences reflect genuine variation in the underlying construct rather than measurement artifacts. Most indicators achieved metric invariance, confirming that the items consistently measure the same construct of belonging across both groups and over time (SB1 and SB2). Several indicators also demonstrated scalar invariance, supporting the interpretation that mean differences represent true differences in latent belonging rather than bias in measurement. These results indicate that the higher SB scores in the experimental group correspond with the intervention condition, reflecting a meaningful statistical link between contextualized instruction and students’ perceived sense of belonging. The longitudinal pattern in Table 5 shows that the experimental group not only began with higher SB scores but also sustained these perceptions over time, whereas the comparison group displayed smaller gains and greater variability. These findings suggest a stable and systematic relationship between the intervention and both the consistency and stability of students’ sense of belonging—a pattern that aligns with sustained motivation, cognitive engagement, and learning endurance. By integrating descriptive statistics, precision estimates, and measurement invariance testing, we present a comprehensive evaluation that reinforces the interpretation of Table 4, demonstrating a robust and reliable correspondence between contextualized instruction and elevated SB scores.
For us to understand how students retained knowledge over time, we examined both immediate and delayed test performance while accounting for initial differences in prior knowledge and sense of belonging. Table 6 shows that the experimental group began with a higher IT score (43.61) compared to the comparison group (37.21), reflecting their initial engagement and comprehension. While these raw scores provide a snapshot of immediate learning, they do not account for differences in baseline performance or students’ initial sense of belonging, both of which can influence delayed retention.
Adjusted Comparison of Learning Scores, Memory Decay, and Effect Sizes Between Experimental and Comparison Groups.
Legend: IT – Immediate test
ODT – One-time delayed test
SD – Standard Deviation
SE – Standard Error.
Using ANCOVA, we examined delayed test scores (ODT) and memory decay, with ODT as the dependent variable, group as a fixed factor, and initial test scores (IT) and baseline sense of belonging (SB1) as covariates. This adjustment allowed us to interpret retention differences relative to each group's starting point, reducing the influence of pre-existing knowledge or initial perceptions of belonging. The adjusted ODT scores show that the experimental group retained higher knowledge levels (Adjusted ODT = 30.12, 95% CI [29.05, 31.19]) than the comparison group (Adjusted ODT = 16.05, 95% CI [15.21, 16.89]). Likewise, standardized mean change (SMC) for memory decay indicates lower forgetting in the experimental group (SMC = 13.49, 95% CI [12.21, 14.77]) compared with the comparison group (SMC = 21.16, 95% CI [19.87, 22.45]). These differences were statistically significant (p < 0.001), with Hedges’ g values of 0.68 and 0.56, representing moderate-to-large and moderate effect sizes, respectively. Post hoc power analysis confirmed that our sample provided over 95% power to detect these differences. Building on this ANCOVA framework, we further examined the potential sequential relationships among instructional group, SB1, IT, and adjusted ODT using a mediation model. In this analysis, SB1 and IT were treated as potential sequential mediators linking instructional group to delayed retention. Standardized coefficients were estimated along each path, and bootstrapped 95% confidence intervals were computed to quantify the precision of indirect effects, providing a complementary perspective on how group differences correspond with variations in students’ perceived belonging and test performance over time. Figure 2 illustrates the standardized path coefficients for the relationships among instructional group, baseline SB1, IT scores, and ODT.
Path diagram of the relationships between instructional group, SB1, IT scores, and ODT.
By adjusting for initial knowledge and baseline SB, we observed that higher retention and reduced memory decay were linked with the experimental instructional condition, independent of pre-existing differences. The comparison group, in contrast, showed smaller retention gains and greater variability, suggesting that conventional instruction was less strongly associated with sustained learning endurance. While we refrain from making causal claims, these patterns indicate a meaningful statistical correspondence between the instructional approach and long-term learning outcomes. The observed trends align with theoretical perspectives emphasizing the role of engagement and connection in learning. Jang et al. (2016) propose that stronger perceptions of relatedness are associated with higher motivation and engagement, which corresponds with greater retention. Likewise, Piagetian theory (Boom, 2009) suggests that integration of new information with prior knowledge is linked to more enduring learning. In this study, the experimental group's instruction was structured to connect content with students’ experiences, reflecting the observed pattern of higher adjusted retention and lower memory decay.
Relationship Between Sense of Belonging and Learning Endurance
A statistical correlation analysis was conducted to assess the degree of association between the sense of belonging and test scores. Table 7 presents the resulting correlation matrix, illustrating the relationships among these variables.
Correlation Between Sense of Belonging, Initial Test Scores, and Delayed Test Performance of the Experimental Group.
Note: Pearson correlation coefficients. ***p < 0.001. Values in brackets are 95% confidence intervals.
For the experimental group (Table 7), we observed a strong correlation between SB1 and IT (r = 0.71), suggesting that students who reported a stronger initial sense of belonging tended to perform better on the immediate test. This pattern aligns with prior research linking belonging with academic engagement and persistence (Davis et al., 2019). Although the correlation between SB1 and ODT (r = 0.41) was positive but weaker, we note that the correlation between SB2 and ODT (r = 0.66) was substantially stronger. These results indicate that students who maintained a higher sense of belonging over time were more likely to retain information, reflecting a potential association between contextualized instruction and students’ perceptions of belonging.
In contrast, our analysis of the comparison group (Table 8) showed a slightly higher correlation between SB1 and IT (r = 0.74), suggesting that students initially performed well on the immediate test. However, the relationship between SB1 and ODT (r = 0.44), combined with the weaker correlation between SB2 and IT (r = 0.31), indicates that early performance did not align as strongly with learning endurance. The moderate correlation between SB2 and ODT (r = 0.52) suggests that some learning was retained over time, yet the overall pattern of associations was less pronounced than in the experimental group. From our perspective, this implies that conventional instructional methods were less consistently associated with sustained learning retention.
Correlation Between Sense of Belonging, Initial Test Scores, and Delayed Test Performance of the Comparison Group.
Note: Pearson correlation coefficients. ***p < 0.001. Values in brackets are 95% confidence intervals.
Across both groups, we observed that SB2 correlated more strongly with ODT than SB1, highlighting that the persistence of perceived belonging over time is more closely linked with learning endurance than initial belonging alone. Moreover, the higher SB2–ODT correlation in the experimental group (r = 0.66) relative to the comparison group (r = 0.52) underscores the relevance of instructional strategies that actively foster and maintain a sense of belonging. We interpret this as indicative of a consistent association between contextualized instruction and sustained perceptions of belonging, which in turn corresponded with higher learning retention.
Our data suggest a clear pattern: belonging is not only associated with immediate academic performance but also positively linked with long-term learning endurance. From our analysis, students who feel connected to the material and their learning environment from the outset tend to show better performance on both immediate and delayed assessments. Importantly, we observe that a sustained sense of belonging exhibits the strongest association with learning retention over time, emphasizing that ongoing engagement with the learning environment is statistically linked with greater cognitive persistence. Our findings support the importance of educational strategies that cultivate a continuous sense of belonging, which aligns with enhanced engagement and long-term learning outcomes.
General Discussion
The study of knowledge retention has largely focused on the amount of information acquired, yet there has been less attention to how long it lasts. While learning acquisition is often emphasized as a driver of growth, the endurance of learning is just as crucial, as it serves as the foundation for continued learning and the transfer of knowledge. The endurance of learning, which is closely related to learning transfer, significantly contributes to meaningful learning (Bennett & Rebello, 2012). Our study stresses that while time is often cited as the primary factor contributing to the forgetting of knowledge, it is the approach to learning—specifically the instructional methods used—that plays an important role in preventing memory decay and ensuring long-term retention.
Contrary to the common notion that time alone dictates memory decay, our findings demonstrate that instructional strategies, particularly those fostering connections with the content, are essential in sustaining learning. Effective teaching approaches must address students’ prior knowledge and align with their prior knowledge and cultural context. Drawing from Piaget's cognitive psychology, which emphasizes the importance of new information fitting into existing mental schemas (Boom, 2009), our study emphasizes the value of activating prior knowledge as a way to enhance retention. Strategies such as advance organizers (Sunasuan & Songserm, 2021), flipped classrooms (Oudbier et al., 2022), and questioning techniques (King & Rosenshine, 1993) have been shown to improve cognitive processing, making new information easier to assimilate and retain.
The study also implies the importance of acknowledging individual differences in learning styles and interests. While instruction that is culturally and contextually relevant plays part in fostering a sense of belonging, the capacity to connect with the content remains deeply personal. Therefore, lessons should not only aim to create a sense of belonging but also consider the emotional and cognitive needs of each student. Affective engagement is crucial for sustaining learning, as students who feel emotionally connected to the lesson are more likely to retain the knowledge presented. In line with this, integrating emotional elements into instruction through activities such as storytelling, reflection, and exploration of personal experiences may enhance the emotional and cognitive connections students have with the content, thereby improving retention (Um et al., 2012; Vuilleumier, 2005).
Furthermore, our study supports the notion that active participation is essential for retaining knowledge. Learning is more enduring when students engage actively with content, through practical activities or integration of cognitive, emotional, and psychomotor domains. Experiential learning approaches, which emphasize “learning by doing,” have been shown to improve retention and comprehension in educational settings (Kolb, 1984; Prince, 2004). Active engagement also fosters a deeper sense of belonging, which can strengthen memory retention (Sridhar et al., 2023).
The broader implications of our study signify that educational instruction should adopt a whole-child approach, recognizing the interconnectedness of cognitive, emotional, and physical engagement in fostering learning endurance. This comprehensive approach, rooted in a sense of belonging, may provide a more sustainable solution to the problem of memory decay, as it focuses on the quality of the instruction rather than solely on the passage of time.
Our study also offers an alternative to the traditional view of the forgetting curve. While previous research has mentioned that continual review and practice are necessary to prevent forgetting, we argue that a sense of belonging-based instruction can effectively mitigate memory decay without the need for constant repetition. When students connect emotionally and cognitively with the lesson, the knowledge they acquire is more likely to endure over time. Thus, fostering a sense of belonging within the classroom is not just about making the content relevant but about ensuring that students feel valued and engaged in the learning process.
It is important to note, however, that our SB2 measure was collected as a retrospective rating one year after the initial instruction. SB2 reflects reconstructed perceptions of engagement and connectedness, rather than students’ immediate classroom experiences, which are more directly related to the encoding of knowledge. As such, SB2 should be interpreted as an indicator of maintenance or enduring perceptions rather than a causal predictor of long-term learning. Potential recall bias must be acknowledged, and any associations observed between SB2 and learning outcomes should be viewed correlationally rather than causally. Establishing causal effects would require longitudinal mediation modeling linking immediate classroom experiences to delayed retention outcomes.
The findings of our study generalizes that learning endurance is not solely dependent on the amount of review or rehearsal, but on the quality of instruction that connects with the student as a whole. Teachers should focus on creating a classroom environment that fosters a sense of belonging, recognizing the role of emotions, prior knowledge, and active engagement in ensuring long-term learning retention. By prioritizing these elements, educators can enhance students’ ability to retain and apply knowledge, ultimately contributing to more meaningful and sustainable learning experiences.
Limitations and Recommendations
The current study identifies five limitations and corresponding recommendations. First, while we have consistently emphasized, supported by prior research, that rehearsal, review, and practice are essential for managing the forgetting curve, (1) we recommend conducting a quasi-experimental study that specifically compares the effects of sense of belonging-focused instruction against reviews and rehearsals. This comparison will provide a clearer understanding of how these two approaches influence learning endurance.
Second, a limitation of this study pertains to the generalizability of the findings. While the experimental and comparison groups were drawn from a single secondary school in Cagraray Island, Philippines, the results may not automatically extend to other schools, regions, or populations with different cultural, socioeconomic, or instructional contexts. Although the participants reflect the local student population and the instructional interventions were designed with contextual relevance in mind, caution must be exercised when extrapolating the effects of sense of belonging-based instruction on learning endurance to broader or distinct populations. (2) Future studies involving diverse schools and communities are recommended to strengthen the external validity of these findings.
Third, our study is limited by its reliance on a single, unannounced delayed test administered one year after the initial instruction, which restricts our ability to assess memory decay during the retention interval period. Therefore, (3) we suggest for a longitudinal study involving multiple delayed assessments extending beyond the initial year. This approach would enable a comprehensive evaluation of learning durability over time and allow for a direct comparison with the effects of rehearsal, review, and practice.
Fourth, we have posited an indirect assumption that integrating affective and psychomotor domains into instructional practices—specifically through activities designed to enhance a sense of belonging—may further enhance learning endurance. To validate this hypothesis, (4) we propose conducting research that examines the combined effects of emotional engagement and psychomotor-based sense of belonging activities on learning endurance. This exploration will clarify the significance of these domains in relation to instructional effectiveness and their potential impact on student learning outcomes.
Lastly, while this study controlled for extraneous instructional variance within the school, it is important to recognize that students may have been exposed to related science content outside the school environment, such as online videos, peer discussions, or independent reading. (5) Future research should consider systematically monitoring both in-school and out-of-school exposures to better isolate the effects of instructional interventions on learning endurance.
Conclusion
Based on the study's findings, we conclude that, alongside well-established interventions like rehearsal and practice, a sense of belonging—integrated as a fundamental principle in instructional design—has a significant impact on learning endurance. This challenges traditional psychological and memory models that primarily rely on time to measure memory decay. Our research demonstrates that time alone is not the sole factor determining learning endurance; instead, the modality of instruction plays a critical role in how knowledge is retained over time. We emphasize that instruction that builds upon students’ existing knowledge fosters better learning retention. Furthermore, by grounding instruction in the students’ prior knowledge, meaningful learning is more likely to occur when students apply what they have learned. Thus, incorporating a sense of belonging in educational practices could support the achievement of educational policy standards and enhance the overall quality of education through improved learning retention.
Supplemental Material
sj-xlsx-1-bel-10.1177_30290805261452826 - Supplemental material for Sense of Belonging: A Fundamental Key to Understanding the Endurance of Learning
Supplemental material, sj-xlsx-1-bel-10.1177_30290805261452826 for Sense of Belonging: A Fundamental Key to Understanding the Endurance of Learning by Michael B. Bibon and Evan Carlo B. Deblois in Belonging
Supplemental Material
sj-docx-2-bel-10.1177_30290805261452826 - Supplemental material for Sense of Belonging: A Fundamental Key to Understanding the Endurance of Learning
Supplemental material, sj-docx-2-bel-10.1177_30290805261452826 for Sense of Belonging: A Fundamental Key to Understanding the Endurance of Learning by Michael B. Bibon and Evan Carlo B. Deblois in Belonging
Footnotes
Acknowledgment
We would like to acknowledge the assistance provided by Ms. Marites V. Bongalon for her support in coordinating with the study's target school and target respondents. Dr. Bibon dedicates this work to Cawayan National High School and to all his students whose hearts he has touched; this research marks the last of its kind from Dr. Bibon within DepEd. Moreover, Dr. Deblois would like to acknolwedge the support of the OIC Dean of Bicol University Gubat Campus, Dr. Roscefe B. Dy, for the inspiration and encouragement to conduct relevant research in the field of education.
Ethical Committee
This study was reviewed and approved by the Institutional Review Board (IRB). While it was not formally evaluated by a separate Research Ethics Review Committee (REC), informed consent for the minors' participation was obtained from their parents or legal guardians.
Dedication
This work is dedicated to Cawayan National High School, Cawayan, Bacacay, Albay.
Informed Consent
Since the participants are minors, parental consent was secured from parents/guardians of the participants.
Author Contribution(s)
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Grant Number
Not applicable.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
Part of the consent we sought from the parents was related to data privacy, as enforced by the Data Privacy Act of the country. The data set is publicly downloadable; however, all identfying information, including the names of the students and parents/legal guardians, has been removed to protect their privacy and confidentiality.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
