Navigating the Forgetting Curve: A Longitudinal Study of Knowledge Retention and Confidence Dynamics in Dental Education

Abstract

Retention of foundational knowledge is critical in dental education, yet the implications of Ebbinghaus forgetting curve are often overlooked. This study examined how well dental students retained core periodontal concepts and how confident in their knowledge evolved over the dental school years. A standardized 20-question multiple-choice exam was administered to 120 students at six points: before and after each of three progressive periodontal courses. Students also rated their confidence for each answer. Data from 113 students were analyzed. During the first-year course, mean scores increased significantly from 42–46% to 76–81%. After a four-month gap, second-year pre-course scores dropped to 66–70% and then rebounded to 84–89% post-course. In the third year, despite a nine-month gap, pre-course scores remained relatively high (78–81%) and rose slightly to 80–83% post-instruction. Performance and confidence improved significantly over time (p < 0.001), and a strong positive correlation was observed between them (r = 0.87, p < 0.001). These findings support the value of repeated reinforcement in promoting long-term knowledge retention and increasing student confidence.

1. Introduction

In healthcare education, acquisition and retention of conceptual knowledge are prerequisites to develop clinical judgement and achieve clinical competency. However, when acquired knowledge is not regularly reviewed or applied in practice, forgetting becomes inevitable (Custers, 2010; Watt, 1987). Ebbinghaus forgetting curve, dating back to the 1880s, highlights how newly acquired knowledge rapidly declines without timely reinforcement, forming the theoretical foundation for modern educational practices, such as longitudinal assessment, spaced repetition, and curriculum scaffolding. Unfortunately, this crucial aspect is often overlooked in dental education due to the substantial volume of knowledge and limitations in curriculum design (Zheng et al., 2022). One such strategy in education, longitudinal assessment, involves the testing of students’ knowledge or skill progression across multiple time points. It offers valuable insights into the long-term effectiveness of educational programs and contributes to the optimization of instructional strategies and curriculum design (Absi et al., 2011; Brittain et al., 2023; Everett et al., 2018; Hausdorfer et al., 2018; Lahoz et al., 2010; Margolis et al., 2020).

Previous studies that employed longitudinal testing utilized assessment methods such as multiple-choice questions (MCQs) and objective structured clinical exams (OSCEs) to evaluate participants (Clerkin, 2020; De Biasio et al., 2016; Dickie et al., 2023; Everett et al., 2018; Houseknecht et al., 2019; Jarvill, 2021; Lahoz et al., 2010; Margolis et al., 2020; Raupach et al., 2013; van Gessel et al., 2003). MCQs are common in healthcare education due to their quick results but often overlook students’ confidence in their answers (McMahan et al., 2014). While MCQs primarily assess correctness, incorporating confidence ratings encourages students to reflect on the certainty of their responses, adding a meaningful dimension to the assessment (Curtis et al., 2013; Curtis et al., 2012; Grazziotin-Soares et al., 2021; McMahan et al., 2014). The Commission on Dental Accreditation (CODA) of the American Dental Association (ADA) includes students’ self-assessment as an important accreditation standard of dental education programs. Tracking students’ self-assessments over time can help educators monitor changes in confidence and better inform curriculum development. Despite its potential, this approach has been underutilized in dental education, with few studies reported (Mays & Branch-Mays, 2016; Schmidt et al., 2022).

In this study, the same multiple-choice test was administered to dental students before and after each of three sequential periodontal courses over three years. The evolution of the students’ performance and confidence levels was tracked. This study aimed to address two central questions: (1) Does a progressive teaching approach reinforcing key periodontal concepts in later courses effectively counteract the forgetting curve? (2) Is there a clear correlation between students’ longitudinal testing performances and their corresponding confidence levels? Our study affirmed both of these hypotheses.

This study builds on prior work by uniquely combining longitudinal multiple-choice testing with confidence tracking over three years within a dental education context. While previous studies have explored these elements separately, their integration remains rare. By addressing this gap, our research contributes new insights into both cognitive and metacognitive aspects of learning, supporting more effective curriculum design and student self-assessment practices in line with the CODA standards.

2. Materials and Methods

The study protocol (#2021-0530) was approved by the Institutional Review Board (IRB) at the University of Mississippi Medical Center. The IRB granted a waiver of written informed consent. An invitation letter was emailed to potential participants, clearly explaining that completion of the tests and post-test surveys would be considered as implied consent to participate. The invitation also emphasized that participation was completely voluntary and that all responses would be kept confidential. Student identities were coded during data analysis, and no personally identifiable information was used in reporting the results.

The predoctoral program’s curriculum in Periodontics at the University of Mississippi Medical Center (UMMC) School of Dentistry comprises three distinct courses that span the initial three years of dental school. The first-year course, titled “Introduction and Concepts of Periodontal Diseases” (Perio I), focuses on foundational aspects, including the anatomy, etiology, and epidemiology of periodontal diseases, as well as an introductory overview of diagnosis and treatment planning. In the second year, the curriculum features “Non-surgical Therapy for Periodontal Diseases” (Perio II), which delves deeper into topics such as diagnosis, microbiology, pathology, and pharmacology related to periodontal diseases, including comprehensive discussions on the rationale, procedures, and outcomes of non-surgical periodontal therapy. In the third year, students complete an advanced course titled “Surgical Therapy for Periodontal Diseases” (Perio III), where they are introduced to the principles of periodontal surgery. It covers a diverse range of surgical procedures, their indications, and the potential consequences associated with each. While each of the three courses highlights distinct facets of periodontics, the fundamental principles, such as biological width/supracrestal attached tissue, attachment loss in periodontitis, and its risk factors, have been progressively introduced, strengthened, and applied. Longitudinal testing was implemented in this study to track the students’ acquisition and retention of core periodontal knowledge from these three courses.

Twenty multiple-choice questions were constructed for this study collaboratively by all the faculty within the Periodontics Department. Eighteen out of 20 questions focused on progressively introduced and reinforced periodontal core concepts across three courses, while the remaining 2 questions covered random dental facts that are not systemically taught in our school. These two questions were designated as control questions, as they were not intentionally revisited. Each of the 20 questions was followed by a survey question about the student’s confidence levels to answer these questions.

Three periodontics courses (Perio I, Perio II, and Perio III) are taught in the first year, the second year, and the third year at the UMMC School of Dentistry. Before and after each of the three courses, the test, including 20 MCQs and 20 confidence level survey questions, was given to the participants using ExamSoft (Figure 1). Hence, each participant was supposed to complete a total of six identical tests, with intervals spanning three years. The participants’ performances were recorded as the percentage of correct answers. The responses to the confidence survey were later transformed to numerical data (quite sure = 3, fairly sure = 2, unsure = 1, don’t know the answer so just randomly pick = 0) (Curtis et al., 2013; Grazziotin-Soares et al., 2021). There was no test review and no exposure of answer keys to the students during the study. If the same fundamental concepts were assessed in high stakes exams, such as mid-terms or finals, the questions were rewritten with different wording and answer options. This approach ensured that the students could not simply memorize the answer keys. Students from three classes graduating in three consecutive years (Class of 2022, Class of 2023, and Class of 2024), totaling 120 participants, took part in the study. Participants were included in the study if they had taken the first test. However, students who did not complete all six tests were excluded. In the end, the analysis included data from 113 students who successfully completed all six tests. The study spanned five years.

The de-identified data, including scores and confidence levels, for each test were subsequently exported from ExamSoft to Microsoft Excel. The data from the 113 participants were then subjected to analysis using SigmaPlot 14. The data passed tests of normality (Shapiro–Wilk test, p = 0.499) and equal variance (Brown–Forsythe test, p = 0.999). A two-way repeated measures analysis of variance (ANOVA) was used to assess the changes on the average scores of the performance and the confidence level over the course periods. Pairwise multiple comparisons were made using Tukey’s honestly significance difference test. In addition, a bivariate correlation analysis was conducted to assess the correlation between performance and confidence levels. All statistical tests were two-sided, and the level of significance was set at 0.05. A p-value less than 0.05 from any statistical tests indicates a significant change or correlation on the variables measured.

During the preparation of this manuscript, the authors made limited use of OpenAI’s ChatGPT to enhance clarity and wording. All content was reviewed and edited by the authors, who take full responsibility for the final version of the manuscript.

3. Results

The test performance trends of the three classes displayed remarkable similarities over time. Using the Class of 2022 as a representation (

Table 1. Students’ performances (test scores) (means ± standard deviations) on 20 multiple-choice questions at six different times over three years. Control questions assessed random dental facts; test questions focused on fundamental periodontal concepts. D1, the first year of dental school; D2, the second year of dental school; D3, the third year of dental school; Pre, pre-course test; Post, post-course test.

Class	Questions	D1 Pre	D1 Post	D2 Pre	D2 Post	D3 Pre	D3 Post
Class 2022	Control	49% ± 30%	42% ± 27%	43% ± 24%	43% ± 21%	37% ± 25%	47% ± 20%
	Test	43% ± 11%	78% ± 12%	66% ± 11%	85% ± 9%	81% ± 10%	83% ± 10%
Class 2023	Control	57% ± 26	42% ± 30%	46% ± 27%	42% ± 27%	39% ± 21%	46% ± 32%
	Test	55% ± 17%	76% ± 11%	69% ± 12%	84% ± 12%	78% ± 8%	80% ± 14%
Class 2024	Control	46% ± 32%	47% ± 27%	35% ± 26%	46% ± 22%	40% ± 20%	47% ± 30%
	Test	42% ± 11%	81% ± 9%	70% ± 9%	89% ± 7%	80% ± 11%	82% ± 14%

), prior to the Perio I course, the initial mean score started at 43% ± 11% (mean ± standard deviation) (D1 pre, baseline). Following the completion of the Perio I course, this score increased to 78% ± 12% (D1 post). A four-month interval before the start of Perio II course led to a decrease in the average score to 66% ± 11% (D2 pre), which subsequently climbed to 85% ± 9% (D2 post) by the course’s conclusion. In the third year, despite an almost 9-month gap, the students achieved scores of 81% ± 10% before the Perio III course (D3 pre), maintaining it at 83% ± 10% afterward (D3 post). All the changes in mean scores between consecutive time points are all significant at the p < 0.001 level. The performances for the Class of 2023 and the Class of 2024 are also detailed in Table 1. When visualized as curves (Figure 2), these data exhibited a pattern resembling “two peaks followed by a high plateau.” This reflects knowledge gains after the first two courses, a decline after two learning gaps, and eventual stabilization in the third year (solid lines in Figure 2). Notably, the peak of the learning curve was reached after the Perio II course, which immersed students in fundamental concepts through activities like case workshops and periodontal clinical rotations. Furthermore, the decrease in retention experienced after the 9-month learning gap following the Perio II course was less prominent than the decline observed after the initial gap (4 months). Conversely, performances on two control questions assessing random facts not covered in the three Perio courses exhibited minimal variation over time, resulting in a flatter curve over time (dash lines in Figure 2).

Remarkably, the students’ confidence levels experienced a strikingly similar fluctuation throughout the three years, resembling the “two peaks followed by a high plateau” pattern (Figure 3). This alignment between the students’ performance and self-assessment highlighted consistent outcomes. The confidence levels of the three classes are detailed in

Table 2. Students’ confidence levels (means ± standard deviations) on answering 20 multiple-choice questions at six different times over three years. Control questions assessed random dental facts; test questions focused on fundamental periodontal concepts. D1, the first year of dental school; D2, the second year of dental school; D3, the third year of dental school; Pre, pre-course test; Post, post-course test.

Class	Questions	D1 Pre	D1 Post	D2 Pre	D2 Post	D3 Pre	D3 Post
Class 2022	Control	0.93 ± 0.61	1.32 ± 0.56	2 ± 0.59	2.03 ± 0.46	2.11 ± 0.48	2.33 ± 0.53
	Test	1.32 ± 0.36	2.32 ± 0.30	1.99 ± 0.29	2.57 ± 0.26	2.30 ± 0.27	2.50 ± 0.26
Class 2023	Control	0.99 ± 0.57	1.24 ± 0.51	1.92 ± 0.30	1.96 ± 0.57	1.99 ± 0.53	2.25 ± 0.69
	Test	1.32 ± 0.31	2.40 ± 0.32	1.71 ± 0.47	2.60 ± 0.29	2.28 ± 0.32	2.41 ± 0.56
Class 2024	Control	0.80 ± 0.55	1.26 ± 0.80	1.64 ± 0.58	1.89 ± 0.69	1.89 ± 0.56	2.34 ± 0.53
	Test	1.23 ± 0.34	2.34 ± 0.29	2.02 ± 0.31	2.52 ± 0.30	2.27 ± 0.37	2.53 ± 0.37

. For both performance and confidence level, significant effects of both pre- and post-course timing (p < 0.001), year (iterations of exposure) (p < 0.001), and the interaction of those two independent variables (p < 0.001) were observed. The presence of the interaction indicates that “pre- and post-course timing” and “year” work together in influencing the outcome of performance and confidence level. In addition, a strong positive correlation between individual student’s confidence levels and their performances on 18 test questions was observed (r = 0.87, n = 113), and the relationship was significant (p < 0.001). Interestingly, in the case of the two control questions, the content of which was not systemically taught, the students’ self-assessment diverged significantly from their actual performances. Despite perceiving an improvement throughout the courses, their actual performance did not align with this perception.

4. Discussion

Following the introductory Perio I course in the first year, the students achieved higher scores (D1 post) compared to the baseline (D1 pre) collected at the beginning of the course, prior to the delivery of any teaching content. However, a mere four-month gap, during which no periodontal instruction took place, resulted in a significant decline in student performance (D2 pre). The Perio I course was taught concurrently with fundamental science courses, such as anatomy and physiology. The students typically employed similar learning strategies for these first-year courses, as they were required to assimilate substantial amounts of new material within a short time frame to pass examinations. Numerous studies in medical education have demonstrated that most of the intricate details of basic science tend to be quickly forgotten shortly after exams (Custers, 2010). As Ebbinghaus forgetting curve demonstrated, without reteaching, revisiting, or practical application in a clinical context, long-term retention of this knowledge is unlikely (Zheng et al., 2022).

In the second-year course, Perio II, core concepts were revisited in greater depth. For instance, the students compared the anatomy of a healthy periodontium to that of a diseased periodontium, healthy dental implants, and peri-implantitis. Patient cases, including periodontal charts and radiographs, were introduced and discussed, providing students with a better understanding of attachment loss and its clinical significance. Clinical rotations commenced, allowing for the students to conduct clinical examinations on one another, manage recall patients with faculty supervision, and assist in various periodontal procedures. The Perio II course prepares students to a level of clinical competency for the diagnostic and nonsurgical treatment concepts. The results of their post-course assessments (D2 post) indicated that their performance had reached the peak of the learning curve. Impressively, this high level of performance was maintained throughout a 9-month gap following the Perio II course, as shown by the D3 pre-course test (D3 pre). Notably, during this interval, although there was no additional didactic instruction, the students began clinical training 2–3 months before the Perio III course. The clinical training at this stage focused on diagnosis and nonsurgical management of periodontal diseases. Although Perio III predominantly concentrated on surgical principles and procedures with less emphasis on fundamental periodontal concepts, the students consistently maintained a high level of performance on the test (D3 post).

The inspiration drawn from Ebbinghaus forgetting curve led to the creation of spaced repetition techniques. This method entails the revisiting and reviewing of information by learners at specific intervals, effectively countering the natural forgetting process and boosting long-term retention (Donker et al., 2022; Kornmeier et al., 2022; Wollstein & Jabbour, 2022). Our progressive teaching approach enabled the students to revisit the essential periodontal concepts at intervals and explore them in the diverse contexts. As a result, we observed a reversal of their forgetting curve on eighteen core-concept questions. However, for the random knowledge sporadically exposed to the students, there was no improvement in retention. Although the conventional Ebbinghaus forgetting curve was not evident on these two control questions, the results indicated that reinforcing information at intervals could enhance knowledge retention.

Nevertheless, the substantial improvement in our students’ performance during the second and third years could also be attributed to the students applying course concepts during their early clinical experiences. This result aligns with the findings of previous studies in medical and dental education. Custers et al. found that basic science knowledge in medical students significantly declined after 1–2 years but greatly improved when knowledge was related to clinical situations (Custers, 2010). Raupach et al. found that factual knowledge increased when medical students were exposed to clinical environments during their practice year (Raupach et al., 2013). Furthermore, Hausdorfer et al. investigated factual and procedural knowledge retention in dental students and found that procedural knowledge remained more stable (Hausdorfer et al., 2018). It was assumed that increased knowledge retention is due to student’s overall expanding of clinical knowledge and increased motivation to learn for patient care needs.

Another significant objective of this study involved tracking the students’ confidence levels during test-taking over time and examining its correlation with their performance. Initially, on the baseline test before Perio I, the students displayed consistently low average confidence levels, with most responses hovering around a score of 1, indicating uncertainty. However, as the learning process progressed, we observed an intriguing pattern in the confidence level data. It closely mirrored the performance curve, “two peaks followed by a high plateau”, on core periodontal concepts. This remarkable consistency has reinforced our confidence in understanding how students acquire, forget, and retain fundamental content. For the two questions assessing random facts, the students paradoxically exhibited increased confidence in their answers over time, despite no corresponding improvement in performance. This disconnect between confidence and accuracy may align with findings in health profession education, where more advanced students sometimes overestimate their knowledge in less reinforced or clinically applied content areas (Kruger & Dunning, 1999; McMahan et al., 2014). These findings highlight the need for educational strategies that develop students’ ability to accurately evaluate their confidence in their learning.

One of the limitations of this study arises from our inability to eliminate the repeated testing effect in the longitudinal assessment. Despite these 20 questions having never been reviewed, their familiarity might prompt students to recall answers during subsequent tests, potentially leading to rote memorization without comprehension. In future studies, questions on the same content but with different wording, distracting options, or even in varied formats should be administered in different years to mitigate this effect. Incorporating improved and additional control questions that accurately represent one-time exposure facts could also reduce this retesting effect (Taveira-Gomes et al., 2015). An additional limitation of this study is that we used averaged and longitudinal analyses to assess the performance of test questions for an entire class, including confidence levels. While this method provided a broad perspective and insights into the long-term correlation between these two factors, it may have overlooked the nuances within individual questions or concepts. Investigating each specific question could potentially reveal more detailed information, such as areas where concepts might have been taught ineffectively or the question could have been better constructed. Exploring these finer details may be a promising avenue for our future research.

While this study was conducted within the context of a specific dental school curriculum, the core findings may have broader relevance to other institutions and healthcare disciplines. Curricular structures may differ in terms of timing, integration of clinical exposure, and assessment strategies; however, the underlying educational principles—such as the importance of spaced reinforcement, clinical application of foundational knowledge, and tracking learner confidence—are applicable across a wide range of health profession education settings. Other programs may adapt these strategies within their own curricular models to enhance long-term knowledge retention and metacognitive development among students.

5. Conclusions

In conclusion, this longitudinal analysis of student performance and confidence levels has provided valuable insights into the learning process. It highlights the importance of progressively reteaching and reinforcing information, especially with varying depth and in different contexts. This holds particular importance for curriculum design after the early years of dental school when a significant volume of foundational knowledge is imparted. Early exposure to clinical environments may contribute to a deeper understanding of foundational concepts from didactic courses and promote long-term retention as well. Moreover, the reinforcement of knowledge is anticipated to improve students’ accountability for their own learning and enhance the accuracy of their self-assessment, aligning more closely with CODA requirements.