Evaluation of the Dunning-Kruger Effects among Dental Students at an Academic Training Institution in UAE

Abstract

Background

The Dunning-Kruger effect (cognitive bias) is a psychological phenomenon that implies that individuals with a lack of knowledge and skills have an unrealistically optimistic image of their abilities compared to others.

Purpose

The study aimed to examine the presence of the Dunning-Kruger effect in fifth (final) year dental undergraduate students at Gulf Medical University, UAE, in clinical domains related to pediatric dentistry.

Methods

A longitudinal cohort study was conducted at Gulf Medical University, College of Dentistry, to evaluate cognitive bias among 5th-year dental students at GMU in the competency domains of communication, diagnosis, and clinical skills in Pediatric dentistry.

Results

Overall, compared to the level of confidence of students in September, there was a decline in the level of confidence in January. With an increase in time, the percentage shifted to a higher level in May across all domains.

Conclusion

Characteristic patterns of fluctuations in students' self-confidence during clinical exposure are observed. The research results prove that the Dunning-Kruger effect, the cognitive bias of the perception pattern, is present in the examined group of dental students.

INTRODUCTION

The Dunning-Kruger effect (DK effect) (cognitive bias) is a psychological phenomenon which implies that individuals with lack of knowledge and skills have an unrealistically optimistic image of their abilities compared to others. This illusionary superiority is a form of cognitive bias. It is often called a "double curse" because individuals lack the ability to perform a task and recognize their incompetence, leading to unrealistic positive self-beliefs (Figure 1).

Figure 1.

A graph depicts how confidence in speaking on a subject grows with experience and knowledge. People tend to overestimate their cognitive ability until/unless their competence increases to the point where they become aware of their shortcomings, according to the Dunning–Kruger effect (Creative Commons CC0 1.0 Universal Public Domain Dedication – Can be copied, modified, distributed, and perform the work, even for commercial purposes, all without asking permission: https://commons.wikimedia.org/wiki/File:Dunning-Kruger_Effect_01.svg).

Dunning and Kruger, social psychologists who discovered the phenomenon claimed that people who have poorly developed skills or lack knowledge have an unrealistically optimistic image or illusion of superiority that develops as a consequence of misperception of others or as an overestimation of their own knowledge (1).

Most of the research that focuses on the DK effect confirms the occurrence of the phenomenon among certain groups of people. A study conducted by a group of authors in the United States confirmed that this phenomenon occurred in aviation students. Students in the examined group had low scores on grammar and professional knowledge tests, and the same students overestimated their knowledge. On the other hand, students who performed well on both tests underestimated their ability, a phenomenon known as the imposter syndrome (2). This pattern of overestimating one's knowledge was also observed among psychology students at an Ecuadorian university in the language (vocabulary) and logic tests (3). Yet another interesting study conducted in the United States of America concluded that individuals with limited knowledge about the causes of autism and its link with vaccines were the ones who were overconfident and showed less support for mandatory vaccination policies (4).

Studies have evaluated ‘perceived stress’ among students as a factor responsible for negatively impacting clinical and academic performance (5). On the contrary, there is limited evidence on the existence of cognitive bias and its impact on students' performance in various health science professions. The existing literature states that the prevalence of the DK effect among students may affect their clinical performance. Research by Fitzmaurice in 2020 shows that a student with poorer results during a continuous assessment shows a higher level of self-confidence than students with better results (6). These observations have significant ethical implications where less-skilled students may accept assignments for which they are unqualified because they are overestimating their skill set. The existence of the DK effect in students will result in poor communication in the student-child-parent communication triangle, misdiagnosis, and poorly performed clinical procedures. It cannot be emphasized enough how important it is to develop confidence in students' patient communication and clinical skills before beginning professional practice (7). Pena in 2010 stated that ‘competence develops after having considerable experience (8)

Considering the minimal research on the DK effect among dental students globally, the present study aimed to examine the presence of the Dunning-Kruger effect in fifth (final) year dental undergraduate students at Gulf Medical University, UAE, in the clinical domains of pediatric dentistry. As per the existing dental curriculum of the universities in the UAE, pediatric dentistry is offered in either the fourth or fifth year. In our university pediatric dentistry is offered in the fifth (final year) only, therefore the authors were convinced to perform the present study only on final-year dental students. The clinical scope of pediatric dentistry includes different procedures such as restorative and preventive treatments (fluoride application and fissure sealants), endodontic treatments (pulpotomy and pulpectomy), and extractions in patients up to the age of 12 years. Furthermore, these students have had a similar experience with the same dental procedures in adult patients in various specialties of dentistry. With the existing cognitive and psychomotor skills, the final year students can be categorized as ‘competent’ in accordance with the Dreyfus model (9).

MATERIAL AND METHODS

The present study is a longitudinal cohort study that was conducted at Gulf Medical University, College of Dentistry, to evaluate cognitive bias among 5^th-year dental students in the competency domains of communication, diagnosis, and clinical skills in Paediatric Dentistry.

Participants in the Study

Two different cohorts of the two academic years batch 2016 (N1= 69) and batch 2017 (N2= 64) with a total number of 133 students were offered to participate in the study, all the students consented to the study. No formal sample size was calculated; rather, all students during the two years study period were included in the analysis. The age range was 21-24 years. The total number of females from both academic years was 84, (N1=46 and N2 =38), and the total number of males was 49 (N1=23 and N2=26). The student participants were from diverse cultural backgrounds that reflect the multidiverse culture existing in the UAE. The study was conducted from September through January until May consecutively in the academic years 2021 and 2022.

Inclusion criteria:

• Year 5 undergraduate dental students who consent to participate in the research

Exclusion criteria:

• Year 1, 2, 3 & 4 undergraduate dental students.

• Undergraduate dental students who do not consent to participate.

Instruments

• Dependent variable:

- Dunning - Kruger effect (cognitive bias)

• Independent variable:

- Level of competency

Procedure

Cognitive bias among 5^th-year dental students at GMU in the competency domains of communication, diagnosis, and clinical skills in pediatric dentistry was evaluated using a questionnaire with three confidence levels [Not confident (L1), Moderately confident (L2), and Highly confident (L3)] (Table 1).

Table 1. Competency Domain Descriptions.

Competency Domain	Competency Description	Level of Competency - trustability (Applicable for all domains)
A - Communication	Thorough verbal communication with patient/parents/caregiver. Uses words, pace, tone, eye contact, and body language that show care and concern. Frequently checks to understand. Structures responses to the patient considering the patient's perspective.	L1 – Not Confident The student did not have adequate skills/knowledge about the procedure and required the faculty to perform the procedure. L2 – Moderately Confident The student has acceptable knowledge about the procedure. The faculty had to guide the student from time to time. L3 – Highly Confident The student performed the procedure safely without any faculty help.
B - Diagnosis	Excellent and precise diagnosis. Highlight all the elements of the pathological condition. All possible options for differential diagnosis have been set. Comprehensive identification/ discussion of signs and symptom
C - Clinical skill - Pulp Therapy	Local anesthesia administration; Access cavity design; Pulp extirpation/removal/conservation; Medicament placement and restoration

There are few possible methodological approaches for analyzing the Dunning-Kruger effect hypothesis test. The original research involving a spectrum of 4 abilities compared the self-assembly of the respondents with specific tasks that would enable a relevant interpretation of the results (6). Over the years, different abilities and skills have been used to analyze this effect (7, 10, 11). The present study used a modified methodology of the original research with a spectrum of 3 variables.

A pilot study was carried out with 15 students to determine the validity and reliability of the questionnaire and approval of the questionnaire was done by two external field experts from Hamdan Bin Mohammed College of Dental Medicine, Dubai, and the University of Nis, School of Medicine, Serbia, Europe.

After obtaining ethical approval for the research from the Institutional Review Board of the university and student consent for participation, the questionnaire (Table 2) was distributed to students at three different time points: the beginning of the IX^th semester (September), at the end of the IX^th semester (January), and the end of the X^th semester (May), the cycle was repeated once among the students of the batch 2016 and once among the students of batch 2017. The Cronbach's alpha was calculated and presented a good performance value of 0.958 for September, 0.979 for January, and 0.922 for May. All batches participated in the questionnaire, and the average age was 20-22 years old.

Table 2. Questionnaire for assessment of the Dunning – Kruger Effect.

Question	Level of Response
Considering your overall experience with the pedodontics clinical training, please rate your self-confidence towards communication skills with pediatric patients/parents. (Domain A)	Level 1 (L1) - Not Confident
	Level 2 (L2) - Moderate Confident
	Level 3 (L3) - Highly Confident
Considering your overall experience with the pedodontics clinical training, please rate your self-confidence towards a proper examination and providing a precise diagnosis. (Domain B)	Level 1 (L1) - Not Confident
	Level 2 (L2) - Moderate Confident
	Level 3 (L3) - Highly Confident
Considering your overall experience with the pedodontics clinical training, please rate your self-confidence towards your overall clinical skills and performance at pulp therapy treatment. (Domain C)	Level 1 (L1) - Not Confident
	Level 2 (L2) - Moderate Confident
	Level 3 (L3) - Highly Confident

Competency domain evaluation was performed in the clinical setting in September, January, and May by two subject experts using a standardized rubric, parallel to the questionnaire distribution. All three domains in the range of level 1 to level 3 were assessed (level of trustability). The results were then correlated with the student’s responses to the questionnaire (Figure 2).

Figure 2.

Study work plan.

STATISTICAL ANALYSIS

Data are presented as frequencies and percentages. The McNemar-Bowker test compares students' September, January, and May responses. The chi squared test and Somer's d correlation coefficient evaluated the association between investigated parameters. The p-value was set at 0.05. Comparisons between groups were made on transformed data using a Chi-squared test with correction for multiple comparisons. Bonferroni correction was applied to take multiple testing into account and p value thresholds were set at 0.002 (0.05/27). Statistical analyses were performed in the R statistical language (version 4.1.3).

RESULTS

In September, approximately every fifth of students was non-confident (21.1%) in communications, 15.8% were non-confident in diagnosis, and 14.3% were non-confident in clinical skills. In January, 38.3% of students were non-confident in communications, 41.4% were non-confident in diagnosis, and 46.6% were non-confident in clinical skills. In May, 7.5% of students were non-confident in communications, 11.3% were non-confident in diagnosis, and 5.3% were non-confident in clinical skills (Figure 3).

Figure 3.

Changes in the confidence levels of students during the period of September, January and May.

A cross-tabulation of responses in September, January, and May detailed the changes in the confidence level of students in communications, diagnosis, and clinical skills (Table 3). The p-values of the McNemar-Bowker`s test for the confidence level of the students for all comparisons in communications, diagnosis, and clinical skills were p <0.001, indicating the difference in responses in September, January, and May was statistically significant and quite large even after correction for multiple comparisons.

Table 3. A cross-tabulation of students’ responses - McNemar-Bowker`s test for the confidence level of the students in all three domains (NC - Non-Confident; MC – Moderate Confident; HC - Highly Confident).

		Communication								p1
		January				May
September	Category	NC	MC		HC	NC		MC	HC
	NC	100%	0		0	35.7%		64.3%	0	S vs. J p<0.001
	MC	39.7%	60.3%		0	0		34.5%	65.5	S vs. M p<0.001
	HC	0	29.8%		70.2%	0		0	100	J vs. M p<0.001
September	Diagnosis
	NC	100%	0		0	71.4%		28.6%	0	S vs. J p<0.001
	MC	70.8%	29.2%		0	0		35.4%	64.6%	S vs. M p<0.001
	HC	0	56.3%		43.8%	0		0	100%	J vs. M p<0.001
September		Clinical skills
	NC	100%	0		0	36.8%		63.2%	0	S vs. J p<0.001
	MC	63.2%	36.8%		0	0		0	100%	S vs. M p<0.001
	HC	0	30.4%		69.6%	0		0	100%	J vs. M p<0.001
	Answers slide to a lower level			The answers match			The answers slide to a higher level

In the competency domain of communication, out of the students who were highly confident in September, 29.8% of students showed a decline in their level of confidence (Moderate confidence) in January, which these students regained by the end of May. Out of the students who were moderately confident in September, 39.7% of students showed a decline in their level of confidence (Not confident) in January. By the end of May, 65.5% of the students who were Moderately Confident in September showed an increase in their level of confidence (Highly Confident).

In the competency domain of Diagnosis, out of the highly confident students in September, 56.3% of students showed a decline in their level of confidence (Moderate confidence) in January, which these students regained by the end of May. Out of the students who were moderately confident in September, 70.8% of students showed a decline in their level of confidence (Not confident) in January. By the end of May, 64.6% of the students who were Moderately Confident in September showed an increase in the level of confidence (Highly Confident).

In the competency domain of Clinical Skills, out of the highly confident students in September, 30.4% of students showed a decline in their level of confidence (Moderate confidence) in January, which these students regained by the end of May. Out of the students who were moderately confident in September, 63.2% of students showed a decline in their confidence level (Not confident) in January. By the end of May, 100% of the students who were MC in September showed an increase in their level of confidence (Highly Confident).

Overall, compared to the level of confidence of students in September, there was a decline in the level of confidence in January. With an increase in time, the percentage shifted to a higher level in May across all domains.

The χ2-test and Somers' D correlation coefficient were done to evaluate the strength and direction of association between the student's confidence level and skill assessment. Somers'D correlation coefficient showed a statistically significant positive association between student level of confidence in communication and diagnosis versus skill assessment in all three timeline measurements (Table 4). The correlation between student level of confidence in the clinical skill domain and skill assessment is also statistically significant. It shows a positive relationship in all three comparisons with a slightly weaker correlation in May.

Table 4. Association between the perception of students` confidence levels and skills assessment.

	Skills assessment
	September		January		May
	Somers`D	p-value	Somers`D	p-value	Somers`D	p-value
Communication	0.752	<0.001	0.782	<0.001	0.908	<0.001
Diagnosis	0.657	<0.001	0.670	<0.001	0.887	<0.001
Clinical skill	0.603	<0.001	0.729	<0.001	0.482	<0.001

The percentage of the highest level of self-confidence is plotted against the highest degree of supervisors' evaluation in Figure 4. Students’ competence constantly increases in the period September-May. High-level self-confidence decreases in the September-January and increases in the period January-May. Figure 4 indicates that a constant increase in all domains of students’ competence is not followed by constant self-confidence in the period September-May.

Figure 4.

Graphical representation of student responses (% of highly confident -HC) and supervisor evaluation (% of level 3 of trustability).

DISCUSSION

Dunning - Kruger effect has been defined as ‘A hypothetical cognitive bias stating that people with low ability at a task overestimate their own ability (self-competence) and people with high ability at a task underestimate their own ability (self-competence)’ (12) or as a ‘failure to recognize incompetence among the incompetent’ (7). With regards to the changing confidence levels of students in the three competency domains, we observed that the number of students who believed that they were highly confident (level 3) in the competency domains of communication, diagnosis, and clinical skills in September (47/133) (62/133) and (46/133), decreased in January to (33/133), (28/133) and (32/133) followed by a significant increase in May (85/100), (95/133) and (114/133) [Graph 1] [p<0.001].

These results can be explained by the Dreyfus model, the theory of cognitive bias, and confirmation bias. Novices (students at the beginning of the 9th semester) who were exposed to the competency domains of communication, diagnosis, and clinical skills in the pediatric clinics for the first time would have overestimated themselves and neglected evidence that contradicted them merely to arrive at a diagnosis and perform the needed clinical procedure refusing to consider an alternative diagnosis, a manifestation of ‘denial’ resulting from sub-optimal social and self-awareness (13, 14). Students at this level exhibit 'illusory superiority' wherein they consider themselves better than others. Mazor and Fleming have described this overconfidence of low performers as a ‘metacognitive deficiency’ (16). As quoted by Confucius, students failed to understand that ‘real knowledge is to know the extent of one’s ignorance (1). Dunning and Kruger reported a similar observation wherein participants who had scored in the bottom quartile (12th percentile) on tests of humour, grammar, and logic grossly overestimated their test performance and ability to be in the 62nd percentile (16). With increased exposure, students were able to recognize their social and intellectual deficits and hence, chose to learn more which can be correlated to the decrease in the number of highly confident students in January in all the three competency domains. With increased learning and clinical exposure in the paediatric clinics, the number of highly confident students increased dramatically by the end of May. On the contrary, there could also be a possibility that high performers would have underestimated themselves and settled for level 2 in the questionnaire (moderately confident), explained as the ‘undue modesty of the top performers’ (6).

The data presented in graph one was cross-tabulated, which determined a statistically significant change in student responses across all domains and time points (p <0.001) [Table 1]. The cross-tabulation of responses in the domain of communication skills indicates that after the initial high level of self-confidence, it decreases significantly in the second measurement to rises sharply again in the last measure in May. (Table 1 – Communication Skills). In our opinion, communication skills are particularly sensitive and largely depend on personality, heritage, and social environment (15). Considering that students have been exposed to patients in the previous years of their curriculum and have had the opportunity to develop communication skills, it is not surprising to observe an initial enthusiasm for their abilities. After exposure to paediatric patients and understanding the complexity of communication in the dentist-patient-parent communication triangle, they become aware of their shortcomings. After weeks of clinical training, students improve by developing special skills that are important for communicating with children and parents, significantly contributed by a standardized protocol for the patient's anamnestic dental data, and a modified Calgary-Cambridge-type checklist used during their clinical training (16).

A similar trend was observed in the domain of diagnostic skills. Initial enthusiasm faded in January with greater student exposure to different clinical cases which required increased learning to arrive at a final diagnosis considering the various differentials (Table 1 Diagnosis) (p <0.001). The longitudinal learning process, which included Team-Based Learning (TBL), and adequate and effective clinical training, significantly contributed to the increase in students’ self-confidence demonstrated in May (17).

In the domain of clinical skills, the same patterns of student responses were observed (Table 1 - Clinical skills). A high level of student self-confidence during the survey in May is noticeable (p<0.001), which can be explained by the intensive clinical training and enhanced psychomotor skills. Yet another factor responsible for the increase in self-confidence is their previous clinical exposure to the different clinical branches of dentistry. Motor (manual) skills are an essential component of the capabilities required of dentists. They must be developed longitudinally with an adequate number of cases that allow crafting this skill (18).

Results show a significant association between students’ communication confidence and skill assessment in all three measurements and between students’ diagnosis confidence and skill assessment (Table 2) (p<0.001). It is known that the age of the respondents can affect the results of the Dunning-Kruger analysis. Older respondents show greater caution when assessing their abilities, which is in line with their greater experience (19). The cohort examined did not have age variability. Therefore, age cannot be attributed to the observed results. The predictable tendency of students' self-confidence to increase during clinical training proves the importance of repeating clinical procedures as an essential part of skills development. Literature research emphasizes that on-site evaluation and real-time feedback by supervisors confirm the efficiency and justification of the current teaching approach (20).

On comparing the student responses (only with a high degree of reliability) and supervisor evaluation (confidence level 3 only) (Graph 2), student responses did not correlate with supervisor evaluation, particularly during the first analysis in September. The highest correlation was observed during the second analysis in January. In May, the measurement followed a similar compliance trend (p<0.001). Students' cognitive bias can explain the research results following the characteristic curve of the Dunning-Kruger test. In our opinion, the previous clinical exposure of students during the 7th and 8th semesters in various specialties of dentistry except paediatric dentistry made students believe that the mode of communication, diagnosis, and clinical skills would be the same while dealing with children and caregivers. This experience influenced their perception of their self-confidence levels.

Understanding the Dunning-Kruger effect is also crucial from the aspect of peer reviewers who must pay special attention to the introductory classes of clinical training. Students overestimating their abilities can sometimes affect the outcome of clinical treatment. Students who show a lower level of performance during clinical assignments are also less likely to accept objective feedback from supervisors. The same behavioral tendency is present in young physicians and dentists who have completed an undergraduate program and tend to overestimate their knowledge and capabilities (21).

Constantly providing constructive feedback to students with clear instructions from the supervisor in a clinical setup must be a mandatory part of the training. In this way, students can develop critical thinking and provoke self-doubts that would aid students in improving their knowledge and skills during clinical training (22). It is understood that inflated self-assessments stem from difficulty recognizing one’s own incompetence (1). Hodges, Regeher, and Martin suggested several modalities to overcome the DK effect such as teaching/testing self-assessment ability during medical school and residency, selection tests of self-assessment capability prior to medical training; and the development of self-assessment skills and self-directed learning during undergraduate and postgraduate education (23). One of the key strategies to overcome the Dunning-Kruger effect is to professionally mentor underperforming students and create a sense of self-doubt which is a critical step toward improved performance (24). Mentoring is the cornerstone for developing knowledge and clinical competencies (25) Regular academic advising, including constructive feedback, would help over-confident students realize the boundaries of their actual competence. In our opinion, awareness sessions on the existence of the Dunning-Kruger effect and its implications on the professional development of students should be conducted in the initial years of the dental curriculum.

Nonetheless, the study had a few limitations such as a limited sample size and was conducted in only one academic institute in the UAE. Additionally, the research relied on results interpreted from the assessment of clinical skills in only one dental specialty. Furthermore, the limited exposure of students to clinical training on pediatric patients could have had an indirect impact on their cognitive bias.

CONCLUSION

The research results concluded that the Dunning-Kruger effect, which is the cognitive bias of the perception pattern, is present in the examined group of dental students.

Students who exhibit this effect may present with attitudes and approaches, resulting in poor communication in the student-child-parents triangle, misdiagnosis, and poor performance in the clinical procedures.

Although fostering students’ exposure to clinical cases is the primary goal in dental education, this can be synergistically augmented with academic advising, mentoring, and real-time constructive feedback and awareness sessions