Item analysis and performance of EMQ, MCQ, and SAQ in dental education and students’ perception of extended-matching questions Authors: Parya Karamdokht, Yasamin Babaee Hemmati, Ideh Dadgaran, Mehran Falahchai Published: 2026 ABSTRACT BackgroundAlthough multiple-choice questions (MCQs), short-answer questions (SAQs), and extended-matching questions (EMQs) are widely used in dental education, simultaneous comparison of the psychometric properties of these formats in one single educational setting has been less investigated. Also, given the more complex structure of the EMQs, the role of students’ perception in their approach and performance still needs further investigation. This study aimed to conduct an item analysis and compare student performance across EMQ, MCQ, and SAQ formats in dental education and to assess students’ perception of extended-matching questions.MethodsThis cross-sectional analytical study was conducted on 29 senior dental students in 2025-2026. The end-of-semester exams of four selected courses included a total of 15 EMQs, 45 MCQs, and 30 SAQs, all of which were designed and standardized based on the Bloom’s taxonomy levels. Perception of EMQs was assessed with a standard questionnaire with confirmed validity and reliability [content validity ratio (CVR) = 0.85, content validity index (CVI) = 0.82, Cronbach’s alpha = 0.81] and a five-point Likert scale (range 15 to 75). Psychometric properties including difficulty, discrimination, and distractor efficiency (DE) of the answer choices were calculated. Data were analyzed by ANOVA, independent t, Mann-Whitney, repeated measures, and Pearson’s correlation tests (α = 0.05).ResultsEMQs were significantly more difficult than MCQs and SAQs (P < 0.001). There was no difference in discrimination among the formats (P > 0.05), and DE was higher in MCQs (P = 0.014). Student performance differed significantly among the formats (P < 0.001), with students obtaining the highest mean scores in SAQs. The mean perception score of EMQs was 44.80 ± 6.92, which was within the average range. Perception did not differ by gender (P = 0.786) or grade point average (GPA) (P = 0.386), but it did differ among the courses (P < 0.001). No significant correlation was found between perception and performance in EMQs (P = 0.211).ConclusionStructural differences in question formats affect student experience and performance, and the mean perception of EMQs is not necessarily related to performance. Appropriate use of formats and strengthening students’ familiarity with EMQs can help improve the quality of assessment. Background Assessment has been reported to be the most effective method for improving the quality of students’ academic performance [1], because most students focus on assessment more than any other component of their academic program [2]. Assessment has significant educational value because it has the potential to motivate, facilitate, and reinforce learning [3]. The assessment method used in an examination is important [4] .It influences how students learn and prepare for the examination, and thus has important educational consequences in terms of the skills and knowledge they acquire [5]. Assessment affects student learning in at least four ways: its content, its format, its timing, and the feedback that is provided to the student afterward [6]. Among assessment tools, written assessments continue to be a major part of dental education due to their ease of administration, standardization, and ability to measure a range of cognitive levels [7, 8]. These tools are generally presented in the form of structured response questions such as short-answer questions (SAQs), multiple-choice questions (MCQs), and extended-matching questions (EMQs) [9]. Each of these formats has a different ability to assess the level of learning: SAQs require students to actively retrieve knowledge and explain clinical reasoning, but their scoring is time-consuming and prone to inconsistency [10]. MCQs allow for the assessment of a large number of concepts in a short time and with high objectivity, but their quality depends on the designer’s skill in generating highly differentiated questions and efficient distractor options [11]. In contrast, EMQs, which are a more advanced form of MCQs, can test the ability to solve the problem, differential diagnosis, and clinical reasoning more carefully and reduce the likelihood of guessing the answer [12]. Psychometric test analysis is a systematic process of a set of defined events for the purpose of data collection and assessment of test quality [13]. In the process of question analysis, the difficulty factor, the discrimination factor of the question between proficient and non-p roficient students, and the efficiency of distractor options are among the factors that determine the quality of the test [14]. The discrimination factor indicates the ability of a question to distinguish between high-performing and low-performing students [15]. Distractor efficiency (DE) indicates the ability of distractor options to distract test takers from the correct answer [16]. A functional distractor is an option that is able to distract test takers from the correct answer and is selected by at least 5% of the participants [16]. The decision to revise questions is made based on the appropriate difficulty factor, the clarity factor, and the effective distractor options [17]. According to studies conducted in recent years, although most students find MCQs easier, they believe that EMQs better assess educational content and lead to more effective learning [6, 18]. However, student performance varied in different studies. According to a study of Nalini et al.,19 students performed better on familiar assessment tools such as MCQs, and in a study by Shehzad et al.,20 the results showed that students performed better on EMQs than on MCQs. To date, most studies [9, 19, 21–25] have either focused on one single question format or have rarely included comprehensive psychometric analysis—including difficulty factor, discrimination factor, and DE. Also, despite the more complex structure of EMQs, the relationship between students’ perceptions of this format and their actual performance has not been systematically investigated. Accordingly, the present study aimed to perform an item analysis and compare student performance across EMQ, MCQ, and SAQ formats and to examine students’ perception of extended-matching questions. Materials and methods This cross-sectional analytical study was conducted at the Faculty of Dentistry, Guilan University of Medical Sciences in 2025-2026. The study protocol was approved by the Ethics Committee of Guilan University of Medical Sciences (IR.GUMS.REC.1404.343). Changes in performance score were used to estimate the sample size. For this purpose, the formula for repeated measures was used. Considering the standard deviation of 26 extracted from a study by Nalini et al.,19 an error level of 0.05, power of 0.80, and effect size of 0.25, a minimum sample size of 28 was obtained (sample size was calculated using G*Power software version 3.1.9.2). To determine the number of questions required in each format, the changes in the discrimination factor in the analysis of variance were used as the basis for justifying the minimum number of items required. Based on the standard deviation of the discrimination factor reported in a study by Frey et al., [11] (standard deviation = 0.113), an error level of 0.05, a power of 0.95, and an effect size of 0.43, the minimum number of items required to achieve a stable estimate of the psychometric indices was calculated to be 15 for the EMQs, 30 for the SAQs, and 45 for the MCQs (calculations were performed using PASS software version 25.0.2). The target population included senior dental students at Guilan University of Medical Sciences in 2025-2026. Due to the limited number of theoretical courses, census sampling was performed on all students who had taken the selected courses in the relevant semester. The inclusion criteria included senior dental students at Guilan University of Medical Sciences in 2025-2026 who had officially taken the selected theoretical courses. The exclusion criteria included absence from the final semester exam, failure to complete the questionnaire, or incomplete questionnaires. From each of the endodontics, pediatric dentistry, prosthodontics, and orthodontics departments of the Faculty of Dentistry, theoretical courses with the highest number of units and high content coverage were selected through purposive sampling after consultation with the head of the departments and faculty members of each department. The reason for selecting the aforementioned departments was that the students’ university admission date was the same, which allowed for better control of the intervening and confounding variables related to educational background. Also, students of this class had the highest number of specialized theoretical courses in the relevant semester. Since they were in their final year of dental education, they had a greater grasp of the content of the selected courses, and their evaluations and opinions could have been more valid. Each test consisted of three question formats: EMQ, MCQ, and SAQ. All questions were standardized based on the Bloom’s taxonomy. To this end, a panel consisting of at least two faculty members from each course and a medical education expert approved this harmonization and standardization before administering the test. The research tool used to examine the students’ perception was the English questionnaire used in a study by Lin et al. [9]. The questionnaire was translated into Persian using the forward-backward translation method by two independent translators. The questionnaire consisted of 15 close-ended questions with a 5-point Likert scale (1 = strongly disagree, 2 = disagree, 3 = no opinion, 4 = agree, 5 = strongly agree). Since the main aim of the present study was to analyze the students’ perception of the EMQ format, only the students’ perception of the EMQ format (in terms of difficulty, clarity, and educational quality) was evaluated. The minimum score obtained was 15 and the maximum was 75. The validity of the questionnaire was determined through consultation with 10 relevant experts, and necessary amendments were made. In the validity study, a content validity ratio (CVR) = 0.85 was obtained, which was confirmed considering the minimum acceptable value of the Lawshe’s Table for 10 experts, which was equal to 0.62. The content validity index (CVI) = 0.82 was also obtained, which was also confirmed considering the minimum acceptable value of 0.79, and in general, the validity of the questionnaire was confirmed. To examine the reliability, the questionnaire was given to 15 people in the target group. In the reliability study, the Cronbach’s alpha coefficient was calculated, which was equal to 0.81, and confirmed considering the minimum acceptable value of 0.70 for the above questionnaire. Demographic information of the students including gender, grade point average (GPA), performance score, and course unit, was also taken into account. The tests were administered in person during the end-of-semester exams. The students’ perception questionnaire was given to the students immediately after the end of each test. Each student was assigned a unique anonymous code that was simultaneously placed on the test sheet and the questionnaire to maintain confidentiality and to allow for linking performance and perception data at the individual level. Student performance on each test was calculated based on the percentage of correct answers to the EMQs, MCQs, and SAQs. For this purpose, the number of correct answers for each student in each question format was calculated and then divided by the total score obtainable in the same question format to obtain the performance percentage. Due to the difference in the number of questions in each format in different courses, all scores were normalized as a percentage score. This method allowed for comparability among students and different courses. In examining the students’ perception of the EMQ format, 29 students entered the study. To clarify the structure of the tests, a total of 15 EMQs, 45 MCQs, and 30 SAQs were used in four selected courses. The number of questions in each format was not exactly the same across courses, but their distribution was proportional to the content volume of each course to maintain sufficient variety of items and appropriate coverage of educational objectives. All EMQs were designed based on the standard structure of this format; each question included a theme, a lead-in, between 2 and 5 stems, and a list of 8 to 10 answer options. The number of answer options in all tests was considered constant or nearly constant to prevent the creation of unwanted clues and to allow for valid psychometric analysis. The MCQs were all four-choice with one correct answer, and the SAQs were also scored 1 to 3. SAQs were scored using a predefined scoring rubric developed by the course instructors prior to the examination. Although rubric-based scoring was used to improve standardization, the authors acknowledge that some degree of examiner-related subjectivity may still exist in constructed-response assessments. Each item was assigned a score ranging from 0 to 3 based on completeness and accuracy of the response. All responses were graded by the course instructor according to the standardized rubric. Psychometric properties including difficulty factor, discrimination factor, and DE were calculated for each question separately, and the results were finally compared. The difficulty factor of a question in MCQs and EMQs indicates the proportion of students who answered that question correctly. Its value is between zero and one, and is expressed as a percentage. The difficulty factor in descriptive questions was calculated based on the mean score obtained compared to the full score [22]. Questions with a difficulty factor less than 30% were considered difficult questions. Questions with a difficulty factor between 30% and 70% were considered medium level, and questions with a difficulty factor greater than 70% were classified as easy questions [26]. A higher value of difficulty factor meant that the question was easier, and a lower value indicated that the question was more difficult [26]. The discrimination factor indicates the ability of a question to correctly distinguish participants with high knowledge or skill levels from those with lower levels. To calculate it, the participants were first divided into two groups, those with the highest and lowest scores. The discrimination factor for the EMQs and MCQs was then calculated as the ratio of the difference in the number of correct answers in the high and low groups to the number of people in the high or low groups. For descriptive questions, the ratio of the difference in the mean scores of the high and low groups to the number of questions was used to determine the discrimination factorn [24]. Questions with a discrimination factor of less than or equal to 0.19 were considered to require removal or substantial modification. Questions with a discrimination factor between 0.20 and 0.29 were considered to require revision, questions with a discrimination factor between 0.30 and 0.39 were considered desirable, and questions with a discrimination factor of 0.40 and above were considered very good questions [27]. DE is the ratio of the number of functional distractors to the total number of distractors in a question. It is a measure of the effectiveness of distractors in attracting respondents. Statistically, an option that was selected by at least 5% of the participants was considered an effective or functional distractor [16]. The Shapiro–Wilk test was used to check the normality of the data, and the Levene’s test was used to assess the homogeneity of variances. The independent samples t-test, Mann-Whitney U test, ANOVA (with the Tukey’s pairwise comparisons), repeated measures test (with the Bonferroni pairwise comparisons), and Pearson’s correlation test were used to analyze the data. SPSS version 27 was used to perform the calculations. The significance level in all tests was set at 0.05. Results In this study, 29 students were evaluated (16 males, 13 females; 55.2% vs. 44.8%, respectively). The mean GPA of the students was 16.22 ± 0.91 out of 20, and the GPA distribution was approximately balanced (≤ 16: 48.3%; >16: 51.7%). The perception questionnaire was completed in person immediately after the tests, and the completion rate among the analyzed individuals was 100% (29/29). The psychometric properties of the questions showed that the difference in the difficulty factor among the types of question formats was statistically significant (ANOVA; F = 11.94, P < 0.001). According to the results obtained from the Tukey’s test, the difficulty factor of the EMQs was significantly different from that of SAQs and MCQs (P < 0.001), such that the difficulty factor of the EMQs was lower than the other two question formats, but there was no significant difference between SAQs and MCQs (P = 0.364). Regarding the difficulty factor classification, the EMQs were placed in the category of medium questions with a mean of 49.26 ± 27.06, and the SAQs and MCQs were placed in the category of easy questions with a mean of 76.69 ± 18.48 and 71.33 ± 22.42, respectively. The difference in the discrimination factor among the formats did not reach the significance threshold (F = 2.91, P = 0.058). Based on the mean discrimination factor of each question format, it can be concluded that the SAQs and MCQs with a mean of 0.29 ± 0.20 and 0.25 ± 0.19, respectively, needed revision due to low discrimination factor. However, the EMQs with a mean discrimination factor of 0.37 ± 0.23 were optimal. For the DE, the comparison of the EMQs and MCQs with the Mann–Whitney test showed a significant difference (P = 0.014), and the MCQs had a higher mean DE (43.63 ± 29.39) than the EMQs (29.95 ± 21.46) (Table 1). Table 1Psychometric properties including difficulty factor, discrimination factor, and DE for each question formatFactorFormatMean ± SDMedianMinimumMaximumP value (statistics)DifficultyEMQ49.26 ± 27.0654.16091.67< 0.001* (11.94)SAQ76.69 ± 18.488027100MCQ71.33 ± 22.42764.17100DiscriminationEMQ0.37 ± 0.230.3200.910.058* (2.91)SAQ0.29 ± 0.200.2500.80MCQ0.25 ± 0.190.2300.83DEEMQ29.95 ± 21.46250750.014** (2.45)MCQ43.63 ± 29.39330100EMQ Extended Matching Question, SAQ Short Answer Question, MCQ Multiple Choice Question, SD Standard Deviation* ANOVA, ** Mann-Whitney U test Table 1Psychometric properties including difficulty factor, discrimination factor, and DE for each question formatFactorFormatMean ± SDMedianMinimumMaximumP value (statistics)DifficultyEMQ49.26 ± 27.0654.16091.67< 0.001* (11.94)SAQ76.69 ± 18.488027100MCQ71.33 ± 22.42764.17100DiscriminationEMQ0.37 ± 0.230.3200.910.058* (2.91)SAQ0.29 ± 0.200.2500.80MCQ0.25 ± 0.190.2300.83DEEMQ29.95 ± 21.46250750.014** (2.45)MCQ43.63 ± 29.39330100EMQ Extended Matching Question, SAQ Short Answer Question, MCQ Multiple Choice Question, SD Standard Deviation* ANOVA, ** Mann-Whitney U test Psychometric properties including difficulty factor, discrimination factor, and DE for each question format EMQ Extended Matching Question, SAQ Short Answer Question, MCQ Multiple Choice Question, SD Standard Deviation * ANOVA, ** Mann-Whitney U test The mean performance of students in the three question formats showed a significant difference (repeated measures test; F = 42.44, P < 0.001). Descriptively, students performed best in the SAQs (78.17 ± 13) and then MCQs (72.84 ± 9.23), and the weakest performance was related to the EMQs (55.87 ± 17.84) (Table 2). Based on pairwise comparisons with the Bonferroni test, the difference between all question formats was significant (P < 0.05). In the analysis of performance based on demographic variables, regarding the GPA variable, a significant difference was observed only in the MCQs (independent t-test; P = 0.009), and students with GPA > 16 obtained higher scores. There was no significant difference in the EMQs (P = 0.319) and SAQs (P = 0.061). Comparison of the performance score according to the question format in the two GPA groups using the repeated measures test revealed that there was a significant difference in the performance score according to the question format in both groups (P < 0.001, Fig. 1). Based on the obtained results, there was no significant difference between the SAQ and MCQ performance scores in GPA ≤ 16 (P = 0.592). However, the difference was significant in the remaining two pairwise comparisons (P < 0.05). In GPA > 16, all pairwise comparisons were significant (P < 0.05). Table 2Comparison of students’ academic performance (score) in response to EMQs, MCQs, and SAQsVariableMean ± SDMedianMinimumMaximumP value (statistics)*EMQ55.87 ± 17.845623.7587.25< 0.001 (42.44)MCQ72.84 ± 9.2373.7550.5089.75SAQ78.17 ± 1381.754694.75EMQ Extended Matching Question, SAQ Short Answer Question, MCQ Multiple Choice Question, SD Standard Deviation* Repeated measures test Table 2Comparison of students’ academic performance (score) in response to EMQs, MCQs, and SAQsVariableMean ± SDMedianMinimumMaximumP value (statistics)*EMQ55.87 ± 17.845623.7587.25< 0.001 (42.44)MCQ72.84 ± 9.2373.7550.5089.75SAQ78.17 ± 1381.754694.75EMQ Extended Matching Question, SAQ Short Answer Question, MCQ Multiple Choice Question, SD Standard Deviation* Repeated measures test Comparison of students’ academic performance (score) in response to EMQs, MCQs, and SAQs EMQ Extended Matching Question, SAQ Short Answer Question, MCQ Multiple Choice Question, SD Standard Deviation * Repeated measures test Fig. 1Students’ academic performance scores in response to EMQs, MCQs, and SAQs based on a GPA less than or equal to 16 and greater than 16 Students’ academic performance scores in response to EMQs, MCQs, and SAQs based on a GPA less than or equal to 16 and greater than 16 In terms of gender, a significant difference was observed only in the SAQs (independent t-test; P = 0.043) and females had higher performance; no difference was observed in the EMQs (P = 0.713) and MCQs (P = 0.150). Comparison of the performance score according to the question format in males and females using repeated measures test indicated that in both groups of males and females, the performance was significantly different according to the question format (P < 0.001; Fig. 2). Based on the results obtained from pairwise comparisons, there was no significant difference between the performance score of SAQs and MCQs in males (P = 0.814). However, there was a significant difference in the other two pairwise comparisons (P < 0.05). In females, all pairwise comparisons were significant (P < 0.05). Fig. 2Students’ academic performance scores in response to EMQs, MCQs, and SAQs by gender Students’ academic performance scores in response to EMQs, MCQs, and SAQs by gender Performance in all three question formats was significantly different based on the course (repeated measures; P < 0.001). Comparison of the performance score in each course based on the question format indicated that there was no significant difference among the performance scores of the three types of questions only in the orthodontics course (P = 0.948). In the remaining three courses, there was a significant difference among the three question formats in terms of performance score (P < 0.05, Table 3). Table 3Summary of the results of determining differences in student performance on three types of questions (EMQ, MCQ, SAQ) based on courseQuestion formatCourseP value (statistics)*Pediatric dentistry (Mean ± SD)Endodontics (Mean ± SD)Prosthodontics (Mean ± SD)Orthodontics (Mean ± SD)EMQ24.83 ± 21.14 Ba64.65 ± 31.70 Aa66.41 ± 32.24 Aa67.59 ± 18.50 A< 0.001 (23.96)MCQ78.65 ± 9.6 Cb68.14 ± 17.96 ABa77.90 ± 12.83 BCa66.65 ± 10.75 A< 0.001 (9.78)SAQ77.17 ± 11.28 Ab79.69 ± 12.31Ab88.21 ± 11.73 Bb67.62 ± 29.24 A< 0.001 (10.49)P value (statistics)*< 0.001 (144.78)0.007 (6.35)< 0.001 (9.28)0.948 (0.03)EMQ Extended Matching Question, SAQ Short Answer Question, MCQ Multiple Choice Question, SD Standard Deviation*Repeated measures test. Values with the same uppercase letters in the same row and those with the same lowercase letters have no significant difference (P > 0.05) Table 3Summary of the results of determining differences in student performance on three types of questions (EMQ, MCQ, SAQ) based on courseQuestion formatCourseP value (statistics)*Pediatric dentistry (Mean ± SD)Endodontics (Mean ± SD)Prosthodontics (Mean ± SD)Orthodontics (Mean ± SD)EMQ24.83 ± 21.14 Ba64.65 ± 31.70 Aa66.41 ± 32.24 Aa67.59 ± 18.50 A< 0.001 (23.96)MCQ78.65 ± 9.6 Cb68.14 ± 17.96 ABa77.90 ± 12.83 BCa66.65 ± 10.75 A< 0.001 (9.78)SAQ77.17 ± 11.28 Ab79.69 ± 12.31Ab88.21 ± 11.73 Bb67.62 ± 29.24 A< 0.001 (10.49)P value (statistics)*< 0.001 (144.78)0.007 (6.35)< 0.001 (9.28)0.948 (0.03)EMQ Extended Matching Question, SAQ Short Answer Question, MCQ Multiple Choice Question, SD Standard Deviation*Repeated measures test. Values with the same uppercase letters in the same row and those with the same lowercase letters have no significant difference (P > 0.05) Summary of the results of determining differences in student performance on three types of questions (EMQ, MCQ, SAQ) based on course EMQ Extended Matching Question, SAQ Short Answer Question, MCQ Multiple Choice Question, SD Standard Deviation *Repeated measures test. Values with the same uppercase letters in the same row and those with the same lowercase letters have no significant difference (P > 0.05) The mean score of students’ perception of the questionnaire used for EMQs (a combination of difficulty, clarity, and educational efficacy) was 44.80 ± 6.92. There was no significant difference in students’ perception of EMQs based on gender (independent t-test; P = 0.786) or GPA (independent t-test; P = 0.386). In contrast, a significant difference was observed based on the course with repeated measures test (F = 6.46, P < 0.001), and the Bonferroni pairwise comparisons showed that questions related to pediatric dentistry had lower perception scores than orthodontics (P = 0.029) and prosthodontics (P = 0.005) questions; while, other comparisons were not significant (P < 0.05, Table 4). Table 4Summary of the results of determining differences in students’ perceptions of EMQs based on courseCourseP value (statistics)Pediatric dentistry (Mean ± SD)Endodontics (Mean ± SD)Prosthodontics (Mean ± SD)Orthodontics (Mean ± SD)Perception of EMQ41.41 ± 7.9142.86 ± 7.7248.48 ± 10.2746.45 ± 10.250.001 (6.46)Pairwise comparisonsOrthodonticsProsthodonticsEndodonticsPediatric dentistryOrthodonticsProsthodontics0.999Endodontics0.4080.099Pediatric dentistry0.0290.0050.999Repeated measures test followed by Bonferroni, SD Standard Deviation Table 4Summary of the results of determining differences in students’ perceptions of EMQs based on courseCourseP value (statistics)Pediatric dentistry (Mean ± SD)Endodontics (Mean ± SD)Prosthodontics (Mean ± SD)Orthodontics (Mean ± SD)Perception of EMQ41.41 ± 7.9142.86 ± 7.7248.48 ± 10.2746.45 ± 10.250.001 (6.46)Pairwise comparisonsOrthodonticsProsthodonticsEndodonticsPediatric dentistryOrthodonticsProsthodontics0.999Endodontics0.4080.099Pediatric dentistry0.0290.0050.999Repeated measures test followed by Bonferroni, SD Standard Deviation Summary of the results of determining differences in students’ perceptions of EMQs based on course Repeated measures test followed by Bonferroni, SD Standard Deviation The Pearson’s correlation between students’ perceptions and their actual performance on the EMQs was not significant (r=-0.239, P = 0.211), indicating that subjective perceptions do not necessarily align with actual performance on this format. Discussion In assessment of the psychometric properties of the question formats, the mean difficulty factor of EMQs was reported to be moderate and significantly lower (more difficult) than the SAQs and MCQs that were classified as easy questions. While, no significant difference was observed between SAQs and MCQs. One possible reason could be the structural nature of EMQs; that is, this format usually includes long clinical scenarios and a set of similar options that require deeper cognitive processing, clinical data analysis, and complex diagnostic reasoning [11]. Also, in EMQs, the probability of guessing the correct answer is lower due to the higher number of distractor options [28]. A similar result was obtained in a study by Frey et al. [11], in which EMQs with moderate difficulty showed a slightly lower mean difficulty factor (more difficult) than other formats that included MCQs. However, this finding was in contrast to the results of some studies. For example, in a study by Kumar et al. [12], both EMQ and MCQ formats were at a similar, moderate level of difficulty. Mee et al. [29], also showed that SAQs were overall more difficult than MCQs with a moderate level of difficulty. These results may be due to differences in study design, the number of questions used in the tests, and the different content of the tests. No significant difference was observed in the discrimination factor among different question formats. Similar studies in this field also show that the discrimination coefficient of questions depends more on the content and writing quality of the items than on the type of question format [12, 30]. According to the mean discrimination factor values in the present study, the EMQs were at the optimal level, but the MCQs and SAQs needed revision. These results are comparable to the mean discrimination values obtained in previous studies [31, 32]. In a study by Kumar et al. [12], similar to the present study, the mean discrimination value was calculated to be higher for EMQs than MCQs, but this difference was not statistically significant. Also, in a study by Mee et al. [29], no difference was observed between the discrimination values of MCQs and SAQs. Assessment of the DE showed that there was a statistically significant difference between the EMQs and MCQs, such that the mean DE of the MCQs was higher than the EMQs. This result is probably related to the more compact structure and limited number of options in MCQs; which allows the question designer to design more attractive options and also reduces the possibility of ineffective distractor options [33]. This finding is consistent with the results of a study by Kumar et al. [12], who reported that EMQs showed lower DE than MCQs. According to the present results, the highest student scores were observed in the SAQs while the weakest performance was observed in the EMQs. However, this finding should be interpreted with caution because performance in SAQs may partly be influenced by examiner-related variability and the inherent subjectivity of constructed-response scoring, even when standardized rubrics are used [28, 29]. The relatively higher scores observed in SAQs may also be related to the absence of distractor options and the opportunity for students to focus directly on the content of the question. While in EMQs, due to the large number of options and the high similarity between them, choosing the correct answer is more difficult, and the need for a higher cognitive load reduces the level of performance [11]. This finding is consistent with the results of Nalini et al. [19], who reported that students performed better in more familiar assessment tools, such as MCQs, compared to less familiar tools, such as EMQs and structured clinical thinking tests. Also, Sam et al. [34] compared SAQs and MCQs, and showed that the mean score of students in SAQs was significantly higher than their score in MCQs. Examining student performance based on contextual variables showed that students with higher GPA performed better on MCQs, but no significant difference was observed on SAQs and EMQs. However, in both groups, the type of question format had an effect on performance, and in the group with higher GPA, this difference was significant in all formats. This result could be because students with higher GPA, due to their stronger memory and better ability to recognize information, perform better on MCQ questions that require more theoretical knowledge and test-taking skills; whereas, in EMQs and SAQs that require answer generation, analysis, and clinical reasoning [9], there was no significant difference in the performance of students with higher GPA. This study was similar to the results of Darmiani et al. [35], in which a direct relationship was observed between students’ scores on MCQ questions and their GPA. In examining the effect of gender on performance of students, no significant difference was observed between the performance of males and females in MCQs and EMQs, but the performance of females was better in SAQs, which could be due to differences in learning styles and organization of responses. Also, the type of question format affected the performance of both male and female students. This finding was in line with previous studies on the effect of gender on student performance in MCQs, in which no significant relationship was observed between gender and mean score of students [36, 37]. In comparison of students’ perceptions of EMQs, gender and GPA did not have a significant effect on students’ level of perception. In contrast, the course played a decisive role in the difference in students’ perception, and a lower perception score was observed in pediatric dentistry than orthodontics and prosthodontics. This finding could be due to the different levels of students’ preparation in different exams, the way the instructors designed the questions for each course, and their different levels of difficulty. Lack of deep understanding of clinical situations and limited practical experience in dealing with the scenarios presented in the form of EMQs could lead to a decrease in the level of perception of these questions by students [20]. In a study conducted by Lin et al. [9], the findings also showed that gender did not have a significant effect on students’ perception of EMQs. In examining the relationship between performance and students’ perceptions of EMQs, the present results indicated that students’ subjective perceptions of the difficulty and quality of EMQs did not necessarily correspond to their actual performance on this type of test. The perception of question quality may be influenced by the level of self-confidence and previous experience in answering similar questions, rather than reflecting actual cognitive ability or the degree of mastery of scientific content. This difference makes the subjective perception of difficulty not necessarily an accurate indicator of performance. Also, test anxiety and psychological stress, and lack of adequate preparation can cause actual performance to be lower than the level of subjective ability, even if the student evaluates the questions as easy or understandable. EMQs usually have multiple and conceptually similar options, and answering them correctly requires critical thinking and information integration [38]. Therefore, a student may initially perceive the question as simple, but due to the sophistication of the options and their conceptual proximity, make a mistake in selecting the correct answer. Other studies, consistent with the results of the present study, were conducted by van Leeuwenkamp et al. [39] and Laprise et al. [40] in which the relationship between students’ perceptions of test quality and learning outcomes was examined. Although the scope of these studies was broader than the present study, and was not limited to a specific type of question, their results showed that perceptions of test quality in general did not have a direct relationship with performance or learning outcomes, and that this relationship was complex and multifactorial. In contrast, the results of a study by Alia et al. [41] showed that there was a significant relationship between students’ perceptions of the test and their performance in three types of tests: announced, immediately after the lesson, and unannounced. The reason for this difference could be the type of tests and the content studied. This study is important in several ways. First, the results of the psychometric analysis showed that the question format plays a decisive role in the level of difficulty and how students perform. The lower difficulty coefficient of the EMQs compared to other formats indicates that these types of questions can be a suitable tool for assessing clinical reasoning skills, complex data analysis, and decision-making ability in real educational situations. Such an approach can improve the quality of evaluation in dental education and move towards measuring deeper learning. On the other hand, the lack of significant difference in the discrimination coefficient among different question formats indicates that the quality of question design plays a more important role than the type of question format in distinguishing students with strong and poor performance. This result emphasizes the need to train and empower faculty members in the field of designing high-quality test items that are in line with educational goals. Also, the findings on the effectiveness of distractor options indicate that more careful design of options can lead to improved overall test quality. Therefore, using psychometric analysis as a feedback tool to improve the quality of questions can help standardize and continuously improve written assessments in dental education. These results can serve as a basis for reviewing educational and testing policies of dental schools. The balanced use of different question formats, according to the educational objective, can lead to the assessment of different dimensions of knowledge, skills, and reasoning, and creation of a fairer and more efficient evaluation system. In addition, the findings can help test designers identify the strengths and weaknesses of questions and take steps to improve the quality of education. A significant limitation of this study was the relatively small sample size, which may limit the generalizability of the results. Also, conducting the study in a dental school and within the framework of a specific educational system may not fully generalize the results to other educational centers with different conditions. The scope of the study also included only a few theoretical units and did not include all specialized areas of dentistry; therefore, the findings do not necessarily reflect the performance and perception of students in all educational fields. In addition, some individual and psychological factors, such as test anxiety, level of preparation, learning motivation, and prior familiarity with different question formats, were not controlled in this study and may have affected the results. Also, the psychometric analysis of the questions was conducted solely on the basis of one single test session, and the stability of the results in similar tests (inter-test reliability) was not assessed; Therefore, generalization of the results to other assessment situations should be done with caution. In addition, although standardized scoring rubrics were used for SAQs, examiner-related variability and the inherent subjectivity of scoring constructed-response questions could not be completely eliminated and may have affected the observed performance differences among assessment formats. Given the existing findings and limitations, future research with a larger sample size and in different dental schools is recommended to increase the external validity of the results. Conducting longitudinal research can provide the opportunity to examine changes in psychometric properties and students’ perceptions and performance at different educational stages. Conclusion According to the present results, question format plays an important role in students’ performance and item difficulty. EMQs were more difficult, whereas students obtained the highest mean scores in SAQs, followed by MCQs. Although the difference in discrimination factor among the formats was not statistically significant, the higher mean discrimination value observed in EMQs may indicate their potential capacity to better differentiate students. DE was significantly higher in MCQs, suggesting more effective distractor performance in this format. Students’ perception of EMQs was at an average level and was not associated with their actual performance. Taken together, these findings suggest that the purposeful use of different assessment formats alongside psychometric analysis may help improve the validity and fairness of assessments in dental education.