Self-oriented perfectionism and psychological well-being in medical students: cross-sectional indirect associations involving perceived cognitive flexibility and shame

Abstract

Background

Medical students encounter sustained academic, evaluative, and emotional demands that are relevant not only to psychological symptoms but also to positive psychological functioning. The present study focused on psychological well-being to capture eudaimonic dimensions of functioning, including self-acceptance, autonomy, purpose in life, and personal growth. Guided by multidimensional perfectionism theory, self-conscious emotion theory, and cognitive-regulatory accounts of rigidity and flexibility, this cross-sectional study examined the indirect associations of self-oriented perfectionism with psychological well-being through perceived cognitive flexibility and shame among Iranian medical students.

Methods

This cross-sectional study was conducted among medical students at Shahid Beheshti University of Medical Sciences from November 21, 2024, to July 21, 2025. Data were collected online using the Porsline platform. After excluding incomplete responses and outliers, 801 participants were retained for the primary analysis. Measures included the self-oriented perfectionism subscale of the Multidimensional Perfectionism Scale, the shame subscale of the Guilt and Shame Proneness Scale, the Cognitive Flexibility Inventory, and the short form of Ryff’s Psychological Well-Being Scale. Data were analyzed using IBM SPSS Statistics 27 and Hayes’ PROCESS macro (Model 4). Pearson correlations and parallel indirect-effects models were estimated using 5,000 bias-corrected bootstrap resamples with 95% confidence intervals and HC4 heteroskedasticity-consistent standard errors.

Results

Self-oriented perfectionism was inversely associated with perceived cognitive flexibility and psychological well-being, and positively associated with shame. In the primary model, the total indirect estimate was significant, B = − 0.135, 95% CI [− 0.175, − 0.096], with significant specific indirect estimates through perceived cognitive flexibility, B = − 0.119, 95% CI [− 0.158, − 0.085], and shame, B = − 0.016, 95% CI [− 0.031, − 0.003]. After controlling for age and gender, the total indirect estimate remained significant, B = − 0.131, 95% CI [− 0.173, − 0.094], as did the indirect estimates through perceived cognitive flexibility, B = − 0.110, 95% CI [− 0.148, − 0.074], and shame, B = − 0.021, 95% CI [− 0.037, − 0.007]. The adjusted model accounted for 47.4% of the variance in psychological well-being.

Conclusions

Self-oriented perfectionism was associated with lower psychological well-being among medical students, with statistically significant indirect components through lower perceived cognitive flexibility and higher shame. The indirect association through perceived cognitive flexibility was larger, whereas the shame pathway was statistically significant but comparatively small. Given the cross-sectional and self-report design, these findings should be interpreted as associations rather than evidence of temporal ordering or causal mechanisms.

Introduction

Medical students occupy a uniquely consequential position in the mental health literature: they are young adults navigating a highly demanding developmental period while simultaneously training for roles that will directly affect the quality, safety, and continuity of future healthcare. In this context, psychological well-being, as conceptualized in Ryff’s eudaimonic framework, extends beyond the absence of distress to encompass positive functioning, including self-acceptance, autonomy, purpose in life, and personal growth [1]. The present study therefore focused on psychological well-being rather than psychological symptoms alone. Symptom measures are indispensable for identifying distress and disorder; however, they do not fully capture whether students experience meaning, competence, growth, autonomy, and self-acceptance during medical training. This distinction is important because students may continue to meet academic expectations while experiencing deterioration in their sense of self, purpose, and inner stability [2]. Concern about medical-student mental health has intensified, as meta-analytic evidence indicates that 27.2% of medical students report depression or depressive symptoms and 11.1% report suicidal ideation, with substantial emotional exhaustion evident even before residency [3, 4]. This pattern persists into professional practice, where burnout remains widespread and suicide risk elevated, particularly among female physicians [5, 6].

Accordingly, research has increasingly examined both structural and individual determinants of medical students’ mental health. Systemic pressures, including heavy workloads, competitive environments, frequent evaluation, and exposure to suffering, intersect with individual differences in shaping students’ well-being [7]. Among these individual differences, personality-related variables are especially salient. In early multicohort study, perfectionism and impostor feelings were more strongly related to psychological adjustment than many demographic variables [8]. More broadly, the medical-education literature has consistently linked perfectionism with depression, anxiety, burnout, psychological distress, and diminished well-being, indicating that perfectionism is not a peripheral trait in this population but a meaningful correlate of student mental health [9].

Within Hewitt and Flett’s multidimensional framework, perfectionism includes self-oriented, other-oriented, and socially prescribed dimensions [10]. The present study focused specifically on self-oriented perfectionism for conceptual and contextual reasons. Self-oriented perfectionism is characterized by internally imposed standards, strong personal striving, and stringent self-evaluation. These features closely correspond to the internal performance pressure, personal ambition, and self-directed evaluation that are often reinforced in medical training. Empirical findings indicate that medical students tend to score highly on this dimension and that it is related to motivation, learning processes, and performance-related outcomes [11, 12]. At the same time, when these internal standards become rigid or self-punitive, self-oriented perfectionism has been linked to depression and anxiety in medical students [12]. Thus, this dimension captures a theoretically important tension between adaptive striving and vulnerability to self-directed distress in a context where high standards are constant but flawless performance is unattainable [13].

The focus on self-oriented perfectionism should not be interpreted as suggesting that other perfectionism dimensions are unimportant. Socially prescribed perfectionism may be particularly relevant to perceived external expectations, hierarchy, and fear of negative evaluation, whereas other-oriented perfectionism may be more closely related to interpersonal expectations, criticism of others, and conflict. Including all three dimensions would have allowed a broader comparative model and might have revealed distinct external-pressure or interpersonal pathways. However, the present study was designed to examine the self-evaluative dimension of perfectionism and its links with self-conscious emotion and perceived cognitive flexibility. This focused approach allowed a more specific test of the internal standards and self-appraisal processes most directly relevant to the selected outcome.

One affective process relevant to this self-evaluative model is shame. Shame is a painful self-conscious emotion organized around the perception that the self is defective, inadequate, or exposed [14]. Shame was selected rather than guilt because shame involves a global negative evaluation of the self, whereas guilt is more commonly tied to a specific behavior and may be more closely connected to reparative action. Perfectionistic individuals often interpret shortcomings as evidence of personal deficiency, which may be associated with greater shame [15]. In medical education, where errors, public evaluation, hierarchy, and clinical performance are highly salient, global self-evaluation may be especially consequential [16]. Available evidence suggests that shame experiences in clerkship and clinical learning environments can erode confidence, reduce motivation, impair engagement in learning, and negatively shape professional identity development [17]. From this perspective, students high in self-oriented perfectionism may be more likely to experience perceived shortcomings as shame-relevant events, which may coincide with lower psychological well-being.

A second, conceptually complementary process is cognitive flexibility. Cognitive flexibility refers to the capacity to adapt one’s thinking and behavior in response to changing demands, shift perspectives, and generate alternative interpretations [18]. We selected cognitive flexibility because it provides a direct theoretical bridge between perfectionistic rigidity and adaptive self-regulation. Medical training requires students to tolerate uncertainty, revise judgments, integrate feedback, and recover from setbacks without becoming fixed in rigid self-appraisals. Evidence suggests that maladaptive perfectionism is characterized by cognitive rigidity, dichotomous thinking, and perseverative self-criticism [19, 20], whereas cognitive flexibility is associated with lower levels of depression and anxiety, greater well-being, and more adaptive coping strategies [21, 22]. In the present study, cognitive flexibility was assessed by self-report; therefore, the construct is interpreted as perceived cognitive flexibility rather than as behavioral performance on an experimental task. Although mindfulness, anger directed toward the self, and other transdiagnostic variables are also relevant, cognitive flexibility was selected because it is most directly aligned with the rigidity-flexibility contrast central to perfectionistic self-evaluation.

Although prior studies have linked perfectionism in medical students to distress, depression, anxiety, burnout, procrastination, and academic processes [9, 23, 24], several gaps remain. First, much of this literature has emphasized negative symptoms or academic outcomes, whereas Ryff’s eudaimonic psychological well-being has received less direct attention as a positive-functioning outcome in relation to perfectionism. Second, perfectionism is often treated broadly or in terms of maladaptive/evaluative concerns, while self-oriented perfectionism requires more focused examination in medical education because it captures internally imposed standards, self-evaluation, and the tension between achievement striving and self-critical rigidity. Third, cognitive flexibility has been linked to coping and psychological well-being [21, 22], and shame has been increasingly recognized as important in medical learning environments [16, 17], but these cognitive-regulatory and self-conscious emotional processes have rarely been evaluated together in the association between self-oriented perfectionism and well-being. Addressing these gaps, the present study evaluated a theoretically specified parallel indirect-effects model in which an affective component (shame) and a cognitive-regulatory component (self-reported cognitive flexibility) were considered simultaneously.

Therefore, the present study examined whether perceived cognitive flexibility and shame were involved in the association between self-oriented perfectionism and psychological well-being among medical students at Shahid Beheshti University of Medical Sciences. Based on theory and prior evidence, it was hypothesized that self-oriented perfectionism would be negatively associated with psychological well-being, negatively associated with perceived cognitive flexibility, and positively associated with shame. It was further hypothesized that perceived cognitive flexibility would be positively associated with psychological well-being, whereas shame would be negatively associated with psychological well-being. Finally, it was expected that both perceived cognitive flexibility and shame would show significant indirect estimates in the association between self-oriented perfectionism and psychological well-being. Because the study was cross-sectional, these hypotheses refer to statistical associations and indirect effects rather than temporal or causal mechanisms.

Methods

Participants and sampling

The study population comprised medical students enrolled at Shahid Beheshti University of Medical Sciences. Data collection was carried out from November 21, 2024, through July 21, 2025. To ensure participation from students across different years of entry, recruitment was coordinated with the university administration and facilitated through student representatives from multiple cohorts. Because MedPower is designed for simple mediation models, the required sample size for the present parallel indirect-effects model was estimated conservatively based on the weaker indirect pathway. Following Kenny’s MedPower logic and joint-significance-based guidance, a sample of approximately 400–500 participants would be sufficient to detect small-to-moderate indirect effects with α = 0.05 and power = 0.80. To enhance the credibility and stability of the findings, and to allow for incomplete or invalid responses, the final target sample size was set at 700 participants. A total of 814 cases were available after data collection, and the demographic characteristics of the final analytic sample are presented in Table 1.

Table 1.

Demographic characteristics of sample (n = 801)

Characteristics	n (%), M (SD)
Age (y)	22.99 (3.37)
Missing	3
Gender
Female	486 (60.7%)
Male	312 (39.0%)
Missing	3 (0.4%)
Marital status
Single	738 (92.1%)
Married	63 (7.9%)

Procedure

The study was conducted online using the Porsline survey platform, an Iranian web-based questionnaire system. The survey link was distributed electronically to eligible students through the university’s communication channels and student representatives. Before accessing the questionnaire, participants were presented with a brief description of the study aims, the voluntary nature of participation, and assurances regarding confidentiality and anonymity. Only those who agreed to participate and provided informed consent proceeded to complete the online questionnaire. The inclusion criteria were current enrollment as a medical student at Shahid Beheshti University of Medical Sciences and willingness to participate in the study. Responses were monitored for completeness, and cases with incomplete questionnaires were excluded during the data-screening stage.

Following data collection, 814 cases were available. Three cases were excluded due to incomplete questionnaire responses (> 20% missing data across all questionnaire items), resulting in 811 cases eligible for data screening. The data were examined for coding accuracy, response quality, and potential outliers. Univariate outliers were assessed using standardized z scores, with values exceeding ± 3 considered potentially influential. This procedure identified three cases: two for PWB (cases 56 and 658) and one for SOP (case 784). In addition, Casewise diagnostics for the regression model predicting PWB identified six cases with standardized residuals greater than ± 3 in absolute value (cases 54, 56, 290, 710, 747, and 758). Visual inspection of the PWB Q-Q plot suggested four distributional outliers (cases 657, 658, 763, and 784). After accounting for overlap across the different screening procedures, a total of 10 unique outlier cases were identified and excluded from the primary analyses: cases 54, 56, 290, 657, 658, 710, 747, 758, 763, and 784. The final analytic sample therefore consisted of 801 participants. The primary analyses were conducted using the outlier-excluded sample. To evaluate whether findings were influenced by outlier removal, sensitivity analyses were conducted using the full screened sample of 811 participants, with all identified outliers retained. Robust regression and winsorization were considered as alternative approaches to limiting the influence of extreme observations. However, the sensitivity analysis retaining all 811 screened observations produced estimates that were virtually identical to those of the outlier-excluded analysis. Accordingly, no observed values were transformed or winsorized, and the full-sample analysis was used to assess the robustness of the findings to case deletion. Multicollinearity diagnostics indicated no serious concerns (VIF = 1.067–1.177).

Measures

Multidimensional perfectionism scale

The Multidimensional Perfectionism Scale was originally developed by Hewitt and Flett as a 45-item instrument designed to assess three dimensions of perfectionism: self-oriented perfectionism, other-oriented perfectionism, and socially prescribed perfectionism [10]. The Tehran Multidimensional Perfectionism Scale is a 30-item Persian measure developed by Besharat, with 10 items for each of the three dimensions [25]. Items are rated on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree), with higher scores indicating higher levels of perfectionism in the corresponding dimension. The Persian measure has demonstrated good psychometric properties, with Cronbach’s alpha coefficients of 0.90 for self-oriented perfectionism, 0.81 for other-oriented perfectionism, and 0.91 for socially prescribed perfectionism [25]. In the present study, only the self-oriented perfectionism subscale was used. This subscale consisted of 10 items, and the total score was calculated based on these items only. The 10-item self-oriented perfectionism subscale demonstrated good internal consistency in the present sample, with a Cronbach’s alpha coefficient of 0.840.

Guilt and Shame Proneness Scale - GASP

Shame was assessed using the shame subscale of the Guilt and Shame Proneness Scale (GASP), originally developed by Cohen et al. [26] to measure individuals’ proneness to experience guilt and shame following personal transgressions. In the present study, only the 8 items related to shame were used. Responses were rated on a 5-point Likert scale ranging from 1 (not at all) to 5 (very much), with higher scores indicating greater shame proneness. Previous research by Jokar and Kamali [27] examined the Persian version of this questionnaire, and Cronbach’s alpha coefficients ranged from 0.61 to 0.71. Furthermore, the overall Cronbach’s alpha value was reported to be 0.65, and content validity was supported [28]. In the present study, the 8-item shame subscale demonstrated acceptable internal consistency, with a Cronbach’s alpha coefficient of 0.751.

Cognitive Flexibility Inventory (CFI)

The Cognitive Flexibility Inventory developed by Dennis and Vander Wal includes 20 items rated on a 7-point Likert scale from strongly disagree to strongly agree. The scale assesses individuals’ perceived capacity to manage difficult situations and adapt to changing circumstances. It evaluates three facets of cognitive flexibility: [1] the tendency to view challenging situations as manageable [2], the capacity to consider alternative explanations and possibilities for human behavior and life events, and [3] the ability to generate multiple solutions for difficult situations [29]. The CFI has shown strong validity and reliability and has been standardized in Iran [30]. The content validity of the CFI was evidenced by content validity ratios greater than 0.60 for all 20 items and item-level validity greater than 0.80. Reliability analysis produced strong internal consistency with a Cronbach’s alpha of 0.88 [31]. In the present study, the scale demonstrated excellent internal consistency, with a Cronbach’s alpha coefficient of 0.891. Because the CFI is a self-report measure, findings involving this variable are interpreted as perceived cognitive flexibility rather than task-based cognitive flexibility.

Ryff’s Psychological Well-Being Scale - short form

Psychological well-being was assessed using the 18-item short form of Ryff’s Psychological Well-Being Scale [32]. The Persian 18-item version has demonstrated an acceptable factor structure and psychometric properties among Iranian university students [33]. Items are rated on a 6-point Likert scale ranging from 1 (strongly disagree) to 6 (strongly agree), with higher scores indicating greater psychological well-being. In the present sample, the original 18-item score showed a Cronbach’s alpha of 0.659. Inspection of the corrected item–total correlations and alpha-if-item-deleted statistics indicated poor performance for Item 3 (“Some people go through life aimlessly, but I am not one of them”; corrected item–total correlation =-0.32) and Item 14 (“I gave up trying to make big improvements or changes in my life a long time ago”; corrected item–total correlation =-0.02). Removal of these items increased Cronbach’s alpha to 0.729.

Because this item removal was post hoc and sample-specific, the resulting 16-item score should not be regarded as a newly validated alternative form of the scale. The 16-item mean score was used in the primary analyses because of its improved internal consistency in the present sample. To examine whether this scoring decision influenced the findings, all analyses were repeated using the original 18-item score. The direction, magnitude, and statistical interpretation of the principal estimates were materially unchanged, as reported in result section. Nevertheless, removal of the two items may limit direct comparability with studies using the original 18-item scale and published normative values.

Data analysis

Data were analyzed using IBM SPSS Statistics 27 and Hayes’ PROCESS macro (Version 5.0). Preliminary data screening included checks for coding accuracy, missing data, response patterns, and outliers. Cases with incomplete responses were excluded prior to analysis. Potential outliers were evaluated using standardized z scores, standardized residuals from Casewise diagnostics, and visual inspection of Q-Q plots. Multicollinearity was assessed using tolerance and variance inflation factor (VIF) values. Descriptive statistics and Pearson’s correlation coefficients were computed for all study variables. To test the hypothesized indirect associations, a parallel indirect-effects analysis was conducted using PROCESS Model 4. Indirect effects were estimated using 5,000 bias-corrected bootstrap resamples and 95% confidence intervals, which are recommended for inference because bootstrap confidence intervals do not assume normality of the indirect-effect sampling distribution. Heteroskedasticity-consistent (HC4) standard errors were used in all regression models. Statistical significance was inferred when the 95% bootstrap confidence interval did not include zero.

To address potential demographic confounding, a separate covariate-adjusted model was conducted with age and gender entered as covariates. Marital status was considered but was not entered into the adjusted model because the distribution was highly imbalanced, with only 7.9% of participants reporting being married in the final analytic sample. Accordingly, marital status is reported descriptively and acknowledged as a limitation rather than modeled as a stable covariate.

Results

Table 1 presents the demographic characteristics of the study sample. In line with common reporting practice, the table summarizes key participant characteristics to help readers understand the composition of the sample and assess the generalizability of the findings.

As shown in Table 1, the final sample consisted of 801 participants with a mean age of 22.99 years (SD = 3.37). Age data were missing for three participants. Regarding gender, 486 participants (60.7%) were female, 312 (39.0%) were male, and 3 (0.4%) had missing gender data. In terms of marital status, the majority of the sample was single (n = 738, 92.1%), whereas 63 participants (7.9%) were married. This pattern indicates that the sample was predominantly young, female, and single. Table 2 presents the descriptive statistics and Pearson’s correlation coefficients for the main study variables.

Table 2.

Descriptive statistics and Pearson’s correlations for study measures (n = 801)

Measure	M	SD	1	2	3	4
1. Self-Oriented Perfectionism	3.35	0.75	-	-0.24**	0.31**	-0.38**
2. Perceived Cognitive Flexibility	4.86	0.82		-	-0.01	0.63**
3. Shame	3.29	0.73			-	-0.14**
4. Psychological Well-Being	3.87	0.64				-

M mean, SD Standard deviation; * p < .05; ** p < .01

As shown in Table 2, self-oriented perfectionism had a mean of 3.35 (SD = 0.75), perceived cognitive flexibility had a mean of 4.86 (SD = 0.82), shame had a mean of 3.29 (SD = 0.73), and psychological well-being had a mean of 3.87 (SD = 0.64). Self-oriented perfectionism was negatively correlated with perceived cognitive flexibility (r = − .24, p < .01) and psychological well-being (r = − .38, p < .01), and positively correlated with shame (r = .31, p < .01). Perceived Cognitive flexibility was positively associated with psychological well-being (r = .63, p < .01), whereas its correlation with shame was not statistically significant (r = − .01, p > .05). Shame was negatively correlated with psychological well-being (r = − .14, p < .01). Overall, the pattern of correlations suggests that higher self-reported cognitive flexibility was associated with better psychological well-being, whereas higher self-oriented perfectionism and shame were associated with lower well-being.

Parallel indirect-effects analyses were conducted using PROCESS V5 Model 4 to examine whether perceived cognitive flexibility and shame were statistically involved in the association between self-oriented perfectionism and psychological well-being. To evaluate the robustness of the findings to demographic adjustment, both unadjusted and covariate-adjusted models were estimated; the adjusted model included age and gender as covariates. Regression coefficients and bootstrap indirect-effect estimate for both models are reported in Table 3. Indirect estimates were tested using 5,000 bootstrap samples and 95% confidence intervals.

Table 3.

Unadjusted and covariate-adjusted parallel indirect-effects results for the association between self-oriented perfectionism and psychological well-being

Path	Unadjusted estimate	95% CI	Adjusted estimate	95% CI
SOP → PWB total	−0.33	[− 0.384, − 0.275]	−0.32	[− 0.380, − 0.269]
SOP → PWB direct	−0.19	[− 0.244, − 0.145]	−0.19	[− 0.244, − 0.142]
SOP → CFI	−0.26	[− 0.335, − 0.185]	−0.24	[− 0.316, − 0.165]
SOP → Shame	0.30	[0.236, 0.356]	0.29	[0.225, 0.345]
Indirect via CFI	−0.12	[− 0.158, − 0.085]	−0.11	[− 0.148, − 0.074]
Indirect via Shame	−0.02	[− 0.031, − 0.003]	−0.02	[− 0.037, − 0.007]

The unadjusted model included 801 participants, and the age- and gender-adjusted model included 798 participants because of missing covariate data. Values are unstandardized coefficients. Confidence intervals for indirect estimates were based on 5,000 bias-corrected bootstrap samples. The ordering of variables represents the prespecified statistical model and does not establish temporal or causal ordering

As shown in Table 3, the pattern of findings was highly similar across the unadjusted and covariate-adjusted models. In the unadjusted model, self-oriented perfectionism was negatively associated with psychological well-being in the total-effect model, B = − 0.33, 95% CI [− 0.384, − 0.275], and this association remained significant after perceived cognitive flexibility and shame were included in the model, B = − 0.19, 95% CI [− 0.244, − 0.145]. Self-oriented perfectionism was also negatively associated with perceived cognitive flexibility, B = − 0.26, 95% CI [− 0.335, − 0.185], and positively associated with shame, B = 0.30, 95% CI [0.236, 0.356]. The total indirect estimate was significant, B = − 0.135, 95% CI [− 0.175, − 0.096], with significant specific indirect estimates through perceived cognitive flexibility, B = − 0.119, 95% CI [− 0.158, − 0.085], and shame, B = − 0.016, 95% CI [− 0.031, − 0.003].

After controlling for age and gender, the results remained substantively unchanged. Due to missing values on the covariates, the adjusted model was estimated with 798 participants. The adjusted total association between self-oriented perfectionism and psychological well-being was significant, B = − 0.324, 95% CI [− 0.380, − 0.269], and the adjusted direct association also remained significant, B = − 0.193, 95% CI [− 0.244, − 0.142]. Self-oriented perfectionism remained negatively associated with perceived cognitive flexibility, B = − 0.241, 95% CI [− 0.316, − 0.165], and positively associated with shame, B = 0.285, 95% CI [0.225, 0.345]. The adjusted total indirect estimate was significant, B = − 0.131, 95% CI [− 0.173, − 0.094], with significant specific indirect estimates through perceived cognitive flexibility, B = − 0.110, 95% CI [− 0.148, − 0.074], and shame, B = − 0.021, 95% CI [− 0.037, − 0.007]. Overall, these findings indicate that the negative association between self-oriented perfectionism and psychological well-being included statistically significant indirect components through lower perceived cognitive flexibility and higher shame, and that the results were robust after adjustment for age and gender.

Among the covariates, gender was significantly associated with perceived cognitive flexibility, B = − 0.134, 95% CI [− 0.248, − 0.020], shame, B = 0.222, 95% CI [0.127, 0.317], and psychological well-being, B = 0.115, 95% CI [0.049, 0.181]. Given that gender was coded as 0 = men and 1 = women, these results indicate that, compared with men, women reported lower perceived cognitive flexibility, higher shame, and higher psychological well-being. Age was significantly and negatively associated with psychological well-being, B = − 0.020, 95% CI [− 0.034, − 0.006], but was not significantly associated with perceived cognitive flexibility or shame. The completely standardized total indirect estimate changed only minimally from the unadjusted model, further supporting the stability of the findings.

Figure 1 presents the covariate-adjusted parallel indirect-effects model. Standardized path coefficients are reported for all direct paths. In the adjusted model, self-oriented perfectionism was negatively associated with perceived cognitive flexibility and positively associated with shame. In turn, perceived cognitive flexibility was positively associated with psychological well-being, whereas shame was negatively associated with psychological well-being. The direct association between self-oriented perfectionism and psychological well-being remained significant after accounting for both indirect variables. The adjusted model explained 6.9% of the variance in perceived cognitive flexibility, 12.4% of the variance in shame, and 47.4% of the variance in psychological well-being.

Fig. 1.

Standardized parameter estimates of the covariate-adjusted mediation model linking self-oriented perfectionism and psychological well-being. All paths represent cross-sectional statistical associations and should not be interpreted as temporal or causal effects. The model was adjusted for age and gender. *p < .05; **p < .01; ***p < .001

Sensitivity analyses were conducted to evaluate the robustness of the findings. First, the full screened sample with identified outliers retained (N = 811) was analyzed. The results closely replicated the primary findings obtained after outlier removal (N = 801), with the total indirect effect remaining significant in both models (primary: b = − 0.135, 95% CI [− 0.175, − 0.096]; sensitivity: b = − 0.135, 95% CI [− 0.177, − 0.097]). The specific indirect effects through perceived cognitive flexibility (primary: b = − 0.119; sensitivity: b = − 0.120) and shame (primary: b = − 0.016; sensitivity: b = − 0.015) also remained significant, indicating that the results were not driven by outlier exclusion. Second, sensitivity analysis using the original 18-item Ryff Psychological Well-Being Scale showed a highly similar pattern to the 16-item version within the same sample (N = 801). The total, direct, and indirect effects remained statistically significant, with only minimal changes in effect sizes, supporting the robustness of the findings across alternative scoring approaches.

Overall, the results consistently showed that self-oriented perfectionism was indirectly associated with lower psychological well-being through lower perceived cognitive flexibility and higher shame. These findings were stable across covariate-adjusted models, outlier-retained analyses, and alternative scoring of the Ryff Psychological Well-Being Scale.

Discussion

The present cross-sectional study identified a pattern of associations among self-oriented perfectionism, perceived cognitive flexibility, shame, and psychological well-being in Iranian medical students. Self-oriented perfectionism was associated with lower psychological well-being, and the statistical model included significant indirect components involving lower perceived cognitive flexibility and higher shame. The specific indirect estimate involving perceived cognitive flexibility was larger than that involving shame, and the general pattern remained similar after adjustment for age and gender.

These findings should not be interpreted as demonstrating that self-oriented perfectionism temporally precedes changes in perceived cognitive flexibility or shame, or that these variables subsequently affect psychological well-being. The ordering of variables was based on the prespecified theoretical model rather than observed temporal evidence. Alternative explanations are equally plausible, including the possibility that lower psychological well-being is accompanied by greater shame and lower perceived cognitive flexibility, or that the associations are reciprocal. The present results should therefore be viewed as cross-sectional and hypothesis-generating.

The negative association between self-oriented perfectionism and psychological well-being is theoretically consistent with Hewitt and Flett’s multidimensional model, in which self-oriented perfectionism reflects not only high personal standards but also stringent self-monitoring and harsh self-evaluation [10]. In this framework, individuals are not only driven to achieve but are also highly sensitive to deviations from their standards, which may promote persistent self-criticism. Complementarily, discrepancy-based self-regulation models provide a process-level interpretation: individuals compare actual performance with internal standards, and perceived discrepancies may be accompanied by negative affect and cognitive distress [34, 35]. In self-oriented perfectionism, these discrepancies may be more frequent and more threatening because standards are rigid and self-worth is strongly contingent on meeting them [36]. Within the evaluative and high-stakes context of medical training, these patterns may be particularly salient. This interpretation is consistent with prior findings linking perfectionism to distress, anxiety, and burnout vulnerability in medical students [37, 38], and suggests that perfectionistic striving, when rigidly maintained, can be associated with lower well-being [39].

A central contribution of this study is the finding that self-reported cognitive flexibility represented the larger indirect component. Prior literature conceptualizes cognitive flexibility as relevant to attentional shifting, emotion regulation, reappraisal, tolerance of ambiguity, and adaptation to corrective feedback [40–42]. In contrast, perfectionism has been consistently associated with cognitive rigidity, including difficulty disengaging from error-focused processing [43]. From an information-processing perspective, individuals high in perfectionism may over-encode mistakes and underutilize corrective feedback, thereby limiting adaptive learning [44]. Students high in self-oriented perfectionism may struggle with these processes when their standards are rigidly held, because setbacks may be less likely to be reframed as manageable and more likely to be interpreted as failures of the self. Prior work also suggests that cognitive reappraisal may moderate the association between perfectionism and cognitive flexibility [45]. The larger indirect estimate through perceived cognitive flexibility is consistent with evidence linking flexibility to better mental health and adaptive coping [46, 47]. However, because the CFI is a self-report measure, this finding reflects students’ perceived rather than objectively assessed cognitive flexibility and should not be interpreted as evidence of a causal cognitive mechanism. The relatively strong correlation between perceived cognitive flexibility and psychological well-being (r = .63) also warrants caution, as conceptual overlap, shared self-report methodology, global self-evaluation, and current affective state may have inflated the association. Thus, the larger indirect effect involving the CFI indicates a stronger cross-sectional statistical relationship, not a stronger objectively measured cognitive mechanism.

Shame provided a statistically distinguishable but substantially smaller indirect component. The unadjusted shame-specific indirect estimate was B = − 0.016, 95% CI [− 0.031, − 0.003], corresponding to a very small standardized effect of approximately β = −0.03. Thus, although the confidence interval excluded zero and the effect was estimated with reasonable precision, its absolute magnitude was very small. Given the large sample, statistical significance should not be equated with practical or clinical importance. The present findings therefore support interpreting shame as a comparatively minor correlate within the observed pattern rather than as a dominant explanatory factor. The association remains theoretically relevant because shame is conceptualized as arising when individuals evaluate the self as globally flawed in response to perceived failure [48, 49]. In medical education, where performance is continuously evaluated, feedback is often hierarchical, and errors are highly salient, this internal attributional style may increase the likelihood of shame rather than more behavior-focused self-evaluative emotions such as guilt. Shame involves a global negative evaluation of the self and is associated with maladaptive regulatory responses, including withdrawal, isolation, reduced help-seeking, and heightened self-criticism [50]. These patterns are particularly relevant in medical training because they may reduce engagement with feedback and limit corrective learning experiences, while also coinciding with lower self-acceptance, personal growth, and environmental mastery [51]. The weak correlation between shame and perceived cognitive flexibility in the present findings suggests that affective self-evaluation and perceived cognitive flexibility may represent distinguishable components of the broader association between perfectionism and well-being. Nevertheless, the present cross-sectional findings do not establish that perfectionistic standards produce shame or that shame subsequently reduces psychological well-being. Reverse and reciprocal associations remain possible.

The indirect association through shame was statistically significant but small in magnitude. Given the large sample size, statistical significance should not be interpreted as evidence of strong practical or clinical importance. Compared with shame, perceived cognitive flexibility appeared to account for a larger portion of the indirect association between self-oriented perfectionism and psychological well-being. Therefore, shame should be interpreted as a distinct but comparatively minor component of the observed pattern rather than as a dominant explanatory factor.

The explained-variance estimates also require cautious interpretation. In the adjusted models, self-oriented perfectionism and the covariates accounted for 6.9% of the variance in perceived cognitive flexibility and 12.4% of the variance in shame. These relatively modest values indicate that most individual differences in perceived cognitive flexibility and shame were associated with factors not represented in the model. Academic stress, feedback experiences, personality, social support, cultural expectations, self-criticism, fear of failure, and other dimensions of perfectionism may all be relevant.

The full outcome model accounted for 47.4% of the variance in psychological well-being. This value reflects the concurrent contribution of all variables in the model and should not be attributed to self-oriented perfectionism alone. It may also partly reflect the strong association and potential measurement overlap between self-reported perceived cognitive flexibility and psychological well-being. Accordingly, the explained variance does not establish causal model fit or confirm the proposed temporal sequence.

The persistence of a significant direct association after including both variables indicates that additional factors are likely involved. From a cognitive-behavioral perspective, rumination, fear of failure, and self-criticism may further account for the association between perfectionism and distress [52, 53]. These processes likely interact with both rigidity and shame, reinforcing maladaptive cycles of evaluation and coping. Future models should examine these factors together with other dimensions of perfectionism, especially socially prescribed perfectionism and other-oriented perfectionism, to clarify whether internal, external, and interpersonal perfectionistic standards show distinct patterns of association with well-being.

This study adds to current knowledge by integrating strands of evidence that have largely developed separately. First, it extends medical-student perfectionism research beyond symptom-focused outcomes by examining eudaimonic psychological well-being, a construct that captures positive functioning, self-acceptance, purpose, autonomy, and personal growth rather than only the presence or absence of distress. Second, it provides a focused analysis of self-oriented perfectionism, which is particularly relevant in medical education because it reflects internally imposed standards and self-evaluation within a culture of high achievement and frequent assessment. Third, it brings together two theoretically distinct processes—shame as a self-conscious affective process and self-reported cognitive flexibility as a cognitive-regulatory process—within a single parallel model. This is important because the findings suggest that perfectionistic self-demands are not associated with well-being through a single psychological route; rather, both diminished perceived cognitive flexibility and greater shame showed statistically significant indirect associations, with the cognitive-flexibility pathway being larger. Fourth, the covariate-adjusted analysis showed that the pattern remained after adjustment for age and gender, supporting the robustness of the observed associations. These contributions are intentionally framed as cross-sectional and self-report based; they do not establish causality, but they identify a theoretically coherent pattern that can be tested more rigorously in longitudinal designs, experimental/task-based assessments of cognitive flexibility, and intervention studies targeting shame-sensitive feedback and flexible self-evaluation.

Limitations and future directions

Several limitations should be considered. First, the cross-sectional design limits causal inference. Although the indirect-effects model is theoretically grounded, temporal ordering cannot be confirmed and reverse or bidirectional associations are possible [54]. For example, lower psychological well-being may contribute to greater shame and lower perceived cognitive flexibility, or reciprocal associations may exist among these constructs. Therefore, the findings should not be interpreted as demonstrating that perfectionism causes changes in perceived cognitive flexibility, shame, or psychological well-being.

Second, all principal constructs were measured using self-report instruments, which may have introduced response biases, global self-evaluation effects, and common method variance [55]. This issue is particularly important for the association between perceived cognitive flexibility and psychological well-being, which was relatively strong (r = .63). The CFI assesses individuals’ perceptions of their flexibility and should not be treated as equivalent to performance on task-switching, set-shifting, or other behavioral measures. The observed association may partly reflect conceptual overlap between positive appraisals of adaptive functioning. Future studies should combine self-report measures with behavioral or neuropsychological indicators and, where possible, multi-informant or temporally separated assessments.

Third, participants were recruited from a single medical school in Tehran, and the sample was relatively homogeneous with respect to age, marital status, and educational context. Most participants were female and single. Institutional characteristics and cultural norms concerning achievement, self-criticism, shame, interpersonal evaluation, and maintenance of social standing may influence the observed associations. The findings therefore should not be assumed to generalize directly to students from other medical schools, other academic disciplines, older or married students, or students in different cultural and educational settings.

Fourth, two items were removed post hoc from the 18-item Psychological Well-Being Scale because of poor item-level performance in the present sample. Although this modification improved internal consistency and analyses using the original 18-item score produced materially similar findings, the 16-item score is sample-specific and has not been independently validated as a separate form. Its use may reduce comparability with studies using the original scale and with published normative values.

Fifth, the primary analyses were conducted after outlier removal. Although the number and criteria for outlier exclusion were reported and sensitivity analyses with the full screened sample showed a similar pattern, outlier decisions can influence estimates and should be considered when interpreting the results.

Sixth, only shame was assessed; guilt, anger directed toward the self, self-criticism, and other self-conscious or self-directed emotional processes were not measured. Seventh, only self-oriented perfectionism was included. Although this focus was theoretically motivated, future research should include socially prescribed and other-oriented perfectionism to determine whether external pressure and interpersonal perfectionistic standards have different associations with shame, perceived cognitive flexibility, and well-being. Finally, marital status was not included in the covariate-adjusted model because the sample was highly imbalanced; future studies with more diverse marital-status distributions should evaluate this variable more fully.

Future research should employ longitudinal and cross-lagged designs to clarify directional and reciprocal associations between perfectionism, perceived cognitive flexibility, shame, and well-being. Multi-center studies are also needed to test generalizability across medical schools and cultural contexts. Expanding the model to include related constructs such as academic stress, fear of failure, self-criticism, self-compassion, guilt, mindfulness, anger directed toward the self, emotion regulation, social support, attachment style, cultural expectations surrounding achievement, and burnout would provide a more comprehensive account of perfectionism-related processes. Finally, intervention studies targeting perceived cognitive flexibility and shame-related experiences, alongside educational changes in feedback culture, could evaluate whether changes in these processes are followed by improvements in medical students’ psychological well-being.

Conclusion

In this cross-sectional sample of Iranian medical students from a single institution, self-oriented perfectionism was associated with lower psychological well-being, and the prespecified statistical model included indirect components involving lower perceived cognitive flexibility and higher shame. The indirect association involving perceived cognitive flexibility was larger, whereas the shame-specific estimate was statistically significant but very small and likely of limited practical importance. Because all variables were assessed concurrently and primarily through self-report, the findings do not establish temporal ordering, causal mediation, or psychological mechanisms. Reverse and reciprocal associations remain plausible, and longitudinal, multi-method, and multi-center research is required to evaluate the proposed relationships more rigorously.

Authors’ contributions

-**AZ** contributed to literature review, data interpretation, and manuscript writing.- **ASK** supervised the study, contributed to methodological design, interpretation of findings, and critical revision of the manuscript.- **AE** assisted in data collection, data entry, and manuscript preparation.- **MAR** contributed to study conceptualization, data collection, and drafting of the manuscript.- **TS** contributed to questionnaire administration, data handling, and initial drafting.- **SZ** assisted in participant recruitment, data organization, and proofreading.- **ESF** contributed to statistical analysis support and manuscript revision. All authors read and approved the final manuscript.

Funding

This study was financially supported by the Student Research Committee and the Research & Technology Chancellor of Shahid Beheshti University of Medical Sciences, Tehran, Iran, under project number 1403/43012765. The funding bodies had no role in the study design, data collection, analysis, interpretation of data, or writing of the manuscript.

Data availability

The datasets generated and/or analyzed during the current study are **available from the corresponding author upon reasonable request**. Corresponding author: *Dr. Amir Sam Kianimoghadam (as.kianimoghadam@gmail.com)*.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Shahid Beheshti University of Medical Sciences under the ethical code IR.SBMU.RETECH.REC.1403.577. All procedures involving human participants were performed in accordance with the Declaration of Helsinki and relevant institutional and national ethical guidelines and regulations. All participants were informed about the study objectives, confidentiality of their responses, and their right to withdraw at any time without any consequences. Written informed consent was obtained from all participants prior to data collection.

Consent for publication

Not applicable. The manuscript does not contain any individual person’s data in any form, including identifiable images, videos, or personal details.

Competing interests

The authors declare no competing interests.