Distinguishing perfectionism and excellencism in graduate students: Contrasting links with performance satisfaction, research self‐efficacy, burnout, and dropout intentions

Abstract

Research on perfectionism in graduate school found inconsistent associations between perfectionistic standards and psychological outcomes. Such unanticipated results led to the Model of Excellencism and Perfectionism (MEP), which differentiates between people pursuing excellence and those pursuing perfection. Recent studies with undergraduate students have shown that excellencism and perfectionism are distinct constructs, differentially associated with achievement and psychological outcomes. In this study, we aimed to offer the first empirical test of the MEP in graduate school with a sample of 376 graduate students (i.e. 81% masters, 19% doctoral). Results of confirmatory factor analyses on the Scale of Perfectionism and Excellencism provided evidence for the conceptual separation of excellencism and perfectionism. Results of multiple regression showed that perfection strivers (compared to excellence strivers) reported higher research self‐efficacy and satisfaction with their research productivity. When accounting for satisfaction with research productivity, perfection strivers experienced more academic burnout and dropout intentions. Perfectionistic students also used more perfectionistic self‐presentation strategies when interacting with their supervisors. Perfectionism was associated with both beneficial and harmful outcomes, which suggests that perfectionism in graduate school is paradoxical and operates like a double‐edged sword. These findings are interpreted in light of the need to help graduate students strike a balance between their academic achievements and psychological adjustment.

Graduate education is a demanding period requiring students to develop specialized skills, adapt to autonomous roles, and frequently manage detailed feedback from peers and professors—all while facing increasing academic demands that can strain their perceived competence and well‐being (e.g. Carless et al., ²⁰²⁴; Schmidt & Hansson, ²⁰¹⁸; Stubb et al., ²⁰¹¹; van Rooij et al., ²⁰²¹). In that context, it remains unclear if pursuing the extremely high personal standards involved in perfectionism offers some advantages for graduate students. Perfectionism in graduate studies is a topic that remains understudied and subject to many contradictory findings. Past studies looking at perfectionism among graduate students did not differentiate perfectionism from the high personal standards involved in a new construct called excellencism (Gaudreau, 2019). In this study, we relied on an innovative approach and performed the first‐ever empirical test of the Model of Excellencism and Perfectionism (MEP; Gaudreau et al., ²⁰²²) in the context of graduate studies. Our goals were to evaluate if excellencism and perfectionism can be distinguished at the conceptual level and to evaluate, at the functional level, the extent to which they are differentially associated with the research productivity, feelings of burnout, and dropout intentions of graduate students.

KNOWNS AND UNKNOWNS ABOUT PERFECTIONISM IN GRADUATE STUDENTS

Perfectionism is a multifaceted construct operationalized and studied in various ways (e.g. Flett & Hewitt, ²⁰²²; Robinson et al., ²⁰²²; Smith et al., ²⁰²²). A central point that unites many conceptual frameworks is the notion that perfectionism involves aiming, pursuing, and evaluating oneself based on extremely high standards (i.e. perfectionistic standards) which are accompanied by recurring concerns, doubts, worries, and appraisals (i.e. perfectionistic concerns, perfectionistic cognitions) and socio‐behavioural expressions (i.e. perfectionistic self‐presentation, other‐oriented perfectionism) that form the unique phenomenological experience of being a person who pursues perfection (e.g. Gaudreau, ²⁰²¹; Hewitt et al., ²⁰¹⁷). This conceptual landscape is broad, and empirical findings vary depending on the specific area researchers have directed their attention to.

Many elements of perfectionism have undisputedly been related to psychological maladjustment (e.g. Limburg et al., ²⁰¹⁷; Smith et al., ²⁰²¹). Studies with graduate students associated perfectionistic concerns and cognitions as well as perfectionistic self‐presentation with negative affectivity and symptoms of depression, anxiety, stress, loneliness, and reduced life satisfaction (e.g. Berry et al., ²⁰²¹; Comerchero & Fortugno, ²⁰¹³; Cowie et al., ²⁰¹⁸; Manova & Khoury, ²⁰²³; Moate et al., ²⁰¹⁹; Onwuegbuzie & Daley, ¹⁹⁹⁹; Rice et al., ^{2012, 2019}; Saddler & Sacks, ¹⁹⁹³; Tigranyan et al., ²⁰²¹). Based on these findings, few doubts remain about the debilitating nature of several aspects of perfectionism.

Results regarding perfectionistic standards are less consistent because this dimension of perfectionism has been conceptualized in a more varied manner. Researchers have frequently used measures in which perfectionistic standards are operationalized using either high standards or an atheoretical blend of high and perfectionistic standards. As a result, the associations of perfectionistic standards with negative affectivity and symptoms of depression, anxiety, stress, loneliness, and reduced life satisfaction have been small (Comerchero & Fortugno, 2013; Cowie et al., ²⁰¹⁸; Manova & Khoury, ²⁰²³), null/negligible (Berry et al., 2021; Foo et al., ²⁰¹⁷; Moate et al., ²⁰¹⁹; Onwuegbuzie & Daley, ¹⁹⁹⁹), and potentially underestimated. Conclusions from the extant literature on perfectionistic standards in graduate school should be approached with caution.

Perfectionism versus Excellencism

Perfectionistic standards remain at the centre of disagreements about the potentially adaptive nature of perfectionism. Many researchers have proposed to retire the label ‘adaptive perfectionism’ from theories and measures of perfectionism (e.g. Gaudreau & Thompson, ²⁰¹⁰; Gotwals & Lizmore, ²⁰²³; Hewitt et al., ²⁰¹⁷; Hill & Madigan, ²⁰¹⁷). First, perfectionistic standards are moderate to strongly correlated with perfectionistic concerns and the many cognitive expressions of perfectionism. Given that, several individuals with perfectionistic standards will experience many expressions of perfectionism typically associated with increased risks of psychological maladjustment. Perfectionistic standards are part of an intricate cognitive‐social‐behavioural network that can compromise the psychological wellness of many individuals (Gaudreau, 2021). Second, the label ‘perfectionistic standards’ should exclude aiming and striving toward high standards (e.g. Blasberg et al., ²⁰¹⁶; Osenk et al., ²⁰²⁰). To that extent, the newly developed MEP has redefined the pursuit of high standards as excellencism to definitively differentiate it from perfectionism (Gaudreau, 2019). More precisely, excellencism is defined as a ‘tendency to aim and strive toward very high yet attainable standards in an effortful, engaged, and determined yet flexible manner’ (Gaudreau, 2019, p. 200). It should be differentiated from perfectionism, which represents ‘a tendency to aim and strive toward idealized, flawless, and excessively high standards in a relentless manner’ (Gaudreau, 2019, p. 200).

At the conceptual level, the MEP treats perfectionism and excellencism as distinct but correlated constructs. They are correlated because a person who pursues perfection not only wants to perform very well but also wants to perform perfectly. In other words, perfectionism is a special case in which the standards exceed those involved in excellencism (see Gaudreau, ²⁰¹⁹, pp. 200–201, for a complete explanation). Moreover, excellencism differs from perfectionism because it is not significantly associated with the many social, cognitive, and behavioural expressions of perfectionism (e.g. perfectionistic concerns) – thus conveying crucial support for discriminant validity (e.g. Bien et al., ²⁰²⁴; Gaudreau, ²⁰²¹; Tape et al., ²⁰²⁴).

At the functional level, excellencism and perfectionism form an integrated system in which their unique effects enable us to compare the outcomes of prototypical cases of non‐perfection/non‐excellence, excellence, and perfection strivers (Gaudreau, 2019).¹ The MEP presents nine scenarios to help interpret the positive, null, and negative effects of excellencism and perfectionism from variable‐centred statistical analyses such as multiple regression (see Figure 2 in Gaudreau & Schellenberg, ²⁰²⁴). Different patterns of excellencism and perfectionism effects hold different theoretical and practical implications. For example, the unique positive effect of excellencism on happiness (controlling for perfectionism) indicates that excellence strivers are happier than non‐excellence/non‐perfection strivers. The unique negative effect of perfectionism on happiness (controlling for excellencism) indicates that perfection strivers are less happy than excellence strivers. Scenarios in which perfection strivers report lower adjustment and higher maladjustment than excellence strivers are typically interpreted as supporting the hypothesis that perfectionism is harmful and unhealthy.

FIGURE 2.

Academic burnout at low, average, and high levels of research productivity satisfaction (RPS). Contrasts are constant/identical across the three levels of research productivity satisfaction, but scores of burnout diminish across the three panels of this figure. Excel, Excellence strivers; Non, non‐excellence/non‐perfection strivers; Perfect, Perfection strivers. Values reported here are taken from Mplus. **p < .01. *p < .05. ^† p = .076.

Studies with graduate students have yet to rely on the MEP. More specifically, researchers have frequently used measures in which perfectionistic standards are operationalized using either high standards or an atheoretical blend of high and perfectionistic standards. As a result, the associations of perfectionistic standards with negative affectivity and symptoms of depression, anxiety, stress, loneliness, and reduced life satisfaction have been small (Comerchero & Fortugno, 2013; Cowie et al., ²⁰¹⁸; Manova & Khoury, ²⁰²³), null/negligible (Berry et al., 2021; Foo et al., ²⁰¹⁷; Moate et al., ²⁰¹⁹; Onwuegbuzie & Daley, ¹⁹⁹⁹), and potentially underestimated. In light of recent conceptual advancements provided by the MEP, conclusions from the extant literature should be approached with caution. New research is needed to distinguish the pursuit of perfectionistic standards more clearly from the pursuit of high standards.

MEP studies have shown that perfectionism and excellencism are distinctively associated with various outcomes. Academic performance (Gaudreau et al., 2022), creative achievement (Goulet‐Pelletier et al., 2022), savouring of positive experiences (Gaudreau et al., 2024), self‐esteem, and openness to experience (Bien et al., 2024) have been positively and negatively associated with excellencism and perfectionism, respectively. The opposite pattern has been observed for symptoms of depression (e.g. Tape et al., ²⁰²⁴; With et al., ²⁰²⁴), psychological distress (With et al., 2024), dropout intentions (Gaudreau et al., 2024), frustration of the need for competence (e.g. Shin & Kim, ²⁰²⁴), feeling like an imposter (Gaudreau et al., 2022) as well as attitudes towards cheating, moral disengagement, and hyper‐competitiveness (Gaudreau & Schellenberg, 2022). Other outcomes such as life satisfaction (e.g. Gaudreau et al., ²⁰²²; Shin & Kim, ²⁰²⁴), psychological well‐being (With et al., 2024), sense of competence (Gaudreau et al., 2022; Shin & Kim, ²⁰²⁴), self‐efficacy (Goulet‐Pelletier et al., 2022), and perceived goal attainment (Gaudreau et al., 2021) were only positively linked to excellencism without being significantly associated with perfectionism. Finally, perfectionism (but not excellencism) was negatively associated with body acceptance (Gaudreau & Schellenberg, 2024) and self‐compassion (e.g. With et al., ²⁰²⁴) while being positively associated with anxiety and stress (Tape et al., 2024), neuroticism (e.g. Bien et al., ²⁰²⁴), fear of failure (e.g. Gaudreau et al., ²⁰²¹), and performance anxiety and negative affect in music (Racine et al., 2025). In summary, several of these findings indicate that excellence strivers experience more positive outcomes than non‐excellence/non‐perfection strivers, whereas perfection strivers experience more negative outcomes than excellence strivers.

In keeping with the tradition of describing perfectionism as a double‐edged sword (Stoeber, 2014) or a case of a successful failure (Madigan, 2019; Missildine, ¹⁹⁶³), the MEP proposes alternative hypotheses to examine whether perfectionism is beneficial, harmful, or unneeded (Gaudreau, 2019; Gaudreau et al., ²⁰²⁴). Arguments suggest that perfectionism may be associated with both positive and negative outcomes among graduate students. On the one hand, environments that emphasize high performance, like graduate school, may foster a person‐environment fit for perfection strivers, who are driven by an obsessive pursuit of achievement (Gaudreau et al., 2022). In these settings, the relentless hard work and over‐striving of perfection strivers are often implicitly valued as essential for success (Bekkouche et al., 2022), with many students feeling pressured to keep up with their peers who appear to be working even harder (Götz, 2019). Graduate students who strive for perfection could turn themselves into high achievers (Morpeth‐Provost et al., 2022; Witcher et al., ²⁰⁰⁷) with a strong sense of self‐efficacy for academic research (Sun et al., 2009). Such tendencies could blend with a ‘publish or perish’ culture, where productivity is attributed to dedication and sacrifice and where ‘good is never good enough’ (Berry et al., 2020). When operating in such contexts, perfection strivers could perform well compared with others.

On the other hand, hypercompetitive cultures are known risk factors for counterproductive behaviours and burnout (Greenbaum et al., 2023). Past studies have found inconsistent results for the association between perfectionistic standards and burnout (e.g. Hill et al., ²⁰¹⁸; Hill & Curran, ²⁰¹⁵), but it can turn positive and significant when excellencism is separated from perfectionism (Gaudreau, 2021). Pressuring and demanding work environments are also the ones from which individuals actively try to disengage (Tan et al., 2024). Results of a recent study with undergraduate students showed that dropout intentions were positively associated with perfectionism but negatively linked to excellencism (Gaudreau et al., 2024). Perfection strivers evaluate themselves using extremely high, unrealistic, and inflexible standards. As a result, their self‐worth is contingent on the highest possible level of achievement (e.g. Egan et al., ²⁰¹⁶; Sturman et al., ²⁰⁰⁹). Altogether, it is reasonable to expect that perfectionism may place a psychological burden on graduate students as their psychological adjustment may suffer from the constant need and pressure to attain and sustain the highest level of academic achievement.

This study

The first goal of this study was to evaluate the factor structure of the Scale of Perfectionism and Excellencism (SCOPE; Gaudreau et al., ²⁰²²) to determine if perfectionism and excellencism can be differentiated among graduate students. Empirical support for the conceptual separability of excellencism and perfectionism has been found in samples of undergraduate students (Gaudreau et al., 2022) and adolescents (Bien et al., 2024; Tape et al., ²⁰²⁴), but validity can vary across contexts and populations (Haynes et al., 1995). We hypothesized that a two‐factor model would provide an acceptable and superior fit compared to an undifferentiated excellencism and perfectionism model.

Our second goal was to evaluate how excellencism and perfectionism are associated with research productivity, feelings of burnout, and dropout intentions. The MEP offers a flexible approach to studying whether the perfectionism of graduate students is beneficial, harmful, or unneeded. We formulated two distinct sets of hypotheses. On the one hand, graduate school can inadvertently reward and reinforce obsession, over‐engagement, and tireless goal‐striving. In that context, perfectionism may confer some achievement advantages to graduate students (e.g. higher perceived productivity, research self‐efficacy) over and above excellencism. On the other hand, it may limit how graduate students define and evaluate their self‐worth (e.g. Egan et al., ²⁰¹⁶; Sturman et al., ²⁰⁰⁹). Real, anticipated, and imagined failures are debilitative for perfection strivers (Flett & Hewitt, 2023). Once achievement is factored in, perfectionism (but not excellencism) could positively relate to academic burnout and dropout intentions. If so, perfection strivers (compared with excellence strivers and non‐excellence/non‐perfection strivers) are potentially more at risk of experiencing feelings of academic burnout and dropout intentions. The intricate role of achievement in perfectionism has been observed in other life domains (e.g. sports, music) that value and reinforce achievement (e.g. Crocker et al., ²⁰¹⁴; Curran & Hill, ²⁰¹⁸; Powers et al., ²⁰⁰⁹; Waleriańczyk et al., ²⁰²²) but have yet to be studied with graduate students.

METHOD

Participants

We recruited a total sample of 376 graduate students on Prolific Academic. The participants were aged between 20 and 50 years old (M = 25.76, SD = 5.28), and they identified as male (56.1%), female (42.8%), or other (1.1%). The use of Prolific Academic allowed us to extend the recruitment to 25 countries with a majority of participants from Portugal (27.4%), United Kingdom (13.8%), Poland (12.2%), Italy (8.5%), United States (8%), Greece (6.6%), Spain (6.1%), Canada (3.5%), Mexico (3.2%), and other countries (10.7%). Collectively, the participants identified as White (78.5%), Latino (7.4%), Black (2.7%), East Asian (2.1%), South Asian (2.4%), Middle Easterner (2.9%), African (1.1%), multicultural (1.6%), or other (1.3%). Most were enrolled in a master's (80.9%) compared to a doctoral (17.9%) or a combined master–doctoral program (1.3%), and the majority were in the first (33.7%) or second (31.8%) year of their program with a minority being beyond the fifth year (3.7%). Each participant was given a small monetary compensation of 3.85 CAD for completing the study. This research was authorized and complied with the University of Ottawa Research and Ethics Board (file H‐06‐20‐5910).

Design and measures

We relied on a cross‐sectional design in which the graduate students completed an online 12–15‐min survey at the end of their academic year (i.e. August 2021). We selected the end of an academic year because it was deemed a relevant time point to have students reflect on their academic experiences in graduate school. The current study was part of a larger survey that included measures not analysed in this report (e.g. laptop use during academic writing, feeling of imposture). Another manuscript looking at imposture feelings is currently under preparation (Benoît et al., 2025).

Excellencism and perfectionism

Excellencism (e.g. ‘… to perform very well’) and perfectionism (e.g. ‘… to perform perfectly’) were measured using the SCOPE (Gaudreau et al., 2022). Participants reported the extent to which each of the 22 statements was representative of the goals they generally pursue in life, using a 7‐point scale from 1 (Not at all agree) to 7 (Totally agree). In this study, the internal consistency was high for both excellencism (ω = .916) and perfectionism (ω = .960).

Perfectionistic self‐presentation

Perfectionistic self‐presentation was measured using the Perfectionistic Self‐Presentation Scale (PSPS; Hewitt et al., ²⁰⁰³). We modified the instructions and some items of the 27 items so that participants would refer to their research supervisor (e.g. ‘I always try to present myself as a picture of perfection to my supervisor’). The participants reported their degree of agreement using a 7‐point scale from 1 (Not at all agree) to 7 (Totally agree). In this study, the internal consistency was good for each of the three subscales: perfectionistic self‐promotion (ω = .943), non‐display of imperfections (ω = .885), and non‐disclosure of imperfections (ω = .871).

Satisfaction with research productivity

Satisfaction with research productivity was assessed using a scale from 1 (Not at all satisfactory) to 7 (Totally satisfactory). Participants answered the following question: ‘Considering the year into your program (and comparing yourself to other students in your field), how would you rate your level of scientific productivity?’

Research self‐efficacy

We used the research self‐efficacy scale (Jöstl et al., 2012) to measure the extent to which graduate students felt capable of completing research‐oriented tasks since the start of their graduate program. Each of the seven items (e.g. ‘So far in my graduate program, I feel like I am capable of publishing regularly in scientific journals’) was rated using a 7‐point scale from 1 (Not at all agree) to 7 (Totally agree). In this study, the internal consistency was good (ω = .881).

Academic burnout

Academic burnout was assessed using the School Burnout Inventory (SBI; Salmela‐Aro et al., ²⁰⁰⁹). Participants referred to their life in graduate school while answering each of the nine items (e.g. ‘I feel overwhelmed by the workload in my graduate program’) using a 7‐point scale from 1 (Not at all agree) to 7 (Totally agree). In this study, the internal consistency coefficient was good for the total score of the SBI (ω = 0.874).

Dropout intention

Dropout intention was measured with three items (e.g. ‘I considered dropping out of graduate school’) similar to those used in previous research (Schellenberg & Gaudreau, 2020). Participants had to answer each item on a scale from 1 (Never) to 6 (Always). In our study, the internal consistency coefficient (ω = .900) was reported to be high.

Finally, we complemented our analysis with four additional items rated on the same frequency scale, looking at one's intention to change program, supervisor, topic, and university. As shown in Table 1, these four items were only moderately correlated, thus supporting our decision to treat each as a single‐item score.

TABLE 1.

Descriptive statistics and standardized factor loadings of the SCOPE.

Items	M	SD	Sk	Ku	WLMSV		BAYES
					E	P	E	P
E1	5.306	1.094	−0.344	−0.161	.775	0	.784	−.017
E2	5.826	1.005	−0.894	1.557	.731	0	.825	−.138
E3	5.368	1.282	−0.518	−0.349	.819	0	.767	.049
E4	5.329	1.243	−0.578	0.234	.810	0	.761	.038
E5	5.719	1.068	−0.863	1.017	.806	0	.785	.020
E6	5.941	1.025	−1.120	1.745	.855	0	.861	−.021
E7	5.421	1.140	−0.636	0.270	.715	0	.721	−.026
E8	5.497	1.148	−0.472	−0.382	.730	0	.743	−.039
E9	5.265	1.320	−0.536	−0.328	.848	0	.766	.084
E10	5.872	1.061	−0.869	0.763	.862	0	.857	.017
E11	5.180	1.311	−0.476	−0.199	.802	0	.765	.028
P1	4.429	1.476	−0.490	−0.032	0	.864	.086	.805
P2	3.787	1.684	−0.043	−0.783	0	.849	−.024	.865
P3	4.225	1.604	−0.373	−0.517	0	.902	.054	.866
P4	3.941	1.688	−0.135	−0.817	0	.823	.065	.782
P5	3.541	1.763	0.078	−0.977	0	.877	−.053	.914
P6	4.632	1.595	−0.602	−0.182	0	.730	.080	.681
P7	3.886	1.748	−0.154	−0.890	0	.939	−.025	.949
P8	4.409	1.669	−0.417	−0.647	0	.885	.045	.864
P9	3.693	1.790	−0.011	−0.970	0	.946	−.058	.969
P10	4.022	1.722	−0.147	−0.840	0	.927	−.002	.926
P11	3.761	1.729	−0.108	−0.961	0	.865	.031	.852

Abbreviations: BAYES, Bayesian for ordered categorical indicators; E, Excellencism; P, Perfectionism; WLSMV, Weighted least square with mean and variance adjusted. Bold indicates primary factor loadings.

Plan of analysis

Analyses. All our analyses were performed in Mplus 10 (Muthén & Muthén, 2024). First, we started with a confirmatory factor analysis (CFA) with the WLSMV estimator to test the suitability of the two‐factor model of the SCOPE with a sample of graduate students. Several fit indices were used to evaluate the suitability of the factor model: Chi‐square test (χ²), Confirmatory Fit Index (CFI), Tucker‐Lewis index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). Based on several recommendations, we relied on cutoff values (CFI and TLI > .95; RMSEA < .06; SRMR < .08) as guidelines rather than rules (e.g. Hu & Bentler, ¹⁹⁹⁹; Marsh et al., ²⁰⁰⁴; Sellbom & Tellegen, ²⁰¹⁹) because they are influenced by specificities of the model such as the distribution of the scores, the size of the factor loadings, the number of items, and the correlations between the factors (e.g. Bollen, ¹⁹⁸⁹; Chen et al., ²⁰⁰⁸; Greiff & Heene, ²⁰¹⁷; McNeish & Wolf, ²⁰²¹). Despite its hypersensitivity, a significant χ² remains indicative of model misfit (Hayduk et al., 2007) and should be followed by a close examination of the potential sources of model misspecifications. To do so, we performed complementary analyses using the Bayesian CFA framework (e.g. Depaoli, ²⁰²¹). The typical CFA fixes all cross‐loadings and correlated residuals to be exactly zero. This is an overly strict and unrealistic specification (Marsh et al., 2013; Muthén & Asparouhov, ²⁰¹²; West et al., ²⁰¹²). Using this specification, it is difficult to determine if model misfit is caused by important psychometric problems (e.g. omission of non‐trivial cross‐loadings) or the sum of trivial and tolerable deviations between the model and the data (e.g. tolerable correlations between residual of items). We relied on Bayesian CFA with informative near‐zero priors to allow cross‐loadings to vary around an interval of −.20 to .20 and correlated residuals to vary between −.20 and .20 (Asparouhov et al., 2015; Depaoli, ²⁰²¹; Muthén & Asparouhov, ²⁰¹²). Typical fit indices are not available for Bayesian CFA with ordered categorical data (Liang & Yang, 2014). Model fit was examined with the posterior predictive p‐value (PPP) and the 95% CI of the difference in observed and replicated χ². Good fit is obtained when the PPP is non‐significant and the 95% CI of the difference in observed and replicated χ² has a negative lower bound (e.g. Asparouhov et al., ²⁰¹⁵). Results from the Bayesian CFA are insightful because they enable a complete assessment of the cross‐loadings and correlated residuals to determine if the sources of the model misfit are substantial or trivial.

Second, we followed with multiple regression to estimate the unique effects of excellencism and perfectionism on a series of external criteria. First, we looked at perfectionistic self‐presentation to evaluate the associations of the SCOPE with another measure of perfectionism (i.e. convergent validity). Second, we looked at the four consequential outcomes in graduate studies' lives to test the MEP hypotheses. The hypotheses of this study were not preregistered, but the multiple regression followed the detailed seven‐step plan of analysis that accompanies the theory elaboration of the MEP (Gaudreau et al., 2024). Excellencism and perfectionism are mean‐centred and included as predictors. The intercept and betas are used to calculate predicted values for theory‐driven prototypical cases of (a) perfection strivers (1SD above excellencism and perfectionism), (b) excellence strivers (1SD above excellencism and 1SD below perfectionism), and (c) non‐excellence/non‐perfection strivers (1SD below excellencism and 1SD below perfectionism). This approach does not divide the sample into subgroups (e.g. median‐split); it uses the multiple regression equation (Cohen et al., 2003) to calculate, graph, and compare predicted values. Predicted values are compared in Mplus in a way that allows for significance testing and effect size estimation (Cohen's d). The data, material, and script to run these analyses are available at https://osf.io/4dkz2/?view_only=81535aaf70f64fbca7b0892c645cd2d0. These predicted values were also graphed using the MEP Shiny App: https://model‐of‐excellencism‐and‐perfectionism.shinyapps.io/Shiny_Version2/.

Sample size determination

First, we aimed to recruit enough participants to reach between a 1:10 (n = 220) to 1:20 (n = 440) items‐to‐participants ratio in CFA (Mundfrom et al., 2005). Second, we also consulted the results of simulation studies (e.g. Wolf et al., ²⁰¹³) to help select a sample size for a two‐factor model with strong factor loadings (= .80). Third, we also relied on the Power4SEM Shiny App (Jak et al., 2021) to estimate the statistical power of a test to determine whether a two‐factor model provided a better fit to the data compared to a one‐factor model. We started by specifying a H0 model based on the average standardized factor loadings of excellencism (= .83) and perfectionism (= .91) and the observed correlation between the two factors (= .59) in the validation study of Gaudreau et al. (2022). We specified an alternative H1 model with the same specifications, except that the correlation between the two factors was fixed at 1 (i.e. this is the equivalent of a one‐factor model with one less degree of freedom than the two‐factor model). We reached a power of 1 to detect a significant Chi‐square difference between the hypothesized two‐factor and the alternative one‐factor model. We reran this power analysis with a correlation of .69 for the H0 model based on a recent estimate obtained in another validation study of the SCOPE (Bien et al., 2024). We reached a power of .998 to detect a significant Chi‐square difference between the two models. Finally, we performed a power analysis in which we compared a H0 model with a correlation of .69 to a H1 model with a correlation of .90 (i.e. assuming a .90 correlation would be considered too strong to support the hypothesis that excellencism and perfectionism are distinct constructs). We reached a power of .882 to detect a significant Chi‐square difference between these two models. In this extreme condition (unlikely given the current knowledge about the SCOPE), the minimum sample size needed to reach a power of .80 was 279. Based on all these analyses, we concluded that a sample above 350 would be acceptable to run the factor analyses and the regression analyses planned for this study (305 was the required sample to reach 80% power for a two‐tailed test (p < .05) of the typical effect size (r = .16, f ² = .026) reported in psychology).

RESULTS

Part 1: Factor structure of the SCOPE

Preliminary analyses

Twenty multivariate outliers (Mahalanobis χ² = 48.27, df = 22, p < .001) were removed from the factor analyses. We examined the distribution of the scores of the 22 items from the SCOPE. Mean, standard deviations, skewness, and kurtosis of the items are presented in Table 1. Average score of excellencism (from 5.18 to 5.94; ‘strongly agree’) were all higher than the average score of perfectionism (from 3.54 to 4.63; ‘slightly to moderately agree’). The absolute value of the skewness was smaller than 1, except for the sixth item of excellencism which displayed the highest average score, skewness, and kurtosis. All items were amenable to factor analysis.

Confirmatory factor analyses

The results of our CFA with the WLSMV estimator showed that the fit of our hypothesized two‐factor model reached many thresholds of acceptable fit: χ² = 892.196, df = 208, p < .001, CFI = .974, TLI = .971, RMSEA = .096 [.090, .103], SRMR = .055. Standardized factor loadings are presented in Table 1. As expected, the excellencism and perfectionism scores were significantly correlated (r = .538, p < .001). In contrast, the fit of the alternative one‐factor model was unacceptable: χ² = 2766.40, df = 209, p < .001, CFI = .901, TLI = .891, RMSEA = .185 [.179, .192], SRMR = .152.

Even if the two‐factor model was superior and acceptable, we inspected potential sources of model misspecification because the χ² was statistically significant. We relied on Bayesian CFA with informative near‐zero priors to allow cross‐loadings to vary around an interval of −.20 to .20 and correlated residuals to vary between −.20 to .20 (Asparouhov et al., 2015; Depaoli, ²⁰²¹; Muthén & Asparouhov, ²⁰¹²). The Bayesian CFA with near‐zero cross‐loadings and correlated residuals provided a good fit to the data: PPP = .358; 95% CI of the difference in observed and replicated χ² = [−56.222, 80.257]. As shown in Table 1, none of the cross‐loadings were statistically significant and they ranged from −.138 to .084 for excellencism items (absolute average = |.043|) and from −.058 to .086 for perfectionism items (absolute average = |.048|). Twenty of 231 correlated residuals were statistically significant, but their value was small to moderate (see Table S1 in our supplementary file). These sources of model misspecifications were largely trivial, and the SCOPE two‐factor model is acceptable to separate excellencism and perfectionism.

Part 2: Multiple regressions

Preliminary analyses

We examined the distribution of the scores of the 13 variables included in our subsequent analyses. Four univariate outliers with low scores on excellencism (Z score < |3|) and nine multivariate outliers (Mahalanobis χ² = 34.53, df = 13, p < .001) were removed from further analyses. Means, standard deviations, and bivariate correlations between all variables are presented in Table 2.

TABLE 2.

Descriptive statistics and bivariate correlations.

Variables	1.	2.	3.	4.	5.	6.	7.	8.	9.	10.	11.	12.	13.
1. Excellencism	—
2. Perfectionism	.46**	—
3. Perfectionistic self‐promotion	.16**	.40**	—
4. Non‐display of imperfection	.08	.25**	.85**	—
5. Non‐disclosure of imperfection	−.01	.17**	.74**	.77**	—
6. Research productivity satisfaction	.29**	.29**	.02	−.12*	−.09	—
7. Research self‐efficacy	.29**	.39**	.07	−.06	.02	.52**	—
8. Academic burnout	−.12*	.03	.29**	.43**	.44**	−.29**	−.16**	—
9. Dropout intentions	−.27**	−.06	.14**	.26**	.33**	−.34**	−.22**	.62**	—
10. Intention to change program	−.10*	.08	.11*	.21**	.24**	−.19**	−.13*	.45**	.56**	—
11. Intention to change supervisor	−.14**	.00	.01	.11*	.15**	−.09	−.03	.32**	.37**	.45**	—
12. Intention to change topic	−.14**	−.01	.08	.19**	.20**	−.20**	−.02	.35**	.37**	.50**	.55**	—
13. Intention to change university	−.11*	.02	.14**	.17**	.19**	−.15**	−.03	.37**	.37**	.49**	.36**	.51**	—
Mean	5.522	4.036	3.761	3.896	3.496	4.061	3.683	3.406	2.597	2.554	2.292	2.497	2.470
SD	0.844	1.414	1.402	1.205	1.078	1.160	1.249	1.285	1.492	1.643	1.607	1.667	1.705

Note: N = 363. **p < .01. *p < .05.

Convergent validity: Perfectionistic self‐presentation

Results of a multiple regression (R ² = .098) showed that perfectionistic self‐presentation (intercept = 3.718) was unrelated to excellencism (B = −0.087, SE = 0.073, p = .23, β = −.064) but positively associated with perfectionism (B = 0.272, SE = 0.046, p < .001, β = .337). Perfection strivers relied on more perfectionistic self‐presentation compared to both excellence strivers (d = 0.674, 95% CI = [0.449, 0.900], p < .01) and non‐excellence/non‐perfection strivers (d = 0.546, 95% CI = [0.298, 0.793], p < .01). Excellence strivers and non‐excellence/non‐perfection strivers did not significantly differ (d = −0.129, 95% CI = [−0.340, 0.082], p = .23). These results, which offered support for the convergent validity of perfectionism, were similar across the three subscales of perfectionistic self‐presentation (see Table S2).

Satisfaction with scientific productivity

Results of a multiple regression (R ²  = .135) showed that satisfaction with scientific productivity (intercept = 4.061) was positively associated with excellencism (B = 0.256, SE = 0.078, p < .01, β = .186) and perfectionism (B = 0.205, SE = 0.051, p < .01, β = .250), and negatively associated with perfectionistic self‐presentation (B = −0.159, SE = 0.053, p < .01, β = −.156). As shown in Figure 1, excellence strivers (d = 0.372, 95% CI = [0.151, 0.593], p < .01) and perfection strivers (d = 0.872, 95% CI = [0.630, 1.114], p < .01) were more satisfied with their scientific productivity compared to non‐excellence/non‐perfection strivers. Furthermore, perfection strivers were more satisfied with their scientific productivity compared to excellence strivers (d = 0.500, 95% CI = [0.255, 0.745], p < .01).

FIGURE 1.

Research productivity satisfaction of non‐excellence/non‐perfection, excellence, and perfection strivers. Exc, Excellence strivers; Non, non‐excellence/non‐perfection strivers; Perf, Perfection strivers. The three predicted values differ at p < .01.

Academic burnout, dropout intentions, and research self‐efficacy

Overview of analyses

We included satisfaction with research productivity as a covariate because it was associated with both excellencism and perfectionism in our previous analyses. We performed three moderated multiple regressions in which excellencism, perfectionism, research productivity satisfaction, and the interaction of research productivity satisfaction × excellencism and research productivity satisfaction × perfectionism were included as predictors of (a) burnout, (b) dropout intentions, and (c) research self‐efficacy. Results are presented in Table 3.

TABLE 3.

Results of the moderated multiple regressions of academic burnout, dropout intentions, and research self‐efficacy.

	Academic burnout			Dropout intentions			Research self‐efficacy
	B	SE	β	B	SE	β	B	SE	β
Intercept	3.423			2.592			3.655
Excellencism (E)	−0.157	0.085	−.103	−0.413**	0.096	−.234	0.054	0.076	.036
Perfectionism (P)	0.146**	0.056	.161	0.139**	0.052	.131	0.224**	0.049	.254
Research productivity satisfaction (RPS)	−0.342**	0.062	−.309	−0.396**	0.072	−.308	0.475**	0.049	.441
E × RPS	0.013	0.078	.011	0.048	0.090	.034	−0.049	0.074	−.041
P × RPS	−0.044	0.047	−.061	−0.018	0.044	−.021	0.090†	0.046	.128
R 2			.109			.159			.346

Note: N = 363. **p < .01. *p < .05. ^† p = .05.

Academic burnout

As expected, satisfaction with research productivity had a negative association with academic burnout (β = −.309). After holding this covariate constant, perfectionism (but not excellencism) was positively associated with academic burnout. When comparing students with the same level of research productivity satisfaction (RPS) (e.g. low, medium, and high), perfection strivers were more likely to experience symptoms of academic burnout compared to both excellence strivers and non‐excellence/non‐perfection strivers. Academic burnout was lower in excellence strivers compared to non‐excellence/non‐perfection strivers, but the difference was not statistically significant (p = .076). As shown in Figure 2, these differences (and their effect size) were constant across low, medium, and high levels of RPS. Academic burnout decreased across levels of satisfaction with research productivity.

Dropout intentions

As expected, satisfaction with research productivity had a negative association with dropout intentions (β = −.308). After holding this covariate constant, perfectionism and excellencism were positively and negatively associated with dropout intentions, respectively. When comparing students with the same level of RPS, perfection strivers had more dropout intentions compared to excellence strivers, who had fewer dropout intentions than non‐excellence/non‐perfection strivers. As shown in Figure 3, these differences (and their effect size) were constant across low, medium, and high levels of RPS. Dropout intentions decreased across levels of satisfaction with research productivity.

FIGURE 3.

Dropout intentions at low, average, and high levels of research productivity satisfaction (RPS). Contrasts are constant/identical across the three levels of research productivity satisfaction, but scores of dropout intentions diminish across the three panels of this figure. Excel, Excellence strivers; Non, non‐excellence/non‐perfection strivers; Perfect, Perfection strivers. Values reported here are taken from Mplus. **p < .01. *p < .05. ^† p = .096.

Research self‐efficacy

The interaction perfectionism × satisfaction with RPS was marginally significant (p = .05). To obtain more precise estimates of the perfectionism effect, we therefore calculated simple slopes of perfectionism at low, medium, and high levels of RPS. Findings indicated that perfectionism was positively and significantly associated with research self‐efficacy across all levels of satisfaction. However, this association became significantly stronger across low (intercept = 3.104, B = 0.120, SE = 0.057, p < .05), medium (intercept = 3.655, B = 0.224, SE = 0.044, p < .01), and high (intercept = 4.206, B = 0.328, SE = 0.064, p < .01) levels of scientific productivity satisfaction. As shown in Figure 4, we can conclude that perfection strivers have higher research self‐efficacy than excellence and non‐excellence/non‐perfection strivers; these differences (and their effect size) are getting stronger with increasing levels of satisfaction with research productivity.

FIGURE 4.

Research self‐efficacy at low, average, and high levels of research productivity satisfaction (RPS). Contrasts are moderated and significantly change across levels of research productivity satisfaction. Excel, Excellence strivers; Non, non‐excellence/non‐perfection strivers; Perfect, Perfection strivers. Values reported here are taken from Mplus. **p < .01. *p < .05.

Complementary analyses

We surveyed students about their intention to change program, supervisor, research topic, and university using four single‐item scores, and we conducted four multiple regressions to explore their associations with excellencism and perfectionism. Complete results are available in the supplementary file (see Table S3). The interaction perfectionism × satisfaction with research productivity was significant for the intention to change program (p < .05) and interpretable for the intention to change research topic and supervisor (p < .07). The results for the intention to change program are shown in Figure 5. The other findings are displayed in Figures S1 to S3 (see supplementary file). The simple slopes of perfectionism were significant at low but not at high levels of satisfaction with research productivity for the (a) intention to change program (Intercept _low = 2.970, B_low = 0.353, SE = 0.089, p < .05 versus Intercept _high = 2.241, B_high = 0.079, SE = 0.100, p = .43), (b) intention to change research topic (Intercept _low = 2.887, B_low = 0.229, SE = .094, p < .05 versus Intercept _high = 2.238, B_high = 0.004, SE = 0.091, p = .96), and (c) intention to change supervisor (Intercept _low = 2.489, B_low = 0.225, SE = 0.091, p < .05 versus Intercept _high = 2.262, B_high = −0.008, SE = 0.095, p = .94). The difference between excellence and perfection strivers was significant at low levels but not at high levels of RPS.

FIGURE 5.

Intention to change program at low, average, and high level of research productivity satisfaction (RPS). Contrasts are moderated and significantly change across levels of research productivity satisfaction. Excel, Excellence strivers; Non, non‐excellence/non‐perfection strivers; Perfect, Perfection strivers. Values reported here are taken from Mplus. **p < .01. *p < .05. ^† p = .055.

DISCUSSION

Consistent with the MEP (Gaudreau et al., 2022), the results of this study showed that pursuing high personal standards (excellencism) and pursuing perfectionistic standards (perfectionism) are two distinct constructs. This study revealed important differences between perfectionism and excellencism in graduate school. First, our findings illustrate an antagonistic pattern in perfectionism, linked to positive performance outcomes and negative psychological experiences among graduate students. Second, excellencism can be seen as optimal because it relates to higher research productivity satisfaction while being associated with lower perfectionistic self‐presentation tactics, academic burnout, and dropout intentions compared to perfection strivers. Overall, the outcomes of perfectionism are complex but become more apparent when viewed through the distinction between perfectionism and excellencism underscored in the MEP.

Key findings

Results of our confirmatory factor analyses supported the factorial validity of the SCOPE and the idea that excellencism and perfectionism should be studied as distinct constructs among graduate students. The SCOPE (Gaudreau et al., 2022) is a newly created instrument previously validated in undergraduate students and adolescents (Bien et al., 2024; Tape et al., ²⁰²⁴). Our findings uniquely demonstrate the potential generalizability of the two‐factor structure of the SCOPE across different levels of education. This broad applicability will enable consistent assessment and comparison of students from secondary to different levels of higher education. Based on the SCOPE, it is now possible to use a series of validated indicators to determine if a student is a perfection striver or an excellence striver. Being ambitious and pursuing high standards is not inherently unhealthy. Drawing a line between excellencism and perfectionism will help service providers in higher education focus on those graduate students whose mental health may be at greater risk during their graduate studies.

Our findings outlined the need to differentiate excellencism and perfectionism at the functional level, as their associations with outcomes differed in many ways. First, excellencism was not significantly associated with perfectionistic self‐presentation, thus conveying information about its discriminant validity and the idea that signature expressions of perfectionism are unique to perfection strivers (Gaudreau, 2021). Second, excellencism was positively associated with satisfaction with research productivity, negatively linked to dropout intentions, and not significantly associated with academic burnout or research self‐efficacy. The pool of outcomes tied to excellencism suggests that those pursuing excellence are less likely to experience a mix of both positive and negative outcomes. Thus, the proverbial double‐edged sword that characterizes the outcomes of perfectionism does not apply to excellencism. Third, both excellencism and perfectionism were positively associated with satisfaction with research productivity. However, none of the outcomes associated with excellencism were contingent (i.e. statistically moderated) on the students' satisfaction with research productivity. The idea that the wellness of graduate students depends on their accomplishments appears limited to perfection strivers. Finally, satisfaction with research productivity was strongly associated with academic burnout and research self‐efficacy. Complementary analyses, in which we did not control for satisfaction with research productivity, showed that excellencism was significantly associated with lower burnout and higher self‐efficacy (see Table S4). Those associations were non‐significant in our main analyses (controlling for satisfaction with research productivity), which suggests that some of the associations involved in excellencism could be mediated by research productivity. Future longitudinal research should investigate the mediating role of research productivity in the psychological benefits of excellencism.

The MEP builds on and extends two foundational principles in perfectionism literature. On the one hand, our findings with graduate students are consistent with the longstanding view of perfectionism as a mixed blessing, presenting both advantages and drawbacks (Madigan, 2019; Missildine, ¹⁹⁶³; Stoeber, ²⁰¹⁴). Even if perfection strivers were more satisfied with their scientific productivity, they experienced more symptoms of burnout and dropout intentions than excellence strivers. Graduate students often feel pressured to do whatever it takes to succeed, so much that they internalize the idea that achievement is the only standard by which they are judged or held accountable (Bartle, 2019). Many graduate programs and research labs have a hyper‐competitive ethos in which many graduate students feel that work should be prioritized over their wellness (Bekkouche et al., 2022). In this context, perfection strivers could easily feel compelled, obliged, and seduced to over‐work, over‐strive, and adhere to the norms and values of an overly pressuring academic culture. Graduate school is inherently challenging, and values and behaviours that expose graduate students to gruelling performance pressure should not be tolerated. Our results provide the impetus to shift the narrative and create the conditions to help students transform their perfectionism into excellencism to optimize their productivity and mental health during and beyond graduate school.

On the other hand, several of our findings supported a second foundational principle from perfectionism research – the idea that the wellness of perfection strivers depends on their accomplishments. For example, perfection strivers reported higher research self‐efficacy compared to excellence strivers, but the strength of this difference significantly increased across levels of scientific productivity satisfaction (see Figure 4). This finding indicates that perfection strivers build their sense of academic self‐efficacy based on their satisfaction with their research productivity. They are increasingly confident about their research skills when they feel like their research outputs are outnumbering those of other graduate students in their discipline at a comparable career stage. Although this elevated self‐efficacy appears like good news for perfection strivers, it also indicates that their research self‐efficacy is more contingent on achievement compared to excellence and non‐excellence/non‐perfection strivers. Perfection strivers pursue extremely high standards and evaluate their self‐worth based on these standards (e.g. Egan & Shafran, ²⁰¹⁸; Hewitt et al., ²⁰¹⁷). Having a contingent self‐worth (e.g. attaching too much importance to performance) is a risk factor for psychological distress (Crocker et al., 2003) and could make them more vulnerable to the frequent ups and downs and repeated rejections inherent to the scientific peer‐review process (Jaremka et al., 2020).

Another illustration of the importance of achievement can be found in our findings regarding the intention to change program, supervisor, and research topic. These findings followed the pattern of a stress diathesis in which the negative effects of perfectionism are assumed to be exacerbated when people experience performance‐related difficulties (Curran & Hill, 2018; Gaudreau et al., ²⁰²⁴; Hewitt & Flett, ¹⁹⁹³). The differences in the intention to change program, supervisor, and research topic between perfection and excellence strivers were statistically significant only at lower levels of research productivity satisfaction. Perfection strivers more frequently doubt their actions, are more concerned about mistakes, and put on a mask to avoid disclosing their imperfections to others (e.g. Gaudreau, ²⁰²¹). It is a well‐established principle that serious doubts – like the ones when perfection strivers are unsatisfied with their research productivity – forge an impulse to disengage and withdraw one's effort from goal‐related intention (Carver & Scheier, 1998). This could explain why perfection strivers who are unsatisfied with their research productivity are more likely than excellence strivers to consider changing program, supervisor, and research topic. Altogether, these findings suggest the need to differentiate the many ways through which perfectionistic students may attempt to leave the scene during their graduate education. They also speak about the importance of creating the conditions needed to help students be satisfied with their research productivity while, at the same time, encouraging the pursuit of more realistic and attainable standards like the ones involved in excellencism.

Finally, graduate students who were the least satisfied with their research productivity reported the highest levels of burnout and dropout intentions. Burnout and dropping out of graduate school can be financially costly for the students, the research supervisor, and their university (Devos et al., 2017; Litalien & Guay, ²⁰¹⁵). When considering two graduate students who were equally satisfied with their productivity, perfection strivers experienced higher academic burnout and dropout intentions than excellence strivers. The strength of these differences was constant across low, medium, and high levels of research productivity satisfaction, which indicates that perfectionism can be considered a risk factor when comparing graduate students with a similar level of performance satisfaction. This finding has methodological implications, as it informs researchers of the need to consider the research productivity of graduate students when assessing the risks associated with their perfectionism.

Limitations and future research

We relied on a cross‐sectional design in which satisfaction with research productivity played an important role in the associations of excellencism and perfectionism with psychological outcomes. Results of Gaudreau et al. (2022) with undergraduate students revealed that good performances are associated with a longitudinal increase in excellencism (but not perfectionism) which, in turn, leads to improvements in performance. Future research should try to replicate these upward developmental effects with graduate students. Motivation for research and well‐being were found to decrease over a graduate program (Jackman et al., 2022; Sverdlik & Hall, ²⁰²⁰). Whether excellencism can protect against the potential deterioration of motivation and mental health of graduate students needs further empirical investigation.

Academic disciplines can have different standards to assess research productivity. The question used to measure satisfaction with research productivity, which implied a social comparison, was meant to ensure that students would compare themselves with students in their discipline at a similar stage in their careers. However, perfection strivers tend to prioritize and make more progress on socially imposed goals at the expense of their personal goals (Gaudreau et al., 2022). Forcing a social comparison may have accentuated this tendency and could explain why perfection strivers perceived themselves as more productive when comparing themselves to others. Excellence strivers are more likely to prioritize intrinsic goals (e.g. learning and self‐improvement), which may explain why they were lower on a performance measure that focused on social comparison. Future studies should use multidimensional measures tapping into different aspects of subjective performance, such as goal progress, goal attainment, task mastery, and normative/social comparisons (e.g. Smyth et al., ²⁰²³). Subjective performance measures should also be complemented with a more objective outlook on graduate students' research productivity. Having at least one publication is mandatory for graduation in several countries (Li, 2016; Moradi, ²⁰¹⁹). Our study did not differentiate first authorship (the student is the real driver of scientific productivity) from co‐authorship (in which the student often makes a smaller contribution) and publications in journals, book chapters, and proceedings of scientific conferences.

Our study was not designed to compare results across countries, fields/disciplines, years into the program, and levels of education (master versus doctoral). Making such comparisons would be preferable in studies with enough participants in each cell to correctly estimate and compare the means of excellencism and perfectionism across subgroups of graduate students. Tracking graduate students over the years of their program would also be desirable. Failure to maintain the same productivity rate across time (when students are expected to publish on their own) could reduce the research productivity satisfaction of perfection strivers. How bumps on the road (e.g. rejected manuscripts) will differentially relate to the self‐efficacy of perfection and excellence strivers and detailed feedback from their supervisor and advisory committee is also a question that awaits future investigations.

Graduate studies are often part of a stressful period marked by many new responsibilities in adult life. Examining the compound effect of stressors inside and outside of school appears needed to better understand the high prevalence of psychological distress among graduate students. Universities may want to consider excellencism as a way to promote healthy ambitions and work‐life balance of graduate students. Studies with undergraduate students showed that excellence strivers have a lower fear of failure, and they experience less conflict when pursuing multiple goals (Gaudreau et al., 2021). They take the time to savour and celebrate their academic success (Gaudreau et al., 2024). When their goals become unattainable, they tend to reengage into new goals that enable them to live meaningful growth‐related experiences (Gaudreau & Chamandy, 2024). Future studies are needed to determine if these psychological benefits of excellencism can inform the development of initiatives to optimize graduate school students' psychological health.

CONCLUSION

Perfectionism researchers have fallen prey to a common fallacy and assumed they could ‘safely rely on the name of a measure to infer its content’ (Lilienfeld & Strother, 2020, p. 282). Our findings showed that perfectionistic standards and excellencism are different constructs that should not be conflated into a single score. Failure to separate them potentially explains the inconsistent findings reported in past studies conducted with graduate students. Consistent with the fundamental postulate of the MEP, excellencism and perfectionism were distinctively associated with graduate students' satisfaction with research productivity, research self‐efficacy, symptoms of burnout, and dropout intentions. Our findings paved the way for a deeper understanding of perfectionism's complex roles in graduate students' research productivity and mental health. They also underscore the potential of promoting excellencism in institutional policies aiming to foster a balance between the productivity and wellness of graduate students.

AUTHOR CONTRIBUTIONS

Patrick Gaudreau: Conceptualization; investigation; funding acquisition; writing – original draft; methodology; visualization; validation; writing – review and editing; formal analysis; project administration; data curation; supervision. Antoine Benoît: Conceptualization; investigation; methodology; validation; visualization; writing – review and editing.

CONFLICT OF INTEREST STATEMENT

We have no known conflict of interest to disclose.

CONSENT

This study has been approved by the research ethics board at the University of Ottawa and all participants provided informed consent.

Supporting information

Data S1.

Gaudreau, P. , & Benoît, A. (2025). Distinguishing perfectionism and excellencism in graduate students: Contrasting links with performance satisfaction, research self‐efficacy, burnout, and dropout intentions. British Journal of Psychology, 116, 907–929. 10.1111/bjop.12798

DATA AVAILABILITY STATEMENT

Data is available on Open Science Framework (see links in the Method sections).