The mediating role of self-compassion in the relationship between trait perfectionism and psychological resilience among Filipino university students

Abstract Perfectionism has been a challenge in university students’ everyday lives. As perfectionism can negatively affect students’ mental health, researchers have suggested ways to manage perfectionism and build students’ resilience towards having a better wellbeing. One way is to practice self-compassion. This study utilized mediation analyses to investigate if self-compassion mediates the relationship between the trait dimensions of perfectionism and psychological resilience among Filipino university students. A sample of 317 undergraduate students from the University of the Philippines Diliman participated and answered scales that measure the variables of interest. Results indicate that when covariates were excluded, self-oriented perfectionism (SOP) and socially prescribed perfectionism influenced psychological resilience indirectly via self-compassion, while SOP and other-oriented perfectionism (OOP) directly affected psychological resilience, independent of self-compassion. When covariates were included, the same findings were found, but OOP also influenced psychological resilience indirectly via self-compassion. Implications of the findings are discussed in this study.


Introduction
The university experience of a student can be a challenging one.Plagued with academic requirements, family conflicts, and social pressures, students are used to feeling stress, which is considered as a normal experience among them (Laidlaw et al., 2016).However, when stress becomes too intense and is not properly managed, especially when adverse or negative circumstances arise, students can have difficulties in recovering or bouncing back from these circumstances.This lack of resilience can unfortunately lead to negative mental health outcomes like depression and anxiety (Cleofe et al., 2018;Wu et al., 2020).According to the World Health Organization (2020), the most vulnerable population globally is those who are 15 to 29 years old with mental health-related deaths as the second leading cause of death among this age group.The University of the Philippines Diliman Psychosocial Services (UPD PsycServ), the mental health task force of UPD, ABOUT THE AUTHOR Jeryl Shawn T. Tan is a licensed psychologist and a senior lecturer from the Department of Psychology at the University of the Philippines Diliman.He finished his Master of Arts degree in Psychology at the said university.also reported that out of the 1,555 students who sought counselling with PsycServ, around 39% had complicated problems of living, around 36% were diagnosed with depressive disorders, and around 24% were diagnosed with anxiety disorders (featured in video of Interactive Learning Center Diliman, 2020).Knowing these statistics, there is a need for students to consider protective and risk factors that affect their resilience.One such factor that has received much attention by researchers is perfectionism.Though studies have looked into the perfectionism-resilience link (e.g., Çerkez, 2017;Klibert et al., 2014), findings have been mixed so far.In addition, studies on factors that mediate its link are still lacking.As such, this study aimed to confirm that selfcompassion is a mediating factor in the relationship between the trait dimensions of perfectionism and psychological resilience among Filipino university students.

Perfectionism and psychological resilience
Perfectionism is commonly associated with striving very hard to reach high unattainable and unrealistic goals or standards.With perfectionism being complex and multidimensional (Frost et al., 1990), many researchers have proposed multidimensional models such as Hewitt and Flett's (1991).They conceptualized trait perfectionism as three different dimensions that reflect the experience of requiring the self or others to be perfect and expecting that others require the self to be perfect.These dimensions include self-oriented perfectionism (SOP), which is the tendency to set exacting standards for oneself and to evaluate and criticize one's behavior in a stringent manner, socially prescribed perfectionism (SPP), which is one's perception that it is crucial to meet the unrealistic expectations set by others and to be perfect as required by others, and other-oriented perfectionism (OOP), which is the tendency to set challenging standards for others and to evaluate and criticize the behavior of others in a stringent manner.
Psychological resilience involves adapting to difficult, challenging experiences or bouncing back or recovering from stress (Smith et al., 2008).Having a sense of resilience among Filipinos is an adaptive coping strategy that allows them to cope with emotional challenges without the need of seeking help from others (Martinez et al., 2020).Past studies have identified significant factors that affect Filipino youth's resilience, including external support, personal strengths, and volunteerism (e.g., Llenares et al., 2020;Salvacion et al., 2018).
As highlighted earlier, findings about the link between perfectionism and psychological resilience are mixed.Çerkez (2017) suggested that perfectionist students tend to have higher resilience given their behavior of setting high targets and positive future expectations, but Saadi et al. (2019) observed that perfectionist students have lower resilience because of their pessimistic thinking and fear of failure.Choo and Prihadi (2019) also reported that individuals with high perfectionistic strivings (typically capture SOP aspects) have higher academic resilience because of their optimism about the future and self-compassion, while individuals with high perfectionistic concerns (typically capture SPP aspects) have lower academic resilience because of their rumination, pessimism, lack of self-compassion, and unwillingness to build their resilience.Using all trait perfectionism dimensions, Klibert et al. (2014) found resilience to have essentially no correlation with OOP but a significant negative correlation with SPP; they reasoned that individuals with SPP are prone to blame themselves for their imperfections or failures, which lead to greater distress symptoms and their inability to navigate through and overcome challenging situations (Hewitt & Flett, 2002).They also found a positive (though nonsignificant) correlation between SOP and resilience as individuals with SOP have conscientiousness characteristics (e.g., purposeful, determined, diligent) that may help in adapting to stressful events (Hill et al., 1997).

Perfectionism and self-compassion
Another factor related to perfectionism is self-compassion, which is "the ability to hold one's feelings of suffering with a sense of warmth, connection and concern" (Raes et al., 2011, p. 250).Neff (2003) identified three major components of self-compassion: self-kindness (versus selfjudgment), common humanity (versus isolation), and mindfulness (versus over-identification). Through self-compassion, individuals can extend intentions of loving-kindness to themselves and acknowledge that their disappointments, failures, and imperfections are part of the human experience (Neff, 2003;Raes et al., 2011).Self-compassion among university students has been shown by studies to improve their psychological health, including increasing their psychological resilience (e.g., Fong & Loi, 2016;Neff & McGehee, 2010).
When it comes to its link with perfectionism, studies have shown consistently that maladaptive perfectionism is negatively associated with self-compassion (e.g., Linnett & Kibowski, 2020;Umandap & Teh, 2020) as it is mainly characterized by excessive self-criticisms when mistakes happen or when performance is unsatisfactory.Though adaptive perfectionism is expected to be positively associated with self-compassion, not all studies agree with that finding.For instance, Yeshua et al. (2019) found a significant negative correlation between unidimensional perfectionism and self-compassion, and Linnett and Kibowski (2020) found adaptive perfectionism to predict higher self-judgment, a component of self-compassion, even after controlling for the suppressor effects of maladaptive perfectionism.These results may be due to the incompatibility between tolerance to perceived inadequacy and the need to exhibit excellent performance and reach very high and stringent standards (Linnett & Kibowski, 2020).Many researchers have also found that all trait perfectionism dimensions have significant negative associations with self-compassion or its related concept of unconditional self-acceptance (e.g., Flett et al., 2003;Stoeber et al., 2020); they reasoned that perfectionist students make their self-worth contingent on their perfectionistic standards or on others' feedback and support, thereby decreasing their self-compassion.On the other hand, both Nisa et al. (2019) and Otrar and Tezcan (2015) reported significant positive correlations between self-compassion and the trait perfectionism dimensions.Self-compassion has been found to be a significant mediator between perfectionism and several mental health outcomes, including depression (Wei et al., 2021) and subjective wellbeing (SWB; Stoeber et al., 2020).However, to the researcher's awareness, there are scarce studies that examine selfcompassion as a mediator in the perfectionism-resilience link, especially among Filipinos.

The present study
In line with the literature review, this study hoped to shed more light on the associations between trait perfectionism dimensions, self-compassion, and psychological resilience among Filipino university students.Particularly, (1) Do the trait perfectionism dimensions significantly correlate with self-compassion?
(2) Do the trait perfectionism dimensions significantly affect psychological resilience directly, independent of self-compassion?and (3) Does self-compassion significantly mediate the link between the trait perfectionism dimensions and psychological resilience?
The researcher hypothesized that (1) all trait perfectionism dimensions significantly correlate with self-compassion in a negative direction, (2) SOP and SPP, but not OOP, significantly affect psychological resilience directly, independent of self-compassion, and (3) self-compassion significantly mediates the relationship between all trait perfectionism dimensions and psychological resilience.

Participant and recruitment
A total of 317 UPD undergraduate students, aged 18 to 30 (M = 20.37;SD = 1.39), answered an online survey.This number of participants is beyond the target sample size identified after conducting Monte Carlo power analysis (Schoemann et al., 2017) with a desired power of .80 and a confidence level of 95% (i.e., at least 281).Participants were recruited by online poster that was posted in different social media pages and by email communications to professors.Participant characteristics are featured in Tables 1 and 2.

Materials and measures
Through Google Forms, participants answered an online survey, which included the following sections: the first section included the informed consent form and questions asking for their Psychology 101 class, professor, and email address, which were needed for the redemption of the 30-minute research credit stub and a list of perfectionism resources as tokens; the second section featured questions asking for their demographic and academic-related data (see Tables 1  and 2).Apart from age and the Coronavirus disease 2019 (COVID-19) perceived stress, these variables were treated as covariates in the analyses as these have been associated with variability in resilience scores among university students (e.g., Erdogan et al., 2015;Findyartini et al., 2021).The next sections included scales for them to complete (see the next paragraphs for the descriptions), and the remaining sections featured a debriefing script and a positive mood induction procedure to induce positive feelings in case they experience any strong feelings from answering the survey.

COVID-19 perceived stress
The COVID-19 Perceived Stress Scale (PSS), which was modified by Pedrozo-Pupo et al. ( 2020) from the original PSS-10 ( Cohen et al., 1983), measures stress related to COVID-19 as perceived by the person during the last month.It consists of 10 items on a 5-point Likert scale from "never" to "very often."The total score, with higher scores indicating higher COVID-19 perceived stress, is determined by reversing the negative subscale items and then computing for the grand total of all items.The modified version has shown validity and good internal consistency with the authors reporting a Cronbach's alpha of .86 among Colombian adults and Oducado et al. (2021) reporting .80among Filipino graduate students.This variable was also treated as a covariate given studies that indicate its significant inverse relationship with resilience among Filipino university students (e.g., Guillasper et al., 2021;Oducado et al., 2021).

Perfectionism
The Multidimensional Perfectionism Scale by Hewitt and Flett (1991) measures the three trait perfectionism dimensions.It consists of 45 items, with 15 items per subscale (perfectionism dimension) on a 7-point Likert scale from "disagree" to "agree."The total score for each subscale is determined by getting the sum of ratings to that subscale.The higher the subscale score, the more a person experiences that perfectionism dimension.The authors reported good reliability for the subscales with Cronbach's alpha ranging from .74 to .88,test-retest reliabilities over 3 months ranging from .65 to .78, and intercorrelations ranging from .25 and .40 for students.The subscales also obtained construct (convergent and discriminant) and concurrent validity.

Self-compassion
The Self-compassion Scale-Short Form by Raes et al. (2011) measures how a person typically acts towards themselves in difficult times.It consists of 12 items on a 5-point Likert scale from "almost never" to "almost always" and has six subscales: self-kindness, self-judgment, common humanity, isolation, mindfulness, and over-identification.The total self-compassion score, with higher scores indicating higher self-compassion, is determined by reversing the negative subscale items and then computing for the grand mean of all six subscale means.The authors reported that the short form has adequate internal consistency with Cronbach's alpha of at least .86 in all samples and is a valid alternative to the long form, especially when using the total score.

Psychological resilience
The Brief Resilience Scale by Smith et al. (2008) measures the ability of a person to bounce back or recover from stress.It consists of six items on a 5-point Likert scale from "strongly disagree" to One participant did not provide their annual family income.As such, multiple imputation was performed to replace the missing value (see preliminary analyses section for more information).b Highest possible general weighted average (GWA) is 1.00, while lowest possible GWA is 5.00.
"strongly agree."After reverse coding half of the items, the mean of the ratings for the six items was calculated.The authors reported good internal consistency with Cronbach's alpha from .84 to .87 and test-retest reliability of .69 for one month among undergraduate students.The scale also had good convergent and discriminant predictive validity.

Procedures
This study obtained ethics approval from the Our Lady of Fatima University Institutional Ethics Review Committee (Reference #2021-IERC1-20128 v2) and the UPD Department of Psychology Research Ethics Committee.Upon viewing the online poster, participants were requested to click on the Google Forms link found in the said poster to answer the online survey.Participants took approximately 20 minutes to answer the survey.After clicking the "Submit" button on the survey's last page, they were brought to the confirmation message page, where they were debriefed and thanked for their participation.At the same time, the survey form automatically sent to them an email receipt that contained the research credit stub and a list of perfectionism resources.

Analysis plan
Preliminary analyses were conducted based on protocols provided by Tabachnick and Fidell (2007).These include inspecting for missing values, detecting outliers, and assessing internal consistency (Cronbach's alphas).Following preliminary analyses, descriptive statistics and Pearson r correlations were calculated for the variables.
Considering that the trait perfectionism dimensions can be correlated with each other and do not necessarily contradict each other (e.g., Hassan et al., 2012;Klibert et al., 2014), these dimensions were analyzed in three separate simple mediation models; analyzing them simultaneously in one model may cancel out each other's direct and/or indirect effects (Hayes, 2017).Additionally, each dimension was analyzed in two separate simple mediation models: a model without the covariates and a model with the covariates.Thus, a total of six simple mediation models were analyzed with bootstrapping (5,000 samples) using Model 4 in Hayes' (2017) PROCESS macro, installed in SPSS Statistics Version 26.

Preliminary analyses
Checking for missing data, one participant did not provide their annual family income.As results from Little's Missing Completely at Random (MCAR) Test in SPSS concluded that the data was indeed MCAR, χ 2 = 5.63, df = 10, p = .845,multiple imputation was performed to replace the missing value.Five imputations were run in sequence to achieve a set of imputed values, but these values were then averaged to arrive at a single value to replace the missing value.For all mediation models without the covariates, no univariate outliers beyond z = ± 3.29 (p < .001)were identified, but there was one multivariate outlier with Mahalanobis distance larger than χ 2 = 20.52 (df = 5; p < .001) that was identified and removed.For all mediation models with the covariates, marital status, a dichotomous variable with 90-10 splits or more between categories, was excluded in the analyses as the scores for the cases in the smaller category (i.e., partnered) have more influence than the scores in the larger category (i.e., single; Rummel, 1970).Additionally, 11 univariate outliers beyond z = ± 3.29 (p < .001)and six multivariate outliers with Mahalanobis distance larger than χ 2 = 36.12(df = 14; p < .001)were identified and removed.

Descriptive statistics, bivariate correlations, and Cronbach's alphas
Descriptive statistics, bivariate correlations, and Cronbach's alphas of all scaled variables are featured in Table 3. Participants reported moderate-to-high SOP and SPP, and moderate OOP, self-compassion, psychological resilience, and COVID-19 perceived stress.There were significant positive correlations between all trait perfectionism dimensions.SOP had a significant negative correlation with self-compassion, a significant positive correlation with COVID-19 perceived stress, and a non-significant correlation with psychological resilience.SPP had similar correlational observations as SOP but had a significant negative correlation with psychological resilience.OOP had a significant positive correlation with psychological resilience and nonsignificant correlations with self-compassion and COVID-19 perceived stress.Cronbach's alphas for all scales displayed acceptable internal reliabilities (from .79 to .91).

Discussion
This study investigated the associations between trait perfectionism dimensions, selfcompassion, and psychological resilience among Filipino university students, including investigating self-compassion as a mediator in the perfectionism-resilience link.As confirmed by the bivariate correlations, both SOP and SPP had significant negative correlations with selfcompassion, which are in line with the results in previous studies (e.g., Flett et al., 2003;Stoeber et al., 2020).Individuals with high SOP are heavily driven to succeed and strive towards satisfying their own goals, while individuals with high SPP believe that they are of value only when they reach others' standards (Sturman et al., 2009).Either way, exhibiting perfectionistic tendencies would strengthen their contingent sense of self-worth and increase their self-criticisms, which can diminish their self-compassion (Flett et al., 2003).On the other hand, OOP had a nonsignificant negative bivariate correlation with self-compassion, but this relationship became significant with the inclusion of the covariates, specifically COVID-19 perceived stress.The role of this particular covariate in the mediation analyses is discussed later.
Mediation analyses with and without the covariates revealed that the effect of SPP on psychological resilience was significantly mediated by self-compassion, where SPP was associated with lower self-compassion which in turn was associated with lower psychological resilience.This finding is expected as studies have shown that individuals with high SPP engage in resiliencedepleting cognitions and behaviors (e.g., self-blame, rumination) and are unwilling to accept imperfections or failures, which decrease their ability to use effective coping strategies amidst stress and challenging situations (e.g., Hewitt & Flett, 2002;Klibert et al., 2014).Similarly, the indirect effect of SOP on psychological resilience through self-compassion was significant.This finding is consistent with Stoeber et al.'s (2020) study that found SOP to negatively predict SWB via low self-compassion but contradicts Choo and Prihadi's (2019) view that individuals with high perfectionistic strivings, which typically capture SOP aspects, experienced more self-compassion, leading towards more resilience.The direct effect of SOP on psychological resilience, after controlling self-compassion, was also significant in the positive direction, suggesting that self-compassion may be one of the many mechanisms through which SOP influences resilience.For instance, noting the findings of Stoeber et al. (2020), it is possible that SOP can both negatively predict resilience via low self-compassion and positively predict resilience via high compassion for others.Having mechanisms with opposing effects on resilience can also confirm that SOP is an ambivalent dimension of perfectionism with positive and negative associations with wellbeing outcomes (Enns & Cox, 2002).
It is interesting that OOP did not significantly influence psychological resilience indirectly via self-compassion without the covariates in the mediation model, but this mediation became significant with the inclusion of the covariates.These findings can primarily be attributed to the effects of the COVID-19 perceived stress covariate in the analyses.In particular, the nonsignificant indirect effect was observed as OOP had a negative correlation with the said covariate (N = 317, r = −.08,p = .138)as compared with the significant positive correlations found for SOP (N = 317, r = .20,p < .001)and SPP (N = 317, r = .32,p < .001).The relationship between OOP and COVID-19 perceived stress is not unexpected: as OOP has been associated with narcissism, characterized by having a greater sense of entitlement and being less interested in helping and supporting others (Stoeber, 2014), it can be inferred that individuals with high OOP may devalue COVID-19 information and protocols (Coleman, 2020) and thus may not experience remarkable distress during the pandemic.Consequently, the indirect effect of OOP on psychological resilience through selfcompassion already became statistically significant as the negative correlation between OOP and self-compassion increased in magnitude and became statistically significant after controlling the COVID-19 perceived stress covariate in the mediation analysis.The direct effect of OOP on psychological resilience, after controlling self-compassion, was also significant in the positive direction, suggesting that other mechanisms are equally important in understanding how OOP affects resilience.For instance, some studies like Stoeber (2014) have shown OOP to be a significant positive predictor of the Dark Triad (i.e., narcissism, Machiavellianism, and psychopathy), which may explain OOP's positive relationship with wellbeing outcomes (Birch et al., 2019).Just like for SOP, understanding other mediating pathways with opposing effects on resilience can also provide clarity to OOP's ambivalent nature (Stoeber, 2014).This study's findings emphasized the benefits of self-compassion in improving wellbeing among Filipino university students, particularly those who identify themselves as perfectionists.Noting the alarming increase of perfectionism (Curran & Hill, 2019) and the rise in cases of negative mental health outcomes or diagnoses among the students (Interactive Learning Center Diliman, 2020), universities need to address these concerns by promoting selfcompassion exercises and interventions.For instance, university mental health professionals can provide cognitive-behavioral therapies and mindfulness-based interventions, with online or Internet-delivered versions of these interventions being the most feasible as the COVID-19 pandemic is still lingering (James & Rimes, 2018;Zetterberg et al., 2019).Faculty can also integrate in their courses or academic curricula mindfulness-based exercises like breathing meditations and diary journals (Klibert et al., 2014;Vidic & Cherup, 2019), which can be given at the beginning of the students' university life when they face the challenge of transitioning from high school and adapting to a new learning and social environment (Robotham & Julian, 2006).Such exercises and interventions can help students reduce or disrupt their ruminative thinking, increase their distress tolerance and emotional regulation, and build on their resilience towards having a better wellbeing.Apart from universities, families can also help in promoting self-compassion among perfectionist students given the criticality of their roles in the development of children functioning.Cook (2012), for example, suggested a familyoriented approach to treating child perfectionism.Parents can be educated on how their cognitions and behaviors impact their children's perfectionism, give love and support to their children while remaining flexible with the rules and standards they set, and guide their children towards acceptance of mistakes and failures by exemplifying these as necessary parts of learning and growth.Some limitations are noted for this study.First, this study adopted a cross-sectional design and thus could not make causal interpretations about the relationships between the variables of interest.Future studies can adopt a longitudinal design to establish consistency of results over time.Second, the sample consisted of students exclusively from UPD. Caution must then be taken into account when generalizing the findings to other Filipino university students (e.g., other UP campuses, non-Metro Manila universities) and non-student Filipino samples.Finally, as the responses to survey questions were self-reported, these may be prone to inaccuracies or response biases.You are free to: Share -copy and redistribute the material in any medium or format.Adapt -remix, transform, and build upon the material for any purpose, even commercially.The licensor cannot revoke these freedoms as long as you follow the license terms.
Under the following terms: Attribution -You must give appropriate credit, provide a link to the license, and indicate if changes were made.You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.

No additional restrictions
You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.

Figure
Figure 2. Model showing selfcompassion as the mediator in the relationship between socially prescribed perfectionism and psychological resilience.Note.Regression coefficients are standardized.Coefficients without the parentheses are for the model without the covariates (N = 316), while coefficients inside the parentheses are for the model with the covariates (N = 300).***p < .001

©
2023 The Author(s).This open access article is distributed under a Creative Commons Attribution (CC-BY) 4.0 license.
and dialog with, expert editors and editorial boards • Retention of full copyright of your article • Guaranteed legacy preservation of your article • Discounts and waivers for authors in developing regions Submit your manuscript to a Cogent OA journal at www.CogentOA.com

Table 1 . Descriptive statistics of participant characteristics I Participant Characteristic n %
This variable is different from year level-for example, a student has been in the university for 4 years but is currently in their third year standing in the program.b Participants came from 60 different undergraduate programs.This table only enumerates the undergraduate programs, where more than 1% of the participants came from.BS = Bachelor of Science; BA = Bachelor of Arts; B = Bachelor.
a Current standing in the program is a designation for the student's progress towards their graduation goal.

Table 2 . Descriptive statistics of participant characteristics II Participant Characteristic M SD Range
a