Associations of self-compassion with shame, guilt, and training motivation after sport-specific daily stress – a smartphone study

ABSTRACT
 By applying a diary study design, we investigated the role of self-compassion during sport-specific daily stress (SSDS) with regard to the negative self-conscious emotions of shame and guilt and training motivation. We hypothesised that self-compassion would protect athletes from certain self-conscious emotions, namely shame, after SSDS. We also predicted that self-compassion would either increase or decrease the relationship between stress and motivation. Ninety-six athletes (Mage  = 22.14, SD = 5.92) reported their level of self-compassion and evaluated their trainings and/or competitions over three weeks in terms of experienced stress, guilt, shame and subsequent training motivation on their smartphones. Multilevel analyses showed that SSDS was associated with more negative self-conscious emotions and reduced training motivation. Moreover, self-compassion weakened the effect of SSDS on shame and was not correlated with training motivation. We discuss the results with regard to sport psychology practice and future research.

In the context of competitive sports, athletes experience situations that may cause stress, such as critical feedback from coaches and teammates, defeats in important games or criticism from the media (Benson & Bruner, 2018;Jones & Hardy, 1990;Oliveira et al., 2021). Some athletes report negative emotions and less motivation after experiencing stress (Kerr & Males, 2010)emotions and motivation being important factors for athletes' performance and wellbeing (Hanin, 2000;Roberts & Treasure, 2012). However, the extent to which stress affects athletes is not the same for everyone (e.g., Kaiseler et al., 2009;Perry et al., 2015), and how people treat themselves seems to play a role in this process (Allen & Leary, 2010). One way of treating oneself in stressful moments is by being supportive and compassionate (i.e., self-compassion, Neff, 2003aNeff, , 2003b. Data from previous studies indicate that self-compassion has a positive effect on athletes, leading to fewer catastrophising thoughts, less negative affect and less rumination in response to hypothetical and recalled severe setbacks in sport (Mosewich et al., 2013;Reis et al., 2015). This study investigated the role of self-compassion in performance athletes' sport-specific daily stress (SSDS). SSDS refers to stress that is triggered by situations that are unique or intrinsic to the sport context and are common in a competitive sport setting. Examples of such situations include making simple mistakes in a training session, performing below one's own expectations and receiving negative feedback. More specifically, we investigated how SSDS affects certain self-conscious emotions (i.e., shame and guilt) and training motivation and examined the role self-compassion plays in these processes.
Self-compassion describes a kind, supportive attitude towards oneself and is defined by the presence of three positive aspects (i.e., self-kindness, common humanity and mindfulness) and the absence of three negative aspects (i.e., self-judgment, isolation, and overidentification, Neff, 2003aNeff, , 2003b. According to Neff (2003aNeff ( , 2003b, self-kindness means treating oneself with kindness and understanding even when feeling inadequate, wishing for one's own wellbeing and adopting an accepting, non-judgmental attitude toward oneself. Common humanity refers to the awareness that unpleasant experiences are shared by all human beings (Neff, 2003a(Neff, , 2003b. Mindfulness is a conscious awareness of one's own suffering, including psychological pain, and involves taking a balanced approach to unpleasant experiences so that painful feelings are not avoided or dramatised (Neff, 2003a(Neff, , 2003b. Self-judgment describes the tendency to display intolerance and self-condemnation in the face of one's own mistakes and difficult experiences, while isolation refers to the feeling of being separated from other people in the face of suffering and weaknesses (Neff, 2003a(Neff, , 2003b. Over-identification describes the tendency to be carried away and overwhelmed by negative experiences (Neff, 2003a(Neff, , 2003b. Given the constant evaluation competitive athletes experience in their lives (Mellalieu et al., 2009), this study investigated the effects of self-conscious emotions, namely shame and guilt. These self-conscious emotions play a role in self-evaluation processes and are a form of internal feedback that a specific goal, expectation or standard has not been achieved (Leary & Tangney, 2003). Shame arises from a negative assessment of the whole self and can therefore be particularly devastating (Dearing & Tangney, 2002). Guilt arises when a specific behaviour is evaluated as a violation of perceived social norms and expectations and therefore usually leads to more adaptive behaviour than shame (e.g., reparative behaviours, Dearing & Tangney, 2002). If an athlete attributes poor performance to their own ability and skills in general, this is more likely to lead to shame, whereas an athlete who attributes poor performance to a lack of effort is more likely to feel guilt (Tracy & Robins, 2006). Research suggests that these two self-conscious emotions should be considered separately (Anolli & Pascucci, 2005). This suggestion is underlined by findings showing that self-compassion in athletes correlates negatively with shame proneness and unrelated to guilt proneness (Mosewich et al., 2011). Since self-compassion offers protection against harsh self-criticism and a global negative evaluation of the self (Neff, 2003b), it is likely that athletes with high self-compassion will feel less shame after SSDS than athletes with low self-compassion. On the other hand, feelings of guilt do not arise based on an overall evaluation of the person, but instead refer to a specific behaviour. Therefore, we do not assume that self-compassion affects the relationship between SSDS and guilt. To the best of our knowledge, no research has yet investigated how self-compassion affects the emotions athletes actually feel after SSDS.
Besides negative self-conscious emotions, this article examines another central factor in competitive sports, namely motivation. More precisely, we investigated the effect of SSDS on athletes' motivation for the following training session. Two hypotheses can be put forward regarding the role of self-compassion in this process: (1) a self-compassionate stance prevents a decrease in training motivation as a response to SSDS, and (2) a selfcompassionate approach to stress leads to reduced motivation for the following training session. The first hypothesis is supported by experimental data from the academic context showing that reminding student participants to be self-compassionate about a test failure led them to spend more time studying for a subsequent test (Breines & Chen, 2012). One possible explanation for this result is that self-compassion could create an internal atmosphere of warmth and understanding, which encourages people to confront their mistakes and weaknesses without either self-deprecation or defensive self-enhancement (Neff, 2003b). We assume that this process could occur in the same way in competitive athletes because both the academic and the sport domains share some commonalities, such as high expectations and performance orientation and, therefore, appear comparable. Accordingly, self-compassion would also offer protection from a decrease in training motivation after SSDS, as it helps athletes not to disengage after unpleasant experiences. Regarding the second hypothesis, which states that a self-compassionate approach to stress leads to reduced motivation for the next training session, existing qualitative research has shown that athletes expect self-compassion to have negative effects on motivational processes and thus assume that harsh self-criticism after a setback is needed to motivate themselves for the next training session (Sutherland et al., 2014). In other words, since self-compassion undermines this "kicking ass" function of self-criticism, the training motivation after SSDS events may also decrease. To the best of our knowledge, the relationship between self-compassion and training motivation after SSDS has not been investigated in the sport context.
This study investigates the effects of self-compassion on shame, guilt and training motivation after experiencing SSDS. We considered that stress experiences in training and competitions would lead to more shame and guilt and less training motivation, and we predicted that self-compassion would protect athletes from feeling shame (but not guilt) after SSDS. Based on the current diverging opinions on the effect of self-compassion on motivation, self-compassion might increase or decrease the stress-motivation relationship. An ecological momentary assessment (EMA) design using smartphones was applied, which made it possible to investigate athletes in their natural environment.

Participants and procedure
Participants were recruited via a mailing list and social media. All data were collected in Switzerland. Eligibility criteria were a minimum age of 16 years and being in possession of a smartphone for data collection. To capture a sufficient number of events in a period that is both reasonable for the athletes and does not act as a disincentive to participation, a minimum of four training sessions and/or competitions per week (e.g., three training sessions and one game) was required. The study was advertised as a diary study investigating the relationships between personality, stress in training and competition, motivation and emotions. Interested athletes could access a website where the study and participants' rights were described. After giving informed consent, participants provided demographic information and completed an online questionnaire assessing self-compassion. In total, 141 athletes completed the baseline questionnaire (M age = 21.52, SD = 5.88; 48% female). Subsequently, these athletes received detailed instructions via email on how to download and install the mquest app (cluetec GmbH, Karlsruhe, Germany) that we used to collect EMA data. All participants provided informed consent and the Institutional Review Board (internal ethics committee) approved the study.
We conducted event-based sampling over three weeks and asked the athletes to evaluate each training session and competition in terms of events that potentially caused stress. These events were subsequently assessed in terms of experienced stress. We also asked them the extent of shame and guilt they experienced and how motivated they were for the next training session. During the study, the participants received two reminders a day on their smartphones that the study was still running. After three weeks, the participants were informed that the study was finished and the app could be deleted. Upon completion of the study, all participants who provided EMA data were informed of the main hypothesis of the study, including an explanation of the self-compassion concept and a rough description of their individual self-compassion and SSDS scores.
To perform multilevel analyses, we included all athletes that rated at least two events, which meant the final sample consisted of 96 athletes (M age = 22.14, SD = 5.92; 55% female; M weeklytraininghours = 12.00, SD = 5.59, 26 different sports, 40% team sports). The three most common sports were soccer (21%), athletics (14%) and floorball (12%). The statement "I can keep up with the national champion in my sport" was used to assess the subjective performance level and resulted in the following answers: 37% absolutely true, 33% almost true, 14% somewhat true, 9% almost not true and 7% absolutely not true. Participants that only completed the online questionnaires and did not start or complete the EMA (i.e., dropouts) did not differ from the other participants regarding weekly training hours and self-compassion (ts < 1.59, ps > .11). However, there was a trend that dropouts were younger (t = 1.82, p = .07) and more likely to be male (t = 2.46, p < .02).

Measures
Questionnaire Self-compassion was assessed using the German version of the 12-item Short Form of the Self-Compassion Scale (SCS-SF, Hupfeld & Ruffieux, 2011;Raes et al., 2011). Items were rated on a five-point Likert scale (1 = strongly disagree to 5 = strongly agree). There is currently a debate concerning the evaluation of the SCS-SF and its long version (Muris & Otgaar, 2020;Neff, 2020), discussing whether an overall score should be considered or whether the positive and negative aspects of self-compassion should be analysed separately. To investigate the possibility of whether positive and negative aspects of self-compassion influence reactions to SSDS differently, we analysed the positive and negative aspects in addition to the SCS-SF total score in corresponding subscales. The total score of the SCS-SF (SCtot) was calculated using the mean of the 12 items after reverse scoring the six items of the negative aspects (i.e., self-judgment, isolation and overidentification; Cronbach's α = .78). The subscale containing the positive aspects of selfcompassion (SCpos) was calculated using the mean of the six items of the positive aspects (i.e., self-kindness, common humanity and mindfulness; Cronbach's α = .66). The subscale containing the negative aspects of self-compassion (SCneg) was calculated using the mean of the six items of the negative aspects (i.e., self-judgment, isolation and overidentification; Cronbach's α = .79). An example item for SCpos is "When I'm going through a very hard time, I give myself the caring and tenderness I need". An example item for SCneg is "I'm disapproving and judgmental about my own flaws and inadequacies".

Ecological momentary assessment measures
To assess SSDS, we asked the athletes how often (i.e., never, 1x, 2x, 3x, 4x, 5x or more) they experienced the following three events in the last training session or competition: (1) Others (e.g., coach, teammates, friends, etc.) pointed out things to me that I should be able to achieve according to my own standards; (2) I made a mistake that I should not make at my level; (3) I performed below my expectations. Provided it happened at least once, we asked how much stress that particular event had caused (0 = not at all to 7 = very much). If no event occurred, we set the value for stress to 0. Our final SSDS score was calculated using the mean of the three event-related stress scores. We assessed the current level of shame and guilt using single items and asked participants to what extent they felt ashamed and guilty, respectively (0 = not at all to 7 = very much). Finally, athletes rated their motivation for the next training session (0 = not at all motivated to 7 = very motivated).

Statistical analysis
We collected multilevel data, with repeated measures nested within individuals. This led to a two-level model with repeated measures at the first level and the individual athletes at the second level. We used maximum likelihood estimation (ML) and the R package "nlme" to run multilevel analyses. The alpha level to determine which results were significant was .05.
Between-subject predictors (i.e., SCtot, SCneg and SCpos) were centred at the grand mean. Perceived stress was not centred prior to the analyses because of the natural zero point of the scale. In the preliminary analyses, we tested the appropriateness of multilevel modelling using intraclass correlations (ICCs) for level-1 variables, as well as increases in model fit when adding random intercepts and random slopes. For our main analyses, we added the level-2 factors (i.e., SCtot, SCneg and SCpos) and the corresponding cross-level interactions to the model. The analyses were run for each level-1 variable (i.e., shame, guilt and training motivation) separately, once with SCtot only, and once with SCneg and SCpos.
The percentage of variability in the repeated variables attributable to between-person influences was 41% in SSDS, 32% in shame, 38% in guilt, and 43% in training motivation. These ICCs suggested that level-2 variance must be considered in the analysis of the present data.

Associations between self-compassion and diary measures
Separate multilevel analyses showed that SCtot was significantly associated with both self-conscious emotions (
The relationship between shame and SSDS is moderated by SCtot (Table 1) and SCneg, but not SCpos (Table 2). A simple slope test (Preacher et al., 2004) for this significant interaction between SCtot and SSDS revealed that among participants with low levels of SCtot (i.e., −1 SD), stress was significantly positively related to shame (slope = .627, z = 7.21, p < .001). Among participants with high levels of SCtot (i.e., +1 SD), there was also a significant, but weaker, positive relation (slope = .299, z = 4.34, p < .001; Figure 1). These findings are consistent with the results of the interaction between SCneg and SSDS (i.e., the higher SCneg, the stronger the stress-shame relationship). A simple slope test for this interaction revealed that among participants with high levels of SCneg (i.e., +1 SD), stress was significantly positively related to shame (slope = .534, z = 9.94, p < .001). Among participants with low levels of SCneg (i.e., −1 SD), there was also a significant, but weaker, positive relationship (slope = .292, z = 4.27, p < .001).
Table 1 (for SCtot) and Table 2 (for both SCpos and SCneg) show the results for the two models on guilt. As expected in both models, on level 1, the degree of SSDS predicted the degree of guilt. However, neither the level-2 trait variables nor the cross-level interactions were significant. The interaction between SCtot x SSDS is barely statistically significant (p = .057). The simple slope tests show a similar pattern to shame, namely, in participants with low levels of SCtot (i.e., −1 SD), stress was significantly positively related to guilt (slope = .525, z = 6.53, p < .001). For participants with high levels of SCtot (i.e., +1 SD), there was also a significant, but weaker, positive relationship (slope = .263, z = 2.80, p < .01). Note. *p < .05, **p < .01, ***p < .001. Figure 1. Cross-level interaction between self-compassion, shame, and sport-specific daily stress.

Discussion
This study investigated the relationship between self-compassion, negative self-conscious emotions and training motivation after SSDS in a sample of competitive athletes. The analysed data were collected directly after training sessions and competitions via smartphones for three weeks, which allowed examination of the experiences of athletes in their day-to-day sport context. We found that (1) SSDS was associated with more negative self-conscious emotions and reduced training motivation, (2) positive and negative aspects of self-compassion showed different associations with SSDS and self-conscious emotions, (3) self-compassion helped to protect athletes from feeling shame after SSDS, and (4) self-compassion was uncorrelated with the motivation to train. The discussion is structured based on these four main findings, and we discuss their scientific and practical implications in each section.
The first main finding of the study shows that the more that athletes experience SSDS due to failure or poor performance, the likelier they are to feel guilty and ashamed, and the more they report reduced motivation for the next training session. These findings suggest that even comparatively harmless events in training and competition can trigger stress, which in turn might have a negative effect on mood and training motivation. From a practical viewpoint, it is therefore important that coaches evaluate training sessions with their athletes and use appropriate debriefing techniques to cushion the blow of such negative effects (Hogg, 2002). In addition, coaches could act as role models for adequate stress management, for example, by being compassionate with themselves and with athletes (Annerstedt & Lindgren, 2014).
By evaluating the positive and negative aspects of self-compassion separately in addition to the total score using the respective subscales, we are able to make some differentiated statements about SSDS and the self-conscious emotions of shame and guilt. The second main finding of our study shows that the negative aspects and the total score, but not the positive aspects, of self-compassion were related to these three negative sensations. These results are in accordance with the results of Brenner et al. (2017), who reported positive connections between negative aspects of self-compassion and stress, anxiety and depression, while the positive aspects of self-compassion were only negatively related to depression. Thus, future studies could investigate whether the two aspects of self-compassion have differential or shared effects on certain/other outcomes. Such findings would help sport psychologists to determine whether to work on the negative or positive aspects of self-compassion or on both. Future studies could also use the long version of the Self-Compassion Scale, which consists of 26 items and six subscales, to investigate the influence of the six subcomponents of self-compassion separately (Neff, 2020).
The third main finding of our study shows that self-compassion buffered the effect of SSDS on shame. This buffer effect is explained by a low level of negative aspects of selfcompassion. These findings suggest that a reduction in negative aspects of self-compassion, such as strong self-condemnation, might lead to less shame after the experience of stress in sport. A reduction in shame could be beneficial for athletes because shame is associated with, for example, poorer wellbeing and lower self-esteem (Velotti et al., 2017). Notably, we found a very similar pattern for guilt as for shame, which suggests that shame and guilt are not as clearly separated as our hypotheses assumed. Future studies could investigate the extent to which self-compassion interventions with athletes reduce shame reactions and whether this, in turn, leads to corresponding changes in, for example, wellbeing. Future studies could also examine the influence of trait shame on coping with failure.
We assumed that self-compassion either cushions or enhances the negative effect of stress on training motivation. However, the fourth main finding from the data in this study shows that self-compassion does not influence training motivation (neither directly nor indirectly via changing the stress-motivation relationship). This finding indicates the athletes' concern that self-compassion is bad for training motivation is not justified. Nevertheless, self-compassion does not seem to protect athletes from a drop in training motivation after SSDS. In the repertoire of sport psychologists, self-compassion could be a valuable intervention. Thus far, no negative side effects have been shown, and there are various empirical indications of the positive effects of self-compassion (Röthlin et al., 2019). In addition to the potential protection from shame as reported here, selfcompassion interventions lead to, for example, increased wellbeing (Zessin et al., 2015) and flow (Carraça et al., 2018) as well as reduced somatic competition anxiety (Röthlin & Leiggener, 2021).
Our study is a first important step in investigating the role of self-compassion in dealing with the everyday stress of athletes. However, this study also has certain limitations. First, it only encompasses a short period, and more research is necessary to examine the long-term effects. Second, since we assessed SSDS, negative self-conscious emotions and training motivation at the same time, we could not ascertain the direction of causality among these variables. Third, the Cronbach's α value of the SCpos scale was slightly low. Finally, we did not cover all forms of SSDS and the situations we covered only triggered weak stress reactions overall. Future studies could investigate a wider range of sports-specific stressors that trigger stronger reactions. However, our study shows that even those situations that only trigger weak stress reactions were associated with more guilt and shame and less motivation, and that self-compassion buffered the effect of stress on shame. Future studies should attempt to replicate in and extend these findings to different countries, cultures and samples (e.g., adolescents, individual vs team sports, different sports, or performance levels).

Data availability statement
The data that support the findings of this study are available from the corresponding author, [PR], upon reasonable request.

Disclosure statement
No potential conflict of interest was reported by the author(s).