The IP was first conceptualized in the clinical field, where Clance and Imes (1978) worked with highly successful women who held prestigious PhDs, were respected experts in their fields, or excellent students – but nevertheless constantly feared getting unmasked as intellectual impostors. Clance (1985a) described six components characterizing the IP: (1) The Impostor Cycle is a vicious circle of unfavorable learning and attribution styles. To deal with challenges, individuals experiencing the IP either procrastinate and balance that by excessive preparation or immediately prepare excessively. When they are successful with these learning styles, they attribute their success to luck or effort. They believe their success is due to those factors rather than their abilities. Consequently, IP sufferers never feel prepared for the next challenge and hence the impostor cycle starts over again. (2) IP sufferers feel an urgent need to be special, i.e., to not only master challenges but to be outstanding. These unrealistically high expectations trigger self-doubts and explain their reduced perception of capability, despite objective evidence suggesting the opposite. (3) Superwoman/Superman aspects indicate that IP sufferers do not only expect themselves to be outstanding, their effort must also be negligible. This accounts for self-doubts and feelings of fraudulence, especially when IP sufferers cannot live up to their unrealistic expectancies. (4) Fear of failure is fostered by IP sufferers’ fear of getting unmasked as unintelligent, resulting in desperate efforts to avoid any kind of failure. (5) Individuals suffering from the IP also typically deny competence and discount praise. They rely on alternative explanations for their success aside from their abilities. (6) Fear of success is grounded in IP sufferers’ doubts about their ability to repeat success and live up to increased expectations of others.

From this description, it is apparent that the IP integrates several usually isolated constructs such as self-concept, attribution, emotion, and achievement motivation that have a strong impact on educational careers. Therefore, in recent years, the IP has been increasingly proposed as an important psychological factor for career development. The IP has been found to reduce career adaptability resources that are positively related to career planning, career exploration, and occupational self-efficacy and negatively related to career decision-making difficulties (Neureiter and Traut-Mattausch, 2017). Previous research points to the IP as a relevant psychological barrier particularly for young academics for several reasons: First, the university context is structured in a way that enhances the IP (McCormick and Barnes, 2008; Klinkhammer and Saul-Soprun, 2009) as it is characterized by many years of evaluation and examination rituals, evoking feelings of deficiency. Due to intense competition, normal crises like writing blocks or motivational problems are rarely communicated and feelings of inadequacy are often kept secret. In addition, constant adaptation to new role expectations coming with new challenges promotes impostor feelings and provides opportunities to get trapped in the impostor cycle. Moreover, the myth of the ingenious scholar with no need for recreational time aligns to superwoman/supermen aspects experienced by IP sufferers and puts impossible demands on young academics (Macha, 1992). Second, the IP is associated with numerous factors that interfere with academic achievement, such as lower self-esteem (Chrisman et al., 1995; Thompson et al., 1998; Sonnak and Towell, 2001), lower research self-efficacy (Jöstl et al., 2012), lower academic self-concept (Leary et al., 2000; Cokley et al., 2015), lower performance expectancies (Cozzarelli and Major, 1990), and self-perceptions of inadequacy and lack of academic preparedness (Craddock et al., 2011). It is further associated with less favorable achievement goals (Kumar and Jagacinski, 2006) as well as higher depression and anxiety (Bernard et al., 2002; McGregor et al., 2008), higher fear of success (Fried-Buchalter, 1997), and higher fear of failure (Thompson et al., 2000). Third, level of faculty rank is negatively related to the IP, indicating that academics suffering from the IP may be likely to drop-out of academia during the early stages of their careers (Topping and Kimmel, 1985). The IP is thus theoretically linked to academic attrition. Central components of the IP such as feelings of intellectual inadequacy, negative emotions, e.g., fear of failure, fear of success, or anxiety, and negative self-perceptions, e.g., self-doubts, self-criticism, or denial of competences are associated with increased drop-out intentions of university students (Szulecka et al., 1987; Langford and Clance, 1993; Bask and Salmela-Aro, 2013; Cortes et al., 2014). This makes especially students suffering from the IP vulnerable to drop-out of their academic programs. The first year at universities is particularly crucial for academic attrition and might be a fertile soil for the IP (e.g., Szulecka et al., 1987; Baars and Arnold, 2014). Adjusting to this unfamiliar learning environment is highly demanding and distressing for first-year undergraduates, especially for those who experience impostor feelings.

From their experiences in clinical treatment, Clance and Imes (1978) conceptualized the IP as a distinctly female experience. They concluded that women are more likely than men to attribute the reason for success outward – either to an external cause (e.g., luck) or a temporary internal quality (e.g., effort) – than to inherent ability. They explained this pattern by internalization of the societal gender stereotype that women are not considered competent (Furnham et al., 2002; Bian et al., 2017). Therefore, high-achieving women must find alternative explanations for their success. However, empirical findings of gender differences in the IP are equivocal: In college and university students, some studies found that women are more affected by the IP than men (King and Cooley, 1995; Kumar and Jagacinski, 2006; McGregor et al., 2008), while others did not find gender differences (Cowman and Ferrari, 2002; Cokley et al., 2013). In faculty members, Jöstl et al. (2012) found higher impostor scores in women, while Topping and Kimmel (1985) found the contrary and Hutchins (2015) found no gender differences. These studies investigated gender differences by simply comparing men and women. However, when examining gender differences, researchers should also consider the extent to which individuals identify with typical characteristics of masculinity and femininity, i.e., their gender-role orientation. Bem (1974) suggests that psychological masculinity and femininity are independent dimensions and that individuals can be high or low on each dimension, irrespective of their gender. Masculinity has been associated with achievement-oriented traits, labeled as agency or instrumentality, whereas femininity has been associated with social-oriented traits, designated as communion or expressivity (Kite et al., 2008). Spence et al. (1975) suggest that androgyny, i.e., a high degree of both, masculinity and femininity, is most desirable as it integrates strength of both dimensions. Possessing few characteristics of femininity and masculinity, i.e., an undifferentiated gender-role orientation, is thus less desirable and associated with less effective functioning compared to the other three types of gender-role orientation. Bem (1974) argues, that the adoption of either gender-role orientation can over-ride one’s gender regarding its impact on psychological functioning. Nevertheless, only limited research has addressed the relationship between gender-role orientation and the IP to date. September et al. (2001) found that students with feminine and undifferentiated gender-role orientation reported higher levels of the IP and lower levels of well-being than those with masculine or androgynous gender-role orientation.

Overall, research supports the assumption that the IP affects academic careers. Imposter experiences cause feelings of intellectual and professional incapability despite objective evidence to the contrary. They entail negative emotions and motivational aspects that interfere with academic achievement and are associated with academic drop-out. Although the IP was originally conceptualized as a distinctly female experience, findings of gender differences in the IP are equivocal and research on gender-role orientation and the IP is sparse. Therefore, we examine gender differences in the IP by focusing on gender and gender-role orientation simultaneously. This approach allows us to investigate the influence of gender and gender-role orientation on the IP, as well as their interaction.

Research Question 1: How are gender and gender-role orientation related to the impostor phenomenon?

Finding resilience factors and interventions against the IP may contribute to preventing young and promising academics from attrition. In our view, self-compassion – an important Buddhist concept becoming increasingly popular in Western psychology – might be a promising psychological resilience factor against the IP. Psychological research on self-compassion is relatively new, but highly relevant to researchers interested in self-concepts and self-attitudes in relation to well-being. In contrast to self-esteem, which is built on evaluations of self-worth and constituted by judgments and comparisons (Crocker and Park, 2004), self-compassion is related to compassion (Neff, 2003a). Compassion entails being empathetic to the suffering of others, and offering non-judgmental understanding to persons who fail or do wrong, so that their actions and behaviors are seen as part of the larger human experience (Gilbert, 2005). Self-compassion therefore involves being sensitive to one’s own suffering and offering non-judgmental understanding to one’s own pain, inadequacies and failures, so that one’s experience is perceived in the context of shared human fallibility (Neff, 2003a,b). Three basic components constitute self-compassion (Neff, 2003a): (1) Self-kindness, i.e., being kind and understanding to oneself rather than judgmental, (2) common humanity, i.e., recognizing that all humans are imperfect, fail, and make mistakes and seeing one’s experiences as part of the larger human experience instead of separating or isolating, and (3) mindfulness, i.e., being aware of one’s present experience in a balanced manner rather than over-identifying with negative aspects of one’s life. The basic components of self-compassion are conceptually distinct and are experienced differently, but also interact by mutually enhancing and engendering one another. On the one hand, a certain amount of mindfulness is needed to keep enough mental distance from negative experiences to allow feelings of self-kindness and common humanity. On the other hand, mindfulness directly contributes to the other two components by reducing self-criticism and increasing self-understanding, thus enhancing self-kindness and by antagonizing egocentrism, entailing feelings of isolation, thus increasing feelings of interconnectedness and common humanity.

Self-compassion is negatively related to depression, anxiety, and stress (Barnard and Curry, 2012; MacBeth and Gumley, 2012), and facilitates resilience by moderating reactions of individuals to negative events. Moreover, self-compassion is linked to increased motivation, and resilient coping (e.g., Allen et al., 2012; Breines and Chen, 2012; Sbarra et al., 2012). According to Neff et al. (2005), students’ self-compassion is positively associated with mastery goals and negatively associated with performance goals. This relationship is mediated by the greater perceived competence and lesser fear of failure of self-compassionate individuals. In case of perceived academic failure, self-compassion is positively related to emotion-focused coping strategies and negatively related to avoidance-oriented strategies. These motivational aspects associated with self-compassion are advantageous in academic contexts.

Evidence suggesting positive associations between self-compassion and well-being set the stage for the development of several interventions tailored to enhance self-compassion. The general therapeutic approach compassion-focused therapy (CFT), helping clients to develop a self-compassionate mind, especially when suffering from shame and self-attack (Gilbert, 2005), or the mindful self-compassion (MSC) program, which significantly increases self-compassion and life satisfaction while reducing depression, anxiety, and stress (Neff and Germer, 2013) are only two out of numerous self-compassion interventions. Moreover, various self-compassion guided meditations and exercises can be accessed freely online (e.g., Neff, 2017). These low-threshold interventions support individuals in developing a new way of relating to themselves by increasing self-compassion.

High-achievers suffering from the IP – an intense, subjective self-perception of phoniness – believe that their success is due to mistake, and live in constant fear of getting unmasked as intellectual impostors. Therefore, there is a clear theoretical link between the IP and self-compassion. IP sufferers are harshly critical and judgmental with themselves depicted in superman/superwomen aspects, their denial of competence, and discount of praise, i.e., they lack self-kindness. They think that they are different from others which is illustrated by their need to be special. They feel that only they are imperfect, fail, and make mistakes, entailing fear of failure and fear of success, i.e., they lack common humanity. IP sufferers further over-identify with negative aspects of themselves, trapping them in the impostor cycle, i.e., they lack mindfulness. Therefore, self-compassion seems to be a promising resilience factor against the IP that can be enhanced by therapeutic interventions. Even though self-compassion is theoretically related to the IP and appears to be a promising resilience factor for the IP, these constructs have not been linked in research to date. Therefore, we aim to bridge this research gap by examining the relation between self-compassion and the IP.

Research Question 2: How is self-compassion related to the impostor phenomenon?

In our study, we aim to gain new insights into psychological barriers for female university careers by focusing on gender, the IP, and self-compassion in first-year undergraduate students. Theoretical considerations suggest negative associations between self-compassion and the IP. Building on equivocal findings of gender differences in the IP (Kumar and Jagacinski, 2006; McGregor et al., 2008; Jöstl et al., 2012), we apply a broader approach and take gender as well as gender-role orientation into account. Empirical research indicates positive associations between feminine gender-role orientation and the IP (September et al., 2001). Gender influences not only the IP but also self-compassion. Males were found to show higher levels of self-compassion than females (Neff et al., 2008; Neff and McGehee, 2010; Yarnell et al., 2015). Research addressing the relationship between self-compassion and gender-role orientation is sparse. Neff (2003b) has indicated that individuals with higher levels of self-compassion are more likely to report that they are equally kind toward themselves as they are to others, whereas individuals with lower levels of self-compassion are more likely to report an imbalance in this context. Therefore, women and men who identify strongly with either masculinity or femininity, are likely to experience low levels of self-compassion for different reasons. Individuals with feminine gender-role orientation prioritize relationships over the self, while individuals with masculine gender-role orientation prioritize the self over relationships. Tatum (2013), however, found positive associations of masculinity and femininity with self-compassion, with masculine gender-role orientation accounting for the greatest proportion of variance in self-compassion. Therefore, the relationship between gender-role orientation and the IP might be mediated by self-compassion. Arguing that self-compassion might be a resilience factor against the IP, we assume that self-compassion mitigates the IP, and consequently reduces the strength of the relationship between gender-role orientation and the IP.

Research Question 3: Is the relationship between gender-role orientation and the IP mediated by self-compassion?

An email invitation to participate in this study was sent to all first-year undergraduates enrolled at four of the five largest universities in Vienna, Austria. We focused on undergraduate students in their first year as the IP is not well researched in this sample, even though impostor experiences are theoretically linked to drop-out intentions and first-year undergraduates are especially vulnerable to academic attrition (e.g., Clance, 1985b; Szulecka et al., 1987; Baars and Arnold, 2014). In this early stage of academic careers, we thus expect that IP experiences begin to unfold which offers opportunities for early interventions. 945 students volunteered to participate in this study, of which 72% completely answered the questionnaire. As high achievement is a prerequisite for the IP, exclusively high-achievers (GPA > 3.0 in the school leaving examination) were included in the final sample (n = 459). Participants’ ages ranged from 17 to 65 years (M = 21, SD = 5.60) and 69% of participants were female, which is representative for this population (Bundesministerium für Wissenschaft, Forschung und Wirtschaft [BMWFW], 2017).

The present study was designed as a cross-sectional online survey. An online consent form informed participants about duration, procedure, and goals of this study. Participants were guaranteed anonymity and confidentiality of their data and were informed that participation was voluntary and could be withdrawn at any point of the questionnaire. After completing the informed consent form, participants answered an online questionnaire spanning demographics (gender, age, and affiliation), GPA, and measures of the IP, gender-role orientation, and self-compassion.

To minimize bias, measures of the IP, gender-role orientation, and self-compassion as well as corresponding items implemented in the questionnaire were presented in random order. We used a 5-point rating scale ranging from 0 = not at all true to 4 = very true for all measures.

The IP was assessed using the German version of the Clance Impostor Phenomenon Scale (CIPS; Klinkhammer and Saul-Soprun, 2009), a 20-item scale (Cronbach α = 0.90) representing components of the IP (sample item: “Sometimes I’m afraid others will discover how much knowledge or ability I really lack”). Higher scores indicate higher levels of the IP. Impostor experiences were grouped into four levels (few, moderate, frequent, and intense) based on the magnitude of the overall score applying the cut-off scores proposed by Clance (1985b).

Gender-role orientation was measured using a short version of the Bem Sex Role Inventory (BSRI; Bem, 1974) in German language, based on Schertzer et al. (2008). A confirmatory factor analysis (CFA) of this 12-item scale was conducted. The two factors represent the subscales feminine and masculine gender-role orientation and fit the data reasonably well [χ²(53) = 271.76, p < 0.001], see Table 1 for factor loadings. Factor loadings were transformed to the same scale by dividing feminine and masculine loadings by their sums. Cronbach’s α of the subscales was α_fem = 0.78 and α_mas = 0.75. Sample items are “I am understanding” (femininity) and “I am ambitious” (masculinity). Using the median-split method (Spence and Helmreich, 1972), four groups of gender-role orientations were obtained as illustrated in Table 2: feminine, masculine, androgynous, and undifferentiated gender-role orientation.

Hupfeld and Ruffieux (2011) German version of the Self-Compassion Scale (SCS-D) was used to assess self-compassion. SCS-D is a 26-item scale (Cronbach α = 0.89) including six subscales that contrast positive and negative characteristics of the three components of self-compassion: self-kindness (sample item “I try to be understanding and patient toward aspects of my personality I don’t like”) vs. self-judgment (sample item “When I see aspects of myself that I don’t like, I get down on myself”’), common humanity (sample item “I try to see my failings as part of the human condition”) vs. isolation (sample item “When I fail at something that’s important to me, I tend to feel alone in my failure”), mindfulness (sample item “When something upsets me I try to keep my emotions in balance”) vs. over-identification (sample item “When something painful happens I tend to blow the incident out of proportion”). Internal consistencies for the subscales were good, with α_sk = 0.88, α_sj = 0.88, α_ch = 0.89, α_i = 0.88, α_m = 0.88, and α_oi = 0.88. Items from subscales addressing negative characteristics were reverse coded so that higher scores indicate higher levels of self-compassion.

We conducted an ANOVA to examine differences in the IP between individuals of different gender and gender-role orientation, see Table 3 for the distribution of types of gender-role orientation in male and female participants. Gender with the categories female and male and gender-role orientation with the categories undifferentiated, feminine, masculine, and androgynous were included as independent variables. Results indicate that gender and gender-role orientation were both statistically detectably associated with the IP [F(1,451)_gender = 7.54, p = 0.006, ηg_gender² = 0.02; F(3,451)_GRO = 6.38, p < 0.001, ηg_GRO² = 0.04]. No statistically detectable interaction effect was found [F(3,451)_{GRO^∗gender} = 0.36, p = 0.76]. Results indicate that male students suffered less intensely from the IP than female students, see Table 4. The relationship between gender-role orientation and the IP is illustrated in Table 5. Masculine (M = 34.3, SD = 15.5) and androgynous students (M = 36.8, SD = 16.4) suffered less intensely from the IP than feminine (M = 39.8, SD = 14.1) or undifferentiated students (M = 42.6, SD = 15.4). Post hoc pairwise t-tests with Bonferroni corrected p-values revealed statistically detectable mean differences between participants with masculine and feminine gender-role orientation (p = 0.03), between participants with masculine and undifferentiated gender-role orientation (p < 0.001), as well as between participants with androgynous and undifferentiated gender-role orientation (p = 0.03). No statistically detectable differences in the IP were found between students with feminine and undifferentiated gender-role orientation as well as between students with masculine and androgynous gender-role orientation (p > 0.05). This indicates that four categories of gender-role orientation are too differentiated. We thus used the scores on the two factors femininity and masculinity specified by CFA (reported above) for further analyses.

To analyze the relationship between self-compassion and the IP, Pearson’s product moment correlation, Welch’s t-tests, and ordered logistic regression were conducted.

Pearson’s product moment correlation indicates that self-compassion is negatively correlated to the IP with a large effect size (r = -0.55, p < 0.001). Kolmogorov–Smirnov tests of the mean of self-compassion in each level of impostor experiences indicate that the requirement for the t-test of normal distribution in each group is met (D_few = 0.08, p = 0.83; D_moderate = 0.04, p = 0.96; D_frequent = 0.05, p = 0.82; D_intense = 0.11, p = 0.72). Conducting t-tests is thus a sensible choice. Welch’s t-tests were used to test for mean differences in self-compassion between every two consecutive levels of IP experiences, accounting for unequal variances. In doing so, three t-tests with Bonferroni correction were conducted, see Table 6 for results. Self-compassion differed statistically detectably between the four IP intensity levels with medium to large effect sizes. With increasing intensity of the IP, self-compassion decreases, see boxplots in Figure 1.

Ordered logistic regression was conducted with self-compassion as independent variable and the IP with four ordered categories (few, moderate, frequent, and intense) as the dependent variable. The resulting odds ration OR = 0.11 fell in the 95% confidence interval CI95% [0.07,0.16] and is thus statistically detectable. Results indicate that the likelihood of high self-compassion to occur in high levels of the IP is 1/10.

Additionally, Pearson product moment correlations of the self-compassion subscales revealed medium negative correlations between the IP and self-kindness (r_sk = -0.29, p < 0.001) as well as mindfulness (r_m = -0.18, p < 0.001). The negative subscales self-judgment, over-identification, and isolation were highly positively correlated with the IP (r_sj = 0.51, p < 0.001; r_oi = 0.53, p < 0.001; r_i = 0.69, p < 0.001). Common humanity was the only subscale without a statistically detectable correlation with the IP (r_ch = 0.09, p = 0.07).

Path analysis was conducted with the R package sem (Fox et al., 2015) to examine our theoretical model spanning gender-role orientation, self-compassion, and the IP. We hypothesized that the relationship between feminine and masculine gender-role orientation (independent variables) and the IP (dependent variable) is mediated by self-compassion.

Following Zhao et al. (2010) recommendation, we estimated 95% confidence intervals using bootstrapping in addition to the point estimates obtained by path analysis. Three path analyses were conducted for (a) the entire sample (n = 459), (b) exclusively female students (n = 315), and (c) exclusively male students (n = 144). The model fits the data reasonably well for all samples [χ²(1,459)_a = 0.04, p = 0.84; χ²(1,315)_b = 0.84, p = 0.36; χ²(1,144)_c = 0.71, p = 0.40]. Results are presented in Figure 2. For an overview of mean values of the IP and self-compassion among gender and gender-role orientation see Table 7.

When all participants were included in the model, all effects were statistically detectable. Feminine and masculine gender-role orientation was positively associated with self-compassion whereas masculine gender-role orientation was associated with self-compassion to a greater extent (de_{fem × SC} = 0.13, p = 0.006; de_{mas × SC} = 0.22, p < 0.001). Self-compassion was further highly negatively associated with the IP (de_{SC × IP} = -0.54, p < 0.001), confirming previous results. As predicted, feminine gender-role orientation was positively associated with the IP (de_{fem × IP} = 0.11, p = 0.006) whereas masculine gender-role orientation was negatively associated with the IP (de_{mas × IP} = -0.08, p = 0.04). Indirect effects of feminine or masculine gender-role orientation through self-compassion to the IP were statistically detectable (ie_{fem × SC × IP} = -0.07, CI95% [-0.01; -0.13]; ie_{mas × SC × IP} = -0.12, CI95% [-0.06; -0.19]), indicating that self-compassion mediates between gender-role orientation and the IP. Direct and indirect effects exist between gender-role orientation and the IP. Whereas for masculinity both effects are negative (i.e., complementary mediation), direct and indirect effects for femininity point in opposite directions indicating a competitive mediation effect (Zhao et al., 2010).

Applying the model to female and male students separately, effects are similar except for a non-statistically detectable direct effect between masculine gender-role orientation and the IP for female students (de = -0.03, p = 0.45), and feminine gender-role orientation and the IP for male students (de = 0.08, p = 0.27). Self-compassion statistically detectably mediated between feminine or masculine gender-role orientation and the IP in all three samples, confirming our hypothesis.

Three limitations of this study should be considered. First, our findings are based on self-report measures. It is thus possible that our data is biased due to response biases, e.g., social desirability. However, participation was anonymous and voluntary, and confidentiality of the data was assured. As this study was designed as an online-survey, participants chose a setting in which they felt comfortable answering the questionnaire. We thus assume that response bias is small and balanced by the large sample size. Second, the cross-sectional design of this study allowed us to gain new insights in the prevalence of the IP in first-year undergraduates, to enhance understanding of the leaky pipeline of female academic careers. The design of future studies should be tailored to the process characteristic of the leaky pipeline of female academics over time. A longitudinal design would be appropriate to grasp the nature of this phenomenon more comprehensively. Third, it should be considered that correlational findings do not imply directionality. Consequently, it is not clear if self-compassion is a resilience factor against the IP or if the IP prevents students from developing self-compassion. Thus, future research is needed to further examine our hypothesis that self-compassion is a resilience factor against the IP. A true experiment evaluating the effectiveness of an intervention designed to increase self-compassion for students experiencing different intensity levels of the IP would be an appropriate study design to further examine this research question.