Coping with violations of educational expectations differs from other expectations, as educational expectations tend to be overly optimistic and particularly resistant to change (Carolan, 2017; Pinquart and Ebeling, 2020; Pinquart and Pietzsch, 2022). Being overly optimistic about one’s abilities and future educational outcomes carries some adaptive consequences that outweigh the inaccuracy of these expectations. Holding positive expectations reduces stress, supports physical and mental health, and increases motivation for exploration and innovation (Sharot and Garrett, 2016). Yet, overly optimistic educational expectations operate in the controversy between these adaptive advantages and disadvantages due to a more frequent need to cope with worse-than-expected outcomes. Therefore, coping with educational expectation violations can be conceptualized as a value-based process to adjust the expectation to embody the highest value (Sharot et al., 2021), without necessarily being a conscious decision (Panitz et al., 2021). Educational expectations tend to be tied closely to an individual’s self-concept (Suárez-Álvarez et al., 2014), which is why the highest accuracy is not always conducive to the most adaptive expectations.

Having optimistic beliefs and expectations about oneself and one’s future has positive effects on the individual’s self-concept (Iovu et al., 2018). If an individual receives expectation-violating information regarding their educational achievement, the advantages of the persistence of the prior expectation are–consciously or unconsciously–compared to the potential advantages of adjusting the expectation. If the advantages of the renewed expectation are perceived to be greater, individuals accommodate and change/adjust the expectation. However, if the integration of the expectation violation into the prior expectation is not seen as beneficial (e.g., because it threatens an individual’s self-concept), individuals immunize against discrepant feedback, and the expectations persists and the self-image remains (Greve and Wentura, 2010) or individuals assimilate and actively change their behavior to avoid future expectation violations. Accommodation tendencies increase when expectation change does not threaten essential parts of a person’s self or their worldview, whereas immunization and assimilation are adaptive to protect self-relevant concepts.

Disconfirming events can be better or worse than expected (Lebois et al., 2016). Contrary to better-than-expected events (positive valence), worse-than-expected events (negative valence) related to educational expectations can threaten individual’s self-concept and individual’s general preference to believe the future is bright rather than dark (Bromberg-Martin and Sharot, 2020). Thus, previous studies found a so-called optimism bias through asymmetric coping depending on the valence of the expectation violation: individuals showed stronger accommodation after better-than-expected events than after worse-than-expected events (Chowdhury et al., 2014; Garrett and Sharot, 2017; Lefebvre et al., 2017; Bromberg-Martin and Sharot, 2020). Adjusting expectations to a lower performance level would potentially be associated with more accurate judgments here, but would also result in the individual suffering from the change in expectation (Sharot et al., 2021). Therefore, individuals often assign more value to expectation persistence and–consciously or unconsciously–cope with immunization or assimilation. Asymmetry is also evident in seeking expectation-confirming information: Individuals are more likely to seek confirmation of positively valued expectations and avoid information seeking that may violate existing positive expectations or confirm negative expectations (Scherer et al., 2012). Even if expectation-violating information is very clear and trustworthy, expectations tend to remain unadjusted if change leads to undesirable outcomes (Eil and Rao, 2011; Kappes and Sharot, 2019). Instead, for affect regulation, individuals preferentially hold on to non-correct expectations associated with positive affect (Sharot and Sunstein, 2020).

The optimism bias is strongly dependent on motivational (motivation to maintain a positive and optimistic view of themselves; Sharot and Garrett, 2016) and emotional factors (feelings predict asymmetric outcomes; Charpentier et al., 2016) and already indicated an asymmetric coping with better-than-expected versus worse-than-expected feedback about educational achievements (Eil and Rao, 2011; Sharot and Garrett, 2016). Nevertheless, there were previous findings with limited evidence for asymmetric coping, because results indicated solely stronger assimilation after experiencing negative rather than positive expectation violations (Henss and Pinquart, 2022), but no differences in accommodation or immunization. Therefore, the influence of valence on coping with expectation violations should be further explored. Individuals seem to consider how coping affects internal states and emotions. A potential threat to the academic self-concept presumably evokes aversive internal states, so negative valence should lead to a stronger tendency toward expectation-persistent strategies. Contrary, positive valence of expectation violation should lead to stronger accommodation because it presumably evokes positive internal states.

Educational expectations are considered to be rather elaborate, certain, and stable constructs with particular significance for an individual’s understanding of the world as they are central elements of the self. Humans rely on their ability to structure information about the world into expectations, schemas, and rules that are simplified models of reality. But individuals differ in their preference for simple or complex models defined as need for cognitive closure (NCC) and therefore in the way they cope with disconfirming information (Webster and Kruglanski, 1994). Higher NCC trait levels should predispose individuals to ignore and resist expectation-inconsistent information (i.e., immunization, assimilation) in order to avoid expectation-discrepant outcomes and protect their models of the world (Neuberg and Newsom, 1993; Schrackmann and Oswald, 2014). This seems to be particularly important for self-relevant characteristics, such as educational expectations. Whereas former studies showed a fairly clear set of findings in which higher NCC was associated with less accommodation or less accommodation-like tendencies (e.g., Dijksterhuis et al., 1996), more recent studies showed strongly context-dependent effects of NCC on coping (Kemmelmeier, 2015; Strojny et al., 2016; Henss and Pinquart, 2022). Higher trait levels may be associated with both stronger assimilation and stronger accommodation because both strategies may reduce uncertainty under some conditions. People are both motivated to hold accurate beliefs and to adapt to changing circumstances but also to defend previously held beliefs and expectations (Schrackmann and Oswald, 2014). Therefore, individuals may sometimes prefer disconfirming information if it is diagnostically more relevant or has a higher utility compared with expectation-confirming information (Schrackmann and Oswald, 2014). On the other hand, individuals may actively search to confirm their beliefs (assimilation) or devalue discrepant information (immunization; Schrackmann and Oswald, 2014) if expectation maintenance is more advantageous (e.g., to academic self-concept) than potentially increasing the accuracy of the expectation.

Overall, this study relies on the assumption that coping processes after expectation violations are not only related to external outcomes, but also to internal states. The respective coping with violated educational expectations does not have to be aimed at producing the most accurate expectations possible, but can also be taken in favor of one’s own states and affects. Individuals evaluate the expected consequence of new information, often even unconsciously, and consider how this new knowledge will influence their psychological well-being when integrating it into their educational expectations. Resulting from this, the likelihood of each coping strategy is determined by costs (e.g., more uncertainty, acceptance of unpleasant self-relevant truths) and benefits (e.g., more accurate expectations). In addition to the situational aspect of valence of expectation violation, it should be noted that individuals have a dispositionally stable tendency as to how much they are willing to search for new information and, if necessary, to integrate it into existing concepts. These cognitive processes, as in NCC, may serve as an explanation for why individuals often do not change educational expectations despite disconfirming evidence (Kappes and Sharot, 2019). Accommodation is more likely when expectation change does not threaten people’s self-concept or the essentials of their view of the world (Kahan, 2017). In Sharot and Sunstein’s model (2020), it is evident that both valence with the associated affects and emotions (will the information induce positive or negative feelings, or have no influence on my affect?) and NCC with the associated cognitive processes (will information improve the ability to comprehend and anticipate my reality?) are crucial in coping with expectation violations (Sharot and Sunstein, 2020). Therefore, it seems reasonable that both predictors interact with each other in addition to their main effect on coping with expectation violations. Both predictors revealed significant but partly surprising results in a previous study (Henss and Pinquart, 2022) and will therefore be further explored in a similar study design. The following research questions were addressed: First, does a negative valence of expectation violation lead to stronger assimilation (similar to our previous study; Henss and Pinquart, 2022) as well as stronger immunization and weaker accommodation compared to positive valence of expectation violation? Second, does higher NCC predict stronger accommodation and assimilation (similar to the study by Henss and Pinquart, 2022) and stronger immunization (as found by Schrackmann and Oswald, 2014)? Third, do valence of expectation violation and NCC interact in predicting accommodation, assimilation, and immunization? Our study aims to shed light on the way in which the situational characteristic valence and the personal disposition NCC explain differences in coping after expectation violations.

Sample size planning via G*Power resulted in a required number of participants of n = 278 (MANOVA: Global Effects, calculated effect size f² = 0.04, alpha = 0.05, ß = 0.80, number of groups: 2, response variables: 3), based on the effect sizes in a comparable study by Henss and Pinquart (2022). Because of expected exclusions, we stopped recruitment after n = 297 participants completed the questionnaire. Participants were recruited via email distribution lists from different German universities. Inclusion criteria were a minimum age of 18 years, very good German language skills, and registration at a university. To compensate for their efforts, the participants could take part in a raffle for two 50€ value vouchers.

The study was conducted online with computer-based randomization to manipulate the valence of expectation violation as a one-factor between-subjects design. The personality variable NCC was assessed as a covariate and the coping strategies accommodation, assimilation, and immunization as dependent variables. Data collection took place from May to July 2022.

The study was advertised with a cover story as an investigation of students’ linguistic abilities in word riddles. The supposed aim of the study was to determine whether the linguistic abilities of academics differ significantly from those of non-academics either at the stage of graduation or at the stage of post-graduate professional practice. The participants were told that their linguistic abilities were determined by their performance in an anagram test. Participants’ demographic data and NCC were collected before participants were given an introduction to their task. To manipulate and induce an expectation violation, anagrams were used in which randomly arranged letters must be rearranged to form a word. Anagrams have been used as an effective means of violating performance expectations in a variety of previous studies (e.g., Boyes and French, 2010; Koppe and Rothermund, 2017). After introduction to the task, all participants completed an identical training task. Participants were told that they would complete a total of four runs of 11 anagrams each. In fact, only three runs were completed, but the misinformation was unavoidable to allow for a valid assessment of coping strategies after the third run.

To manipulate the valence of expectation violation, we systematically varied between the positive versus negative valence of expectation violation at three points: test-taking information, anagram solvability, and standardized performance feedback differed between the positive and negative valence groups, similar to the successful manipulation of valence in a previous study (Henss and Pinquart, 2022).

To calculate main statistical effects of the manipulated variable valence and the personality trait NCC, as well as their interaction with respect to the three coping strategies, a MANCOVA was performed for the overall model with valence as an independent variable and NCC as covariate as well as their interaction. The MANCOVA was calculated at a significance level of 5%.

We checked data of all n = 297 participants for univariate outliers via Box-Whisker plots and for multivariate outliers via Mahalanobis Distance (p < 0.001). Whereas univariate outliers were checked for plausibility and if necessary, only excluded pairwise, we excluded three data sets from further analysis because they were identified as multivariate outliers. Furthermore, 26 data sets of participants were excluded because the participants had either detected the experimental manipulation or had not experienced an expectation violation. The former was indicated qualitatively by a free response field at the end of the study, where participants could indicate what they thought the background of the study was. Here, exclusion occurred when subjects indicated that the study goal was to analyze coping with expectation violations and additionally indicated that they did not find the manipulation credible (e.g., „I assume that the feedback was not related to my actual performance, but to investigate coping with violated expectations“). The second was indicated quantitatively by specifying that participants in the negative valence group had the expectation of solving at least six anagrams and that participants in the positive valence group had the expectation of solving a maximum of six anagrams.

The final sample consisted of n = 268 participants. The participants were mainly young adults (M = 23.87, SD = 4.98), female (n_female = 215, n_male = 73, n_diverse = 9), bachelor students (n_bachelor = 204,n_master = 93) in the fields of psychology (n = 108), social sciences (n = 83), natural sciences (n = 63), teaching (n = 22) or others (n = 25).

To verify that expectations differed between groups, we compared the reported expectations before the manipulation of the valence of expectation violation. There should be significantly lower expectations in the positive valence group compared with the negative valence group. Independent t-tests revealed that participants’ expectations differed significantly between the groups (t₁(266) = −10.54, p < 0.001; t₂(215) = − 25.14, p < 0.001; t₃(266) = − 35.14, p < 0.001) and that the difference between the groups increased with each run (M_1p = 5.31, SD_1p = 1.79 vs. M_1n = 7.66, SD_1n = 1.86; M_2p = 3.72, SD_2p = 1.14 vs. M_2n = 8.34, SD_2n = 1.78; M_3p = 3.30, SD_3p = 1.25 vs. M_3n = 8.88, SD_3n = 1.35). It can be concluded that the manipulation of the valence of expectation violation was successful, and significantly lower expectations were induced in the positive valence group compared with the negative valence group. Expectations solidified over time and, on average, corresponded in the run before the expectation violation to the expectation values that participants received in advance as information about prior performance.

To evaluate both the main effects of NCC and valence of expectation violation and their interaction effect regarding the dependent variables accommodation, assimilation, and immunization, we conducted a MANCOVA. The main effects analysis showed statistical significance for both valence and NCC. Valence of expectation violation was a significant predictor in the overall model (Wilk’s Λ = 0.96, F(3, 262) = 3.26, p = 0.02, η_p² = 0.03) and revealed a significant effect on accommodation (F(1, 264) = 9.66, p < 0.01, η_p² = 0.04) and assimilation (F(1, 264) = 4.19, p = 0.04, η_p² = 0.02). Positive valence of expectation violation was associated with less accommodation (M_p = 4.09, SD_p = 1.09 vs. M_n = 4.24, SD_n = 1.10), and negative valence with stronger assimilation (M_p = 3.95, SD_p = 0.93 vs. M_n = 4.23, SD_n = 0.88). However, there was no significant effect of valence of expectation violation on immunization (F(1, 264) = 0.10, p = 0.77; M_p = 4.28, SD_p = 0.94 vs. M_n = 4.69, SD_n = 0.84). The covariate NCC was also a significant predictor in the overall model (Wilk’s Λ = 0.97, F(3,262) = 2.99, p = 0.03, η_p² = 0.03). Here, statistically significant effects were found on accommodation (F(1, 264) = 6.99, p < 0.01, η_p² = 0.03) and assimilation (F(1) = 6.37, p = 0.01, η_p² = 0.02), but not on immunization (F(1, 264) = 0.07, p = 0.79, η_p² = 0.00). Individuals with higher NCC reported stronger accommodation and assimilation. As well, the interaction of the two independent variables valence and NCC was significant (F(3, 262) = 4.50, p < 0.01, η_p² = 0.05) and showed a statistical effect on accommodation (F(1, 264) = 11.76, p < 0.01, η_p² = 0.04) and assimilation (F(1, 264) = 6.85, p < 0.01, η_p² = 0.03), but again not on immunization (F(1, 264) = 0.17, p = 0.68, η_p² = 0.00). Whereas higher NCC was associated with slightly less accommodation in the positive valence group, the opposite was true in the negative valence group: the higher a participant’s NCC, the more accommodation was reported (see Figure 1).

With regard to assimilation, the same pattern emerged: in the positive valence group, participants with higher NCC reported slightly less assimilation, whereas in the negative valence group, with increasing NCC, more assimilation was reported (see Figure 2). Overall, analysis of the total MANCOVA model revealed medium effect sizes for accommodation (Wilk’s Λ = 0.95, F(3, 264) = 5.84, p = 0.001, η_p² = 0.06) and assimilation (F(3, 264) = 4.76, p = 0.001, η_p² = 0.06) and small effect size for immunization (F(3, 264) = 4.67, p < 0.01, η_p² = 0.05) according to Cohen’s taxonomy (Cohen, 1988).

Educational expectations tend to be overly optimistic and do not necessarily need to be accurate to be adaptive (Garrett et al., 2018). Expectation violations might be costly in certain situations, but advantageous in others when their benefits outweigh their costs (McKay and Dennett, 2009). This cost–benefit trade-off should be particularly important when considering the valence of the expectation violation and should lead to differences in coping: former studies indicated an optimism bias, because individuals integrated information into expectations asymmetrically based on the valence and therefore desirability of the information. We suspected that especially for self-relevant beliefs like educational expectations, individuals protect their academic self-concept through coping related to expectation maintenance after negative valence of expectation violation (Greve and Wentura, 2010) and positively adjust their academic self-concept through expectation update after positive valence.

Our first finding was not in line with an optimism bias: individuals reported less accommodation after an expectation violation with positive valence. This could possibly be related to the measurement of accommodation, which states, among others, that the respondent will try to form more realistic expectations in the future. However, individuals might have been encouraged by the positive feedback to be more optimistic, which would also be consistent with the theoretical approach of overly optimistic educational expectations.

Nevertheless, our second finding was in line with asymmetric coping after expectation violation and the result of our former study: individuals reported more assimilation after an expectation violation with negative valence. Thus, when individuals are confronted with a reality in which their positive achievement expectations are not met, they report active behavioral tendencies aimed at fulfilling their future expectations despite worse-than-expected present feedback. Increasing effort to compensate for a worse-than-expected performance can be considered as adaptive and in line with theoretical assumptions of self-concept protection and stability of educational expectations. Expectation update is more likely for uncertain expectations and less likely for certain and elaborated expectations such as educational expectations in university (Spicer et al., 2020).

It is nevertheless surprising that, after worse-than-expected feedback, individuals reported stronger assimilation, but not stronger immunization, which is also considered to be an expectation-maintaining strategy for self-concept protection (Greve and Wentura, 2010). The lack of significance could have both theoretical and methodological reasons: First, the dependence of the effect of negative valence on immunization on a high degree of expectation violation found in a previous study could explain the absence of the effect in this study (Henss and Pinquart, 2022). If the feedback in our study was not perceived as “expectation-violating enough” because we did not include information about a high degree of expectation violation, participants probably had no incentive to immunize. Performance feedback could be adjusted by clearly indicating that performance was strongly above or below expectations. Second, it should be noted that immunization as an unconscious and non-intentional process is difficult to capture by an explicit self-report measure. Recent literature suggests that immunization as an automatic process might be adequately assessed via indirect measures (Rief et al., 2022) or that the different facets of immunization might be captured via open questions and qualitative analysis (Kube et al., 2022).

Previous studies have suggested that the effects of NCC on coping are context-dependent and may promote both expectation update and expectation maintenance. Our results support this assumption and are perfectly in line with Sharot and Sunstein’s thoughts (2020): information can enhance or reduce individual’s view to comprehend their environment and disconfirming information challenges people’s existing models and schemata (Sharot and Sunstein, 2020). In the present study, simply ignoring the discrepancy between expected and actual achievement would not be the best way of coping for individuals with high NCC as they believed to participate in a fourth run immediately thereafter which could provide additional expectation violations. Therefore, immunization did not differ between individuals with higher vs. lower NCC. Individuals strive to make accurate predictions and therefore integrated performance feedback through expectation change or behavior adjustment to ensure that expectations are less likely to be violated in the future. Thus, adjustment of expectation in the direction of the formerly unexpected feedback (accommodation) or active behavior of creating a reality that conforms to prior expectations (assimilation) are more likely to fulfill their need for clarity and structure, and to the avoidance of uncertainty compared with denial or devaluation (immunization).

Accommodation is related to the improvement of existing expectations, schemata and models with new information to improve the fit between expectation and reality (Sharot and Sunstein, 2020). Stronger accommodation after disconfirming information leads to more comprehension of the world and, therefore, fulfills the needs of an individual with high NCC.

Assimilation is related to the active avoidance of information that is suspected to weaken the understanding of the world. Stronger assimilation after disconfirming information promotes a fit between the internal representation of the expectation and reality by actively changing the reality of which individuals are aware (Sharot and Sunstein, 2020).

Interestingly, in our study, accommodation and assimilation are not mutually exclusive, but positively correlated with each other (see also Henss and Pinquart, 2022). According to the ViolEx model, it is assumed that expectation violations may lead to accommodation (which can be both expectation update but also “only” expectation destabilization), which can in turn motivate stronger assimilation to restore confidence. Moreover, the combination of modest expectation adjustment and efforts to meet expectations may potentially be an adaptive approach to minimize the magnitude of future expectation violations. Individuals might accept a new reality, but nevertheless strive to meet the prior expectation. This conclusion seems especially plausible when considering the interaction of both predictors in the following paragraph.

As effects of NCC seem to be context-dependent and may be determined by the assigned value or advantage of beliefs, effects of NCC likely depend on the valence of expectation violation (i.e., will accepting a new reality be advantageous or will maintaining existing schemata despite less accuracy be advantageous?). The evaluation of information is a non-intentional, unconscious cognitive process and strongly sensitive to motivational influences like valence of the expectation violation (Schrackmann and Oswald, 2014). Indeed, there were differences between positive and negative valence: stronger accommodation and assimilation of individuals with high NCC were only found after experiencing a negative valence of expectation violation. No difference was found for positive valence, and this pattern was shown for both accommodation and assimilation. NCC seems to depend on valence, and the trait NCC seems to be particularly important for coping with expectation violations when individuals experience a worse-than-expected reality. Possibly, a negative violation of expectations causes a stronger need for regulation due to stress, which in turn could lead to a stronger impact of personality dispositions like NCC on behavior.