اهداف دست نیافتنی آموزشی: طرح خروج از حضور مجدد با اهداف جایگزین و عواقب بهزیستی ذهنی

Abstract Introduction Is it always positive, in terms of well-being, to be highly committed to one's goals? Even if giving up on goals is most often seen as an undesirable response to difficulties, some researchers have begun to demonstrate the benefits of goal disengagement and reengagement with another goal when faced with unattainable goals. Objective This study aims to test the impact of goal commitment, goal disengagement and goal reengagement on several indicators of well-being in the higher education context, where in first year, a relatively large proportion of students may perceive their initial educational goal as unattainable. Method Some 357 students with secondary school leaving qualifications were surveyed in a first wave of data collection; 186 of these also participated in a second wave. Results Results show that the positive impact of goal commitment on well-being disappears, or even becomes negative, when the goal is perceived as unattainable. Moreover, disengagement from an unattainable goal was found to have beneficial effects on self-mastery. However, this disengagement was not enough to reduce depressive feelings; it must be accompanied by reengagement with an alternative goal. Conclusion These results are discussed and proposals for future research are put forward.

Introduction The transition from secondary school to post-secondary education is a key moment in students’ educational trajectories, as it requires them to make important career choices. Increasingly, advisers try to develop interventions to guide students in forming their educational goals. Most of these interventions are aimed at helping students define an educational goal for themselves and increasing their commitment to this goal. This work is influenced by the generally acknowledged positive role of goal commitment. The positive consequences of being highly committed to an educational goal have been extensively demonstrated empirically (e.g., Germeijs and Verschueren, 2007). But high commitment can also have negative consequences (Pomerantz et al., 2000), particularly when students have goals that they perceive as unattainable. Despite this, this negative side to high goal commitment is rarely investigated. A notable exception is the work of some researchers (e.g., Brandtstädter and Rothermund, 2002 and Wrosch et al., 2003a) who have begun to demonstrate the benefits of two processes in the self-regulation of perceived unattainable goals. These are goal disengagement and goal reengagement. However, these processes have generally been studied as trait-like personal characteristics. There is still a need for research on the self-regulation of a specific type of unattainable goal, namely educational goals. During their first year at college, a considerable number of students will feel that their educational goal is unattainable. This study aimed to investigate the consequences of high commitment to an educational goal on students’ subjective well-being when this goal is perceived as unattainable. The processes of disengagement from the unattainable educational goal and reengagement with another goal are also analysed for their impact on subjective well-being. 1.1. The consequences of goal commitment Goal commitment— the extent to which a particular goal is associated with a strong sense of determination and with the willingness to invest effort in attaining it —and its consequences have been investigated in several empirical studies (see Pomerantz et al., 2000, for a summary). These studies have identified numerous positive consequences of a strong commitment to a goal, including an improvement in subjective well-being. People who are very committed to their goals see their lives as exceptionally meaningful (McGregor and Little, 1998 and Wrosch et al., 2003a), perceive themselves as being in control of their lives (e.g., Ryff and Keyes, 1995) and experience a good deal of positive emotion (feeling proud, happy and joyful) (e.g., Emmons, 1986). These emotions in turn give rise to few depressive symptoms. The more specific consequences of a high commitment to an educational goal — defined as the goal the student is pursuing in choosing his or her study programme — were investigated by Germeijs and Verschueren (2007). Their results suggest that, within the career decision-making process, commitment to the goal is the most important predictor of choice satisfaction, choice stability and adjustment within the chosen option, and is therefore an indirect predictor of performance. However, to the best of our knowledge, the consequences of a high commitment to an educational goal on students’ well-being have not been investigated. All these results support the general impression that strong commitment to personal goals is important, and therefore that giving up on a goal is an undesirable response to difficulty (Wrosch et al., 2003a). However, some authors have begun to investigate the proposition that strong commitment to a goal can also increase psychological distress. Among them are Brandtstädter and Rothermund (2002), who propose that a continuing commitment to unfeasible goals may result in the individual becoming trapped on a barren path. This experience of slow progress toward goal attainment or even failure may have a negative impact on subjective well-being (Carver and Scheier, 1990). 1.2. Unattainable educational goals People who perceive themselves as unable to attain a specific goal whatever their efforts are likely to perceive this goal as unattainable (Wrosch et al., 2003a). During the first year in higher education, a certain proportion of students will come to perceive their educational goal as much more difficult to attain than they had imagined when they chose their study programme. Some of these students will even begin to perceive their educational goal as becoming out of reach. There can be various reasons for educational goals to be seen as unattainable. Some students have focused, sometimes from an early age, on a vocational dream, but have not thought about the educational requirements of this dream; they are pursuing a goal that does not match their competences (Miller and Brickman, 2004). Moreover, the transition from secondary school to post-secondary education involves a lot of change: the picture students had of the academic requirements at the end of secondary school may be quite different from the actual demands made of them at university (Tinto, 1997). In this new setting, even if students had thought about the requirements of their educational goal when they chose it, unexpected failure in academic tasks occurs quite frequently (Wrosch et al., 2003a), which can give rise to perceived goal unattainability. Moreover, the attainable or unattainable character of the goals is especially evident in most educational systems, due to the frequent and clear positive or negative feedback given to students on their progress. It is therefore particularly interesting to investigate the self-regulation of perceived unattainable goals in this educational context. When facing problems in the pursuit of a goal, people can react in two different ways: either they tenaciously continue to be committed to their goal even in the face of obstacles and try to modify the situation to meet the requirements of the goal (assimilative tendency); or they disengage from the goal and adjust their aims to the constraints of the situation (accommodative tendency) (Brandtstädter and Renner, 1990 and Brandtstädter and Rothermund, 2002). Based on the expectancy-value model (Eccles and Wigfield, 2002), people who doubt their ability to attain their goal should be more likely to abandon it. However, not everyone gives up his or her goal equally easily. This is a relatively difficult task and people who have a strong sense of personal control and efficacy may be less ready to adjust their goal to the situation (Brandtstädter and Renner, 1990 and Wrosch et al., 2007). Moreover, not all goals are equally easy to disengage from. The more important a goal is perceived to be, the more difficult the disengagement is likely to be (Carver and Scheier, 1998). The influence of people's reactions to unattainable goals on their well-being has already been investigated. Brandtstädter and Renner (1990) summarised the evidence on the occurrence of depressive symptoms as follows: “The onset, duration, and severity of depressive episodes depend not only on the degree of perceived control over personally important developmental domains [as postulated by learned helplessness theories] but as well on the ability or willingness to disengage from unfeasible goals and to build up new commitments and developmental perspectives” (pp. 64–65). Being aware of one's inability to attain a goal to which one is strongly committed has been shown to be likely to give rise to depressive feelings. Disengagement from the impossible goal and the construction of a new personal goal may put an end to this depressive phase (Brandtstädter and Rothermund, 2002). However, accommodative processes (as a synonym for resignation) often have negative connotations, and these processes were neglected for a long time by research on motivation. The empirical studies conducted by Wrosch and his colleagues (Wrosch et al., 2007, Wrosch et al., 2003a and Wrosch et al., 2003b) on goal disengagement and goal reengagement are a notable exception. 1.3. Goal disengagement Goal disengagement is defined as the ease with which people report being able to reduce their effort and relinquish commitment towards goals they perceive as unattainable (Wrosch et al., 2003b). When the obstacles are seen as too great to overcome, Wrosch et al. (2003a) argue that giving up goal commitment allows the individual to stop trying to attain something that appears to be impossible, prevents repeated failure, and therefore preserves subjective well-being by avoiding the potential distress associated with commitment to an unattainable goal. The capacity to relinquish unattainable goals should be a process as central to adaptive self-regulation as goal pursuit (O’Connor et al., 2009). Some studies focusing on specific goals (e.g., becoming a mother, developing an intimate relationship) have provided support for this idea, and even shown an association between disengagement from perceived unattainable goals and high levels of subjective well-being (e.g., Heckhausen et al., 2001 and Wrosch and Heckhausen, 1999). Wrosch et al. (2003b) complemented this examination to specific goals by investigating goal disengagement as a trait-like personal characteristic. They propose differentiating between people who generally found it easy to disengage from unattainable goals, and people for whom this disengagement was much more difficult. Their results confirm that individual differences exist in people's general tendencies to adjust to unattainable goals, irrespective of the specific unattainable goal encountered. Moreover, studies conducted by Wrosch et al. (2003b) showed that people's general tendencies to disengage from unattainable goals are associated with subjective well-being (e.g., high levels of self-mastery, low levels of depressive symptoms). Other studies showed that physical health is positively influenced by adaptive goal disengagement and that this impact is mediated by changes in subjective well-being (Wrosch et al., 2007). It should be noted that the disengagement studied by Wrosch and his colleagues was an envisaged or projected disengagement; further research investigating the consequences of actual disengagement could prove an interesting complement to this work. 1.4. Reengagement with alternative goals If individuals have disengaged from a previously held goal, they will then be faced with two possibilities: either they will adopt a new goal, or they won’t. There may be various ways of reengaging with an alternative goal (Wrosch et al., 2003a), such as scaling back to a more limited goal in the same area (e.g., a student who has disengaged from the goal of becoming a surgeon and who reengages with the goal of becoming a general practitioner) or forming a new goal in another area (e.g., a student who has disengaged from the goal of becoming a teacher and who reengages with the goal of becoming a lawyer, having discovered that he or she prefers or feels more able to protect children than to have an ongoing role in their education). Wrosch et al. (2003b) were among the first to investigate reengagement with a valued alternative goal as part of the self-regulation of perceived unattainable goals. Goal reengagement is defined as the extent to which people can identify and commit to alternative goals when they are confronted with perceived unattainable goals (Wrosch et al., 2003a). Wrosch et al. (2003a) suggested that goal disengagement is “an adaptive response when it leads to the taking up of other goals” (p. 7). More specifically, Wrosch et al. (2003b) hypothesised that goal reengagement enhances subjective well-being because it is likely to shift the person's focus away from the failure associated with the initial goal onto the positive aspects of the new goal. Wrosch et al. (2003b) showed that young adults’ general tendency to reengage in an alternative goal when they realised their original goal was unattainable was associated with high levels of self-mastery and purpose in life, and low levels of perceived stress and intrusive thoughts, independent of the effect of goal disengagement. Further research could complement this work by investigating the consequences of reengagement to a specific goal (e.g., an educational goal). In addition to these main effects of disengagement and reengagement on subjective well-being, Wrosch et al. (2003b) also showed an interaction effect between the two factors in predicting self-mastery, perceived stress, and affect balance. The positive effect of goal reengagement on subjective well-being was particularly marked when the young adult had difficulty disengaging from unattainable goals. Reengagement seems to be less crucial to young adults who can disengage from the perceived unattainable goal. A similar interaction effect was observed by Wrosch et al. (2007), who showed that goal reengagement tendencies were only (positively) related to subjective well-being and physical health among young adults who had poor disengagement capacities. However, this interaction effect varies depending on the population being studied. Wrosch et al. (2003b) observed that among older adults goal disengagement needs to be accompanied by reengagement to alternative meaningful goals in order to positively influence well-being. For older adults who have difficulty defining alternative goals, it might be better to stay committed to their initial unattainable goal than not be committed to any goal. However, for young adults, disengagement from the unattainable goal seems sufficient to positively influence well-being, even if it is not immediately followed by reengagement with a new goal. Wrosch et al. (2003b) suggest that any distress potentially caused by disengaging from an unattainable goal without finding a new goal to pursue is probably reduced, among young adults, by their characteristically positive expectations about their future. They are more optimistic than older adults about their ability to commit to another goal later on. This result for young adults differs from Wrosch and his colleagues’ general assumption of the need to combine goal disengagement and reengagement (Wrosch et al., 2003a). It is therefore necessary to further explore the interaction between the two processes and to verify the specific pattern observed by Wrosch and his colleagues among young adults (Wrosch et al., 2007 and Wrosch et al., 2003b). Finally, it is interesting to note that the above mentioned studies (mainly conducted by Wrosch and his colleagues) have approached well-being through several indicators. Well-being can be defined as an optimal psychological functioning and experience (Ryan and Deci, 2001). Two indicators in particular were frequently the focus of the studies reviewed above, namely people's depressive feelings and their self-mastery. Depressive feelings are characterised by intrusive negative ideas about the world, the future, and the self (Beck et al., 1996). Self-mastery is defined as the degree to which individuals possess perceived personal control over life outcomes (Pearlin and Schooler, 1978). Lucas et al. (1996) have highlighted the value of examining a variety of global evaluations of well-being. Indeed, they showed that the different indicators of well-being used in the literature are not synonymous, and can be distinguished from each other. This review of the existing literature has highlighted the need to study the consequences of educational goal commitment, disengagement and reengagement on well-being, more specifically on different indicators of well-being, to investigate actual disengagement, and to further explore the interaction between disengagement and reengagement. The study reported below explores these issues. 1.5. The current study The current study proposes to apply the emerging literature on adaptive self-regulation of unattainable goals (Wrosch et al., 2007, Wrosch et al., 2003a and Wrosch et al., 2003b) to a context where goal unattainability is frequently encountered. As explained above, in their first year at university, a relatively large proportion of students will begin to perceive their initial educational goal as unattainable (Miller and Brickman, 2004 and Tinto, 1997). In this educational context, as in many other ones, goal commitment and persistence are largely acknowledged as something to encourage. However, Wrosch and his colleagues’ work shows the potential negative side of goal commitment and the potential positive side of goal disengagement, particularly when faced with an unattainable goal. The aim of this study is to investigate the impact on students’ well-being: (1) of remaining committed to an unattainable educational goal; (2) of disengaging from such a goal; (3) and of reengaging to an alternative educational goal. The potential influences of goal commitment, disengagement and reengagement will more specifically be investigated on two specific indicators of subjective well-being: self-mastery and depressive feelings. These two specific indicators were chosen because, in the literature, most of the results relating to the consequences of goal commitment, disengagement and reengagement appeared using these indicators of well-being (Brandtstädter and Renner, 1990, Ryff and Keyes, 1995, Wrosch et al., 2007 and Wrosch et al., 2003b). To investigate the consequences of goal commitment, disengagement and reengagement on well-being, we have formulated three moderation hypotheses. Hypotheses 1 and 2 postulate that the size, or even direction, of the impacts of goal commitment and goal disengagement on subjective well-being depends on the perceived goal attainability. Hypothesis 1: when the educational goal is perceived as highly attainable, goal commitment positively influences subjective well-being. When the educational goal is perceived as unattainable, this positive impact disappears, or even becomes negative (Brandtstädter and Renner, 1990, Brandtstädter and Rothermund, 2002 and Carver and Scheier, 1990). Hypothesis 2: when the educational goal is perceived as highly attainable, goal disengagement negatively influences subjective well-being. When the educational goal is perceived as unattainable, this negative impact disappears, or even becomes positive (Wrosch et al., 2003a and Wrosch et al., 2003b). Moreover, Hypothesis 3 concerns only low levels of perceived goal attainability. Based on the interactions found by Wrosch and his colleagues in young adults, it postulates that the size of the impact of goal reengagement on subjective well-being depends on goal disengagement. Hypothesis 3: when the educational goal is perceived as unattainable, the positive effect of goal reengagement on subjective well-being is stronger if goal disengagement is low (Wrosch et al., 2003b).

Results Three participants had some outliers (±3 standard deviations from the mean) and were excluded from the analyses. Table 2 shows the means and standard deviations provided by the t-tests: (1) for the total sample; (2) and distinguishing between the 317 students who continued their courses and the 40 students who dropped out, in order to compare them. Actual goal disengagement is not included in Table 2 because it is a dichotomous variable. At T1, the comparison logically showed that the students who dropped out were less committed to the goal they had been pursuing by choosing their study programme, and therefore more disengaged from this goal, than the students who continued their courses. These 40 students also perceived their goal as less attainable. Moreover, the students who dropped out reported more depressive feelings, and felt a lower level of self-mastery. These results are in line with the widely-shared assumption that goal commitment is positively associated with subjective well-being, whereas goal disengagement is negatively associated with it. At T2, the only significant difference concerns goal commitment: the students who had dropped out became slightly more committed at the end of their training than the students who continued their courses. Moreover, their level of well-being is not any lower than that of the students who persevered. Table 2. Means and standard deviations for the total sample, the students who continued their courses and those who dropped out. M (SD) Variables Total sample (n = 357) Students who continued their courses (n = 317) Students who dropped out (n = 40) t(df) p-values Educational goal commitment T1 3.97 (0.33) 4.03 (0.28) 3.55 (0.31) 9.85 (291) < 0.001 Educational goal commitment T2 4.07 (0.25) 4.06 (0.26) 4.18 (0.18) –1.98 (225) < 0.05 Envisaged educational goal disengagement T1 1.50 (0.27) 1.46 (0.25) 1.74 (0.25) –6.70 (290) < 0.001 Perceived educational goal attainability T1 3.53 (0.59) 3.60 (0.56) 3.13 (0.58) 4.85 (277) < 0.001 Depressive feelings T1 2.31 (0.72) 2.23 (0.69) 2.77 (0.70) –4.50 (284) < 0.001 Depressive feelings T2 2.25 (0.69) 2.28 (0.67) 1.96 (0.79) 1.93 (218) 0.054 Self-mastery T1 3.60 (0.83) 3.67 (0.81) 3.15 (0.79) 3.78 (284) < 0.001 Self-mastery T2 3.69 (0.78) 3.67 (0.78) 3.90 (0.74) –1.26 (218) 0.21 Table options Table 3 gives the correlations between the measures. Pearson's correlation coefficient was used, except for those relationships implying actual educational goal disengagement (i.e., a dichotomous variable), which were measured using the point-biserial correlation coefficient. This first approach of our data supports the general assumption of a positive role of goal commitment and a negative role of goal disengagement. At T1 and at T2, educational goal commitment is positively related to self-mastery and negatively related to depressive feelings. To the opposite, both indicators of educational goal disengagement are negatively related to self-mastery and positively related to depressive feelings. Table 3. Correlations between the scales. Variables 1 2 3 4 5 6 7 8 9 1. Ed. goal commitment T1 1.00 2. Ed. goal commitment T2 0.38*** 1.00 3. Envisaged ed. goal disengagement T1 –0.64*** –0.28*** 1.00 4. Actual ed. goal disengagement –0.50*** 0.13* 0.37*** 1.00 5. Perceived ed. goal attainability T1 0.47*** 0.07 –0.38*** –0.28*** 1.00 6. Depressive feelings T1 –0.35*** –0.12 0.29*** 0.26*** –0.63*** 1.00 7. Depressive feelings T2 –0.09 –0.33*** 0.11 –0.13 –0.33*** 0.56*** 1.00 8. Self-mastery T1 0.26*** 0.19* –0.18** –0.22*** 0.44*** –0.67*** –0.35*** 1.00 9. Self-mastery T2 0.03 0.31*** –0.02 0.09 0.27*** –0.46*** –0.66*** 0.56*** 1.00 Ed: educational. *p < 0.05; **p < 0.01; ***p < 0.001. Table options 3.1. Impact of goal commitment on well-being Hypothesis 1, that the impact of goal commitment on subjective well-being is moderated by the perceived attainability of the goal, was tested at T1. To test this moderated effect, two multiple regressions were conducted to predict each of the two aspects of subjective well-being (i.e., depressive feelings and self-mastery). Each regression included goal commitment as an independent variable, perceived goal attainability as a moderator variable and the product of the independent and the moderator variables (i.e., the interaction) as a third predictor3. If the interaction is significant, this indicates that the moderator hypothesis is supported (Baron and Kenny, 1986). The results of these regressions are presented in Table 4. Table 4. Multiple regressions to predict depressive feelings T1 and self-mastery T1, with educational goal commitment T1, perceived educational goal attainability T1, and the product of educational goal commitment T1 and perceived educational goal attainability T1 as predictors. R 1 (n = 279) Depressive feelings T1 (DV) R 2 (n = 279) Self-mastery T1 (DV) B SE B β B SE B β Constant 2.34 0.04 3.54 0.05 Ed. goal commitment T1 (IV) –0.19 0.11 –0.09 0.18 0.15 0.07 Perceived ed. goal attainability T1 (MV) –0.76 0.07 –0.62*** 0.63 0.09 0.44*** IV x MV –0.37 0.15 –0.12* 0.66 0.19 0.18** R2 0.42 0.23 R: regression; DV: dependent variable; IV: independent variable; MV: moderator variable; Ed.: educational. *p < 0.05; **p < 0.01; ***p < 0.001. Table options The two dimensions of subjective well-being were both significantly predicted by perceived goal attainability (depressive feelings: β = –0.62, p < 0.001; self-mastery: β = 0.44, p < 0.001). Moreover, as postulated by Hypothesis 1, both dimensions were significantly influenced by the interaction between goal commitment and perceived goal attainability (depressive feelings: β = –0.12, p < 0.05; self-mastery: β = 0.18, p < 0.01). These interaction effects were probed using the Johnson-Neyman (J-N) technique to gain a better understanding of the conditions under which the impact of goal commitment on subjective well-being is strong or weak, or positive or negative. This technique overcomes the limitations of two other approaches widely used in the literature for analysing interactions, namely the subgroup analysis and the pick-a-point approach. One of the main problems with these two traditional methods is that they involve arbitrary cut-off procedures to determine the subgroups or to select values of the moderator at which to probe the interaction (Hayes and Matthes, 2009 and Stone-Romero and Anderson, 1994). The J-N technique identifies regions within the range of the moderator variable where the effect of the independent variable on the dependent variable is statistically significant and regions where it is not (Hayes and Matthes, 2009). It avoids the potential arbitrariness of the traditional methods by mathematically deriving the point or points along the continuum of the moderator that delimit(s) these regions. The results of the J-N analysis for depressive feelings showed that a value of perceived goal attainability of 3.62 defined the limit of the region of significance for goal commitment. Above this level of perceived goal attainability, depressive feelings decreased significantly with goal commitment, but when goal attainability was at or below 3.62, the impact of goal commitment on depressive feelings was non-significant. For the prediction of self-mastery, there were two regions where goal commitment had a significant effect. When perceived goal attainability was higher than 3.73, goal commitment had a significant positive impact on self-mastery, and when perceived goal attainability was lower than 2.57, goal commitment had a significant negative impact on self-mastery. Between these two values, the impact of goal commitment was non-significant. These results confirm that the impact of goal commitment on subjective well-being is moderated by perceived goal attainability, as postulated by Hypothesis 1. More specifically, when the educational goal is perceived as relatively attainable, depressive feelings decrease with goal commitment. When the goal is perceived as unattainable, the impact of goal commitment disappears. Even more strikingly, the direction of the impact of goal commitment on self-mastery changes depending on the level of perceived goal attainability. This impact is significantly positive when the goal is perceived as relatively attainable, but it is significantly negative when the goal is perceived as relatively unattainable. 3.2. Impact of goal disengagement on well-being The second hypothesis, that the impact of goal disengagement on subjective well-being is moderated by perceived goal attainability, was also tested at T1. Two indicators of goal disengagement were used: envisaged and actual disengagement. Four multiple regressions were needed to test the moderated effect postulated by Hypothesis 2: each regression predicted one of the two aspects of subjective well-being using one of the two indicators of goal disengagement as the independent variable. Perceived goal attainability (i.e., the moderator variable) and the product of the independent variable and the moderator variable were included in each regression. The results of these regressions are presented in Table 5 and Table 6. Table 5. Multiple regressions to predict depressive feelings T1 and self-mastery T1, with envisaged educational goal disengagement T1, perceived educational goal attainability T1, and the product of envisaged educational goal disengagement T1 and perceived educational goal attainability T1 as predictors. R 1 (n = 278) Depressive feelings T1 (DV) R 2 (n = 278) Self-mastery T1 (DV) B SE B β B SE B β Constant 2.32 0.04 3.56 0.05 Envisaged ed. goal disengagement T1 (IV) 0.21 0.13 0.08 –0.09 0.18 –0.03 Perceived ed. goal attainability T1 (MV) –0.76 0.06 –0.62*** 0.65 0.08 0.45*** IV x MV 0.28 0.18 0.07 –0.58 0.24 –0.13* R2 0.42 0.22 R: regression; DV: dependent variable; IV: independent variable; MV: moderator variable; Ed.: educational. *p < 0.05; ***p < 0.001. Table options Table 6. Multiple regressions to predict depressive feelings T1 and self-mastery T1, with actual educational goal disengagement, perceived educational goal attainability T1, and the product of actual educational goal disengagement and perceived educational goal attainability T1 as predictors. R 1 (n = 279) Depressive feelings T1 (DV) R 2 (n = 279) Self-mastery T1 (DV) B SE B β B SE B β Constant 2.32 0.03 3.59 0.05 Actual ed. goal disengagement (IV) 0.33 0.11 0.16** –0.33 0.15 –0.14* Perceived ed. goal attainability T1 (MV) –0.76 0.06 –0.62*** 0.59 0.08 0.42*** IV x MV 0.43 0.17 0.14* –0.27 0.22 –0.08 R2 0.42 0.21 R2: regression; DV: dependent variable; IV: independent variable; MV: moderator variable; Ed.: educational. *p < 0.05; **p < 0.01; ***p < 0.001. Table options Table 5 shows the results of the two regressions using envisaged goal disengagement as the independent variable. The two dimensions of subjective well-being were significantly predicted by perceived goal attainability (depressive feelings: β = –0.62, p < 0.001; self-mastery: β = 0.45, p < 0.001). However, only one of these dimensions, namely self-mastery (β = –0.13; p < 0.05), was significantly predicted by the interaction between envisaged goal disengagement and perceived goal attainability. More specifically, the J-N technique showed two regions of significance for the impact of envisaged goal disengagement. When perceived goal attainability was higher than 4.23, envisaged goal disengagement had a significant negative impact on self-mastery. However, when perceived goal attainability was lower than 2.11, envisaged goal disengagement had a significant positive impact on self-mastery. Between these two values, the impact was not significant. Table 6 shows the results of the two regressions using actual goal disengagement as the independent variable. The two dimensions of subjective well-being were significantly predicted by perceived goal attainability (depressive feelings: β = –0.62, p < 0.001; self-mastery: β = 0.42, p < 0.001) and by actual goal disengagement (depressive feelings: β = 0.16, p < 0.01; self-mastery: β = –0.14, p < 0.05). This main effect of actual goal disengagement is in line with the results of the comparison between the students who dropped out and those who continued their courses ( Table 2). Depressive feelings were significantly predicted by the interaction between actual goal disengagement and perceived goal attainability (β = 0.14; p < 0.05). The J-N technique showed one region of significance for the impact of actual goal disengagement: when perceived goal attainability was above 3.22, depressive feelings increased significantly with actual goal disengagement, but at or below this level the impact of actual goal disengagement on depressive feelings was not significant. Taken together, these results confirm that the impact of goal disengagement on subjective well-being is moderated by perceived goal attainability, as postulated by Hypothesis 2. More specifically, this moderated impact was shown: (1) for the prediction of self-mastery using envisaged goal disengagement as the independent variable; (2) and for the prediction of depressive feelings using actual goal disengagement as the independent variable. When the goal was perceived as highly attainable, depressive feelings increased significantly with actual goal disengagement, but there was no such impact when the goal was perceived as unattainable. The effect of envisaged goal disengagement on self-mastery was even more marked, and indeed it changed direction depending on the level of perceived goal attainability. The moderating effect of goal disengagement on each one of the dimensions of subjective well-being therefore depended on which indicator of goal disengagement was used. Finally, Table 5 and Table 6 showed a main effect of actual disengagement on the two dimensions of subjective well-being, but no significant main effects of envisaged disengagement. It seems that actual disengagement is a stronger predictor of subjective well-being than envisaged disengagement. 3.3. Interaction between goal disengagement and goal reengagement The third hypothesis concerned the potential interaction between goal disengagement and goal reengagement in predicting subjective well-being. This hypothesis only applied to students perceiving their educational goal as unattainable. To assess students’ reengagement, we used the measure of goal commitment at T2. Given that we are focusing on those students who perceived their educational goal as unattainable at T1, this measure indicates students’ degree of commitment to the educational goals they are pursuing by choosing their subject for the coming academic year after encountering what they perceived as an unattainable goal. Hypothesis 3 postulates that subjective well-being at T2 depends on the interaction between three factors: perception of goal attainability at T1, goal disengagement at T1 and goal commitment at T2. Multiple regressions including this triple interaction were conducted to predict well-being at T2 controlling for its level at T1. More specifically, four multiple regressions were run, two predicting depressive feelings at T2 (one using envisaged disengagement as a predictor (Table 7), and the other using actual disengagement (Table 8)) and another two predicting self-mastery at T2 (again, one using envisaged disengagement as a predictor (Table 9), and the other using actual disengagement (Table 10)). Table 7. Multiple regression to predict depressive feelings T2, controlling for depressive feelings T1, with envisaged educational goal disengagement T1, perceived educational goal attainability T1, educational goal commitment T2, the three 2 × 2 products of these variables, and the product of the three variables as predictors. Depressive feelings T2 (DV) B SE B β Constant 2.23 0.04 Depressive feelings T1 (CV) 0.57 0.07 0.57*** Ed. goal commitment T2 (IV) –0.72 0.18 –0.26*** Perceived ed. goal attainability T1 (MV1) –0.05 0.09 –0.04 Envisaged ed. goal disengagement T1 (MV2) –0.44 0.07 –0.16* IV × MV1 0.46 0.30 0.10 IV × MV2 –1.13 0.66 –0.11 MV1 × MV2 0.14 0.21 0.04 IV × MV1 × MV2 2.47 0.81 0.19** R2 0.48 DV: dependent variable; CV: control variable; IV: independent variable; MV: moderator variable; Ed.: educational. *p < 0.05; **p < 0.01; ***p < 0.001. n = 173. Table options Table 8. Multiple regression to predict depressive feelings T2, controlling for depressive feelings T1, with actual educational goal disengagement T1, perceived educational goal attainability T1, educational goal commitment T2, the three 2x2 products of these variables, and the product of the three variables as predictors. Depressive feelings T2 (DV) B SE B β Constant 2.24 0.04 Depressive feelings T1 (CV) 0.61 0.07 0.61*** Ed. goal commitment T2 (IV) –0.41 0.18 –0.15* Perceived ed. goal attainability T1 (MV1) –0.06 0.09 – 0.05 Actual ed. goal disengagement T1 (MV2) –0.73 0.18 –0.33*** IV × MV1 0.77 0.33 0.16* IV × MV2 2.20 0.95 0.20* MV1 × MV2 0.14 0.33 0.04 IV × MV1 × MV2 2.07 1.94 0.11 R2 0.51 DV: dependent variable; CV: control variable; IV: independent variable; MV: moderator variable; Ed.: educational. *p < 0.05; **p < 0.01; ***p < 0.001. n = 173. Table options Table 9. Multiple regression to predict self-mastery T2, controlling for self-mastery T1, with envisaged educational goal disengagement T1, perceived educational goal attainability T1, educational goal commitment T2, the three 2x2 products of these variables, and the product of the three variables as predictors. Self-mastery T2 (DV) B SE B β Constant 3.66 0.05 Self-mastery T1 (CV) 0.44 0.06 0.47*** Ed. goal commitment T2 (IV) 0.69 0.21 0.23** Perceived ed. goal attainability T1 (MV1) 0.18 0.09 0.13 Envisaged ed. goal disengagement T1 (MV2) 0.52 0.20 0.18* IV × MV1 0.24 0.35 0.05 IV × MV2 1.07 0.75 0.10 MV1 × MV2 –0.55 0.24 –0.14* IV × MV1 × MV2 –1.73 0.92 –0.12 R2 0.43 DV: dependent variable; CV: control variable; IV: independent variable; MV: moderator variable; Ed.: educational. *p < 0.05; **p < 0.01; ***p < 0.001. n = 173. Table options Table 10. Multiple regression to predict self-mastery T2, controlling for self-mastery T1, with actual educational goal disengagement T1, perceived educational goal attainability T1, educational goal commitment T2, the three 2 × 2 products of these variables, and the product of the three variables as predictors. Self-mastery T2 (DV) B SE B β Constant 3.65 0.05 Self-mastery T1 (CV) 0.49 0.06 0.52*** Ed. goal commitment T2 (IV) 0.33 0.21 0.11 Perceived ed. goal attainability T1 (MV1) 0.17 0.09 0.13 Actual ed. goal disengagement T1 (MV2) 0.50 0.20 0.21* IV × MV1 –0.02 0.38 –0.00 IV × MV2 –1.83 1.11 –0.16 MV1 × MV2 –0.39 0.38 –0.10 IV × MV1 × MV2 –2.62 2.27 –0.13 R2 0.43 DV: dependent variable; CV: control variable; IV: independent variable; MV: moderator variable; Ed.: educational. *p < 0.05; **p < 0.01; ***p < 0.001. n = 173. Table options Only the regression predicting depressive feelings using envisaged disengagement as a predictor showed a significant triple interaction (Table 7). Three main effects were revealed: a positive impact of the control variable, namely depressive feelings at T1 (β = 0.57; p < 0.001), a negative impact of goal commitment at T2 (β = –0.26; p < 0.001), and a negative impact of envisaged goal disengagement at T1 (β = –0.16; p < 0.05). Moreover, as postulated by Hypothesis 3, the level of depressive feelings at T2 was significantly predicted by the interaction between goal commitment at T2, envisaged goal disengagement at T1 and perceived goal attainability at T1 (β = 0.19; p < 0.01). We could not directly explore this complex interaction using the J-N technique because this technique is only designed to test simple moderated effects. Therefore, we first had to use a more traditional approach for probing interactions, namely the subgroup approach ( Hayes and Matthes, 2009). This approach involves splitting the data file into two (or more) subsets defined by one (or more) values of the moderator and repeating the analysis on these subsets. Our sample was split into two subsamples defined by the median of perceived goal attainability, 3.60. A multiple regression analysis was then conducted on each subsample to predict depressive feelings at T2. According to Hypothesis 3, we should observe an interaction effect between disengagement at T1 and commitment at T2 in the subsample characterised by a low level of perceived goal attainability (< 3.60). The results are presented in Table 11. Table 11. Multiple regression to predict depressive feelings T2, controlling for depressive feelings T1, with envisaged educational goal disengagement T1, educational goal commitment T2, and the product of envisaged educational goal disengagement T1 and educational goal commitment T2 as predictors: subgroup analysis. Depressive feelings T2 (DV) SG1: low-level of perceived ed. goal attainability (< 3.60) n = 139 SG2: high-level of perceived ed. goal attainability (≥ 3.60) n = 140 B SE B β B SE B β Constant 2.17 0.07 2.25 0.07 Depressive feelings T1 (CV) 0.64 0.10 0.56*** 0.57 0.10 0.50*** Ed. goal commitment T2 (IV) –0.75 0.26 –0.26** –0.85 0.27 –0.33** Envisaged ed. goal disengagement T1 (MV) –0.52 0.26 –0.18* –0.42 0.24 –0.16 IV × MV –3.52 1.01 –0.31** –0.27 0.82 –0.03 R2 0.44 0.40 DV: dependent variable; CV: control variable; IV: independent variable; MV: moderator variable; Ed.: educational. *p < 0.05; **p < 0.01; ***p < 0.001. Table options In the subgroup of students who had perceived their goal as being unattainable at T1 (Subgroup 1), we found a main effect of all three predictors: a positive impact of the control variable, namely depressive feelings at T1 (β = 0.56; p < 0.001), a negative impact of goal commitment at T2 (β = –0.26; p < 0.01), and a negative impact of envisaged goal disengagement at T1 (β = –0.18; p < 0.05). Moreover, as postulated by Hypothesis 3, the interaction between goal commitment at T2 and goal disengagement at T1 was significant in this subgroup (β = –0.31; p < 0.01). In the subgroup characterised by a high level of perceived goal attainability at T1 (Subgroup 2), two main effects were revealed: a positive impact of the control variable, namely depressive feelings at T1 (β = 0.50; p < 0.001) and a negative impact of goal commitment at T2 (β = –0.33; p < 0.01). The interaction was not significant in this subgroup. The significant simple interaction observed in Subgroup 1 was explored using the J-N technique. This technique showed one region of significance: when envisaged goal disengagement was above 1.44, depressive feelings at T2 decreased significantly with goal reengagement, but at or below this level the impact of goal reengagement on depressive feelings was not significant. These results confirmed that goal disengagement and goal reengagement interact in their effect on subjective well-being when the goal is perceived as unattainable, but did not support the specific interaction effect postulated by Hypothesis 3. It was hypothesised that when the educational goal was perceived as unattainable, the positive effect of goal reengagement on subjective well-being was stronger if goal disengagement is low. What was observed in this study is quite different. Our results showed that, when the educational goal was perceived as unattainable, a positive effect of goal reengagement on subjective well-being was only observed if goal disengagement was high (above 1.44).