Fear of failure and student athletes’ interpersonal antisocial behaviour in education and sport
Abstract
Background. The link between fear of failure and students’ antisocial behaviour has received scant research attention despite associations between fear of failure, hostility, and aggression. Also, the effect of sport experience on antisocial behaviour has not been considered outside of the sport context in adult populations. Further, to date, sex differences have not been considered in fear of failure research.
Aims. To examine whether (a) fear of failure and sport experience predict antisocial behaviour in the university and sport contexts in student athletes, and whether this prediction is the same in males and females; and (b) sex differences exist in antisocial behaviour and fear of failure.
Sample. British university student athletes (n= 176 male; n= 155 female; Mage= 20.11 years).
Method. Participants completed questionnaires assessing fear of failure, sport experience, and antisocial behaviour in both contexts.
Results. (a) Fear of failure and sport experience positively predicted antisocial behaviour in university and sport and the strength of these predictions did not differ between males and females; (b) females reported higher levels of fear of devaluing one's self-estimate than males whereas males reported higher levels of fear of important others losing interest than females. Males engaged more frequently than females in antisocial behaviour in both contexts.
Conclusions. Fear of failure and sport experience may be important considerations when trying to understand antisocial behaviour in student athletes in education and sport; moreover, the potential effect of overall fear of failure and of sport experience on this frequency does not differ by sex. The findings make an important contribution to the fear of failure and morality literatures.
Background
Fear of failure is the motive to avoid failure in achievement contexts, and involves cognitive, behavioural, and emotional experiences. It is prominent in both sexes and across levels of students’ actual and perceived ability (Covington, 1992; Elliot & Sheldon, 1997). Fear of failure is a dispositional avoidance-oriented achievement motive that typically orients individuals high in fear of failure to avoid engaging in achievement related tasks (Elliot & Sheldon, 1997). This achievement motive is activated in situations where individuals perceive failure to perform adequately to be a threat to their achieving personally meaningful goals (Conroy & Elliot, 2004).
Fear of failure is a motive deeply rooted in self-evaluative disposition that is socialized during childhood (between ages 5 and 9 years; McClelland, 1958). As such, individuals high in fear of failure will feel this way across achievement contexts (Conroy, 2003; Conroy & Elliot, 2004; Elliot & Thrash, 2004). It is a shame-based achievement motive as it involves the capacity to anticipate negative affect after failure such as shame and embarrassment, and loss of social status (McGregor & Elliot, 2005). While early achievement motivation theorists (e.g., Atkinson, 1964) viewed fear of failure as a uni-dimensional construct (positioning shame at its core), contemporary theorists view it as multi-dimensional (e.g., Conroy, Willow, & Metzler, 2002). The recent multi-dimensional model of fear of failure proposed by Conroy et al. (2002) provided the theoretical framework for the present study.
Conroy et al. (2002) view fear of failure as the dispositional tendency to appraise threat to the achievement of personally meaningful goals when one fails. Individuals high in fear of failure have learnt to associate failure with aversive consequences and typically perceive failure in evaluative situations as threatening, and believe that aversive consequences will occur after failure. Conroy et al.'s model proposes that fear of failure consists of five beliefs about the consequences of failure that are associated with threat appraisal and feared. These are: (a) experiencing shame and embarrassment upon failure, which is related to beliefs of self-presentational failure and personal diminishment; (b) having an uncertain future, which is related to beliefs of losing future opportunities; (c) devaluing one's self-estimate, which is related to beliefs of having poor ability and control over one's performance; (d) important others losing interest, which is related to beliefs of losing social value and influence in the performance domain; and (e) upsetting important others, which is related to beliefs of others’ disapproval and loss of affection following failure. Thus, Conroy et al. (2002) conceptualized fear of failure as a function of one's interaction with others and as reflecting one's desire to avoid the lowering of one's evaluation (by the self and by others).
Individuals high in fear of failure adopt self-regulatory strategies (e.g., self-handicapping, defensive pessimism, adoption of avoidance achievement goals) that are often harmful to their academic and sporting performance (e.g., poor grades, low intrinsic motivation), well being (e.g., decreased satisfaction and enjoyment in achievement pursuits, high anxiety; Conroy & Elliot, 2004; Elliot & Sheldon, 1997; Martin & Marsh, 2003), and interpersonal behaviour (e.g., become less sociable and tolerant to people, seek isolation; Sagar, Lavallee, & Spray, 2009). One extreme strategy is rage, which is adopted by individuals to diffuse their fear and anticipation of failure and its consequences (Gilbert & McGuire, 1998). Rage elicits hostility, anger, and aggression and can lead to attacking (verbally or physically) those who have threatened the person's status or witnessed their social exposure when they have failed (Fournier, Moskowitz, & Zuroff, 2002; Gilbert, Pehl, & Allan, 1994). Thus, fear of failure can lead to problematic interpersonal behaviour.
Two studies have examined the impact of fear of failure on interpersonal problems. In the first study (Wright, Pincus, Conroy, & Elliot, 2009), some high fear of failure individuals employed aggressive self-regulatory strategies that were associated with hostility and domineering behaviour (e.g., arrogance, avoidance); in the second study, individuals who were high in fear of failure engaged in either hostile-dominant or submissive behaviours (Conroy, Elliot, & Pincus, 2009). Moreover, fear of failure positively predicted the interpersonal problem of being vindictive (i.e., wanting to get revenge against people). Taken together with research indicating that individuals high in fear of failure adopt rage as an interpersonal regulatory strategy (Gilbert & McGuire, 1998), these findings suggest that some high fear of failure individuals are more likely to engage in acts that are harmful to others.
Two achievement contexts, in which fear of failure and its associated interpersonal behaviours can be expressed, are education and sport. These are social contexts in which students interact with other students and athletes, and in which performances are evaluated by the self and by important others (e.g., teachers, coaches, peers, parents). Incompetence can be exposed in either context because they both involve the possibility of failure. In individuals who fear failure, this can elicit shame and devaluation of the self, leading to undesirable social behaviours (Conroy et al., 2009; Stuewig, Tangney, Heigel, Harty, & McCloskey, 2010; Wright et al., 2009). A class of social behaviours that has been extensively investigated in the sport domain is antisocial behaviour. This has been defined as behaviour intended to harm or disadvantage another individual or group of individuals (Sage, Kavussanu, & Duda, 2006), and examples in sport are trying to injure, provoking, or verbally abusing another player. Such behaviours can have negative consequences for others’ rights and well being and, therefore, have relevance to morality (see Turiel, 1983). Morality is concerned with social relations and regulates people's behaviour and engagement in fair and just social interactions (Janoff-Bulman, Sheikh, & Hepp, 2009). Thus, from a moral perspective, infrequency of engagement in antisocial behaviour reflects one's ability to function effectively in social contexts. Fear of failure should, therefore, predict antisocial behaviour in the education and sport contexts because fear of failure is grounded in self- and social-evaluative aspects of competence. Further, both contexts expose one's incompetence upon failure, and aggression serves as an interpersonal regulatory strategy to diffuse this fear (Conroy et al., 2009; Wright et al., 2009).
To date, however, no study has investigated antisocial behaviours such as those described above among student athletes in the university context, or examined whether fear of failure predicts antisocial behaviour in the education or in the sport context. The only relevant studies are by Wright et al. (2009) and Conroy et al. (2009). However, although these researchers examined the impact of fear of failure on interpersonal problems (e.g., arrogance, avoidance, vindictiveness), they did not examine antisocial behaviour as defined above in the education or in the sport context, and they did not study how the same individuals behave in two different achievement contexts.
Sport participation has been consistently associated with lower levels of morality in research in sport. Specifically, in studies involving children, adolescents, and university athletes, participation in medium or high contact sports has been positively associated with self-reported tendencies to be aggressive, judgments that aggressive sport acts are legitimate, and lower levels of moral functioning (e.g., Bredemeier, Weiss, Shields, & Cooper, 1986; Conroy, Silva, Newcomer, Walker, & Johnson, 2001; Kavussanu & Ntoumanis, 2003). To date, however, no study has investigated whether sport participation is linked with more frequent antisocial behaviour in student athletes in the education and sport contexts.
Sex differences in fear of failure have also not been investigated previously. Nevertheless, sex-role differences in shaping perceptions of success and failure, and in achievement motivation, have long been examined in psychological and educational research. For example, Atkinson's (1964) expectancy-value theory (which relates to the motive for success and the probability and incentive value of success) emphasizes sex differences in the tendency to approach achievement activity, indicating that males and females differ in their concerns about failure. Research has generally shown that male and female perceptions of success and competence/ability-related beliefs and behaviours follow sex-role socialization and stereotypes (see Meece, Glienke, & Burg, 2006, for a review). Finally, male athletes typically are more competitive and have a greater desire to win than female athletes (see Gill, 2002, for a review). Accordingly, variations in threat appraisal and fears of failure between males and females are likely, and thus merit examination.
Sex differences in variables reflecting morality in non-sport and sport contexts are also evident in the literature. For example, male university students have been shown to engage more frequently in physically (e.g., reciprocally hitting another person) and verbally (e.g., arguing with people following disagreements) aggressive acts than female students (Buss & Perry, 1992), while male athletes have reported greater aggressive tendencies (Bredemeier, 1994), unsportsmanlike attitudes (e.g., cheating; Duda, Olson, & Templin, 1991), judgments regarding the perceived legitimacy of aggressive and injurious acts (Duda et al., 1991; Gardner & Janelle, 2002; Kavussanu & Roberts, 2001), and antisocial sport behaviour (Kavussanu, Stamp, Slade, & Ring, 2009) than female athletes. To date, however, no study has investigated sex differences in antisocial interpersonal behaviour of student athletes in the sport and education contexts using one sample concurrently.
The present study
The present study had two aims. The first aim was to investigate whether overall fear of failure and sport experience predict antisocial behaviour in the university and sport contexts, and whether sex moderates this prediction. We hypothesized that fear of failure (Conroy et al., 2009; Wright et al., 2009) and sport experience (Conroy et al., 2001; Kavussanu & Ntoumanis, 2003) would positively predict antisocial behaviour in both contexts. However, we formed no hypotheses regarding sex as a moderating variable due to the lack of indicative evidence.
We also expected fear of failure and sport experience to be positively related. As sport experience increases, so does the level at which individuals compete, and the resulting pressure to achieve top performance and win can bring with it an increase in fear of failure (Hosek & Man, 1989), because fear of failure is activated in situations where failure to perform adequately is perceived to be a threat to the accomplishment of one's goal (Conroy & Elliot, 2004). Accordingly, we assumed that individuals who have been involved in their sport longer may place more importance on competitive outcomes and experience fear of failure as a result. We therefore expected these two variables to be positively related.
The second study aim was to examine sex differences in antisocial behaviour and fear of failure. We hypothesized that males would engage more frequently than females in antisocial behaviour in both contexts (e.g., Buss & Perry, 1992; Kavussanu & Roberts, 2001; Kavussanu et al., 2009). Although interpersonal antisocial behaviours such as trying to injure, criticizing, provoking, or verbally abusing another person have not previously been examined in student athletes within the education/university context, we based our hypothesis with respect to this context on evidence of sex differences in general physical and verbal aggression in everyday life (Buss & Perry, 1992). We hypothesized that female students would report higher levels of fear of de-valuating one's self-estimate than males as this fear relates to competence beliefs (i.e., of having low ability and control over one's performance) based on evidence that sex-role differences exist in concerns about failure, perceptions of success, and competence/ability-related beliefs (Atkinson, 1964; Gill, 2002; Meece et al., 2006). No other sex differences in the dimensions of fear of failure were hypothesized.
Method
Participants and procedure
Participants
A sample of 176 male and 155 female students was recruited from two British universities. At the time of data collection, participants’ average age was 20.11 years (SD= 1.65 years; range = 18–29 years) and they were reading for undergraduate (n= 306) and postgraduates (n= 25) degrees in: Human Sciences (n= 113); Social Sciences (n= 54); Engineering (n= 46); Business Studies (n= 58); other (n= 60). They were competing in rugby (n= 82); hockey (n= 56); football (soccer; n= 66); basketball (n= 38); American football (n= 38); netball (n= 29); and lacrosse (n= 22) university teams. These sports were selected because they are classified as medium or high contact sports with significant potential to raise moral issues (see Bredemeier et al., 1986). At the time of data collection, participants had been playing for their university team for an average of 1.80 years (SD= 0.91 years; range = 1–5 years), and had been competing in their respective sports for an average of 7.76 years (SD= 4.21 years; range = 1–16 years). The highest levels at which they competed at the time of data collection was university (47%); club (22%); regional/county (11%); national (9%); and international (11%).
Procedure
After obtaining approval from the Ethics Committee and permission from the athletic bodies and head coaches of both universities, student athletes were approached by the first author during the final month of their sporting season (and in the middle of their academic year) either before or after their sport training sessions. Participants were informed of the study's aims, its voluntary nature, that honesty in responses was vital, and that data would be used only for research purposes and kept strictly confidential. After signing an informed consent form, each student completed the questionnaire that contained the measures described below.
Measures
Fear of failure
Fear of failure was measured using the Performance Failure Appraisal Inventory (PFAI; Conroy et al., 2002), which consists of 25 items measuring beliefs associated with aversive consequences of failure. The PFAI has five subscales capturing fear of experiencing shame and embarrassment (seven items; e.g., ‘When I am failing, it is embarrassing if others are there to see it’), fear of having an uncertain future (four items; e.g., ‘When I am failing, it upsets my ‘plan’ for the future’), fear of devaluing one's self-estimate (four items; e.g., ‘When I am failing, I blame my lack of talent’), fear of important others losing interest (five items; e.g., ‘When I am not succeeding, people are less interested in me’), and fear of upsetting important others (five items; ‘When I am failing, people who are important to me are disappointed’). Participants were asked to indicate how often they believed each of these consequences was true for them in their achievement pursuits (e.g., sport, work, education). Items were answered on a five-point scale that included response options of 0 (do not believe it at all), 2 (believe it 50% of the time), and 4 (believe it 100% of the time), thus indicating the strength of their belief that each consequence was likely to occur to them after failure. Scores were computed by averaging across items. When assessing overall fear of failure all 25 items are used. Conroy et al. (2002) have provided evidence for the construct validity of the PFAI, as well as the reliability of its sub-scales (α range .74–.81).
Antisocial behaviour in sport
The antisocial behaviour sub-scales of the Pro-social and antisocial Behaviour in Sport Scale (PABSS; Kavussanu & Boardley, 2009) were used to assess reported antisocial sport behaviours. Participants were presented with 13 items describing sport behaviours and were asked to report how often they had engaged in each behaviour in the current season on a scale anchored by 1 (never) and 5 (very often). The two sub-scales measured antisocial behaviour towards team-mates (five items) and opponents (eight items). Kavussanu and Boardley (2009) have provided evidence for the discriminant and concurrent validity, as well as the reliability of the antisocial team-mate (α= .83) and opponent (α= .86) sub-scales. Finally, a very strong positive relationship (r= .70) between observed and reported antisocial behaviours similar to those assessed here, has been reported in previous research (Kavussanu, Seal, & Phillips, 2006). We computed the mean of all items to produce a single score for overall antisocial behaviour because (a) we did not expect that fear of failure would differentially predict the two types of antisocial sport behaviour, (b) the two behaviour types were strongly and positively related (r= .62), and (c) the differentiation between team-mate and opponent behaviour that forms the separate sub-scales in this measure was not present in the scale assessing antisocial behaviour in university.
Antisocial behaviour in university
An adapted version of the antisocial behaviour subscales of the PABSS (Boardley, Kavussanu & Ring, 2010) was used to assess reported antisocial behaviour in university. The measure contains the same number of items (i.e., 13) as the original subscales, and item content was chosen to match as closely as possible the items from the PABSS. Thus, the only major change in this version was that the words ‘team-mate’ and ‘opponent’ were replaced by the word ‘student’. Other minor adjustments to item content were made where necessary to ensure the behaviours were relevant to the university context. For example, the item ‘deliberately fouled an opponent’ was changed to ‘deliberately hurt a student’. Participants were presented with the items and asked to report how often they had engaged in each behaviour in the current academic year on a scale anchored by 1 (never) and 5 (very often). The mean for all items was computed and used in the main analyses.
Evidence is available to support the construct validity of the antisocial university behaviour measure. Regarding the factorial validity of the scale, an exploratory principal-axis factor analysis on the 13 items resulted in the extraction of a single factor (eigenvalue = 6.52; extraction based on deflection point on scree plot; Cattell, 1966), thus demonstrating the anticipated uni-dimensional factorial structure. A confirmatory factor analysis using a separate sample of 372 student athletes (Boardley et al., 2010) demonstrated a good fit, χ2 (65) = 302.09, p < .001, CFI = .933, RMSEA = .09, SRMR = .08, therefore confirming this uni-dimensional structure. With respect to the criterion validity of the scale, scores obtained with this measure have shown a strong positive correlation with moral disengagement, a weak positive relationship with academic ego goal orientation, and a weak negative association with academic task goal orientation (Boardley et al., 2010). These results provide evidence for the scale's criterion validity as the relationships described are supported by theory. Collectively, the above results provide support for the construct validity of the scale.
Sport experience
Sport experience was measured using a single item. Specifically, once participants had indicated their main sport, they were asked to indicate how many years they had been participating competitively in this sport. Responses were recorded in years and months. Similar approaches have been used previously to assess the extent of athletes’ sport participation (e.g., Bredemeier et al., 1986; Kavussanu & Ntoumanis, 2003).
Results
Psychometric properties of the scales
The measures used in the present study demonstrated satisfactory to excellent levels of internal consistency. Cronbach's (1951) alpha coefficients for the antisocial behaviour scales were .92 for university and .88 for sport behaviour. With respect to the fear of failure scales alpha coefficients were .84 for fear of shame and embarrassment, .71 for fear of devaluing one's self-estimate, .84 for fear of important others losing interest, .77 for fear of upsetting important others, .85 for fear of having an uncertain future, and .93 for overall fear of failure.
Descriptive statistics and sex differences
Descriptive statistics for antisocial behaviour in university and sport and the five dimensions of fear of failure are presented separately for males and females in Table 1. It can be seen that, on average, antisocial university behaviour was reported to occur rarely for males and never to rarely for females, whereas antisocial behaviour in sport was reported to occur rarely to sometimes for males and rarely for females. With respect to the five fears of failure, fear of experiencing shame and embarrassment was the strongest fear reported for both sexes. The weakest fear reported by males was the fear of devaluing one's self-estimate, whilst for females it was the fear of important others losing interest. Finally, overall fear of failure was experienced almost 50% of the time by both sexes and on average males and females had considerable experience in their sport although this did vary widely and to a similar extent in both sexes.
| Variable | Males | Females | Univariate F(1, 329) | ||||
|---|---|---|---|---|---|---|---|
| M | SD | Range | M | SD | Range | ||
| Antisocial behaviour context | |||||||
| University | 2.21 | 0.76 | 1.00–4.15 | 1.69 | 0.61 | 1.00–3.85 | 51.36*** |
| Sport | 2.65 | 0.69 | 1.00–4.77 | 2.10 | 0.62 | 1.00–3.77 | 60.67*** |
| Fear of failure dimensions | |||||||
| FSE | 3.09 | 0.76 | 1.00–4.86 | 3.22 | 0.89 | 1.00–5.00 | 3.61 |
| FDSE | 2.43 | 0.87 | 1.00–4.67 | 2.71 | 0.85 | 1.00–5.00 | 12.09** |
| FIOLI | 2.74 | 0.79 | 1.00–4.80 | 2.45 | 0.84 | 1.00–4.80 | 8.58** |
| FUIO | 2.90 | 0.73 | 1.00–5.00 | 2.73 | 0.81 | 1.00–4.80 | 2.29 |
| FUF | 2.64 | 0.91 | 1.00–5.00 | 2.58 | 0.96 | 1.00–5.00 | 0.34 |
| Overall fear of failure | 2.83 | 0.64 | 1.28–4.60 | 2.82 | 0.69 | 1.04–4.48 | – |
| Sport experience | 7.64 | 4.74 | 1.0–16.0 | 7.52 | 3.59 | 1.0–15.0 | – |
- Note. FSE = fear of shame and embarrassment; FDSE = fear of devaluing one's self-estimate; FIOLI = fear of important others losing interest; FUIO = fear of upsetting important others; FUF = fear of an uncertain future. **p < .01; ***p < .001.
A multi-variate analysis of variance was conducted to examine sex differences in antisocial behaviour and the five fears of failure. A significant multi-variate main effect emerged, F(7, 323) = 16.25, p < .001, 
. Follow-up univariate analyses of variance (ANOVA) revealed several significant differences between male and female athletes (see Table 1). Specifically, compared to females, males reported higher frequency of antisocial university ( p < .001; 
) and sport ( p < .001; 
) behaviour, lower fear of devaluing one's self-estimate ( p < .01; 
) and higher fear of important others losing interest ( p < .01; 
). A further ANOVA showed no difference between males and females in overall fear of failure, F(1, 329) = .02, p > .05, 
.
Correlational analyses
Pearson correlations between all study variables are presented in Table 2 separately for males and females. Several significant weak-to-moderate relationships were identified between the five fears of failure and antisocial behaviour. For males, all fears had significant positive relationships with both types of antisocial behaviour. For females, the fears of experiencing shame and embarrassment, of devaluing one's self-estimate, and of important others losing interest had significant positive correlations with antisocial sport behaviour. However, only the fears of important others losing interest and of upsetting important others had such relationships with antisocial university behaviour. The five fears of failure were all positively inter-related in both sexes, with the strength of these relationships ranging from moderate to strong. Overall fear of failure had a moderate positive correlation with antisocial behaviour in both contexts in males, but was related to antisocial behaviour in the university context only in females. Similarly, sport experience was weakly and positively linked to antisocial behaviour in both contexts in males, but only in the sport context in females. Finally, the two types of antisocial behaviour were positively and strongly inter-related in both sexes.
| Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | |
|---|---|---|---|---|---|---|---|---|---|---|
| 1. | Antisocial university behaviour | – | .60** | –.06 | .14 | .19* | .21* | .06 | .12 | .08 |
| 2. | Antisocial sport behaviour | .48** | – | .13 | .15 | .25** | .23** | .17* | .24** | .23** |
| 3. | Fear shame and embarrassment | .16* | .17* | – | .44** | .57** | .56** | .39** | .85** | –.06 |
| 4. | Fear devaluing one's self-estimate | .19* | .23** | .42** | – | .31** | .37** | .42** | .60** | –.01 |
| 5. | Fear important others losing interest | .32** | .27** | .68** | .39** | – | .64** | .41** | .79** | –.01 |
| 6. | Fear upsetting important others | .24** | .24** | .66** | .40** | .64** | – | .41** | .80** | .01 |
| 7. | Fear uncertain future | .29** | .24** | .55** | .58** | .59** | .62** | – | .67** | .05 |
| 8. | Overall fear of failure | .28** | .28** | .87** | .64** | .84** | .83** | .81** | – | –.02 |
| 9. | Sport experience | .18** | .21** | .16* | .07 | .24** | .19* | .22** | .21** | – |
- Note. Correlations for males are below the diagonal whereas those for females are above the diagonal.*p <.05; **p <.01.
Fear of failure, sport experience, and antisocial behaviour
The main purpose of the study was to examine (a) whether fear of failure and sport experience predicted antisocial behaviour in university and sport and (b) whether sex moderated this prediction. This purpose was examined using structural equation modelling (SEM). Specifically, the first part of this purpose was examined using the approach recommended by Anderson and Gerbing (1988), whilethe second part was tested using multi-sample SEM.
All SEM analyses were conducted using the EQS 6.1 statistical package with the robust least squares estimator (Bentler & Wu, 2002). It is usual when reporting SEM results to provide fit indices as indicators of model fit. In accordance with the guidelines of Bentler (2007), the indicators used to help assess model fit were: the Satorra–Bentler Chi-square (χ2); the robust comparative fit index (CFI); the standardized root mean square residual (SRMR); and the robust root mean square error of approximation (RMSEA). A good model fit is achieved when CFI values approach or exceed .95, the SRMR approaches or is below .08, and the RMSEA approaches or is less than .06 (Hu & Bentler, 1999).
Testing the measurement model
The first step of the Anderson and Gerbing (1988) approach entails testing the measurement model, that is, the relationships of the observed items to their posited factors. The initial measurement model consisted of all items (N= 51) measuring antisocial university (n= 13) and sport (n= 13) behaviours, and overall fear of failure (n= 25). Specification of this model resulted in a reasonable fit, χ2 (1,221) = 3,873.12; CFI = .908; RMSEA = .081; SRMR = .085. As the CFI value did not approach the target criterion for good fit, a number of problematic items were eliminated to improve the factorial structure of the model. These items were identified through scrutiny of the standardized residual matrix as the fit of the hypothesized model to the data can be quantified using standardized residuals (Hofmann, 1995). The removal of items occurred through an iterative process with model fit re-evaluated following the removal of the most problematic item from the previous model. The removal of items to improve the factorial structure of a model is considered legitimate as it retains the general structure of the originally hypothesized model, whilst utilizing only the best available indicators (Hofmann, 1995). This process resulted in a final model that consisted of 36 items as indicators of three latent variables. This model fitted the data very well, χ2 (591) = 1,186.33; CFI = .967; RMSEA = .055; SRMR = .063. Factor loadings ranged from .40 to .88 (M= .67). All of the items from the final measurement model can be found in the Appendix.
Testing the structural model
The second step recommended by Anderson and Gerbing (1988) is to test the structural model. Thus, a model where overall fear of failure and sport experience predicted antisocial behaviour in the two contexts was specified. Positive relationships between fear of failure and sport experience and between the two antisocial behaviours were also specified. This model fitted the data well, χ2 (624) = 1,315.45 ( p <.05); CFI = .931; RMSEA = .058; SRMR = .059, offering support for the posited model (see Figure 1). Fear of failure and sport experience was moderate and weak positive predictors of antisocial behaviour, respectively, in both university and sport.
Final structural model. The first value in each pair of values refers to the value for females and the second for males. A * indicates non-invariance between values for females and males. All values are significant ( p < .05) with the exception of the covariance between fear of failure and sport experience in females.
Sex as moderating variable
To examine whether sex moderates the prediction of antisocial behaviour, we tested a series of sequentially constrained models to determine the measurement and structural invariance of the model across sex using multi-sample SEM. If any of the structural parameters in the model were non-invariant across males and females then moderation by sex of the specific parameter/s in question would be evidenced. To test for invariance between models, the CFI for the first combined model (i.e., configural invariance model) was compared with the models tested subsequently; ΔCFI values of less than −0.01 indicate invariance between models (see Cheung & Rensvold, 2002). Differences in model fit were tested using ΔCFI in preference to Δχ2 because the latter is influenced by sample size whereas ΔCFI is not (Cheung & Rensvold, 2002).
The series of models tested were based on the recommendations of Byrne (2006) for testing the equivalence of structural models. The fit statistics for all models tested are presented in Table 3. The first models tested were baseline models for males and females and determined whether the model fitted the data well for both groups individually. Good model fit for both males and females confirmed that this was the case. Next, we tested for configural invariance, which involved testing the model for invariance across the two groups simultaneously without placing any invariance constraints on the model. Good model fit here indicates that all items are indicators of the same factors in both groups, which was confirmed in the current analysis. Following this, we tested for metric invariance by constraining the factor loading of each item on its respective factor to be equal across the two groups. This was to determine whether factor loadings were equivalent across males and females. This analysis resulted in a ΔCFI of −.007 between the configural and metric invariance models, thus demonstrating metric invariance.
| Model | df | χ2 | CFI | SRMR | RMSEA |
|---|---|---|---|---|---|
| Baseline males | 624 | 956.44 | .935 | .072 | .055 |
| Baseline females | 624 | 857.10 | .939 | .076 | .049 |
| Configural invariance | 1,248 | 1,829.63 | .937 | .074 | .053 |
| Metric invariance | 1,281 | 1,929.93 | .930 | .083 | .055 |
| Equivalence construct variance and covariance | 1,284 | 2,235.16 | .897 | .084 | .067 |
| ECVCAB Only | 1,283 | 1,860.48 | .938 | .083 | .052 |
| Structural invariance | 1,287 | 1,805.11 | .944 | .086 | .049 |
- Note. df, degrees of freedom; χ2, Satorra–Bentler scaled Chi-square; CFI, robust comparative fit index; SRMR, standardized root mean square residual; RMSEA, root mean square error of approximation; ECVCAB Only, equivalence construct variance and covariance antisocial behaviour constrained only.
We then tested for the equivalence of construct variance and covariance across males and females by constraining the variances and covariances of all constructs to be equal across the two groups. This examined whether the variances and covariances of all constructs were equivalent across males and females. The resultant ΔCFI was −.040 and, therefore, demonstrated non-invariance. In order to determine the specific constraint/s that led to the reduction in fit, we sequentially released each invariance constraint and retested the model. This process determined that the covariance constraint between fear of failure and sport experience had caused the reduction in model fit and was therefore non-invariant across sex. Releasing this constraint and retesting the model resulted in a ΔCFI of .001 between this model and the configural invariance model. As a result of its non-invariance, the constraint in question remained released when testing the subsequent model. Finally, we tested for structural invariance by constraining causal paths to be equal across males and females. This model determined whether the causal model was invariant across the two sexes and therefore whether sex moderated the prediction of antisocial behaviour by fear of failure and sport experience. The ΔCFI for this model was .007 compared to the configural invariance model and therefore indicated structural invariance. Overall, these analyses demonstrated that the measurement and structural models were invariant across sex with the exception of the covariance between fear of failure and sport experience.
The final model explained 12% of the variance in antisocial sport behaviour and 12% of the variance in antisocial university behaviour for females, and 14% of the variance in antisocial sport behaviour and 11% of the variance in antisocial university behaviour for males. Fear of failure and sport experience had no relationship in females, but had a significant weak-to-moderate positive correlation in males. In both sexes antisocial behaviour was strongly and positively related across the university and sport contexts.
Discussion
The link between fear of failure and sport experience and student athletes’ interpersonal antisocial behaviour in education and sport has received scant attention in the literature. In addition, sex differences in antisocial behaviour in the education and sport contexts and the five dimensions of fear of failure have not been examined previously. Therefore, the present study aimed to examine whether (a) fear of failure and sport experience predicted antisocial behaviour in the university and sport contexts, and whether this prediction was the same in male and female student athletes; and, (b) sex differences existed in the levels of the five dimensions of fear of failure and the frequency of antisocial behaviour in the university and sport contexts.
Model testing showed that fear of failure moderately and positively predicted students’ antisocial behaviour in both the sport and university contexts. This is consistent with recent research that suggests that individuals adopt hostile and aggressive behaviour as a regulatory strategy to deal with their fear of failure (Conroy et al., 2009; Wright et al., 2009). Individuals high in fear of failure adopt such behaviours as self-regulatory strategies to diffuse their fear and the anticipation of failure and of its aversive consequences. These strategies are often harmful to their interpersonal behaviour (Conroy et al., 2009; Elliot & Thrash, 2004; Sagar et al., 2009; Wright et al., 2009) and performance (e.g., academic, sporting; Conroy & Elliot, 2004; Elliot & Sheldon, 1997). Although the link between fear of failure and harmful conduct has been examined previously, it has not been examined in two different achievement contexts concurrently using one sample. Our findings show that fear of failure may contribute to more frequent student engagement in antisocial behaviour in the university and sport contexts. This finding supports the claim that fear of failure is a motive that is deeply rooted in self-evaluative disposition (Conroy & Elliot, 2004) and, therefore, transfers across achievement domains. This is a novel finding that enhances knowledge on the link between fear of failure and antisocial behaviour in the university and sport contexts.
Model testing also showed that sport experience positively predicted antisocial behaviour in both contexts. Although researchers have not previously investigated sport experience and student athletes’ antisocial behaviour in sport and university concurrently, children's participation in sport and their aggressive tendencies in sport and everyday life have been studied. Specifically, Bredemeier et al. (1986) found that participation of girls in medium contact sports and of boys in high contact sports was positively linked with physical and non-physical aggression both in sport and everyday life contexts. Thus, our findings provide initial evidence that the link established by Bredemeier et al. (1986) in children, between sport involvement and harmful conduct in sport and non-sport contexts, continues into early adulthood in student athletes. This is clearly a salient finding with regard to student athletes’ overall interpersonal functioning.
We also found a strong positive correlation between the frequencies of antisocial behaviour in both contexts. Thus, those who reported engaging more frequently in antisocial behaviour in one context were also likely to report more frequent antisocial conduct in the other context. This finding supports research that has shown children's aggressive action tendencies to be positively related across contexts. Specifically, Bredemeier (1994) found that aggressive action tendencies were strongly and positively related across sport and daily life contexts in fourth-to-seventh-grade summer camp attendees. Therefore, an important finding of the current study is that student athletes who engage in antisocial behaviour more frequently than others on the university sport fields are also more likely to do so during their day-to-day educational activities.
As part of the primary aim of the current study, multi-sample analyses were conducted to determine whether sex moderated the prediction of antisocial behaviour by fear of failure and sport experience. Overall, these analyses showed that the model was largely invariant. Specifically, the prediction of antisocial behaviour in both contexts by fear of failure and sport experience and the strength of the relationship between the two context-specific antisocial behaviours did not differ significantly between males and females. This suggests that any causal links between student athletes’ fear of failure and/or sport experience and their frequency of antisocial behaviour in university and sport may exist equally in both males and females. Further, such cross-validation of the causal model provides additional support for its validity (Byrne, 2006).
Multi-sample analyses identified a significant difference in the relationship between fear of failure and sport experience. A weak-to-moderate positive relationship between these two variables was evident in males, but no significant relationship existed in females. This demonstrates that fear of failure was stronger the more years male student athletes had played their sport. However, this was not evident in female student athletes. This is another novel finding because previous research on fear of failure in athletes has not examined sport experience or sex differences in relation to fear of failure (e.g., Sagar, Lavallee, & Spray, 2007; Sagar et al., 2009; Sagar & Stoeber, 2009).
The fact that male athletes have been found to be more competitive than female athletes and to have a greater desire to win (Gill, 2002) may help to explain the sex differences in the relationship between sport experience and fear of failure. It is likely that as sport experience increases, so does the level at which athletes participate. This increased level of participation has the potential to interact with males’ competitive nature, leading to greater fear of failure in males who have participated in their sport longer due to increased importance placed on competitive outcomes at higher levels of competition. The increased pressure at high levels of competition to achieve top performance and win can bring with it an increase in fear of failure (Hosek & Man, 1989). Accordingly, it is important that psychologists working with male student athletes who have participated in medium to high contact sports for some time are acutely aware of the potential for increased levels of fear of failure in these student athletes.
The current findings suggest that sex differences in frequency of antisocial behaviour exist in student athletes’ other life domains as well as in sport. The positive link identified between sport experience and fear of failure in males, but not in females, may in part explain such differences. Other factors (e.g., empathy; see Kavussanu et al., 2009) that may explain these sex differences have been identified in sport-based research. Thus, in addition to the increases in fear of failure associated with increased sport experience in males but not in females, similar factors to those identified by Kavussanu et al. (2009) may also help explain sex differences in student athletes’ antisocial behaviour in the university context.
The second aim of the study was to examine sex differences in levels of the five dimensions of fear of failure and in the frequency of antisocial behaviour in the university and sport contexts. Findings revealed that sex differences existed in the levels of two fears. One difference was that male students reported higher levels of FIOLI than female students. This finding suggests that male student athletes have a stronger belief and anticipation than female student athletes that failure will lead to aversive relational consequences such as losing social value and influence in the performance domain, as well as losing the interest of others. Thus, male students perceive failure to be a threat to their relationships with important others and, therefore, fear it more than females. This novel finding indicates that maintaining social value and influence in the performance domain may be more salient to male than to female students. Having social status and value in a group (e.g., through success) reflects domination and validation of oneself through achieving recognition and popularity among group members or peers, and enables the development of relationships and bonds (Wentzel, 1998). The perceived value of social status in a group differs between males and females as a result of socialization differences (Ellis & Zarbatany, 2007). Thus, it is possible that a greater focus on achieving social status and value through success in males compared to females may explain the higher levels of the FIOLI in males compared to females.
Sex differences were also apparent in the FDSE. Specifically, females reported higher levels of this fear than males. This suggests that female student athletes have a stronger belief that failure indicates having poor ability and control over one's performance than male student athletes and, therefore, fear failure and the subsequent de-valuation of their self-estimate more than males. Research generally has shown that male and female ability-related beliefs and behaviours and perceptions of success and failure in achievement motivation follow sex-role stereotypes, and that males score higher than females on perceived competence (see Meece et al., 2006, for a review). Therefore, it is possible that the generally lower competence beliefs in females compared to males may explain why females were more fearful of devaluing their self-estimate further when failing compared to males. As such, psychologists working with female student athletes should be particularly aware of the potential for high levels of this fear.
Sex differences in reported antisocial behaviour were also evident. Males reported more frequent antisocial behaviour in both the university and sport contexts than females. Our results accord with previous research that has consistently shown that males engage more frequently than females in antisocial sport behaviour (e.g., Kavussanu et al., 2009; Kavussanu & Roberts, 2001). Previous research has reported that male university students engaged more frequently than female students in physically aggressive behaviours such as reciprocally hitting another person and verbally aggressive acts including arguing with people following disagreements (Buss & Perry, 1992). Thus, although previous research has reported sex differences in antisocial behaviour in sport and in non-sport contexts, our study is the first to show sex differences in antisocial behaviour in student athletes in sport and non-sport contexts concurrently. Further, our findings converge with previously reported cross-context sex differences in other morally relevant variables (Bredemeier, 1994). However, Bredemeier (1994) did not investigate differences in frequency of antisocial behaviour within an adult population. As such, our findings enhance knowledge on sex differences in antisocial behaviour among student athletes in the university and sport contexts.
Limitations and directions for future research
The present study revealed some novel findings, which need to be interpreted in light of the study limitations. Firstly, the study is cross sectional and, as such, does not allow for the direction of causality in our findings to be determined. In addition, the findings do not inform us as to whether student athletes’ behaviours fluctuate over time in accordance with events in the academic year (e.g., exam pressure) and sporting season (e.g., major/minor competitions, selection events). Future research should employ a longitudinal design to address these issues. Our findings are also limited to university student athletes and, thus, cannot be generalized to individuals in other settings. Future research should attempt to replicate the present findings with other populations such as those involved in recreational rather than competitive sport. It is also important to note that the frequency of antisocial behaviour was lower in the university context than in the sport context. Although fear of failure and sport experience predicted antisocial behaviour similarly across the two contexts, the actual frequency of antisocial conduct in the university context was low. Finally, researchers should examine whether fear of failure leads to interpersonal antisocial behaviour through other personality factors that have been associated with fear of failure (e.g., perfectionistic tendencies; Conroy, Kaye, & Fifer, 2007; Sagar & Stoeber, 2009).
Conclusions
The findings of the present study suggest that fear of failure and sport experience may increase the frequency of antisocial behaviour both in and out of sport in student athletes, thus extending morality and fear of failure research. Further, this study identifies sex differences in two aspects of fear of failure (i.e., fears of important others losing interest and of devaluing self-estimate) and shows higher frequency of antisocial behaviour in male student athletes compared to females in university and sport concurrently. These findings have implications for the understanding of student athletes’ interpersonal antisocial behaviour in the university and sport contexts, and extend knowledge in this area as well as on the fear of failure construct more generally.
Appendix
Items used in final structural model
- •
Antisocial university behaviour items retained: Argued with a student; retaliated after being hurt by a student; undermined a student; tried to wind up a student; swore at a student; tried to injure a student; intentionally distracted a student during a class; intentionally broke the rules of the university; showed frustration at a student's poor performance; physically intimidated a student.
- •
Antisocial sport behaviour items retained: Deliberately fouled an opponent; verbally abused a team-mate; retaliated after a bad foul; criticized a team-mate; tried to wind up an opponent; swore at a team-mate; tried to injure an opponent; intentionally distracted an opponent; showed frustration at a team-mate's poor play; intentionally broke the rules of the game; physically intimidated an opponent.
- •
Fear of failure items retained: When I am failing, it is often because I am not smart enough to perform successfully; when I am failing, my future seems uncertain; when I am failing, it upsets important others; when I am failing, I lose the trust of people who are important to me; when I am not succeeding, I am less valuable than when I am succeeding; when I am not succeeding, people are less interested in me; when I am failing, I am worried about it affecting my future plans; when I am not succeeding, people seem to want to help me less; when I am failing, important others are not happy; when I am not succeeding, people tend to leave me alone; when I am failing, important others are disappointed; when I am failing, I believe that everybody knows I am failing; when I am not succeeding, some people are not interested in me anymore; when I am failing, I feel that my doubters feel that they were right about me; when I am not succeeding, my value decreases for some people.
