• No results found

Strategic behavior in the classroom : the role of causal attribution dimensions in self-handicapping in a Dutch adolescent sample

N/A
N/A
Protected

Academic year: 2021

Share "Strategic behavior in the classroom : the role of causal attribution dimensions in self-handicapping in a Dutch adolescent sample"

Copied!
37
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

(eds

Strategic behavior in the classroom

The role of causal attribution dimensions in self-handicapping in a Dutch adolescent sample

Master’s thesis Educational Sciences University of Amsterdam L. M. van der Sluijs Supervised by dr. J. A. Schuitema Amsterdam, October 2013

(2)

Strategic behavior in the classroom

The role of causal attribution dimensions in self-handicapping in a Dutch adolescent sample

Master’s thesis Educational Sciences University of Amsterdam L. M. van der Sluijs Supervised by dr. J. A. Schuitema Amsterdam, October 2013

(3)

Abstract

This study examined what types of attributions, in terms of both outcome (i.e., success or failure) and perceived locus, stability, and controllability, predicted academic self-handicapping in a sample of 310 Dutch secondary school students. In addition, the effects of attributional bias in perceived locus and stability of success and failure attributions were investigated. The Academic Self-Handicapping Strategies Scale (Midgley et al., 2000) and the revised Causal Dimension Scale (McAuley, Duncan & Russell, 1992) were used to assess self-handicapping and perceived causality, respectively. As hypothesized, internal attributions for failure predicted greater self-handicapping tendencies, regardless of school type, gender, or ethnic background. Internal attributions for success, on the other hand, predicted lower self-handicapping tendencies for students with a foreign background. Results also supported the hypothesis that students with a negativity bias in perceived locus would report greater self-handicapping tendencies than other students. The effects of attributional bias in perceived stability differed by ethnic background. Contrary to expectations, controllable attributions for failure predicted greater self-handicapping tendencies. Further research is required to determine the mechanisms associated with this effect. The findings from this study demonstrate the importance of distinguishing between success and failure attributions, as the effects of causal attribution dimensions may depend on the outcome for which the attribution was made. The results also imply that the effects of success and failure attributions cannot fully be established without taking both attributions into account.

In deze studie is onderzocht welke soorten attributies, in termen van zowel uitkomst (d.w.z., succes of mislukking) als ervaren locus, stabiliteit en controleerbaarheid, zelfondermijnend leergedrag voorspelden in een steekproef van 310 Nederlandse middelbare scholieren. Ook is gekeken naar de effecten van ervaren locus en stabiliteit van ieders attributie voor succes en mislukking gecombineerd. Zelfondermijnend leergedrag en de door de leerling ervaren oorzakelijkheid zijn gemeten met respectievelijk de Academic Self-Handicapping Strategies Scale (Midgley et al., 2000) en de gereviseerde Causal Dimension Scale (McAuley, Duncan & Russell, 1992). Zoals verwacht voorspelden interne attributies voor mislukkingen een grotere mate van zelfondermijnend leergedrag, ongeacht schoolniveau, sekse, of afkomst. Interne attributies voor succes voorspelden juist een kleinere mate van zelfondermijnend leergedrag, maar alleen voor leerlingen van niet-Nederlandse afkomst. Leerlingen die mislukkingen meer aan zichzelf toeschreven dan succes, scoorden zoals verwacht hoger op zelfondermijnend leergedrag. De effecten van ervaren stabiliteit van attributies voor succes en mislukkingen gecombineerd waren afhankelijk van de afkomst van leerlingen. In tegenstelling tot onze verwachtingen, voorspelden controleerbare attributies voor mislukkingen een grotere mate van zelfondermijnend leergedrag. Nader onderzoek zal moeten uitwijzen welke mechanismen verantwoordelijk zijn voor dit effect. De bevindingen van deze studie tonen aan dat het belangrijk is om onderscheid maken tussen attributies voor succes en attributies voor mislukkingen, aangezien de effecten van causaliteitsdimensies kunnen afhangen van de uitkomst waarvoor de attributie is gemaakt. Daarbij laten de resultaten zien dat de effecten van attributies voor succes en mislukkingen niet geheel kunnen worden vastgesteld zonder beide attributies in ogenschouw te nemen.

(4)

Table of contents

1. Introduction ... 5

2. Theoretical framework ... 7

2.1 Defining self-handicapping ... 7

2.2 Attributions for prior performance outcomes as elicitors of self-handicapping ... 8

2.3 The role of locus, stability, and controllability of success and failure attributions ... 10

2.4 Attributions for success and failure combined: the role of attributional bias ... 12

2.5 Present study and relevance ... 14

2.6 Hypotheses ... 15 3. Method ... 16 3.1 Design ... 16 3.2 Participants ... 16 3.3 Measures ... 16 3.4 Procedure ... 18 3.5 Analyses ... 18 4. Results ... 20 4.1 Self-handicapping ... 20 4.2 Correlates of self-handicapping ... 20

4.3 Predicting self-handicapping from locus, stability, and controllability of success and failure attributions ... 21

4.4 Predicting self-handicapping from attributional bias ... 23

5. Discussion ... 26

6. References ... 31

(5)

Introduction

Students can go to great lengths to protect their sense of self-worth and maintain an able and worthy appearance to others. Sometimes they may resort to the use of self-handicapping strategies (Jones & Berglas, 1978), like procrastination and deliberate reduction of effort. The created impediment will conveniently deflect attention away from their ability, should subsequent performance be unsatisfactory (Covington, 1984, 2009). Self-handicapping comes with a price, however: not only is self-handicapping negatively related to actual performance (Martin, Marsh & Debus, 2001, 2003; McCrea & Hirt, 2001; Midgley, Arunkumar & Urdan, 1996; Midgley & Urdan, 1995; Murray & Warden, 1992), poor performance, in turn, appears to lead to an increased tendency to self-handicap over time (Gadbois & Sturgeon, 2011). Furthermore, it eventually undermines students’ self-esteem (Zuckerman & Tsai, 2005; Zuckerman, Kieffer & Knee, 1998), academic self-efficacy (Gadbois & Sturgeon, 2011; Midgley & Urdan, 1995) and sense of self-worth (Midgley & Urdan, 1995).

In what case, then, would students sometimes rather self-handicap than accept the challenge of doing their best? Jones and Berglas (1978) proposed that individuals are especially likely to use self-handicapping strategies when they are uncertain about their competence and want to avoid being judged incompetent. In academic achievement situations, the feeling one might fail an important test could prompt the ‘need’ to self-handicap. Attributions for prior performance outcomes seem to affect future success expectancies (Weiner, 1979, 2010). Causal attribution dimensions (i.e., locus, stability and controllability) are of particular importance. For instance, perceiving the cause of failure as relatively stable could lower subsequent performance outcome expectations (Weiner, 1985). The feeling one might fail could, in turn, motivate students to minimize the implications of that outcome through self-handicapping (Covington, 2009; Jones & Berglas, 1978; Urdan & Midgley, 2001). Other types of attributions, such as internal and stable attributions for success and external and unstable attributions for failure, allow individuals to remain positive about their ability to succeed and are less likely to elicit self-handicapping (Weiner, 1985).

The presumed relationship between self-handicapping and attributions has been confirmed in several empirical studies (e.g., Berglas & Jones, 1978; Greenlees, Jones, Holder & Thelwell, 2006; Martin et al., 2001, 2003; Murray & Warden, 1992; Rhodewalt, Morf, Hazzlet & Fairfield, 1991; Tice, 1991; Warner & Moore, 2004). However, most of these studies have focused on three or four specific attributions (e.g., ability, effort, luck, context), from which causal dimensions were inferred. This has been termed “the fundamental attribution researcher

(6)

error” (Russell, 1982, p. 1137), considering that the researcher does not necessarily perceive causes in the same way as the attributor (Hau & Salili, 1993; Weiner, 1979). Research on the relationship between attributions and self-handicapping has generally paid little attention to how students themselves perceive their attribution in terms of the causal dimensions. Moreover, studies investigating the way in which attributions affect self-handicapping have not distinguished between success and failure attributions. This is problematic since a given attribution could elicit quite different reactions, depending on the outcome for which it was made (Cheng & Furnham, 2001; Graham, 1991; Weiner, 1985). One purpose of the present study was to examine what types of attributions, in terms of both outcome (i.e., success or failure) and perceived locus, stability, and controllability, would predict academic self-handicapping. First, the effects of success and failure attributions were examined separately. Because we expected contrasting effects of perceived locus and stability of success and failure attributions, we felt it would also be interesting to investigate the effects of perceived locus and stability of success and failure

attributions combined. Therefore, a related purpose of this study was to determine whether students who displayed a negativity bias, that is, students who perceived their attribution for general academic success as less internal and stable than their attribution for general academic failure, would self-handicap more often than other students. To our knowledge, no study to date has tested this hypothesis.

(7)

Theoretical framework

2.1 Defining self-handicapping

Over the last three decades, numerous behaviors and dispositions have been suggested as examples of self-handicapping (Higgins, Snyder & Berglas, 1990; Hirt, Deppe & Gordon, 1991; Jones & Berglas, 1978; Urdan & Midgley, 2001; Zuckerman & Tsai, 2005). These include procrastination, lack of preparation and effort, drug and alcohol use, lack of sleep, illness and overinvolvement in after-school activities. Research has generally distinguished between behavioral handicapping (e.g., strategic reduction of effort, drug use) and reported handicapping (e.g., claimed test anxiety). Unlike behavioral handicapping, reported self-handicapping does not involve actively creating an impediment and is therefore less likely to actually impede performance (Hirt et al., 1991). In this study, the focus is on behavioral handicapping. The strategic nature of any given behavior is believed to be the essence of self-handicapping (Urdan & Midgley, 2001). That is, students who procrastinate or are overinvolved in after-school activities are only considered to be self-handicapping if they primarily engage in this behavior to use it as an explanation for poor performance. In order to clarify this, we need to address the self-worth theory of achievement motivation (Covington, 1984, 2009). This theory assumes that students are particularly motivatedto protect their sense of self-worth, for which ability perceptions are critical. Ability is deemed essential to succeed (i.e., obtaining a desired outcome) and inability is generally seen as the primary cause of failure (i.e., not obtaining a desired outcome). Hence, failure poses a threat to perceptions of ability and sense of self-worth.

The obvious way to handle this threat is to avoid failure and strive for success instead. However, rather than being concerned with avoiding failure, self-handicappers are primarily concerned with how they would appear to others if they performed poorly (Urdan & Midgley, 2001). In the face of failure, the self-handicapper’s primary motive is to avoid the implications of that outcome, namely, “that they are stupid and hence unworthy” (Covington, 1992, p. 74). By creating an impediment that is in itself sufficient to explain poor performance, self-handicappers hope to protect their sense of self-worth and maintain a worthy and able appearance to others. For example, a student could stay out late the night before an important exam so that poor performance can be attributed to tiredness and lack of preparation, rather than lack of ability. If the student was able to succeed in spite of the handicap, one could infer that the student is highly able (Kelley, 1973; Urdan & Midgley, 2001).

(8)

2.2 Attributions for prior performance outcomes as elicitors of self-handicapping

Regardless of potential self-enhancement, behavioral self-handicapping appears to serve primarily as a self-protecting strategy (Urdan & Midgley, 2001). Therefore, it is likely to be called upon when students feel they need to protect perceptions of their ability – that is, when they are uncertain about their competence or expect to fail. Indeed, whether or not students are inclined to self-handicap, depends largely on their performance outcome expectations (Jones & Berglas, 1978; Urdan & Midgley, 2001). These expectations appear to be determined by attributions for prior performance outcomes – something that is elaborated in Weiner’s (1985, 2010) attributional theory of achievement motivation. This theory argues that not only an outcome itself, but also its causal ascription generates cognitive, affective and behavioral reactions. An outcome in itself gives rise to general emotions (e.g., happy for success and sad for failure), whereas its causal ascription evokes more distinct reactions. The cognitive, affective and behavioral reactions that attributions may evoke are best understood in light of their underlying properties. Weiner (1979) identified three causal attribution dimensions: locus, stability and controllability. A conventional classification of attributions according to the three dimensions is illustrated in Table 1. Ability, for example, is typically seen as an internal, stable and uncontrollable cause, whereas luck is generally considered to be external, unstable and uncontrollable. Although this classification may suggest otherwise, each dimension is a continuum and not a dichotomy (Weiner, 1985).

Several studies (e.g., Martin et al., 2001, 2003; Warner & Moore, 2004) have established that some attributions are more likely than others to elicit self-handicapping as a behavioral reaction. In each of these studies, attributions for general academic performance were measured – no distinction between success and failure attributions was made. The researchers focused on three or four specific attributions (e.g., ability, effort, luck, context), from which causal

dimensions were inferred according to the classification in Table 1. Martin and colleagues (Martin

et al., 2001, 2003) reported that attributing general academic performance to external causes (i.e.,

luck, context) was a strong positive predictor of self-handicapping. Similarly, Warner and Table 1

Attributions, Classified According to Locus, Stability and Controllability

Internal External

Stable Unstable Stable Unstable

Uncontrollable Ability Mood Task difficulty Luck

Controllable Typical effort Immediate effort Usual help from others

Unusual help from others

Note. Adapted from “A theory of motivation for some classroom experiences”, by B. Weiner, 1979, Journal of Educational Psychology, 71, 1, p. 7.

(9)

Moore (2004) found that attributing general academic performance to luck positively predicted handicapping. In addition, they found that attributions to ability positively predicted self-handicapping.

The finding that attributions to external causes positively predicted self-handicapping is “in some respects […] surprising” (Martin et al., 2003, p. 24), considering that, when performance outcomes are attributed to external causes, there would be little self-relevant implications and therefore little need to self-handicap. Indeed, if students see their performance as externally caused, self-handicapping would seem rather pointless since its primary purpose is to externalize attributions for failure or at least discount ability attributions for that outcome (Covington, 1992; Jones & Berglas, 1978; Urdan & Midgley, 2001). In explaining this result the researchers argued that, when students see their performance outcomes as externally caused, they perceive little control over their ability to maintain success or avoid failure, as a response to which they are inclined to self-handicap (Martin et al., 2003; Warner & Moore, 2004).

Although certainly possible, this argument is based on the assumption that the researcher perceives causes in the same way as the attributor – a phenomenon that Russell (1982) has called “the fundamental attribution researcher error” (p. 1137). The conventional classification, as illustrated in Table 1, is not absolute; the dimensional location of any specific attribution may vary over time, between people and situations (Russell, 1982; Weiner, 1979). For example, ability can be perceived as more or less stable, depending on whether or not learning is thought to enhance ability (Dweck, 1991; Dweck & Elliot-Moskwa, 2010). Thus, whether participants indeed perceived luck and context to be external and uncontrollable causes remains uncertain, for causal dimensions were inferred by the researchers rather than rated by participants themselves. In a more appropriate methodology, the perceived dimensional location of participants’ causal attributions would be measured separately as to avoid misinterpretation (Weiner, 1983). That is, participants would be asked to rate the attribution in terms of perceived locus, stability, and controllability. McAuley, Duncan and Russell (1992) argued that a

methodology that is faithful to the attribution process would also allow participants to provide an open-ended attribution for an outcome, rather than focus on three or four specific causes. Such a methodology is less likely to affect the respondent’s perception of the situation (Hau & Salili, 1993). Accordingly, the present study used a free response-format in which participants were asked to provide their own causal explanations, of which causal dimensions were measured – not inferred.

The findings of previous studies (Martin et al., 2001, 2003; Warner & Moore, 2004) were particularly puzzling in the sense that both external (i.e., luck and context) and internal (i.e.,

(10)

ability) attributions positively predicted self-handicapping. If both external and internal attributions would elicit self-handicapping, it would not matter how students perceive their attributions in terms of locus for it would make no difference in self-handicapping tendencies. Perhaps these attributions share a common underlying property, such as perceived

uncontrollability, which is responsible for the greater self-handicapping tendencies. It is also possible that participants considered the three attributions to be equally internal, in which case perceived locus of attributions does matter. As the dimensional location of attributions was not rated by participants themselves, this remains uncertain. Another explanation lies in the fact that no distinction between success and failure attributions has been made. This is problematic since success and failure attributions are likely to evoke quite different cognitive and affective

reactions, and may consequently bring about different behavior (Cheng & Furnham, 2001; Graham, 1991; Weiner, 1985). Indeed, ability attributions for failure would imply that the student is lacking ability, whereas ability attributions for success would suggest the opposite, namely, that the student has high ability (Perry, Hechter, Menec & Weinberg, 1993). We believe that a given attribution for failure, such as ability, may elicit self-handicapping, while that same attribution for success may make students less inclined to self-handicap. For that reason, the present study examined the effects of both success and failure attributions.

2.3 The role of locus, stability, and controllability of success and failure attributions

In any case, causal attribution dimensions, rather than specific attributions, may be of particular importance in explaining self-handicapping. Each causal dimension is uniquely related to a set of cognitive and affective reactions, the nature of which depends on the outcome (i.e., success or failure) for which the attribution was made (see Graham & Williams, 2009; Weiner, 1985 for an overview of empirical studies). Some of these reactions could make students inclined to self-handicap (Covington, 1984, 2009; Urdan & Midgley, 2001). This is elaborated in the following three paragraphs.

The locus dimension refers to whether the cause reflects an aspect of oneself. The perceived locus of attributions determines the extent to which an outcome is seen to have self-relevant implications. It is associated with changes in self-esteem, academic self-efficacy and sense of self-worth, the magnitude of which depends on the extent to which the cause is felt to be internal (Weiner, 1985, 2010). Failure that is generally perceived to be due to internal factors may be an important incentive for self-handicapping. Indeed, the more students perceive their attributions for failure as internal, the larger the negative impact on their sense of self-worth would be (Weiner, 1985, 2010). This could fuel their urge to protect it, consequently making

(11)

them more inclined to self-handicap (Covington, 1984, 2009; Zuckerman & Tsai, 2005). In contrast, when failure is generally perceived to be due to external factors there would be little impact on students’ sense of self-worth and therefore little need to self-handicap. Thus, the more internal the causes of failure are perceived to be, the more students may be inclined to self-handicap. With respect to perceived locus of attributions for success, we suggest the opposite. Success that is not attributed to the self would have little – if any – positive self-relevant implications (Weiner, 1985, 2010). In this case, the negative implications of failure may be particularly threatening to students’ sense of self-worth, considering that success would have (too) little positive self-relevant implications to make up for it. Thus, the less internal the causes of success are perceived to be, the more students may be inclined to self-handicap in order to limit the negative implications of failure. Success that is generally perceived to be due to internal factors is less likely to elicit self-handicapping because of the large positive self-relevant

implications it entails.

The stability dimension reflects whether the cause can change over time and is associated with (shifts in) performance outcome expectations (Weiner, 1985, 2010). Failure that is generally attributed to stable causes may be a major elicitor of self-handicapping, for it would make future failure seem likely. Indeed, if the cause of failure is deemed to prevail in the future, then so would the outcome (Weiner, 1985, 2010). The more stable the attribution for failure is perceived, the more students would expect to fail and the more they would be motivated to avoid the implications of that outcome rather than accept the challenge of doing their best (Covington, 2009; Jones & Berglas, 1978; Urdan & Midgley, 2001). Unstable attributions for failure, on the other hand, have little influence on performance outcome expectations and may therefore be less likely to induce self-handicapping. We believe the opposite may be true for perceived stability of attributions for success. Success that is generally attributed to stable causes would make future success seem likely, in which case students may feel safe to do their best, rather than self-handicap. In contrast, success that is generally attributed to causes that are considered unstable could evoke uncertainty – unstable attributions for success make it difficult to estimate the likelihood of future success – and is therefore more likely to elicit

self-handicapping.

The controllability dimension is concerned with whether the cause is under one’s own volitional control. We do not expect opposite effects of perceived controllability of success and failure attributions, as we did for perceived locus and stability. Failure that is generally attributed to causes that are deemed uncontrollable could give rise to uncertainty about one’s ability to avoid failure and obtain success instead, which may motivate students to self-handicap (Martin et

(12)

al., 2003). Similarly, success that is generally attributed to uncontrollable causes could give rise to

uncertainty about one’s ability to maintain success and avoid failure. In contrast, success that is attributed to controllable causes could increase students’ confidence in their ability to maintain success and avoid failure, in which case there would be little need to self-handicap. Failure that is attributed to controllable causes may elicit guilt (Weiner, 1985), but may also allow the individual to remain confident about their ability to obtain success because they feel they can control and therefore change the main cause of the outcome. In other words, perceiving the cause of failure as controllable may decrease the extent to which the outcome lowers performance outcome expectations, consequently diminishing the need to self-handicap (Perry et al., 1993).

2.4 Attributions for success and failure combined: the role of attributional bias

Attributions do not necessarily correspond to the actual causes of performance

outcomes. In fact, individuals are likely to be biased in their attribution process. A meta-analysis of 266 studies showed that people make substantially more stable and internal attributions for positive events than for negative events, a phenomenon known as positivity bias (Mezulis, Abramson, Hyde & Hankin, 2004). Such a positivity bias is not necessarily present in every individual – some students may perceive success and failure attributions as equally stable and internal, whereas others may display a negativity bias and perceive the causes of failure as more internal and stable than the causes of success (Boyer, 2006; Graham & Williams, 2009; Rowe & Lockhart, 2005).

A positivity bias in perceived locus and stability of success and failure attributions appears to serve an adaptive function. Several studies have reported that an attenuated or absent positivity bias is detrimental to people’s physical and psychological well-being (see Mezulis et al., 2004 for an overview of empirical studies). Students with a negativity bias appear to have lower self-esteem (Chan & Wong, 2013) and perform worse in school than students with a positivity bias (Boyer, 2006; Rowe & Lockhart, 2005). We believe that a negativity bias may be associated with greater self-handicapping tendencies, something that has also been related to poorer academic performance (e.g., Gadbois & Sturgeon, 2011; McCrea & Hirt, 2001; Zuckerman et al., 1998), lower self-esteem and poorer well-being (e.g., Midgley & Urdan, 1995; Zuckerman & Tsai, 2005). Recall that stable and internal attributions for failure and unstable and external

attributions for success may increase self-handicapping tendencies. Therefore, perceiving the causes of failure as stable and internal while at the same time perceiving the causes of success as external and unstable (i.e., a negativity bias) may considerably increase students’ self-handicapping

(13)

tendencies. The present study sought to determine whether students who displayed a negativity bias would indeed self-handicap more often than other students.

With respect to perceived locus of success and failure attributions, the present study compared students who displayed a negativity bias, students who displayed a positivity bias, students who displayed a bias toward internal attributions and students who displayed a bias toward external attributions with one another. Students who displayed a bias toward internal or external attributions rated both success and failure attributions as relatively internal or external, respectively. It should be noted that this does not exclude the possibility of a (less pronounced) positivity or negativity bias. Students who clearly displayed a positivity bias in perceived locus rated their success attribution as relatively internal and their failure attribution as relatively external. We shall refer to these students as self-serving, since attributing success more to internal causes while attributing failure more to external causes is generally referred to as a self-serving bias (Campbell & Sedikides, 1999; Miller & Ross, 1975; Mezulis et al., 2004). Students who displayed a negativity bias in perceived locus rated their success attribution as relatively external and their failure attribution as relatively internal. Attributing success more to external causes while attributing failure more to internal causes has been referred to as a self-defeating bias (Chan & Wong, 2013; Lloyd, Walsh & Shehni Yailagh, 2005; Wichmann, Coplan & Daniels, 2004). Hence, students who displayed a negativity bias in perceived locus of success and failure attributions shall be referred to as self-defeating. The self-defeating nature of their attributional bias may be reflected in lower levels of self-esteem, considering that success that is attributed to external factors would have too little self-relevant implications to compensate for the large negative relevant implications of failure that is attributed to the self. Therefore, self-defeating students may be more motivated than other students to prevent further damage to their sense of self-worth through self-handicapping (see Covington, 1992). To our knowledge, no study to date has tested this hypothesis.

As to perceived stability of success and failure attributions, we compared students who displayed a negativity bias, students who displayed a positivity bias, students who displayed a bias toward stable attributions and students who displayed a bias toward unstable attributions with one another. Students who displayed a bias toward stable or unstable attributions rated both success and failure attributions as relatively stable or unstable, respectively. Again, this does not exclude the possibility of a (less pronounced) positivity or negativity bias. Students who clearly displayed a positivity bias in perceived stability rated their success attribution as relatively stable and their failure attribution as relatively unstable. We shall refer to these students as optimistic, considering that they expect the cause of success, but not the cause of failure to prevail in the

(14)

future. Students who displayed a negativity bias in perceived stability rated their success attribution as relatively unstable and their failure attribution as relatively stable. They shall be referred to as pessimistic, for they expect the cause of failure, but not the cause of success to prevail in the future. The labels pessimistic and optimistic have been used to refer to a negativity or positivity bias in a broader sense (see Graham & Williams, 2009). In the present study, they were chosen to reflect the association between perceived stability of one’s success and failure attributions and performance outcome expectations (see Weiner, 1985). Because of their rather low performance outcome expectations, pessimistic students may be particularly motivated to avoid the implications of failure through self-handicapping, rather than accept the challenge of doing their best (see Covington, 2009; Jones & Berglas, 1978; Urdan & Midgley, 2001). To our knowledge, no study to date has tested this hypothesis.

2.5 Present study and relevance

The present study was based on two presumptions. Recall that perceived locus, stability, and controllability of the same attribution may vary considerably over time, between people and situations. Therefore, we believe that not attributions themselves, but rather their causal

dimensions – as perceived by the attributor – may be of key importance to explain self-handicapping. To what extent the way students themselves perceive their attributions indeed explains self-handicapping, has not yet been investigated. In addition, we believe that what cognitive, affective and behavioral reactions an attribution is likely to evoke depends largely on whether the attribution was made for success or failure. As noted earlier, this distinction between implications of success and failure attributions has not been acknowledged in previous studies on self-handicapping. The purpose of the present study was to examine what types of attributions, in terms of both outcome (i.e., success or failure) and causal dimensions, predict academic self-handicapping. Perceived locus, stability, and controllability were measured, not inferred, and participants were allowed to provide open-ended attributions. The relationship between attributions and self-handicapping was studied only in terms of their causal dimensions and the effects of success and failure attributions were examined separately. Another, related purpose of the present study was to determine whether students who displayed a negativity bias, that is, students who perceived their attribution for general academic success as less internal or stable than their attribution for general academic failure, would self-handicap more often than other students. Therefore, we also examined the effects of perceived 1) locus and 2) stability of participants’ success and failure attributions combined. The present study adds to a knowledge

(15)

base on (predictors of) self-handicapping, from which an evidence-based approach to prevent self-handicapping could eventually be developed.

2.6 Hypotheses

The following hypotheses were advanced. 1. Locus

a. Perceived locus of attributions for general academic success [Locus Success] negatively predicts self-handicapping.

b. Perceived locus of attributions for general academic failure [Locus Failure] positively predicts self-handicapping.

c. Self-defeating students report greater self-handicapping tendencies than other students.

2. Stability

a. Perceived stability of attributions for general academic success [Stability Success] negatively predicts self-handicapping.

b. Perceived stability of attributions for general academic failure [Stability Failure] positively predicts self-handicapping.

c. Pessimistic students report greater self-handicapping tendencies than other students. 3. Controllability

a. Perceived controllability of attributions for general academic success [Controllability Success] negatively predicts self-handicapping.

b. Perceived controllability of attributions for general academic failure [Controllability Failure] negatively predicts self-handicapping.

Previous studies (e.g., Midgley et al., 1996; Midgley & Urdan, 1995; Warner & Moore, 2004) have reported gender and ethnic background differences in (the processes associated with) self-handicapping. Self-handicapping tendencies may also vary across school type. Therefore, we also sought to explore school type, gender, and ethnic background differences in

(16)

Method

3.1 Design

The purpose of the present study was to examine what types of attributions, in terms of both outcome (i.e., success or failure) and perceived locus, stability, and controllability, predict academic self-handicapping. A second purpose of this study was to determine whether students who displayed a negativity bias, that is, students who perceived their attribution for general academic success as less internal or stable than their attribution for general academic failure, would self-handicap more often than other students. Accordingly, a quantitative, predictive correlational design was employed (see Creswell, 2012).

3.2 Participants

Three hundred ten students from three secondary schools in a large municipality in the province of South Holland participated in the study. All students were in their third year of either senior general secondary education [HAVO] (n = 186) or pre-university education [VWO] (n = 124). Approximately 56% (n = 173) of the participants were female and 44% (n = 137) were male. Most of the participants were of Dutch origin, 17% (n = 53) of the participants had a foreign ethnic background. The ratio of male to female students was approximately equal across ethnic backgrounds. Types of parental education were merged into three parental education levels: low (i.e., none, primary school, lower vocational education [LBO], pre-vocational education [VMBO], intermediate general secondary education [MAVO]), medium (i.e., secondary vocational education [MBO], senior general secondary education [HAVO], pre-university education [VWO]) and high (i.e., higher professional education [HBO], university education [universiteit]).

Participants were assigned the highest level that at least one of their respective parents

completed. Parental education level was low for 16% (n = 50) of the participants, medium for 19% (n = 60) of the participants and high for 52% (n = 160) of the participants. For 13% (n = 40) of the participants, the parental education level could not be established. It should be noted that many participants remarked that they did not exactly know what level of education their parents had completed, but answered the question nevertheless. Parental education level was therefore excluded from further analyses.

3.3 Measures

For this study we developed a questionnaire, largely based on existing measures. The complete questionnaire is presented in Appendix A. All participants completed the following

(17)

scales.

The revised Causal Dimension Scale [CDSII] (McAuley et al., 1992), from which three subscales were drawn and translated into Dutch, was used to measure how participants perceived their attributions for academic success and failure. The three subscales represent the dimensions of locus, stability, and controllability and comprised a total of nine semantic differentials (e.g., “Do you feel the cause is something permanent/temporary?”). For continuity purposes, the original 9-point response scale was transformed into a 7-point scale. Higher scores reflect higher levels of perceived internality, stability and controllability, respectively. A confirmatory factor analysis with varimax rotation was conducted for both success and failure attributions to test the proposed three-factor structure of the scale. These analyses confirmed that the three factors corresponded to the three causal dimensions, although there were some cross-loading items in the locus and stability subscales. Factor loadings of the individual items are shown in Table 2. The three factors accounted for 72% to 75% of the total variance. The subscales were internally consistent, with average Cronbach’s alphas of .77 (locus), .75 (stability), and .88 (controllability).

The Academic Self-Handicapping Strategies Scale [ASHSS] (Midgley et al., 2000), which was also translated into Dutch, was used to assess to what extent students used self-handicapping strategies. The original ASHSS consists of six Likert items, each describing a self-handicapping strategy (e.g., “Some students put off their school work until the last minute so that if they don’t do well on their work they can say that is the reason”). Some of the items were felt to describe two strategies, and were split up into two separate items. An item regarding social media was added. The resulting scale consisted of nine items. The original 5-point response scale was thought to leave to little room for differentiation and was transformed into a 7-point response Table 2

Factor Loadings of the Causal Dimension Items

Attributions for success Attributions for failure

Item Locus Stability Controllability Locus Stability Controllability

1 .794 --- --- .658 .439 --- 2 --- --- .845 --- --- .880 3 .448 .608 --- --- .775 --- 4 --- --- .881 --- --- .906 5 --- .761 --- --- .831 --- 6 .751 .371 --- .718 .445 --- 7 --- .868 --- --- .771 --- 8 --- --- .885 --- --- .900 9 .784 --- --- .871 --- ---

Note. Factor loadings below .3 were suppressed. The three factors contain the following items (see

(18)

scale, with end points labeled “never” (1) and “always” (7). Higher scores reflect a more frequent use of self-handicapping strategies. Because the strategic nature is believed to be the essence of self-handicapping, participants were specifically asked to rate how often they engaged in such behavior for the stated reason. The scale showed good internal consistency with a Cronbach’s alpha of .84.

3.4 Procedure

The study was introduced as research on learning strategies. Students were informed that participation in the study was voluntary, anonymous and confidential. They were also told that there were no right or wrong answers. The printed questionnaires were completed during class in approximately fifteen minutes. Participants were asked to provide an open-ended attribution for 1) general academic success and 2) general academic failure and to complete the three subscales drawn from the CDSII for both attributions. They also completed the ASHSS.

3.5 Analyses

Mean scores were calculated for participants who had filled in at least two thirds of the items of the (sub)scale in question. These scores were used in subsequent statistical analyses, for all of which an alpha level of .05 was used. A three-way factorial ANOVA was conducted to test whether self-handicapping tendencies varied across school type (i.e., HAVO vs. VWO), gender (i.e., male vs. female), and ethnic background (i.e., Dutch vs. foreign). Simple main effects analyses were employed to examine significant interaction effects.

Bivariate correlations were calculated to examine the magnitude and direction of the relationship between self-handicapping and perceived locus, stability, and controllability of success and failure attributions for the sample as a whole. Hierarchical regression analyses were conducted to test the hypotheses that Locus Success, Stability Success, Controllability Success and Controllability Failure would negatively predict self-handicapping, and Locus Failure and Stability Failure would positively predict self-handicapping. We also explored school type, gender, and ethnic background differences in self-handicapping tendencies. Inclusion of interaction terms enabled us to explore whether the relationship between attributions and self-handicapping differed by school type, gender, or ethnic background. To reduce multicollinearity between interaction terms and their constituents, continuous predictor variables were centered around their respective means (see Smith & Sasaki, 1979). Interaction terms were created by computing the product of each centered continuous predictor variable and the school type, gender, and ethnic background variable, respectively. Extraneous variables that showed no significance in the regression analyses were excluded from further analyses.

(19)

Causal dimensions for success and failure attributions were transformed into dichotomous categories (i.e., high vs. low scores), using the median as a cutoff point. By combining the two outcomes, four independent attributional bias categories were created for each causal dimension. The classification of attributional bias in perceived locus and stability of success and failure attributions is summarized in Table 3. For clarity purposes, we used

contrasting labels (e.g., internal, external). It should be noted, however, that placement in any of the categories is relative to the other participants in the sample. Participants with a bias toward external attributions, for example, did not necessarily perceive both success and failure attributions as external – rather, they perceived both attributions as less internal than did the other participants in the sample. Dependent samples t tests were conducted to test the

presumption that defeating and pessimistic students displayed a negativity bias, whereas self-serving and optimistic students displayed a positivity bias. We also sought to determine whether a positivity or negativity bias was present in the other categories. To test the hypotheses that participants with a negativity bias in perceived locus and stability would self-handicap more often than other participants, two ANOVAs were conducted, in which attributional bias and

extraneous variables (i.e., school type, gender, ethnic background) that showed significance in the regression analyses were included as a factor. Significant interaction effects were examined with simple main effects analyses. Tukey’s HSD post-hoc tests were used as a follow-up to significant main effects, to analyze which categories differed from one another.

Table 3

Attributional Bias Categories for Perceived Locus and Stability of Success and Failure Attributions

Locus Stability

Failure – High Failure – Low Failure – High Failure – Low

Success – High Internal Self-serving Success – High Stable Optimistic

(20)

Results

4.1 Self-handicapping

Responses on the ASHSS indicated that frequent self-handicapping was uncommon in this sample (M = 2.98, SD = 1.17). Nevertheless, mean self-handicapping scores of a

considerable proportion of the sample (17%) exceeded the mid-point of the 7-point response scale, which was anchored at “never” and “always”. In addition, each individual strategy was endorsed (i.e., scored above the mid-point) by 11% to 38% of the participants, with an average of 23%. Self-handicapping strategies most endorsed were letting oneself get distracted by social media (38%), overinvolvement in after-school activities (31%) and procrastination (29%). No significant main effects or two- or three-way interaction effects of school type, gender, or ethnic background on self-handicapping tendencies were evident.

4.2 Correlates of self-handicapping

This study sought to examine what types of attributions, in terms of both outcome (i.e., success or failure) and causal dimensions, predict academic self-handicapping. As a first step, Pearson product-moment correlations were computed. The results of these bivariate correlations are presented in Table 4. In line with expectations, a significant, negative correlation was found between self-handicapping and Locus Success (r = -.138), whereas a significant, positive

correlation was found between self-handicapping and Locus Failure (r = .182). Contrary to expectations, a significant, positive correlation was found between self-handicapping and Controllability Failure (r = .189). No significant correlations were found between

self-handicapping and Stability Success, between self-self-handicapping and Stability Failure, and between self-handicapping and Controllability Success.

Table 4

Pearson Product-Moment Correlations (n = 304)

Variable .1 .2 .3 .4 .5 .6 .7 1. Self-handicapping 2. Locus Success -.138* 3. Stability Success -.094 .528** 4. Controllability Success -.045 .317** .251** 5. Locus Failure .182** .095 -.022 -.132* 6. Stability Failure .070 .035 .205** -.130* .566** 7. Controllability Failure .189** .071 .043 .275** .313** -.133* * p < .05; ** p < .01; *** p < .001.

(21)

4.3 Predicting self-handicapping from locus, stability, and controllability of success and failure attributions

To determine which variables significantly predicted self-handicapping for the sample as a whole and to determine whether self-handicapping tendencies and the relation with attributions differed by school type, gender, or ethnic background, we used hierarchical regression analysis. In the first step of the analysis all predictor variables, including school type, gender, and ethnic background, were entered into the model simultaneously. Next, all of the School Type

✕ Predictor Variable, Ethnic Background ✕ Predictor Variable and Gender ✕ Predictor Variable

interactions were entered into the model, using a stepwise procedure. In addition, interaction terms for School Type ✕ Ethnic Background, School Type ✕ Gender and Ethnic Background ✕

Gender were included in the second step of the analysis. The results of the regression analysis are presented in detail in Table 5.

Two variables, Locus Success (β = -0.14, p = .043) and Controllability Failure (β = 0.18, p = .012), emerged as significant predictors of self-handicapping in the first step of the regression analysis. This step was statistically significant (F(9, 292) = 3.15, p = .001) and accounted for 8.8% of the variance in self-handicapping. With the inclusion of the Ethnic Background ✕ Locus

Success interaction term in the second step (β = -0.19, p = .003), the main effect of Locus Success lost significance (β = -0.08, p = .272). No other interaction terms were significant predictors of self-handicapping. The resulting prediction model was statistically significant (F(10, 291) = 3.84, p = .000), and accounted for 12% of the variance in self-handicapping. It contained nine predictors, two of which significantly predicted self-handicapping. To create a more

parsimonious prediction model, all nonsignificant variables except for the constituents of the significant Ethnic Background ✕ Locus Success interaction term were removed in a stepwise

procedure, starting with the lowest t values. As a result of the removal of nonsignificant predictor variables, Locus Failure became a significant predictor (β = 0.17, p = .005). The regression results from the final analysis are presented in Table 6. The resulting prediction model was statistically significant (F(5, 296) = 7.12, p = .000) and accounted for 11% of the variance in

handicapping. The model contained five predictors, three of which significantly predicted self-handicapping.

(22)

Table 5

Initial Hierarchical Regression Analysis: Predicting Self-handicapping from Locus, Stability, and Controllability of Success and Failure Attributions (n = 302)

Step 1 Step 2 Predictor variable B SE B β B SE B β Step 1 School typea -0.19 0.14 -0.08 -0.22 0.14 -0.09 Genderb 0.02 0.14 0.01 0.03 0.14 0.01 Ethnic backgroundc 0.16 0.17 0.05 0.24 0.17 0.08 Locus Success -0.15 0.07 -0.14* -0.08 0.08 -0.08 Stability Success -0.03 0.07 -0.03 -0.03 0.07 -0.03 Controllability Success -0.02 0.06 -0.02 -0.00 0.06 -0.00 Locus Failure 0.11 0.07 0.13 0.13 0.07 0.14 Stability Failure 0.02 0.07 0.02 0.02 0.07 0.02 Controllability Failure 0.13 0.05 0.18* 0.11 0.05 0.15* Step 2

Ethnic BackgroundcLocus Success -0.54 0.18 -0.19** F-value 3.15** 3.84*** Intercept 3.02 3.04 R .30 .34 R2 .09 .12 ΔR2 .09** .03**

a School type was coded as HAVO = 0; VWO = 1. b Gender was coded as male = 0; female = 1. c Ethnic background was coded as Dutch background = 0; foreign ethnic background = 1.

* p < .05; ** p < .01; *** p < .001.

Table 6

Final Hierarchical Regression Analysis: Predicting Self-handicapping from Locus, Stability, and Controllability of Success and Failure Attributions (n = 302)

Step 3 Predictor variable B SE B β Step 3 Ethnic backgrounda 0.21 0.17 0.07 Locus Success -0.11 0.06 -0.10 Locus Failure 0.15 0.05 0.17** Controllability Failure 0.10 0.04 0.13*

Ethnic Backgrounda Locus Success -0.52 0.18 -0.18**

F-value 7.12***

Intercept 2.97

R .33

R2 .11

ΔR2 -.01

a Ethnic background was coded as Dutch background = 0; foreign ethnic background = 1.

(23)

4.4 Predicting self-handicapping from attributional bias

Dependent samples t tests confirmed the presumption that self-defeating students displayed a negativity bias, whereas self-serving students displayed a positivity bias in perceived locus of success and failure attributions. It was revealed that self-serving students were not the only students to display a positivity bias; a smaller positivity bias was also present in the internal and external bias categories. Sample sizes, means, standard deviations and the results of the dependent samples t tests are presented in Table 7. Cohen’s d effect sizes indicate that all bias categories represented a large bias, although the bias represented by the self-serving bias category was particularly large (Cohen, 1992).

The presumption that pessimistic students displayed a negativity bias, whereas optimistic students displayed a positivity bias in perceived stability of success and failure attributions was also confirmed. Much like the internal and external bias categories, the stable and unstable bias categories were revealed to represent a less pronounced positivity bias. Sample sizes, means, standard deviations and the results of the dependent samples t tests are presented in Table 8. Cohen’s d effect sizes indicate that the optimistic, pessimistic and unstable bias categories represented a large bias – although the bias represented by the optimistic bias category was particularly large – and that the stable bias category represented a moderate bias (Cohen, 1992).

To examine the effects of ethnic background and attributional bias in perceived 1) locus and 2) stability on self-handicapping, two 2 (ethnic background) ✕ 4 (attributional bias) factorial

ANOVAs were conducted. School type and gender showed no significance in the regression analyses and were not included as a factor. Mean self-handicapping scores and standard deviations for each bias category are presented in Table 9.

A two-way ANOVA was conducted to examine the effects of ethnic background and attributional bias in perceived locus on self-handicapping. No (interaction) effect of ethnic background was found. A main effect was found for bias in perceived locus (F(3, 294) = 4.85, p = .003, ηp2 = .047). As hypothesized, Tukey’s post-hoc tests showed that students with a

self-defeating bias scored significantly higher on self-handicapping than students with an internal (p = .036), external (p = .011), or self-serving (p = .002) bias. Other pairwise comparisons were not significant.

Another two-way ANOVA was conducted to examine the effects of ethnic background and attributional bias in perceived stability on self-handicapping. No main effect of ethnic

background was found. A main effect was found for bias in perceived stability (F(3, 294) = 3.41,

p = .018, ηp2 = .034). Nevertheless, none of the pairwise comparisons using Tukey’s post-hoc

(24)

Table 7

Perceived Locus of Success and Failure Attributions by Attributional Bias: Sample Sizes, Means, Standard Deviations and Dependent Samples t Tests (n = 304)

Locus Success Failure

n M SD M SD t d Internal 77 6.04 0.57 5.42 0.89 6.00*** 0.83 Self-serving 63 5.93 0.51 2.87 0.79 24.00*** 4.60 Self-defeating 71 4.19 0.78 5.01 0.63 -6.11*** -1.17 External 93 4.15 0.71 3.20 0.77 8.27*** 1.27 * p < .05; ** p < .01; *** p < .001. Table 8

Perceived Stability of Success and Failure Attributions by Attributional Bias: Sample Sizes, Means, Standard Deviations and Dependent Samples t Tests (n = 304)

Stability Success Failure

n M SD M SD t d Stable 103 5.14 0.76 4.55 0.95 5.91*** 0.68 Optimistic 61 5.16 0.71 2.24 0.59 24.31*** 4.49 Pessimistic 70 3.38 0.70 4.30 0.87 -6.27*** -1.17 Unstable 70 3.21 0.81 2.16 0.66 10.42*** 1.43 * p < .05; ** p < .01; *** p < .001. Table 9

Attributional Bias Categories: Mean Self-handicapping Scores and Standard Deviations

Locus Stability

Attributional bias M SD Attributional bias M SD

Internal 2.92 1.26 Stable 2.91 1.24

Self-serving 2.72 1.19 Optimistic 2.82 1.13

Self-defeating 3.44 1.23 Pessimistic 3.32 1.23

External 2.88 0.94 Unstable 2.90 0.99

background was also found (F(3, 294) = 3.08, p = .028, ηp2 = .030). With respect to students

with a Dutch background, simple main effects analyses confirmed that pessimistic students (M = 3.24, SD = 1.28) scored significantly higher on self-handicapping than optimistic students (M = 2.75, SD = 1.09; p = .027) or students who displayed a bias toward unstable attributions (M = 2.82, SD = 0.99; p = .050). Pessimistic students with a Dutch background did not report significantly greater self-handicapping tendencies than Dutch students who displayed a bias toward stable attributions (M = 3.04, SD = 1.23; p = .312). As to students with a foreign ethnic background, we found that pessimistic students (M = 3.66, SD = 0.95) reported significantly greater self-handicapping tendencies than students who displayed a bias toward stable

(25)

1.29; p = .383) or students who displayed a bias toward unstable attributions (M = 3.34, SD = 0.90; p = .505). It was also revealed that foreign students who displayed a bias toward unstable attributions reported significantly greater self-handicapping tendencies than foreign students who displayed a bias toward stable attributions (p = .032). Other pairwise comparisons were not significant. The results regarding students with a foreign ethnic background should be

interpreted with some degree of caution, considering that sample sizes were rather small (i.e., n ranges from 9 to 20).

(26)

Discussion

The present study was conducted for two related purposes. The first was to examine what types of attributions, in terms of both outcome (i.e., success or failure) and perceived locus, stability, and controllability, would predict academic self-handicapping. Specifically, we

investigated the relationship between self-handicapping and perceived locus, stability, and controllability of attributions for general academic success and failure in a sample of 310 Dutch secondary school students. The effects of success and failure attributions were examined

separately. A related purpose of this study was to determine whether students who displayed a negativity bias, that is, students who perceived their attribution for general academic success as less internal and stable than their attribution for general academic failure, would self-handicap more often than other students. Therefore, we also examined the effects of perceived locus and stability of participants’ success and failure attributions combined. In addition, we explored school type, gender, and ethnic background differences in self-handicapping tendencies, as well as in the relationship between attributions and self-handicapping. Eight hypotheses were advanced; three for locus and stability and two for controllability. Which of these hypotheses were supported and which were not, is discussed in the following paragraph.

The hypothesis that Locus Success would negatively predict self-handicapping was supported for participants with a foreign ethnic background, but not for participants with a Dutch background – something that is discussed later in this section. The less success was attributed to the self, the more foreign students self-handicapped. The hypothesis that Locus Failure would positively predict self-handicapping was supported for the sample as a whole. The less failure was attributed to the self, the more students self-handicapped – regardless of school type, gender, or ethnic background. The hypothesis that self-defeating students (i.e., students who displayed a negativity bias in perceived locus of success and failure attributions) would report greater self-handicapping tendencies than other students was also supported. The first two hypotheses regarding the stability dimension were not supported; neither Stability Success, nor Stability Failure significantly predicted self-handicapping. The combination of the two

attributions did yield significant effects, although the hypothesis that pessimistic students (i.e., students who displayed a negativity bias in perceived stability of success and failure attributions) would report greater self-handicapping tendencies than other students was not fully supported. The effects of attributional bias in perceived stability were found to differ by ethnic background. These results are discussed later in this section. None of the hypotheses regarding the

(27)

self-handicapping. Contrary to expectations, it was revealed that Controllability Failure positively predicted self-handicapping. This finding was considered most surprising and is discussed in the next paragraph.

It has been suggested that perceived control over the causes of failure may allow students to remain confident in their ability to avoid failure and obtain success instead, for if the cause of an outcome can be changed, then so might the outcome (Martin et al., 2003; Perry et al., 1993). Therefore, we expected perceived control over failure attributions to be associated with lower self-handicapping tendencies. Nevertheless, a relationship in the opposite direction was found. A possible explanation for this result lies in the notion that the relationship between attributions and self-handicapping is reciprocal. That is, self-handicapping appears to be induced by some types of attributions (Martin et al., 2001, 2003; Warner & Moore, 2004), whereas

self-handicapping also provides the basis for an attribution, as it enables students to attribute failure to the handicap rather than to lack of ability (Berglas & Jones, 1978; Greenlees et al., 2006; Rhodewalt et al., 1991; Tice, 1991) or effort (Murray & Warden, 1992). Because this handicap is created by students themselves, they should perceive this cause to be under their own volitional control (Jones & Berglas, 1978; Urdan & Midgley, 2001). The finding that Controllability Failure was associated with greater self-handicapping tendencies possibly reflects that self-handicapping provides the basis for failure attributions to a controllable cause, namely, the handicap.

In this respect, students may be less inclined to self-handicap over time because

perceived control over the causes of failure may allow them to remain confident in their ability to avoid failure and obtain success instead (Martin et al., 2003; Perry et al., 1993). Nevertheless, over time self-handicappers appear to handicap more rather than less (Gadbois & Sturgeon, 2011; Zuckerman & Tsai, 2005). If self-handicapping is the reason that students perceive control over the cause of failure, it makes sense that this perception of control would not reduce the need to self-handicap. Changing the outcome by altering the cause would indeed require that the

handicap is no longer present, but that is not all it requires. To think that the absence of this handicap means that students consider success to be well within reach may be a bit too optimistic – if the handicap was perceived to be the only factor that kept them from succeeding, students would not have needed to self-handicap in the first place. Students may still feel that success is unlikely, due to other, less controllable causes (e.g., lacking ability or work ethic) and may therefore be inclined to self-handicap nonetheless.

Another possible explanation for the finding that Controllability Failure positively predicted self-handicapping may be that the guilt associated with attributing failure to

(28)

Marshall, 2001; Tennen & Herzberger, 1987), which could make them more inclined to self-handicap (Covington, 1984, 2009; Zuckerman & Tsai, 2005). Further research is needed to determine the mechanisms associated with this effect. Considering that self-handicapping tendencies have been found to increase over time (Gadbois & Sturgeon, 2011; Zuckerman & Tsai, 2005), it may be possible that failure attributions to controllable causes and

self-handicapping reinforce each other. Therefore, we recommend longitudinal research on the relationship between attributions and self-handicapping, that includes associated variables such as performance outcome expectations and students’ feelings about the self (e.g., sense of self-worth, self-esteem, self-efficacy).

Interestingly, the results of the present study suggest that how students perceive their attributions in terms of locus may affect self-handicapping tendencies to a greater extent than how students perceive their attributions in terms of stability. Indeed, neither Stability Success, nor Stability Failure significantly predicted self-handicapping, whereas Locus Success did for students with a foreign ethnic background, and Locus Failure did for the sample as a whole. In addition, self-defeating students reported greater self-handicapping tendencies than other students, whereas pessimistic students did not. Like the effects of perceived locus of success attributions, the effects of attributional bias in perceived stability of success and failure

attributions differed by ethnic background. With respect to students with a Dutch background, pessimistic students reported significantly greater self-handicapping tendencies than optimistic students and than students who displayed a bias toward unstable attributions, but not than students who displayed a bias toward stable attributions. As to students with a foreign ethnic background, pessimistic students reported significantly greater self-handicapping tendencies than students who displayed a bias toward stable attributions, but not than optimistic students or students who displayed a bias toward unstable attributions. No other ethnic background, school type, or gender differences were found. Due to the fact that only 53 students with a foreign ethnic background participated in the study, we were not able to differentiate between the diverse (e.g., Western European, Eastern European, Asian, African and American) ethnic backgrounds. This severely complicates explaining the ethnic background differences found in the present study. Further research is needed to examine whether these differences persists, and in what specific ethnic backgrounds and (cultural) circumstances.

One of the major strengths of this study is that it distinguished between success and failure attributions. This distinction has provided some new insights into the relationship

between attributions and self-handicapping. Unlike previous studies (i.e., Martin et al., 2001, 2003; Warner & Moore, 2004), the present study did not find that external attributions positively

(29)

predicted self-handicapping. In fact, students who displayed a bias toward external attributions reported lower self-handicapping tendencies than self-defeating students, similar to self-serving students and students who displayed a bias toward internal attributions. With respect to

individual attributions, perceived locus of success and failure attributions had contrasting effects on self-handicapping tendencies. External attributions for success were associated with greater self-handicapping tendencies, whereas external attributions for failure were associated with lower self-handicapping tendencies. These results support our notion that what cognitive, affective, and behavioral reactions a given attribution is likely to evoke depends largely on the outcome for which it was made. Investigating the effects of attributions for general performance, without distinguishing between success and failure attributions, may very well generate inaccurate conclusions.

The results of the present study also demonstrate that it is essential to examine the effects of success and failure attributions combined, for the effects of one cannot fully be established without taking the effects of the other into account. Indeed, neither Stability Success, nor Stability Failure significantly predicted self-handicapping, whereas the combination of perceived stability of success and failure attributions into bias profiles did. Considering that students experience not only success or failure, but both, it makes sense that the combination of the two determines students’ cognitive, affective, and behavioral reactions. This applies to the locus dimension as well. Locus Success did not independently predict self-handicapping for students with a Dutch background, which suggests that how internal they attribute their success had little influence on the cognitive (e.g., performance outcome expectations) and affective (e.g., sense of self-worth) reactions associated with the behavioral reaction of self-handicapping. Nevertheless, the finding that self-defeating students – who attributed success less to the self than they did failure – reported greater self-handicapping tendencies than other students, regardless of ethnic background, indicates that perceived locus of success attributions cannot be disregarded altogether. Taken together, the results of the present study suggest that how internal students with a Dutch background attribute their success has little influence on self-handicapping tendencies as long as they do not display a self-defeating bias.

This study is not without limitations. Most importantly, the effects of causal attribution dimensions have been investigated separately. And although locus, stability, and controllability are essentially separate constructs, every attribution can be placed in a three-dimensional causal space (see Table 1). That is, an attribution is not just stable, or internal, or uncontrollable – it is all these things together and the effects of one dimension may depend on the other. For example, failure that is ascribed to an uncontrollable cause could elicit either shame when perceived as

(30)

internal, or frustration when perceived as external (Weiner, 1985). Nevertheless, cognitive, affective, and behavioral reactions have generally been documented for each individual causal dimension; how these dimensions interact remains largely unknown. Another limitation is that students’ performance outcome expectations and their feelings about the self (e.g., sense of self-worth, self-esteem, self-efficacy) have not been included in the present study. Although these variables have been associated with both causal attribution dimensions (e.g., Graham & Williams, 2009; Weiner, 1985, 2010) and self-handicapping (e.g., Covington, 1984, 2009; Jones & Berglas, 1978; Zuckerman & Tsai, 2005), it would have been interesting to investigate how causal

attribution dimensions, self-handicapping, students’ performance outcome expectations and their feelings about the self interact.

The findings from the present study make several contributions to the current literature. The distinction between success and failure attributions has provided some new insights into the relationship between attributions and self-handicapping, most importantly that the effects of causal attribution dimensions may depend on the outcome for which the attribution was made. In addition, the present study has demonstrated that the effects of success and failure

attributions cannot fully be established without taking both attributions into account. For

students with a Dutch background, it appears that perceived locus of attributions for success has little influence on self-handicapping tendencies as long as they do not display a self-defeating bias. The results of the present study suggest that attribution retraining programs should focus on altering students’ attributional bias, rather than their success or failure attributions alone. And although attribution retraining programs have generally focused on altering perceived stability of success and failure attributions (Graham & Williams, 2009; Perry et al., 1993), the results of the present study indicate that perceived locus of success and failure attributions may affect

self-handicapping tendencies to a greater extent than perceived stability of success and failure attributions. Hence, the current findings suggest that attribution retraining programs should place a greater emphasis on altering students’ attributional bias in perceived locus of success and failure attributions.

Referenties

GERELATEERDE DOCUMENTEN

Modern users of the language expenence this use of doen as somehow strange, they would not use It m thts context themselves, but rather prefer Laten But they

Using longitudinal data from two samples of 218 oncology care providers and 967 teachers, the three models discussed above were compared to each other and to a fourth model (our

Clearly this DAG does not satisfy frugality, however it satisfies the P-minimality assumption since it is Markovian and entails a CI statement that is not entailed by the true DAG

To get a better understanding of the implementation process of e-HRM, the focus in this paper lies on two dimensions which will both be from the users’ point of view, namely on (1)

Main effect and interaction effect between review platform and valence on causal attribution H 2a posits that consumers will be more likely to attribute positive online

All tests demonstrated that there was no significant influence of physical store presence as moderator between the dimensions of e-reputation (online brand characteristics, quality

Neglecting the extra delay and the additional subband filter taps strongly limits the convergence of the adaptive filters and leads to a residual undermodelling error.. Extending

It is generally assumed that permission is the original meaning of laten, the causative uses bemg denved later If that is correct, the change must defimtely have occurred before