- Email: [email protected]
Belief in luck and luckiness: Conceptual clarification and new measure validation
Personality and Individual Differences 54 (2013) 501–506
Contents lists available at SciVerse ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Belief in luck and luckiness: Conceptual clarification and new measure validation q Edmund R. Thompson a,⇑, Gerard P. Prendergast b,1 a b
School of Management, University of Bath, Bath BA2 7AY, United Kingdom School of Business, Hong Kong Baptist University, Kowloon Tong, Hong Kong
a r t i c l e
i n f o
Article history: Received 20 January 2012 Received in revised form 4 September 2012 Accepted 20 October 2012 Available online 11 December 2012 Keywords: Belief in luck Luckiness Conceptualization Measurement Scale validation Personality
a b s t r a c t Research on the dimensionality and measurement of luck beliefs has yet to produce a clear conceptual and metrical consensus. This research theorizes a bidimensional model of luck beliefs that is tested through a series of studies (total n = 1205) validating the new Belief in Luck and Luckiness Scale. Unlike existing conceptualizations and measures, this new model is applicable to both believers and non-believers in luck, and reveals belief in luck and personal luckiness to be discrete, uncorrelated, and respectively unidimensional constructs. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction Personality researchers have devoted increasing theoretical, empirical, and measurement attention to irrational beliefs about luck (André, 2006; Bridgstock, Marais, & Sturgess, 2011; Darke & Freedman, 1997a; Day & Maltby, 2003; Maltby, Day, Gill, Colley, & Wood, 2008; Wiseman & Watt, 2004; Young, Chen, & Morris, 2009). As this research interest has grown, so too have disparate conceptions about the dimensionality of luck beliefs and the means to measure these. For example, Darke and Freedman (1997b) propose a unidimensional conceptualization and measure, while André (2006) proposes a 6-dimensional, and Maltby et al. (2008) a 4-dimensional, conception and measure. Despite these developments in theoretical and metrical nuance, luck beliefs research has still to produce a clear conceptual and measurement consensus. Moreover, Maltby et al. (2008) hint that the precise nature of luck beliefs’ dimensionality may yet be obscured by two limitations in existing measures: (a) items possibly producing artifactual components unreflective of true underlying constructs and (b) unexplored potential differences between luck believers and disbelievers.
q The authors thank Colin Cooper and two anonymous reviewers for their constructive comments and suggestions. ⇑ Corresponding author. Tel.: +44 (0) 1225 386742; fax: +44 (0) 1225 386473. E-mail addresses: [email protected] (E.R. Thompson), gerard@hkbu. edu.hk (G.P. Prendergast). 1 Tel.: +852 3411 7570; fax: +852 3411 5586.
0191-8869/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.paid.2012.10.027
We build on existing research by proposing and examining a new bidimensional conceptualization and measure of luck beliefs designed to reduce potential artifactual components and to account for both luck believers and disbelievers. Unlike prior studies, ours finds support for just two discrete dimensions, belief in luck and personal luckiness, that are each themselves unidimensional, uncorrelated, and differently correlated with personality and individual difference variables. Our systematically theorized and validated new measure contributes a parsimonious tool to investigate the separate effects of, respectively, belief in luck and personal luckiness constructs on psychology and behavioral differences that have hitherto been examinable only with measures that either erroneously conflate or spuriously subdivide these two distinct luck constructs. 2. Conceptualizations and measures of luck beliefs 2.1. Belief in good luck Darke and Freedman (1997b) propose a unidimensional conceptualization of irrational belief in luck, running from belief that good luck deterministically favors particular people at one end of a continuum, to the view that luck is simply random chance at the other. To capture this conceptualization they developed the often-cited 12-item Belief in Good Luck Scale (BIGLS; 1997b). However, they found that their scale both fails to distinguish between those believing themselves lucky or unlucky (1997a), and produced a multidimensional rather than unidimensional solution when factor analyzed (1997b, p. 493, fn.3). Prendergast and
502
E.R. Thompson, G.P. Prendergast / Personality and Individual Differences 54 (2013) 501–506
Thompson (2008) confirmed this multidimensionality and additionally found the scale’s two predominant and psychometrically usable sub-scales, a general belief in the existence of deterministic luck and a belief in being personally lucky, are uncorrelated and differently predict criterion variables. 2.2. Multidimensional beliefs about luck André (2006), noting both the multidimensional nature of Darke and Freedman’s scale and other luck-related constructs like illusion of control (Wiseman & Watt, 2004), developed an 18-item 6-dimensional measure exploring positive and negative luck beliefs. While she finds good and bad luck beliefs are empirically discrete, she also finds they correlate relatively highly, suggesting their conceptual closeness. This also suggests the possibility that their empirical separation may stem from the tendency of positively and negatively worded items sometimes to produce artifactually separate components even when conceptually discrete underlying constructs do not exist (Spector, Van Katwyk, Brannick, & Chen, 1997). 2.3. Beliefs around luck Maltby et al. (2008) observe that André’s (2006) measure encapsulates only some of the beliefs around luck contained in Darke and Freedman’s scale. To address this, they developed a 22-item scale and found empirical support for a 4-dimensional model: general belief in luck; rejection of belief in luck; being lucky; and being unlucky. However, they note their own 4-dimesional model ‘provides only a reasonable fit to the data’, forms two closely conceptually related pairs, and may be ‘attributable to an artifact of scoring’ (2008, p. 659). This latter suggestion is plausible because many of their scale’s positive and negative items are duplicative except for the positive/negative valence of one or a few words (e.g. good/bad or lucky/unlucky), a circumstance likely to produce ‘item direction factors’ (Spector et al., 1997, p. 661). Additionally, Maltby et al. (2008) suggest their model’s particular dimensionality could simply reflect their sample comprising both luck believers and disbelievers. They accordingly advise that ‘some definitive studies are needed to test this explanation’ (p. 659), hinting that alternative models with fewer than four dimensions may exist.
3. A new bidimensional model: belief in luck and luckiness 3.1. Belief in luck We propose that one clear unidimensional component of luck beliefs is whether individuals believe or disbelieve in the existence of luck as a deterministic phenomenon in the first place. We conceptualize belief in luck as encompassing both good and bad luck. This accords with Maltby et al. (2008) finding no support for their proposition that discrete belief in good luck and belief in bad luck constructs exist. Hence, with irrational belief in luck (good and bad) at one end of a bipolar continuum, the conceptual issue becomes what should be at the opposite end. Conceptualizing, as Darke and Freedman (1997b) do, the antithesis of belief in luck to be recognition of random chance’s existence is illogical: Keren and Wagenaar (1985) found individuals can irrationally believe in deterministic luck while concurrently recognizing the separate existence of random chance. We hence conceptualize belief in luck to have the straightforward bipolar opposite of disbelief in luck. In this we accept that the empirically discrete separation of a general belief in luck and a rejection of belief in luck Maltby et al. (2008)
found is indeed likely the scoring artifact they suggest it might be. We therefore hypothesize: H1. Deterministic luck is something individuals believe or disbelieve in to greater or lesser degrees on a unidimensional continuum. 3.1.1. Construct validity with personality and individual differences 3.1.1.1. Personality. Maltby et al. (2008) found no association between the five-factor personality model and acceptance or rejection of belief in luck. However, because belief in luck is irrational and irrationality correlates with neuroticism (Hart & Hope, 2004), we expect luck belief may positively correlate with neuroticism. 3.1.1.2. Locus of Control (LoC). Darke and Freedman (1997b) found belief in good luck was positively related to the chance and powerful others dimensions of LoC. Maltby et al. (2008) found internal LoC correlated positively with rejection of belief in luck and negatively with general belief in luck. Hence, we expect luck belief may correlate positively with chance and powerful others, but negatively with internal LoC. 3.1.1.3. Wellbeing. Internal LoC and emotional stability (neuroticism) are suggested to form part of a higher order construct of core self- and life-evaluation broadly constituting wellbeing (Judge, Erez, Bono, & Thoresen, 2002). We therefore expect luck belief may negatively associate with wellbeing. 3.1.1.4. Demographics. André (2006) found a negative relationship between some luck beliefs and age. Accordingly, we expect belief in luck may negatively correlate with age. No prior luck belief research has examined sex differences. However, educational achievement is found likely to be attributed to luck more by males than females (Stipek & Gralinski, 1991), hence we expect males may believe in luck more than females. 3.2. Personal luckiness We propose a second component of luck beliefs is personal luckiness, but that its dimensionality may differ depending on whether or not individuals believe or disbelieve in luck to begin with. Our proposition here accords with self-consistency (Nail, Misak, & Davis, 2004) and cognitive dissonance (Festinger, 1957) theories. These suggest, respectively, that individuals seek to maintain congruence between their beliefs and to avoid the adverse psychological effects of holding incompatible beliefs. 3.2.1. Luck believers Only if individuals have an irrational belief in deterministic luck in the first place is it reasonable, in line with selfconsistency and cognitive dissonance theories, to suppose that they might also believe themselves personally to have deterministic luck in some degree. Because Maltby et al. (2008) found no support for discrete constructs of beliefs in, respectively, good and bad luck, luck believers are unlikely to assess discretely their personal good and bad luck. The correlations between the discrete personal good and bad luck components André (2006) and Maltby et al. (2008) each found support for this supposition, as does the possibility that artifactual scoring effects alone produced their discrete personal luckiness components. Hence we hypothesize: H2. For luck believers, belief in personal luckiness will form a unidimensional continuum running from lucky to unlucky.
E.R. Thompson, G.P. Prendergast / Personality and Individual Differences 54 (2013) 501–506
3.2.2. Disbelievers On the face of it, the only logically consistent way for disbelievers in luck to respond to items about whether or not they are personally lucky or unlucky is to disagree that they are either. For them to think otherwise would constitute ‘the existence of nonfitting cognitions’ likely to prompt ‘psychologically uncomfortable’ feelings of dissonance (Festinger, 1957, p.3). Hence, for luck disbelievers the notion of personally having deterministic luck cannot exist as a conceptually meaningful or psychometrically valid construct. Accordingly, attempting to measure personally having luckiness in populations of disbelievers ought to be nonsensical, and multidimensional measures going beyond the single dimension of belief in luck ought in populations containing large proportions of disbelievers to be equally nonsensical and incapable of producing internal consistency reliability. However, this line of reasoning manifestly runs contrary to the findings of acceptable internal consistency reliabilities demonstrated by André (2006) and Maltby et al. (2008) for their subscales relating to luckiness. To explain what might be happening, we suggest that disbelievers have a semantically differentiated dual understanding of the term luck that facilitates measurement of a conceptually meaningful and psychometrically valid construct of personal luckiness. In common parlance luck is not exclusively understood to mean an irrational deterministic phenomenon. As Darke and Freedman (1997b) highlight, personal luckiness can be taken simply to mean how fortunate individuals consider themselves to be. In this sense, those considering themselves lucky do not think they personally have positive deterministic luck with predictable ‘implications for the future’ (Darke & Freedman, 1997b, p. 499). Instead, they consider themselves simply fortunate that outcomes largely dependent on random chance have, by happy coincidence, generally turned out favorably for them. Because the personal possession of deterministic luckiness is a logically incoherent notion for disbelievers, we propose they are likely to take questions about personal luckiness to be questions about fortunateness. This is especially likely when scale items do not overtly address personally possessing deterministic luck but rather appear to refer to luckiness as fortunateness, which is predominantly the case in existing measures. For example, the item I consider myself a lucky person (André, 2006; Darke & Freedman, 1997b; Maltby et al., 2008) is likely to be read and understood by disbelievers straightforwardly as I consider myself a fortunate person. To do otherwise would cause disbelievers to suffer cognitive dissonance and self-inconsistency. Conceptualizing personal luckiness (fortunateness) as a unidimensional continuum running from fortunate to unfortunate would seem consonant with the more rational nature of luck disbelievers. Hence we hypothesize: H3. For luck disbelievers, belief in personal luckiness (fortunateness) will form a unidimensional lucky/unlucky continuum. However, Maltby et al. (2008) suggest the discrete personal luckiness and unluckiness components they derived perhaps mirror the bidimensionality of wellbeing constructs, such as optimism, that are found to comprise discrete, possibly non-artifactual, positive and negative dimensions (Kam & Meyer, 2012). If this is indeed the case, we suggest a competing hypothesis:
503
3.2.3. Construct validity with personality and individual differences 3.2.3.1. Personality. Maltby et al. (2008) found belief in being unlucky correlated negatively with extraversion, conscientiousness and openness, but positively with neuroticism, with opposite signs, albeit not always significant, for belief in being lucky. They also found positive signs for being unlucky and agreeableness, although their small sample was perhaps insufficient for significance. As Maltby et al.’s sample comprised both believers and disbelievers, we expect luckiness to reflect this pattern of associations for believers and disbelievers both together and separately, and whether or not luckiness proves unidimensional or bidimensional. 3.2.3.2. LoC. Darke and Freedman (1997b) found their unlucky items correlated positively with chance and powerful others, but negatively with internal LoC. Again, we expect luckiness to reflect this pattern of associations for believers and disbelievers, and whether or not luckiness proves unidimensional or bidimensional. 3.2.3.3. Wellbeing. Maltby et al. (2008) found belief in being lucky correlated positively with wellbeing variables, with signs simply reversed for belief in being unlucky. Accordingly, we expect luckiness to reflect this for believers and disbelievers, and whether or not luckiness proves unidimensional or bidimensional. 3.2.3.4. Demographics. André (2006) found a negative relationship between bad fortune and age. Accordingly, we expect luckiness may positively correlate with age. As males more than females ascribe failings to chance (Stipek & Gralinski, 1991), assuming sexes fail equally, we speculate males may believe themselves unluckier than females. 4. Study one: Dimensionality of luck beliefs 4.1. Participants Participants comprised 315 volunteer students, 159 female, aged 18–59 years (mode age category 20–24, 79%), 152 from three universities in Britain, 163 from an English-language international university in Japan. 4.2. Measure Drawing on existing luck beliefs scales, we developed 6 belief in luck and 6 personal luckiness items, half for each construct negatively worded. To help minimize the possibility of producing artifactual constructs, we differentiated all items by more than just the negative or positive valence of a single word. We drafted belief in luck items to reflect or imply belief in both good and bad luck. To enhance research practicality, we made items short. We sought to facilitate international research applicability with nonnative English-speakers by making all items maximally etic and minimally emic through avoiding colloquialisms, like I often feel like it’s my lucky day (Darke & Freedman, 1997b). To reduce method variance caused by priming effects, we followed Scheier, Carver, and Bridges (1994) by adding distracter items, thereby constructing what we shall henceforth call the Belief in Luck and Luckiness Scale (BILLS; see Appendix). 4.3. Analyses and results
H4. For luck disbelievers, personal luckiness (fortunateness) will constitute a bidimensional positive–negative construct, regardless of careful item wording designed to reduce possible artifactual constructs.
4.3.1. BILLS bidimensionality Following Maltby et al. (2008), we subjected the BILLS to principal component analysis using parallel analysis to determine
504
E.R. Thompson, G.P. Prendergast / Personality and Individual Differences 54 (2013) 501–506
factor retention (Table 1), with a set at .01 to minimize potential over retention (Dickie, Surgenor, Wilson, & McDowell, 2012). This derived a 2-factor model directly reflecting the discrete personal luckiness and belief in luck dimensions we propose. Confirmatory factor analyses (CFA) found this bidimensional model produced a significantly better fit than a unidimensional model (Table 1). The Belief in Luck and Personal Luckiness subscales have internal consistency reliabilities, respectively, of .85 and .88, and are uncorrelated (.07, p = .25), further supporting their representation of the two discrete belief in luck and personal luckiness constructs we propose. We tested the structural invariance of the BILLS across native and non-native English-speakers and found no significant difference between the constrained and unconstrained structural models (Dv2 1.84, p = .61), supporting the BILLS’s cross-cultural bidimensionality and applicability. 4.3.2. Belief in luck dimensionality Parallel analysis of just the belief in luck items indicated a single component, and CFA found a bidimensional model separating belief and disbelief items fitted insignificantly better than a unidimensional model of all items (Table 1). Hence our hypothesis (H1) that belief in luck is unidimensional is supported. 4.3.3. Personal luckiness dimensionality In examining hypotheses 2, 3 and 4, to classify, respectively, believers and disbelievers we split the sample using a summated mean score above the neutral mid-point on the Belief in Luck subscale. For luck believers, parallel analysis produced a single component and CFA found a unidimensional model fitted well, lending support to our hypothesis (H2) that personal luckiness is unidimensional for luck believers (Table 1). Nevertheless, CFA indicated a bidimensional model produced a marginally statistically, although not practically, significantly better fit. For luck disbelievers, parallel analysis suggested personal luckiness constitutes a single component, lending support to H3. However, CFA found a bidimensional model of luckiness and unluckiness items fitted very significantly better than a unidimensional model (Table 1), lending support to our competing hypothesis (H4) that personal luckiness for disbelievers might comprise discrete lucky and unlucky components.
5. Study two: Dimensionality confirmation and construct validity 5.1. Participants Participants were 431 volunteer students, 314 female, aged 18–59 years (mode age category 20–24, 73%), from an Englishlanguage university in Hong Kong. 5.2. Personality and individual differences measures 5.2.1. Personality We used the 40-item lexical International English Big-Five Mini-Markers (Thompson, 2008): Cronbach’s alphas were .89 for extraversion; openness, .84; neuroticism, .80; conscientiousness, .86; and agreeableness, .79. 5.2.2. LoC Following Darke and Freedman (1997b), we used Levenson’s (1981) 24-item LoC measure, Cronbach’s alphas: powerful others, .83; chance, .77; internal, .70.
5.2.3. Wellbeing Like Maltby et al., we used a range of extensively validated wellbeing measures: Diener, Emmons, Larson, et al.’s (1985) 5-item Satisfaction with Life Scale, alpha .85; the 10-item International Positive and Negative Affect Short-Form, alphas: PA .73, NA .79 (Thompson, 2007); and Scheier et al.’s (1994) 6-item revised Life Orientation Test, alpha .78.
5.3. Dimensionality Parallel analysis and CFA of the BILLS in Study 2 very closely reflected results for Study 1 (Table 1), confirming our overall proposal of a bidimensional belief in luck and luckiness model. The Belief in Luck subscale (a .79) is again unidimensional supporting our hypothesis (H1), and is uncorrelated ( .02, p = .72) with the Luckiness subscale (a .89). Luckiness for believers again constituted a unidimensional component supporting our hypothesis (H2), although once more CFA indicated a bidimensional model was marginally statistically, if not practically, significantly better fitting (Table 1). However, for disbelievers, both parallel analysis and CFA indicated a bidimensional model, lending support to our hypothesis (H4) that for disbelievers luckiness comprises discrete lucky and unlucky components rather than a unidimensional construct (H3). The positive and negative luckiness items, respectively, formed reliable subscales, Lucky (a .90) and Unlucky (a .84). These correlate with each other significantly negatively and strongly for the whole sample ( .61, p < .01) and for believers ( .76, p < .01), but also for disbelievers ( .48, p < .01), tending to suggest luckiness is in fact a unidimensional construct for believers and disbelievers alike.
5.4. Construct validity The Belief in Luck subscale correlated as expected with neuroticism, powerful others, chance, life satisfaction, negative affect and optimism (Table 2), but was uncorrelated with internal LoC, positive affect, and demographic variables. For the whole sample, and believer and disbeliever subsamples, the Luckiness subscale correlated similarly and significantly as anticipated with all the personality, LoC, wellbeing and demographic measures, with the exception of age for each sample and openness for disbelievers. The Lucky and Unlucky subscales for the whole sample exhibited similar patterns of correlations except with reversed signs (Table 2), indicating these subscales represent mirror images of each other. Splitting the sample into believers and disbelievers broadly replicated these mirrored correlations for Lucky and Unlucky subscales. The pattern of significant correlations for believers and disbelievers is identical except for openness, chance, and sex, but even here sign directions are the same, indicating that Lucky and Unlucky scales straightforwardly represent positively and negatively the same luckiness construct for both believers and disbelievers.
5.5. Temporal stability Two months after initial completion, the BILLS was again completed by 155 of the original Study 2 participants. Correlations between the test and retest were .56 (p < .01) for Belief in Luck, and .75 (p < .01) for Personal Luckiness, indicating satisfactory temporal stability for each subscale.
505
E.R. Thompson, G.P. Prendergast / Personality and Individual Differences 54 (2013) 501–506
6. Study 3. Convergent validity
7. Discussion
To examine convergent validity, we administered the BILLS, first, with Maltby et al.’s (2008) measure to 101 university nonacademic employees in England (64 female, mode age category 35–39) and, second, with Darke and Freedman’s (1997b) BIGL and Young et al.’s (2009) Belief in Fleeting Luck (BFL) measures to 203 postgraduate students in England (126 female, mean age 28.9). Belief in Luck correlated with Maltby et al.’s most equivalent subscale, General Belief in Luck, highly (.70, p < .01), suggesting convergent validity. Personal Luckiness correlated with the subscales Belief in Being Lucky and Belief in Being Unlucky highly (.66 and .55, respectively, ps < .01), again indicating convergent validity. Belief in Luck correlated with BIGL (.68, p < .01) and BFL (.52, p < .01), and Personal Luckiness correlated with BIGL (.47, p < .01), suggesting convergent validity. Personal Luckiness is uncorrelated with BFL ( .05, p = .67), suggesting BFL does not capture personal luckiness.
This research lends support both to our conceptualization of a new bidimensional belief in luck and luckiness model and to the BILLS, whose bidimensionality is structurally invariant across cultures, and whose subscales are uncorrelated, have internal consistency reliability, convergent validity, and temporal stability. We find consistent empirical support for belief in luck being unidimensional, and no evidence for belief and disbelief in luck being separate constructs. That the Belief in Luck subscale correlates positively with neuroticism, powerful others, chance, and negative affect, but negatively with life satisfaction, and optimism, is consonant both with luck belief being indicative of low personal agency and with Maltby et al.’s (2008) suggestion it is a construct representing irrationality and maladaptivity. Non-correlation with age or sex suggests luck belief is distributed evenly across both. On balance, our findings suggest personal luckiness is also unidimensional. Our hypotheses that both the nature and dimensionality of personal luckiness might differ for luck believers and disbelievers find little support. Whether believers and disbelievers
Table 1 Confirmatory factor analyses. Parallel analysis determined factors
Study 1 BILLS Belief in Luck items Personal Luckiness items Believers, n = 129 Disbelievers, n = 186 Study 2 BILLS Belief in Luck items Personal Luckiness items Believers, n = 189 Disbelievers, n = 242
Model v2/CFI
Dv2
1-Factor
2-Factor
2 1
1101.30/.451 5.10/.995
238.55/.894 4.63/.996
862.75* 0.47
1 1
31.50/.964 111.75/.794
15.87/.992 13.99/.993
15.33* 97.76*
2 1
1008.40/.590 26.64/.984
425.16/.844 13.34/.991
583.24* 13.30*
1 2
55.45/.942 194.61/.774
13.23/.993 37.24/.964
42.22* 157.37*
Note: CFI = Comparative Fit Index. * p < .01.
Table 2 Correlations between luck and personality and individual difference variables. All Belief in Luck
*
Believers Personal Luckiness
Lucky
Unlucky
Personal Luckiness
Disbelievers Lucky
Unlucky
Personal Luckiness
Lucky
Unlucky
Personality Extraversion Openness Neuroticism Conscientiousness Agreeableness
.08 .07 .13* .02 .02
.28* .16* .28* .26* .21*
.24* .15* .20* .25* .13*
.26* .14* .30* .22* .25*
.36* .22* .24* .27* .19*
.32* .19* .16 .24* .12
.35* .21* .29* .27* .24*
.19* .10 .32* .25* .23*
.18* .12 .27* .27* .13
.14* .04 .29* .15* .26*
Locus of Control Powerful Others Chance Internal
.09 .32* .06
.33* .24* .19*
.22* .09 .22*
.36* .34* .13*
.31* .28* .22*
.23* .18 .22*
.36* .34* .20*
.34* .21* .17*
.23* .04 .23*
.36* .32* .05
Wellbeing Life Satisfaction Positive Affect Negative Affect Optimism
.15* .01 .11 .10
.47* .31* .39* .59*
.41* .30* .28* .48*
.43* .26* .42* .58*
.57* .38* .39* .59*
.55* .37* .27* .52*
.52* .35* .46* .58*
.37* .22* .40* .60*
.29* .22* .32* .47*
.34* .16 .37* .57*
Demographics Male Age
.03 .02
.15* .03
.15* .01
.12* .07
.15 .05
.18 .01
.10 .11
.15 .02
.12 .01
.13 .04
p < .01. p < .05.
506
E.R. Thompson, G.P. Prendergast / Personality and Individual Differences 54 (2013) 501–506
are looked at separately or together, the patterns of Luckiness, Lucky and Unlucky subscale correlations with personality, LoC, wellbeing and demographic variables are essentially similar. The opposite-signed but otherwise identical patterns of correlations for Lucky and Unlucky, plus the high negative correlation they have with each other, suggest they are simply positive and negative representations of precisely the same personal luckiness construct. The similar correlation patterns for believers and disbelievers also suggest these both construe personal luckiness in essentially the same way. Because the largest correlations here are uniformly with wellbeing variables, both believers and disbelievers would each appear to take luckiness straightforwardly to mean fortunateness. Maleness correlating negatively with luckiness is consonant with lower male scores on other wellbeing measures like optimism (Scheier et al., 1994). We therefore suggest that whether random chance, deterministic luck or something else determines personal luckiness (fortunateness), it seems to amount to the same wellbeing-related construct equally for believers and disbelievers, implying they both make the same everyday semantic differentiation between luck (a) as a deterministic phenomenon and (b) a synonym of fortunateness. Because wellbeing-related constructs’ dimensionality is an ongoing debate (Kam & Meyer, 2012), further research on personal luckiness’ dimensionality might be fruitful. With no compelling theory suggesting luckiness and unluckiness should be conceptually distinct, we speculate the modest statistical evidence we find for such a distinction is simply item-valency artifactuality. Additional research on this, and to extend the BILLSs’ validity, will be useful. More broadly, now that a clear conceptual distinction between belief in luck and personal luckiness has been demonstrated, luck researchers can now move beyond studies that inappropriately combine or unjustifiably subdivide these two distinct unidimensional constructs. For example, health, happiness, confidence, gambling, risk-taking, and consumption have all been linked to luck beliefs. But how they relate to beliefs, respectively, in deterministic luck or personal luckiness, either separately, additively, or interactively, has been unresearched, and indeed unresearchable without suitable instrumentation. The development of the BILLS now provides a valid and reliable instrument to address this research lacuna. Appendix A. Belief in Luck and Luckiness Scale Question: To what extent do you personally agree or disagree with the following statements? Items: I believe in good and bad lucka [I try hard to be niceb] I mostly have bad luckc,⁄ There is no such thing as good or bad lucka,⁄ [It’s hard to be niceb] I’m not luckyc,⁄ Good and bad luck really do exista I generally have good luckc [I’m nice if I tryb] Luck doesn’t affect what happens to mea,⁄ I consider myself a lucky personc Belief in luck is completely sensiblea
[It’s nice to try hardb] Bad luck happens to me oftenc,⁄ Luck only exists in peoples’ mindsa,⁄ I’m usually luckyc Interval measure: 1 (Strongly disagree) to 5 (Strongly agree). Note. aBelief in Luck items. bDistracter items. cPersonal Luckiness items. ⁄Reverse coded. References André, N. (2006). Good fortune, luck, opportunity and their lack: How do agents perceive them? Personality and Individual Differences, 40, 1461–1472. Bridgstock, M., Marais, I., & Sturgess, K. (2011). The structure of superstitious action – A further analysis of fresh evidence. Personality and Individual Differences, 50, 795–798. Darke, P. R., & Freedman, J. L. (1997a). Lucky events and beliefs in luck: Paradoxical effects on confidence and risk-taking. Personality and Social Psychology Bulletin, 23, 378–388. Darke, P. R., & Freedman, J. L. (1997b). The belief in good luck scale. Journal of Research in Personality, 31, 486–511. Day, L., & Maltby, J. (2003). Belief in good luck and psychological wellbeing: The mediating role of optimism and irrational beliefs. Journal of Psychology, 137, 99–110. Dickie, L., Surgenor, L. J., Wilson, M., & McDowell, J. (2012). The structure and reliability of the Clinical Perfectionism Questionnaire. Personality and Individual Differences, 52, 865–869. Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The satisfaction with life scale. Journal of Personality Assessment, 49, 71–75. Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA: Stanford University Press. Hart, K. E., & Hope, C. W. (2004). Cynical hostility and the psychosocial vulnerability model of disease risk: Confounding effects of neuroticism (negative affectivity) bias. Personality and Individual Differences, 36, 1571–1582. Judge, T. A., Erez, A., Bono, J. E., & Thoresen, C. J. (2002). Are measures of self-esteem, neuroticism, locus of control, and generalized self-efficacy indicators of a common core construct? Journal of Personality and Social Psychology, 83, 693–710. Kam, C., & Meyer, J. P. (2012). Do optimism and pessimism have different relationships with personality dimensions? A re-examination. Personality and Individual Differences, 52, 123–127. Keren, G., & Wagenaar, W. A. (1985). On the psychology of playing blackjack – Normative and descriptive considerations with implications for decisiontheory. Journal of Experimental Psychology: General, 114, 133–158. Levenson, H. (1981). Differentiating among internality, powerful others, and chance. In H. Lefcourt (Ed.). Research with the locus of control construct (Vol. 1, pp. 15–63). New York: Academic Press. Maltby, J., Day, L., Gill, P., Colley, A., & Wood, A. M. (2008). Beliefs around luck: Confirming the empirical conceptualization of beliefs around luck and the development of the Darke and Freedman beliefs around luck scale. Personality and Individual Differences, 45, 655–660. Nail, P. R., Misak, J. E., & Davis, R. M. (2004). Self-affirmation versus self-consistency: A comparison of two competing self-theories of dissonance phenomena. Personality and Individual Differences, 36, 1893–1905. Prendergast, G. P., & Thompson, E. R. (2008). Sales promotion strategies and belief in luck. Psychology & Marketing, 25, 1043–1062. Scheier, M. F., Carver, C. S., & Bridges, M. W. (1994). Distinguishing optimism from neuroticism (and trait anxiety, self-mastery, and self-esteem): A re-evaluation of the Life Orientation Test. Journal of Personality and Social Psychology, 67, 1063–1078. Spector, P. E., Van Katwyk, P. T., Brannick, M. T., & Chen, P. Y. (1997). When two factors don’t reflect two constructs: How item characteristics can produce artifactual factors. Journal of Management, 23, 659–677. Stipek, D. J., & Gralinski, J. H. (1991). Gender differences in children’s achievementrelated beliefs and emotional response to success and failure in mathematics. Journal of Educational Psychology, 83, 361–371. Thompson, E. R. (2007). Development and validation of an internationally reliable short-form of the positive and negative affect schedule (PANAS). Journal of Cross-Cultural Psychology, 38(2), 227–242. Thompson, E. R. (2008). Development and validation of an International English BigFive Mini-Markers. Personality and Individual Differences, 45, 542–548. Wiseman, R., & Watt, C. (2004). Measuring superstitious belief: Why lucky charms matter. Personality and Individual Differences, 37, 1533–1541. Young, M. J., Chen, N., & Morris, M. W. (2009). Belief in stable and fleeting luck and achievement motivation. Personality and Individual Differences, 47, 150–154.
Contents lists available at SciVerse ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Belief in luck and luckiness: Conceptual clarification and new measure validation q Edmund R. Thompson a,⇑, Gerard P. Prendergast b,1 a b
School of Management, University of Bath, Bath BA2 7AY, United Kingdom School of Business, Hong Kong Baptist University, Kowloon Tong, Hong Kong
a r t i c l e
i n f o
Article history: Received 20 January 2012 Received in revised form 4 September 2012 Accepted 20 October 2012 Available online 11 December 2012 Keywords: Belief in luck Luckiness Conceptualization Measurement Scale validation Personality
a b s t r a c t Research on the dimensionality and measurement of luck beliefs has yet to produce a clear conceptual and metrical consensus. This research theorizes a bidimensional model of luck beliefs that is tested through a series of studies (total n = 1205) validating the new Belief in Luck and Luckiness Scale. Unlike existing conceptualizations and measures, this new model is applicable to both believers and non-believers in luck, and reveals belief in luck and personal luckiness to be discrete, uncorrelated, and respectively unidimensional constructs. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction Personality researchers have devoted increasing theoretical, empirical, and measurement attention to irrational beliefs about luck (André, 2006; Bridgstock, Marais, & Sturgess, 2011; Darke & Freedman, 1997a; Day & Maltby, 2003; Maltby, Day, Gill, Colley, & Wood, 2008; Wiseman & Watt, 2004; Young, Chen, & Morris, 2009). As this research interest has grown, so too have disparate conceptions about the dimensionality of luck beliefs and the means to measure these. For example, Darke and Freedman (1997b) propose a unidimensional conceptualization and measure, while André (2006) proposes a 6-dimensional, and Maltby et al. (2008) a 4-dimensional, conception and measure. Despite these developments in theoretical and metrical nuance, luck beliefs research has still to produce a clear conceptual and measurement consensus. Moreover, Maltby et al. (2008) hint that the precise nature of luck beliefs’ dimensionality may yet be obscured by two limitations in existing measures: (a) items possibly producing artifactual components unreflective of true underlying constructs and (b) unexplored potential differences between luck believers and disbelievers.
q The authors thank Colin Cooper and two anonymous reviewers for their constructive comments and suggestions. ⇑ Corresponding author. Tel.: +44 (0) 1225 386742; fax: +44 (0) 1225 386473. E-mail addresses: [email protected] (E.R. Thompson), gerard@hkbu. edu.hk (G.P. Prendergast). 1 Tel.: +852 3411 7570; fax: +852 3411 5586.
0191-8869/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.paid.2012.10.027
We build on existing research by proposing and examining a new bidimensional conceptualization and measure of luck beliefs designed to reduce potential artifactual components and to account for both luck believers and disbelievers. Unlike prior studies, ours finds support for just two discrete dimensions, belief in luck and personal luckiness, that are each themselves unidimensional, uncorrelated, and differently correlated with personality and individual difference variables. Our systematically theorized and validated new measure contributes a parsimonious tool to investigate the separate effects of, respectively, belief in luck and personal luckiness constructs on psychology and behavioral differences that have hitherto been examinable only with measures that either erroneously conflate or spuriously subdivide these two distinct luck constructs. 2. Conceptualizations and measures of luck beliefs 2.1. Belief in good luck Darke and Freedman (1997b) propose a unidimensional conceptualization of irrational belief in luck, running from belief that good luck deterministically favors particular people at one end of a continuum, to the view that luck is simply random chance at the other. To capture this conceptualization they developed the often-cited 12-item Belief in Good Luck Scale (BIGLS; 1997b). However, they found that their scale both fails to distinguish between those believing themselves lucky or unlucky (1997a), and produced a multidimensional rather than unidimensional solution when factor analyzed (1997b, p. 493, fn.3). Prendergast and
502
E.R. Thompson, G.P. Prendergast / Personality and Individual Differences 54 (2013) 501–506
Thompson (2008) confirmed this multidimensionality and additionally found the scale’s two predominant and psychometrically usable sub-scales, a general belief in the existence of deterministic luck and a belief in being personally lucky, are uncorrelated and differently predict criterion variables. 2.2. Multidimensional beliefs about luck André (2006), noting both the multidimensional nature of Darke and Freedman’s scale and other luck-related constructs like illusion of control (Wiseman & Watt, 2004), developed an 18-item 6-dimensional measure exploring positive and negative luck beliefs. While she finds good and bad luck beliefs are empirically discrete, she also finds they correlate relatively highly, suggesting their conceptual closeness. This also suggests the possibility that their empirical separation may stem from the tendency of positively and negatively worded items sometimes to produce artifactually separate components even when conceptually discrete underlying constructs do not exist (Spector, Van Katwyk, Brannick, & Chen, 1997). 2.3. Beliefs around luck Maltby et al. (2008) observe that André’s (2006) measure encapsulates only some of the beliefs around luck contained in Darke and Freedman’s scale. To address this, they developed a 22-item scale and found empirical support for a 4-dimensional model: general belief in luck; rejection of belief in luck; being lucky; and being unlucky. However, they note their own 4-dimesional model ‘provides only a reasonable fit to the data’, forms two closely conceptually related pairs, and may be ‘attributable to an artifact of scoring’ (2008, p. 659). This latter suggestion is plausible because many of their scale’s positive and negative items are duplicative except for the positive/negative valence of one or a few words (e.g. good/bad or lucky/unlucky), a circumstance likely to produce ‘item direction factors’ (Spector et al., 1997, p. 661). Additionally, Maltby et al. (2008) suggest their model’s particular dimensionality could simply reflect their sample comprising both luck believers and disbelievers. They accordingly advise that ‘some definitive studies are needed to test this explanation’ (p. 659), hinting that alternative models with fewer than four dimensions may exist.
3. A new bidimensional model: belief in luck and luckiness 3.1. Belief in luck We propose that one clear unidimensional component of luck beliefs is whether individuals believe or disbelieve in the existence of luck as a deterministic phenomenon in the first place. We conceptualize belief in luck as encompassing both good and bad luck. This accords with Maltby et al. (2008) finding no support for their proposition that discrete belief in good luck and belief in bad luck constructs exist. Hence, with irrational belief in luck (good and bad) at one end of a bipolar continuum, the conceptual issue becomes what should be at the opposite end. Conceptualizing, as Darke and Freedman (1997b) do, the antithesis of belief in luck to be recognition of random chance’s existence is illogical: Keren and Wagenaar (1985) found individuals can irrationally believe in deterministic luck while concurrently recognizing the separate existence of random chance. We hence conceptualize belief in luck to have the straightforward bipolar opposite of disbelief in luck. In this we accept that the empirically discrete separation of a general belief in luck and a rejection of belief in luck Maltby et al. (2008)
found is indeed likely the scoring artifact they suggest it might be. We therefore hypothesize: H1. Deterministic luck is something individuals believe or disbelieve in to greater or lesser degrees on a unidimensional continuum. 3.1.1. Construct validity with personality and individual differences 3.1.1.1. Personality. Maltby et al. (2008) found no association between the five-factor personality model and acceptance or rejection of belief in luck. However, because belief in luck is irrational and irrationality correlates with neuroticism (Hart & Hope, 2004), we expect luck belief may positively correlate with neuroticism. 3.1.1.2. Locus of Control (LoC). Darke and Freedman (1997b) found belief in good luck was positively related to the chance and powerful others dimensions of LoC. Maltby et al. (2008) found internal LoC correlated positively with rejection of belief in luck and negatively with general belief in luck. Hence, we expect luck belief may correlate positively with chance and powerful others, but negatively with internal LoC. 3.1.1.3. Wellbeing. Internal LoC and emotional stability (neuroticism) are suggested to form part of a higher order construct of core self- and life-evaluation broadly constituting wellbeing (Judge, Erez, Bono, & Thoresen, 2002). We therefore expect luck belief may negatively associate with wellbeing. 3.1.1.4. Demographics. André (2006) found a negative relationship between some luck beliefs and age. Accordingly, we expect belief in luck may negatively correlate with age. No prior luck belief research has examined sex differences. However, educational achievement is found likely to be attributed to luck more by males than females (Stipek & Gralinski, 1991), hence we expect males may believe in luck more than females. 3.2. Personal luckiness We propose a second component of luck beliefs is personal luckiness, but that its dimensionality may differ depending on whether or not individuals believe or disbelieve in luck to begin with. Our proposition here accords with self-consistency (Nail, Misak, & Davis, 2004) and cognitive dissonance (Festinger, 1957) theories. These suggest, respectively, that individuals seek to maintain congruence between their beliefs and to avoid the adverse psychological effects of holding incompatible beliefs. 3.2.1. Luck believers Only if individuals have an irrational belief in deterministic luck in the first place is it reasonable, in line with selfconsistency and cognitive dissonance theories, to suppose that they might also believe themselves personally to have deterministic luck in some degree. Because Maltby et al. (2008) found no support for discrete constructs of beliefs in, respectively, good and bad luck, luck believers are unlikely to assess discretely their personal good and bad luck. The correlations between the discrete personal good and bad luck components André (2006) and Maltby et al. (2008) each found support for this supposition, as does the possibility that artifactual scoring effects alone produced their discrete personal luckiness components. Hence we hypothesize: H2. For luck believers, belief in personal luckiness will form a unidimensional continuum running from lucky to unlucky.
E.R. Thompson, G.P. Prendergast / Personality and Individual Differences 54 (2013) 501–506
3.2.2. Disbelievers On the face of it, the only logically consistent way for disbelievers in luck to respond to items about whether or not they are personally lucky or unlucky is to disagree that they are either. For them to think otherwise would constitute ‘the existence of nonfitting cognitions’ likely to prompt ‘psychologically uncomfortable’ feelings of dissonance (Festinger, 1957, p.3). Hence, for luck disbelievers the notion of personally having deterministic luck cannot exist as a conceptually meaningful or psychometrically valid construct. Accordingly, attempting to measure personally having luckiness in populations of disbelievers ought to be nonsensical, and multidimensional measures going beyond the single dimension of belief in luck ought in populations containing large proportions of disbelievers to be equally nonsensical and incapable of producing internal consistency reliability. However, this line of reasoning manifestly runs contrary to the findings of acceptable internal consistency reliabilities demonstrated by André (2006) and Maltby et al. (2008) for their subscales relating to luckiness. To explain what might be happening, we suggest that disbelievers have a semantically differentiated dual understanding of the term luck that facilitates measurement of a conceptually meaningful and psychometrically valid construct of personal luckiness. In common parlance luck is not exclusively understood to mean an irrational deterministic phenomenon. As Darke and Freedman (1997b) highlight, personal luckiness can be taken simply to mean how fortunate individuals consider themselves to be. In this sense, those considering themselves lucky do not think they personally have positive deterministic luck with predictable ‘implications for the future’ (Darke & Freedman, 1997b, p. 499). Instead, they consider themselves simply fortunate that outcomes largely dependent on random chance have, by happy coincidence, generally turned out favorably for them. Because the personal possession of deterministic luckiness is a logically incoherent notion for disbelievers, we propose they are likely to take questions about personal luckiness to be questions about fortunateness. This is especially likely when scale items do not overtly address personally possessing deterministic luck but rather appear to refer to luckiness as fortunateness, which is predominantly the case in existing measures. For example, the item I consider myself a lucky person (André, 2006; Darke & Freedman, 1997b; Maltby et al., 2008) is likely to be read and understood by disbelievers straightforwardly as I consider myself a fortunate person. To do otherwise would cause disbelievers to suffer cognitive dissonance and self-inconsistency. Conceptualizing personal luckiness (fortunateness) as a unidimensional continuum running from fortunate to unfortunate would seem consonant with the more rational nature of luck disbelievers. Hence we hypothesize: H3. For luck disbelievers, belief in personal luckiness (fortunateness) will form a unidimensional lucky/unlucky continuum. However, Maltby et al. (2008) suggest the discrete personal luckiness and unluckiness components they derived perhaps mirror the bidimensionality of wellbeing constructs, such as optimism, that are found to comprise discrete, possibly non-artifactual, positive and negative dimensions (Kam & Meyer, 2012). If this is indeed the case, we suggest a competing hypothesis:
503
3.2.3. Construct validity with personality and individual differences 3.2.3.1. Personality. Maltby et al. (2008) found belief in being unlucky correlated negatively with extraversion, conscientiousness and openness, but positively with neuroticism, with opposite signs, albeit not always significant, for belief in being lucky. They also found positive signs for being unlucky and agreeableness, although their small sample was perhaps insufficient for significance. As Maltby et al.’s sample comprised both believers and disbelievers, we expect luckiness to reflect this pattern of associations for believers and disbelievers both together and separately, and whether or not luckiness proves unidimensional or bidimensional. 3.2.3.2. LoC. Darke and Freedman (1997b) found their unlucky items correlated positively with chance and powerful others, but negatively with internal LoC. Again, we expect luckiness to reflect this pattern of associations for believers and disbelievers, and whether or not luckiness proves unidimensional or bidimensional. 3.2.3.3. Wellbeing. Maltby et al. (2008) found belief in being lucky correlated positively with wellbeing variables, with signs simply reversed for belief in being unlucky. Accordingly, we expect luckiness to reflect this for believers and disbelievers, and whether or not luckiness proves unidimensional or bidimensional. 3.2.3.4. Demographics. André (2006) found a negative relationship between bad fortune and age. Accordingly, we expect luckiness may positively correlate with age. As males more than females ascribe failings to chance (Stipek & Gralinski, 1991), assuming sexes fail equally, we speculate males may believe themselves unluckier than females. 4. Study one: Dimensionality of luck beliefs 4.1. Participants Participants comprised 315 volunteer students, 159 female, aged 18–59 years (mode age category 20–24, 79%), 152 from three universities in Britain, 163 from an English-language international university in Japan. 4.2. Measure Drawing on existing luck beliefs scales, we developed 6 belief in luck and 6 personal luckiness items, half for each construct negatively worded. To help minimize the possibility of producing artifactual constructs, we differentiated all items by more than just the negative or positive valence of a single word. We drafted belief in luck items to reflect or imply belief in both good and bad luck. To enhance research practicality, we made items short. We sought to facilitate international research applicability with nonnative English-speakers by making all items maximally etic and minimally emic through avoiding colloquialisms, like I often feel like it’s my lucky day (Darke & Freedman, 1997b). To reduce method variance caused by priming effects, we followed Scheier, Carver, and Bridges (1994) by adding distracter items, thereby constructing what we shall henceforth call the Belief in Luck and Luckiness Scale (BILLS; see Appendix). 4.3. Analyses and results
H4. For luck disbelievers, personal luckiness (fortunateness) will constitute a bidimensional positive–negative construct, regardless of careful item wording designed to reduce possible artifactual constructs.
4.3.1. BILLS bidimensionality Following Maltby et al. (2008), we subjected the BILLS to principal component analysis using parallel analysis to determine
504
E.R. Thompson, G.P. Prendergast / Personality and Individual Differences 54 (2013) 501–506
factor retention (Table 1), with a set at .01 to minimize potential over retention (Dickie, Surgenor, Wilson, & McDowell, 2012). This derived a 2-factor model directly reflecting the discrete personal luckiness and belief in luck dimensions we propose. Confirmatory factor analyses (CFA) found this bidimensional model produced a significantly better fit than a unidimensional model (Table 1). The Belief in Luck and Personal Luckiness subscales have internal consistency reliabilities, respectively, of .85 and .88, and are uncorrelated (.07, p = .25), further supporting their representation of the two discrete belief in luck and personal luckiness constructs we propose. We tested the structural invariance of the BILLS across native and non-native English-speakers and found no significant difference between the constrained and unconstrained structural models (Dv2 1.84, p = .61), supporting the BILLS’s cross-cultural bidimensionality and applicability. 4.3.2. Belief in luck dimensionality Parallel analysis of just the belief in luck items indicated a single component, and CFA found a bidimensional model separating belief and disbelief items fitted insignificantly better than a unidimensional model of all items (Table 1). Hence our hypothesis (H1) that belief in luck is unidimensional is supported. 4.3.3. Personal luckiness dimensionality In examining hypotheses 2, 3 and 4, to classify, respectively, believers and disbelievers we split the sample using a summated mean score above the neutral mid-point on the Belief in Luck subscale. For luck believers, parallel analysis produced a single component and CFA found a unidimensional model fitted well, lending support to our hypothesis (H2) that personal luckiness is unidimensional for luck believers (Table 1). Nevertheless, CFA indicated a bidimensional model produced a marginally statistically, although not practically, significantly better fit. For luck disbelievers, parallel analysis suggested personal luckiness constitutes a single component, lending support to H3. However, CFA found a bidimensional model of luckiness and unluckiness items fitted very significantly better than a unidimensional model (Table 1), lending support to our competing hypothesis (H4) that personal luckiness for disbelievers might comprise discrete lucky and unlucky components.
5. Study two: Dimensionality confirmation and construct validity 5.1. Participants Participants were 431 volunteer students, 314 female, aged 18–59 years (mode age category 20–24, 73%), from an Englishlanguage university in Hong Kong. 5.2. Personality and individual differences measures 5.2.1. Personality We used the 40-item lexical International English Big-Five Mini-Markers (Thompson, 2008): Cronbach’s alphas were .89 for extraversion; openness, .84; neuroticism, .80; conscientiousness, .86; and agreeableness, .79. 5.2.2. LoC Following Darke and Freedman (1997b), we used Levenson’s (1981) 24-item LoC measure, Cronbach’s alphas: powerful others, .83; chance, .77; internal, .70.
5.2.3. Wellbeing Like Maltby et al., we used a range of extensively validated wellbeing measures: Diener, Emmons, Larson, et al.’s (1985) 5-item Satisfaction with Life Scale, alpha .85; the 10-item International Positive and Negative Affect Short-Form, alphas: PA .73, NA .79 (Thompson, 2007); and Scheier et al.’s (1994) 6-item revised Life Orientation Test, alpha .78.
5.3. Dimensionality Parallel analysis and CFA of the BILLS in Study 2 very closely reflected results for Study 1 (Table 1), confirming our overall proposal of a bidimensional belief in luck and luckiness model. The Belief in Luck subscale (a .79) is again unidimensional supporting our hypothesis (H1), and is uncorrelated ( .02, p = .72) with the Luckiness subscale (a .89). Luckiness for believers again constituted a unidimensional component supporting our hypothesis (H2), although once more CFA indicated a bidimensional model was marginally statistically, if not practically, significantly better fitting (Table 1). However, for disbelievers, both parallel analysis and CFA indicated a bidimensional model, lending support to our hypothesis (H4) that for disbelievers luckiness comprises discrete lucky and unlucky components rather than a unidimensional construct (H3). The positive and negative luckiness items, respectively, formed reliable subscales, Lucky (a .90) and Unlucky (a .84). These correlate with each other significantly negatively and strongly for the whole sample ( .61, p < .01) and for believers ( .76, p < .01), but also for disbelievers ( .48, p < .01), tending to suggest luckiness is in fact a unidimensional construct for believers and disbelievers alike.
5.4. Construct validity The Belief in Luck subscale correlated as expected with neuroticism, powerful others, chance, life satisfaction, negative affect and optimism (Table 2), but was uncorrelated with internal LoC, positive affect, and demographic variables. For the whole sample, and believer and disbeliever subsamples, the Luckiness subscale correlated similarly and significantly as anticipated with all the personality, LoC, wellbeing and demographic measures, with the exception of age for each sample and openness for disbelievers. The Lucky and Unlucky subscales for the whole sample exhibited similar patterns of correlations except with reversed signs (Table 2), indicating these subscales represent mirror images of each other. Splitting the sample into believers and disbelievers broadly replicated these mirrored correlations for Lucky and Unlucky subscales. The pattern of significant correlations for believers and disbelievers is identical except for openness, chance, and sex, but even here sign directions are the same, indicating that Lucky and Unlucky scales straightforwardly represent positively and negatively the same luckiness construct for both believers and disbelievers.
5.5. Temporal stability Two months after initial completion, the BILLS was again completed by 155 of the original Study 2 participants. Correlations between the test and retest were .56 (p < .01) for Belief in Luck, and .75 (p < .01) for Personal Luckiness, indicating satisfactory temporal stability for each subscale.
505
E.R. Thompson, G.P. Prendergast / Personality and Individual Differences 54 (2013) 501–506
6. Study 3. Convergent validity
7. Discussion
To examine convergent validity, we administered the BILLS, first, with Maltby et al.’s (2008) measure to 101 university nonacademic employees in England (64 female, mode age category 35–39) and, second, with Darke and Freedman’s (1997b) BIGL and Young et al.’s (2009) Belief in Fleeting Luck (BFL) measures to 203 postgraduate students in England (126 female, mean age 28.9). Belief in Luck correlated with Maltby et al.’s most equivalent subscale, General Belief in Luck, highly (.70, p < .01), suggesting convergent validity. Personal Luckiness correlated with the subscales Belief in Being Lucky and Belief in Being Unlucky highly (.66 and .55, respectively, ps < .01), again indicating convergent validity. Belief in Luck correlated with BIGL (.68, p < .01) and BFL (.52, p < .01), and Personal Luckiness correlated with BIGL (.47, p < .01), suggesting convergent validity. Personal Luckiness is uncorrelated with BFL ( .05, p = .67), suggesting BFL does not capture personal luckiness.
This research lends support both to our conceptualization of a new bidimensional belief in luck and luckiness model and to the BILLS, whose bidimensionality is structurally invariant across cultures, and whose subscales are uncorrelated, have internal consistency reliability, convergent validity, and temporal stability. We find consistent empirical support for belief in luck being unidimensional, and no evidence for belief and disbelief in luck being separate constructs. That the Belief in Luck subscale correlates positively with neuroticism, powerful others, chance, and negative affect, but negatively with life satisfaction, and optimism, is consonant both with luck belief being indicative of low personal agency and with Maltby et al.’s (2008) suggestion it is a construct representing irrationality and maladaptivity. Non-correlation with age or sex suggests luck belief is distributed evenly across both. On balance, our findings suggest personal luckiness is also unidimensional. Our hypotheses that both the nature and dimensionality of personal luckiness might differ for luck believers and disbelievers find little support. Whether believers and disbelievers
Table 1 Confirmatory factor analyses. Parallel analysis determined factors
Study 1 BILLS Belief in Luck items Personal Luckiness items Believers, n = 129 Disbelievers, n = 186 Study 2 BILLS Belief in Luck items Personal Luckiness items Believers, n = 189 Disbelievers, n = 242
Model v2/CFI
Dv2
1-Factor
2-Factor
2 1
1101.30/.451 5.10/.995
238.55/.894 4.63/.996
862.75* 0.47
1 1
31.50/.964 111.75/.794
15.87/.992 13.99/.993
15.33* 97.76*
2 1
1008.40/.590 26.64/.984
425.16/.844 13.34/.991
583.24* 13.30*
1 2
55.45/.942 194.61/.774
13.23/.993 37.24/.964
42.22* 157.37*
Note: CFI = Comparative Fit Index. * p < .01.
Table 2 Correlations between luck and personality and individual difference variables. All Belief in Luck
*
Believers Personal Luckiness
Lucky
Unlucky
Personal Luckiness
Disbelievers Lucky
Unlucky
Personal Luckiness
Lucky
Unlucky
Personality Extraversion Openness Neuroticism Conscientiousness Agreeableness
.08 .07 .13* .02 .02
.28* .16* .28* .26* .21*
.24* .15* .20* .25* .13*
.26* .14* .30* .22* .25*
.36* .22* .24* .27* .19*
.32* .19* .16 .24* .12
.35* .21* .29* .27* .24*
.19* .10 .32* .25* .23*
.18* .12 .27* .27* .13
.14* .04 .29* .15* .26*
Locus of Control Powerful Others Chance Internal
.09 .32* .06
.33* .24* .19*
.22* .09 .22*
.36* .34* .13*
.31* .28* .22*
.23* .18 .22*
.36* .34* .20*
.34* .21* .17*
.23* .04 .23*
.36* .32* .05
Wellbeing Life Satisfaction Positive Affect Negative Affect Optimism
.15* .01 .11 .10
.47* .31* .39* .59*
.41* .30* .28* .48*
.43* .26* .42* .58*
.57* .38* .39* .59*
.55* .37* .27* .52*
.52* .35* .46* .58*
.37* .22* .40* .60*
.29* .22* .32* .47*
.34* .16 .37* .57*
Demographics Male Age
.03 .02
.15* .03
.15* .01
.12* .07
.15 .05
.18 .01
.10 .11
.15 .02
.12 .01
.13 .04
p < .01. p < .05.
506
E.R. Thompson, G.P. Prendergast / Personality and Individual Differences 54 (2013) 501–506
are looked at separately or together, the patterns of Luckiness, Lucky and Unlucky subscale correlations with personality, LoC, wellbeing and demographic variables are essentially similar. The opposite-signed but otherwise identical patterns of correlations for Lucky and Unlucky, plus the high negative correlation they have with each other, suggest they are simply positive and negative representations of precisely the same personal luckiness construct. The similar correlation patterns for believers and disbelievers also suggest these both construe personal luckiness in essentially the same way. Because the largest correlations here are uniformly with wellbeing variables, both believers and disbelievers would each appear to take luckiness straightforwardly to mean fortunateness. Maleness correlating negatively with luckiness is consonant with lower male scores on other wellbeing measures like optimism (Scheier et al., 1994). We therefore suggest that whether random chance, deterministic luck or something else determines personal luckiness (fortunateness), it seems to amount to the same wellbeing-related construct equally for believers and disbelievers, implying they both make the same everyday semantic differentiation between luck (a) as a deterministic phenomenon and (b) a synonym of fortunateness. Because wellbeing-related constructs’ dimensionality is an ongoing debate (Kam & Meyer, 2012), further research on personal luckiness’ dimensionality might be fruitful. With no compelling theory suggesting luckiness and unluckiness should be conceptually distinct, we speculate the modest statistical evidence we find for such a distinction is simply item-valency artifactuality. Additional research on this, and to extend the BILLSs’ validity, will be useful. More broadly, now that a clear conceptual distinction between belief in luck and personal luckiness has been demonstrated, luck researchers can now move beyond studies that inappropriately combine or unjustifiably subdivide these two distinct unidimensional constructs. For example, health, happiness, confidence, gambling, risk-taking, and consumption have all been linked to luck beliefs. But how they relate to beliefs, respectively, in deterministic luck or personal luckiness, either separately, additively, or interactively, has been unresearched, and indeed unresearchable without suitable instrumentation. The development of the BILLS now provides a valid and reliable instrument to address this research lacuna. Appendix A. Belief in Luck and Luckiness Scale Question: To what extent do you personally agree or disagree with the following statements? Items: I believe in good and bad lucka [I try hard to be niceb] I mostly have bad luckc,⁄ There is no such thing as good or bad lucka,⁄ [It’s hard to be niceb] I’m not luckyc,⁄ Good and bad luck really do exista I generally have good luckc [I’m nice if I tryb] Luck doesn’t affect what happens to mea,⁄ I consider myself a lucky personc Belief in luck is completely sensiblea
[It’s nice to try hardb] Bad luck happens to me oftenc,⁄ Luck only exists in peoples’ mindsa,⁄ I’m usually luckyc Interval measure: 1 (Strongly disagree) to 5 (Strongly agree). Note. aBelief in Luck items. bDistracter items. cPersonal Luckiness items. ⁄Reverse coded. References André, N. (2006). Good fortune, luck, opportunity and their lack: How do agents perceive them? Personality and Individual Differences, 40, 1461–1472. Bridgstock, M., Marais, I., & Sturgess, K. (2011). The structure of superstitious action – A further analysis of fresh evidence. Personality and Individual Differences, 50, 795–798. Darke, P. R., & Freedman, J. L. (1997a). Lucky events and beliefs in luck: Paradoxical effects on confidence and risk-taking. Personality and Social Psychology Bulletin, 23, 378–388. Darke, P. R., & Freedman, J. L. (1997b). The belief in good luck scale. Journal of Research in Personality, 31, 486–511. Day, L., & Maltby, J. (2003). Belief in good luck and psychological wellbeing: The mediating role of optimism and irrational beliefs. Journal of Psychology, 137, 99–110. Dickie, L., Surgenor, L. J., Wilson, M., & McDowell, J. (2012). The structure and reliability of the Clinical Perfectionism Questionnaire. Personality and Individual Differences, 52, 865–869. Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The satisfaction with life scale. Journal of Personality Assessment, 49, 71–75. Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA: Stanford University Press. Hart, K. E., & Hope, C. W. (2004). Cynical hostility and the psychosocial vulnerability model of disease risk: Confounding effects of neuroticism (negative affectivity) bias. Personality and Individual Differences, 36, 1571–1582. Judge, T. A., Erez, A., Bono, J. E., & Thoresen, C. J. (2002). Are measures of self-esteem, neuroticism, locus of control, and generalized self-efficacy indicators of a common core construct? Journal of Personality and Social Psychology, 83, 693–710. Kam, C., & Meyer, J. P. (2012). Do optimism and pessimism have different relationships with personality dimensions? A re-examination. Personality and Individual Differences, 52, 123–127. Keren, G., & Wagenaar, W. A. (1985). On the psychology of playing blackjack – Normative and descriptive considerations with implications for decisiontheory. Journal of Experimental Psychology: General, 114, 133–158. Levenson, H. (1981). Differentiating among internality, powerful others, and chance. In H. Lefcourt (Ed.). Research with the locus of control construct (Vol. 1, pp. 15–63). New York: Academic Press. Maltby, J., Day, L., Gill, P., Colley, A., & Wood, A. M. (2008). Beliefs around luck: Confirming the empirical conceptualization of beliefs around luck and the development of the Darke and Freedman beliefs around luck scale. Personality and Individual Differences, 45, 655–660. Nail, P. R., Misak, J. E., & Davis, R. M. (2004). Self-affirmation versus self-consistency: A comparison of two competing self-theories of dissonance phenomena. Personality and Individual Differences, 36, 1893–1905. Prendergast, G. P., & Thompson, E. R. (2008). Sales promotion strategies and belief in luck. Psychology & Marketing, 25, 1043–1062. Scheier, M. F., Carver, C. S., & Bridges, M. W. (1994). Distinguishing optimism from neuroticism (and trait anxiety, self-mastery, and self-esteem): A re-evaluation of the Life Orientation Test. Journal of Personality and Social Psychology, 67, 1063–1078. Spector, P. E., Van Katwyk, P. T., Brannick, M. T., & Chen, P. Y. (1997). When two factors don’t reflect two constructs: How item characteristics can produce artifactual factors. Journal of Management, 23, 659–677. Stipek, D. J., & Gralinski, J. H. (1991). Gender differences in children’s achievementrelated beliefs and emotional response to success and failure in mathematics. Journal of Educational Psychology, 83, 361–371. Thompson, E. R. (2007). Development and validation of an internationally reliable short-form of the positive and negative affect schedule (PANAS). Journal of Cross-Cultural Psychology, 38(2), 227–242. Thompson, E. R. (2008). Development and validation of an International English BigFive Mini-Markers. Personality and Individual Differences, 45, 542–548. Wiseman, R., & Watt, C. (2004). Measuring superstitious belief: Why lucky charms matter. Personality and Individual Differences, 37, 1533–1541. Young, M. J., Chen, N., & Morris, M. W. (2009). Belief in stable and fleeting luck and achievement motivation. Personality and Individual Differences, 47, 150–154.