Understanding the relationship between mood and creativity:A meta-analysis qMark A.Davis *Department of Management,University of North Texas,P.O.Box 305429,Denton,TX 76203-5429,USAa r t i c l e i n f o Article history:Received 28August 2007Accepted 4April 2008Available online 21May 2008Accepted by John Schaubroeck Keywords:Mood Affect Creativitya b s t r a c tA meta-analysis of 62experimental and 10non-experimental studies was conducted to evaluate the positive-mood-enhances-creativity generalization.While the results demonstrate that positive mood enhances creativity,the strength of that effect is contingent upon the comparative or referent mood state (i.e.,neutral or negative mood)as well as the type of creative task.Further,the pattern of effect sizes sup-ports a curvilinear relationship between affective intensity and creative performance.In general,a con-textual perspective of mood–creativity relations is supported.Ó2008Elsevier Inc.All rights reserved.A large body of theoretical and empirical work testifies to the keen interest in how affective states influence work-related cogni-tion and behavior (Forgas &George,2001).One domain that has re-ceived considerable attention is the relationship between affect and creativity.The current emphasis that many firms place on enhancing creativity,a key to organizational effectiveness and competitive advantage (Amabile,1996;Woodman,Sawyer,&Griffin,1993),is a strong impetus for creativity research.Further,there is general agreement that tasks of creative thinking are mood sensi-tive.Unfortunately,whether positive or negative moods facilitate or inhibit creativity is an ongoing debate in the literature.On the one hand,a number of studies support the view that positive mood facilitates creativity across a range of tasks (e.g.,Forgas,2000;Hirt,1999;*Isen,Daubman,&Nowicki,1987)leading some researchers to conclude that ‘‘pleasant moods promote original thinking”(Lyubomirsky,King,&Diener,2005).On the other hand,findings that contradict the positive mood—enhanced creativity perspective have been reported.In fact,some studies demonstrate that positive moods can inhibit creativity and negative moods can facilitate it (George &Zhou,2002;Kaufmann &Vosburg,2002).Consequently,Kaufmann,2003contends that a positive-mood-promotes-creativ-ity generalization is premature;rather,researchers should adopt a contingency view that incorporates contextual characteristics and conditions likely to moderate mood–creativity relationships.To shed some light on this debate,this paper presents a meta-analytic review of mood–creativity research.Given the conflicting perspectives on the relationship,a primary objective of the research is to provide evidence for the direction and mag-nitude of mood effects on creativity,and investigate moderator variables with the potential for altering those effects.The criteria for including or excluding studies for any meta-analysis necessi-tate a coherent definition of the focal constructs.In fact,varying theoretical and empirical definitions employed in mood–creativ-ity studies may explain some of the inconsistent results in this literature.Accordingly,the paper begins with an overview of cre-ativity.Next,concepts and definitions of mood are stly,theory and hypotheses linking mood and creativity are presented.Conceptualizations of creativityWhen discussing creativity,it is useful to distinguish creative outcomes from creative processes.Amabile (1983,1996)and others define creative performance in terms of the creative out-come:‘‘Creativity means a person’s capacity to produce new or original ideas,insights,restructuring,inventions,or artistic objects,which are accepted by experts as being of scientific,aes-thetic,social,or technological value (Vernon,1989,p.94).”This definition describes creativity in terms of results,which are judged on the dual standards of (1)novelty or uniqueness and (2)usefulness or value.Although standards for recognizing crea-tive outcomes are essential,an obvious question remains ‘‘what processes bring about creative outcomes?”Runco and Chand (1995)posit a two-tiered componential model of creative think-ing that may be useful for understanding these processes.The0749-5978/$-see front matter Ó2008Elsevier Inc.All rights reserved.doi:10.1016/j.obhdp.2008.04.001qI thank Philip Bobko and two anonymous reviewers for their helpful and constructive comments;and William McDowell and Zhen Li for assistance in data coding.*Fax:+19405654394.E-mail address:davism@Organizational Behavior and Human Decision Processes 108(2009)25–38Contents lists available at ScienceDirectOrganizational Behavior and Human Decision Processesj o ur na l h om e pa ge :w w w.e ls e v ie r.c o m/lo c at e/ob hdpprimary tier includes three controlling components of creative thinking:problemfinding,ideation,and evaluation.Knowledge (declarative and procedural)and motivation(intrinsic and extrinsic)are components comprising the secondary tier.In ef-fect,knowledge and motivation are seen as contributing rather than controlling factors in creative thinking.Runco and Chand relegate knowledge and motivation because each can be depen-dent on the primary factors.For instance,intrinsic motivation is often contingent upon certain judgments or appraisals(Lazarus, 1991).Nonetheless,Runco and Chand regard motivation and knowledge as critical to creative thinking and suggest that empirical research can‘‘treat them as adjacent to or in-line with the primary components(p.246).”With respect to the primary tier,problemfinding involves iden-tifying,defining,and working toward a solution to the problem. Problem identification and problem definition are particularly important subprocesses.The former implies that the individual recognizes there is a hurdle or challenge to overcome,while the latter involves defining and redefining the problem into a form that allows solution.Ideation involves a family of skills includingfluency(production of ideas),originality(uniqueness of ideas),andflexibility(variety of ideas).Three tasks that rely upon this ideation component ap-pear extensively in mood–creativity research:divergent thinking, categorization and remote associates tasks.Divergent thinking tasks emphasizefluency,originality,andflexibility,whereas cate-gorization tasks deal primarily with cognitiveflexibility.For in-stance,categorization tasks frequently require sorting concepts into categories or listing similarities and differences among concepts(e.g.,*Hirt,Melton,McDonald,&Harackiewicz,1996; *Murray,Sujan,Hirt,&Sujan,1990).Finally,the Remote Associates Test(RAT;Mednick,1962),the prototypical remote associate task, gauges the ability to discern relationships among remote ideas.Gi-ven much creativity research relies upon tests of divergent think-ing(Runco&Chand,1995),it is worth mentioning that the value of divergent thinking is subject to intense debate.Opposing per-spectives view divergent thinking as either synonymous with or completely unrelated to creativity.Between these camps,some researchers characterize divergent thinking as an estimate of the potential for creativity.In large part,divergent thinking tasks cap-ture the standard of novelty,but neglect the criterion of usefulness espoused by many creativity theorists.This exclusive focus on nov-elty probably accounts for much of the criticism surrounding these tasks.According to Runco and Chand(1995)evaluation or appraisal is the most neglected and misunderstood component of creativity. Interestingly,this component provides an important link to the standard of usefulness.Runco and Chand argue that evaluation works in concert with ideation to ensure that ideas are both origi-nal andfitting or useful.The creativity literature employs a number of creative problem-solving tasks such as the Dunker Candle Prob-lem(Duncker,1945)and Maier’s Two String Problem(Maier, 1931).The significance of these insight problems is that the tasks approximate both standards for creative performance:novelty and usefulness.That is,individuals may offer unique approaches to problem solution without actually solving the problem,but only approaches that solve the problem constitute creative perfor-mance.It stands to reason that evaluation or judgment is particu-larly important for problem-solving tasks as the individual must determine whether the ideas generated are useful,or at least have potential for solving the problem.In sum,creativity can be described in terms of problemfinding, ideation,and evaluation processes.Measures of divergent thinking or associative processes,which dominate much creativity research, largely reflect the ideation component of creativity(i.e.,fluency,flexibility,and originality).By comparison,insight problem-solving tasks likely incorporate ideation and evaluation components of creativity inasmuch as creative solutions invoke the dual standards of novelty and usefulness.Conceptualizations of moodFor purposes of this review,the term affect denotes a generic or superordinate category of phenomena that encompasses the con-cepts of mood and emotion.According to Frijda(1994)there are two primary bases upon which one can distinguish affective phe-nomena:whether the phenomenon implies a relationship with a particular object and whether it refers to a response state or an enduring disposition.Other distinctions include the specificity, intensity,and duration of the affective experience.In practice, defining the concept of mood involves distinguishing moods from emotions.To begin,emotions typically involve a relationship with some object or event in the individual’s environment that directs atten-tion and encourages action.In addition,numerous specific emo-tions exist in nature and each is related to a relatively narrow range of responses.Thus,one is angry at someone,happy about something,or afraid of somebody.In contrast,moods are more dif-fuse or generalized affective states that are not usually directed at any particular object or event.In addition,most theorists view moods as typically less intense than emotions(Morris,1989). The lack of specificity,a defining characteristic for some research-ers(e.g.,Isen,1993),suggests a self-regulatory function in which mood conveys ongoing information about one’s general state of being.In line with this self-regulatory function,empirical studies frequently discern two broad,independent dimensions of affective experience,positive and negative.Although some researchers have posited additional dimensions(e.g.,four rather than two)and more elaborate,hierarchical structures that feature specific feeling states nested under both positive(i.e.,joviality,attentiveness,and self-assurance)and negative(i.e.,fear,guilt,and hostility)moods,the majority of the mood–creativity literature applies the two-factor perspective(Hullett,2005).Some researchers maintain that moods are experienced over a longer period than emotions(e.g.,Watson&Clark,1994).For in-stance,an angry episode may last a few seconds to a few minutes, whereas an irritable mood may persist for several hours or a few days.Yet,Lazarus(1994)argues that duration may not be a suit-able basis for distinguishing moods and emotions.For instance, some moods can be experienced as brief orfleeting and emo-tional episodes can persist over time with more acute and less acute phases of affective experience succeeding one another. Applying the duration criterion is complicated by the fact that emotions and moods can mutually influence each other.For example,a strong emotion of joy can prolong itself and create a positive mood over several days.Similarly,an irritable mood can predispose one to react angrily to petty frustrations. Undoubtedly,deciding when an emotion ends and a mood begins is problematic.Furthermore,moods as well as emotional episodes can be prolonged.In brief,the duration of affective experiences is a common,but imperfect criterion for distinguishing moods and emotions.The notion that moods are experienced over longer periods of time brings to mind the view of mood as a disposition or trait.In general,emotional states(e.g.,state-anxiety)have been defined as transitory responses or feelings;whereas emotional traits(e.g., trait-anxiety)refer to stable individual differences in the likelihood that a person would experience particular emotions.Strictly speak-ing,moods and emotions should be regarded as response states (Lazarus,1994).Nevertheless,empirical research evidences signif-icant individual differences in the tendency to experience negative or positive moods.In addition,studies demonstrate that these26M.A.Davis/Organizational Behavior and Human Decision Processes108(2009)25–38differences persist over time and across different life situations (Watson&Clark,1984).In sum,moods,in contrast to emotions,are relatively diffuse, generalized affective states that typically lack a particular object relation that stimulates an action-orientation.The duration of these transitory affective states can vary.Consistent with the self-regulatory view of moods,empirical work has focused on two general states,positive and negative.Theory and hypothesesAccording to Isen(1999),‘‘Much research using varied affect inductions and varied measures of creativity andflexibility,now supports the conclusion that positive affect fosters cognitive com-plexity,creativity,and innovation(p.530).”Different explana-tions have been offered to account for the facilitative effects of positive mood on creativity.Isen(1993)offers a cognitive priming explanation that focuses on the content of information activated in memory when one is in a happy mood.Ostensibly,positive moods elicit more positive material from memory.Given positive material is more abundant and extensive than other material (e.g.,Cramer,1968),individuals in a positive mood,have access to a fuller,more diverse range of information.Therefore,the indi-vidual should be able to make more diverse connections among ideas as well as discern more differences among the items or con-tent.Hence,positive moods promote divergent thinking and cog-nitivelyflexibility.In support of the cognitive priming perspective,Isen and her colleagues(Isen&Daubman,1984;Isen et al.,1987;*Isen,Johnson, Mertz,&Robinson,1985)found that subjects in whom positive affect had been induced performed better than neutral mood con-trols on a variety of creative tasks.Creativity among negative mood subjects was comparable to neutral mood controls.Elsewhere,Isen (2002)cites evidence that negative affect is a noticeably less effec-tive retrieval cue for material in memory,which may limit though not necessarily impair effects on creative thinking.Of course,the widespread view of mood as a single continuum ranging from po-sitive to negative(Hullett,2005)encourages expectations of mood congruence where negative moods are assumed to be unrelated if not detrimental to creativity.Nonetheless,as referent moods for gauging the impact of positive mood on creativity,neutral and neg-ative moods are clearly dissimilar states.In fact,asymmetric effects of positive and negative mood on information processing do occur (e.g.,Forgas,1995).For instance,one may observe no differences when comparing the effects of positive and negative moods on cre-ativity,while a comparison of positive and neutral mood states may yield sizeable differences in creativity.Consequently,failure to distinguish estimates of the positive-mood-enhanced creativity relationship derived from neutral rather than negative mood con-trol groups could obscure any asymmetric influence.1Accordingly, this meta-analysis provides independent estimates of the positive-mood-promotes-creativity relationship using neutral mood and neg-ative mood reference points.Hypothesis 1.Given a positive versus neutral mood contrast, positive mood enhances creative performance.Hypothesis2.Given a positive versus negative mood contrast,posi-tive mood enhances creative performance.Mood-incongruent effectsThe debate over the positive-mood-enhances-creativity hypoth-esis stems from evidence that contradicts the relationship.What might explain these conflicting results?Task effects and motiva-tional influences are distinct,though complementary accounts for mood-incongruentfindings.In regard to task effects,contradictory results appear with some frequency in studies that employ insight problem-solving tasks (e.g.,Jausovec,1989;Kaufmann&Vosburg,1997).Schwarz’s (1990)feelings-as-information theory provides one explanation for these task effects.Schwarz contends that positive and negative moods act as signals that elicit different processing strategies when approaching creative tasks.Specifically,positive moods sig-nal a state of well-being and evoke a relaxed,playful approach to tasks,whereas,negative mood states indicate danger or distress, which evokes more effortful,systematic thinking in accomplishing tasks.Arguably,a processing strategy that involves playful combi-nations of ideas and exploration of novel solutions should facilitate creative performance,especially divergent thinking.At the same time,the more effortful,detail-oriented processing strategy fostered by negative moods would be advantageous when one seeks the optimized solutions required for insight problem solving tasks.In effect,the feelings-as-information model implies that negative rather than positive mood may boost performance on insight prob-lem solving tasks.Russ(1993)makes a similar argument that incongruent results occur because mood effects are contingent upon the cognitive processes required by the task.She contends that positive mood facilitates performance on measures of diver-gent thinking because individuals relax their category boundaries and are disposed toward accepting that divergent concepts are connected.Yet,the critical assessment or faultfinding fostered by negative mood may actually be detrimental to suchflexibility (Runco&Chand,1995).A contrary pattern emerges when individ-uals must solve creative problems.Inasmuch as negative moods encourage more stringent evaluations of potential solutions,the likelihood offinding optimal solutions increases.However,if posi-tive mood produces contentment with the status quo,individuals will more readily adopt satisficing,not optimizing solutions to a creative problem.With respect to motivational influences,George and Zhou (2002)reason that negative moods can cause individuals to be more critical and discerning.As such,they may drive themselves to produce ideas that are novel and useful.Likewise,Verhaeghen, Joorman,and Khan(2005)argue that self-reflective rumination, which accompanies negative mood(depression),encourages seri-ousness about creative endeavors.In turn,seriousness affects the effort exerted on creative activity.Furthermore,the signaling function of positive moods indicates that everything is going well.Hence,individuals may be less inclined to exert the effort needed to produce original and effective ideas.This motivational account is consistent with the mood-as-input model posited by Martin and Stoner(1996).A unique variation of the more general feelings-as-information model,this model emphasizes the context-dependent feature of mood information.In this model, individuals consider the motivational implications of their moods in particular contexts(Martin,2001).For instance,if the context involves creative performance,mood can inform the individual’s qualitative(‘‘Is this idea creative?”)and quantitative(‘‘Do I have a sufficient number of ideas?”)judgments of creativity.Both the task-dependent and motivational accounts concern the potential advantages of negative moods in comparison with positive moods.Neither directly challenges the benefits to crea-tivity from positive mood vis-à-vis neutral mood.Ultimately, the predictions from each account converge with respect to divergent thinking and insight problem-solving tasks.While the1All the studies comprising this meta-analysis employed between subjectscomparisons when assessing mood effects,however,grounds for identifying thereferent mood state also apply to within-subject comparisons.M.A.Davis/Organizational Behavior and Human Decision Processes108(2009)25–3827former stresses the ideation component of creativity(i.e.,fluency, originality,orflexibility);the latter requires both ideation and evaluation components of creativity because creative perfor-mance will be judged by the dual standards of novelty and use-fulness.That is,a solution implies a‘‘useful”or‘‘fitting”answer to the problem.Thus,whether due to more effortful,systematic processing or an increased drive to produce a genuine remedy, negative moods stimulate the behaviors needed for recognizing and achieving(useful)problem solutions.In contrast,the in-creased cognitiveflexibility and the relaxed,playful processing that characterizes positive moods should foster creative perfor-mance on divergent thinking,remote associates,and categoriza-tion tasks.In sum,for creative tasks emphasizing ideation skills,positive rather than negative mood should enhance performance.However, insight problem solving tasks involve ideation and evaluation com-ponents of creativity.This raises the possibility of incompatible influences.That is,while positive mood facilitates cognitive idea-tion,the accompanying state of well-being may also diminish the effort and stringent evaluation needed to solve the problem.Con-versely,while negative moods can stimulate effort and critical evaluation,adverse effects onfluency andflexibility(i.e.,ideation) could hinder the creative thinking needed to generate novel ideas. Accordingly,Hypothesis3.Given a positive versus negative mood contrast, positive mood enhances creative performance on ideation tasks. Hypothesis4.Given a positive versus negative mood contrast,cre-ative performance on problem-solving tasks does not differ as a function of mood.Experimental research involving affect and creativity character-istically uses a mood induction procedure(MIP)to manipulate in-tended mood states.Westermann,Spies,Stahl,and Hesse(1996) present evidence that these procedures differ in strength,while other studies suggest the duration of the mood effects generated by these procedures varies as well.For instance,some studies re-veal a diminishing influence for positive mood after10to20min (e.g.,Frost&Green,1982;Isen,Clark,&Schwartz,1976;Isen& Gorgoglione,1983).By comparison,negative moods,and presum-ably their effects,appear to last longer than positive moods(Frost &Green,1982).Given that dissipation of positive mood states is least pronounced with tasks of brief duration,the benefits of posi-tive(vis-à-vis negative)mood for creative performance should be strongest for short-lived tasks.By comparison,longer tasks in-crease the chances of accruing performance benefits from negative mood.First,the signaling function of negative mood could moti-vate individuals to exert more effort or to delay the decision to quit.Second,these benefits could persist long after comparative ef-fects of positive mood have waned.Although the time allotted to creative tasks is not always reported,researchers commonly state whether those tasks are timed or untimed.Timed tasks naturally restrict the performance interval that follows an induction proce-dure,whereas,untimed tasks allow participants to continue task performance indefinitely.Given their limited duration,the pros-pect for observing enhanced creativity due to positive mood should increase with timed tasks.Conversely,performance benefits asso-ciated with negative mood are more likely to emerge with untimed tasks of indefinite duration.In sum,the performance advantages of positive moods are most likely to occur with tasks of short duration,whereas tasks of longer duration increase the chances of accruing performance benefits from negative mood.On the whole,timed tasks diminish the dura-tion of performance opportunities vis-à-vis untimed tasks. Therefore,Hypothesis 5.Given a positive versus negative mood contrast, positive mood enhances creative performance on timed tasks. Hypothesis 6.Given a positive versus negative mood contrast, negative mood enhances creative performance on un timed tasks.Mood attributionsIn addition to the signaling function of mood,Schwarz’s(1990) feelings-as-information theory holds that individuals alerted to the source of their mood will adjust the informational value or mean-ing of their mood state.This often has the effect of attenuating mood influences on task performance.In contrast,when individu-als are unaware of or misattribute the source of their mood,the affective state provides a basis for judgment and any mood influ-ences on task performance are strengthened.In fact,prior research indicates that simply cueing individuals to the source of their feel-ings,whether positive or negative,eliminates the influence of mood on many types of evaluative judgments such as reports of happiness,evaluations of other persons,and satisfaction with con-sumer goods(Clore,Schwarz,&Conway,1994).For this reason, most experimental studies employ a ruse or cover story to intro-duce the mood manipulation procedure.A typical ruse is to inform participants that they will participate in two or more separate tasks or experiments(e.g.,*Martin,Ward,Achee,&Wyer,1993). The‘‘first”experiment might involve viewingfilm clips or imagin-ing experiences with different emotional qualities.To assess the effectiveness of the manipulation,researchers often embed mood-related items within a questionnaire or rating scale required as part of the experiment.In some instances,a brief distractor task is employed prior to introducing the‘‘second”experiment.The aim of the cover story is to ensure that misattribution occurs to avoid subjects’discounting the information value of their mood state during task performance.Based on the feelings-as-information theory,one should observe an augmenting effect of mood among studies that employ a cover story designed to promote mood misattribution.Accordingly: Hypothesis7.Studies that incorporate a(mood induction)‘‘cover story”yield a stronger positive-mood-enhanced-creativity effect than studies that exclude a cover story.Mood intensityMoods vary in intensity;and affective intensity can have an effect on cognitive and motivational arousal(Martindale,1981; Schwarz,1990;Weis&Cropanzano,1996).Martindale(1981) contends that moderate arousal facilitates creativity,whereas ex-tremely low and high levels of arousal impede it.This perspec-tive is congruent with the Yerkes and Dodson(1908)theory of physiological arousal,which posits a relationship between arou-sal and performance that conforms to an inverted-U shaped function.Furthermore,evidence suggests that both intense posi-tive and intense negative emotion generate disordered thinking that can divert attention from or preclude the kind of thinking required for the generation,analysis,and elaboration of ideas that give rise to creativity(e.g.,Eysenck,1995;Isen et al., 1987).In contrast,extremely low levels of affective intensity may not possess the threshold levels of arousal needed to pro-mote creative thinking.Given the potency of mood induction varies with the procedure employed by the researcher(Westermann et al.,1996),affective intensity and arousal will depend on the strength of the induction procedure.It follows that the manipulation strength of induction procedures governs the influence that mood states have on crea-28M.A.Davis/Organizational Behavior and Human Decision Processes108(2009)25–38tive performance.According to the curvilinear perspective,excep-tionally weak mood inductions fail to produce the baseline level of arousal needed to observe hypothesized effects,whereas exces-sively potent induction procedures generate levels of arousal that negate mood effects.Ultimately,the likelihood of mood enhancing effects on creative performance is greatest with mood inductions of moderate potency.Accordingly:Hypothesis8.Studies that employ mood induction procedures of moderate strength yield stronger mood influences on creative performance than studies characterized by induction procedures of low or high strength.The hypotheses comprising this study dictate a principal focus on experimental evidence pertinent to the mood–creativity rela-tionship.Experimental approaches offer some important advan-tages over non-experimental studies(Cook&Campbell,1979). For instance,the manipulation of mood states permits more di-rect inferences of causation because the experimenter can more clearly discern the presence or absence of immediate mood states that precede creative performances.Non-experimental ap-proaches tend to rely upon participants’recall of average nega-tive and positive mood states experienced over a particular time period in the past(e.g.,George&Zhou,2002,2007; *Madjar,Oldham,&Pratt,2002).Inasmuch as mood is construed as a background phenomenon that is rarely the focus of con-scious awareness(Lazarus,1994),recalling the experience of these background states is undoubtedly subject to error.In par-ticular,the individual’s recollection of an affective experience introduces causal ambiguity if the recalled mood state reflects an outcome rather than an antecedent of a creative episode.Fur-ther,James,Brodersen,and Eisenberg(2004)argue that mis-matches in time frame can make the logical connection between affect and creative performance problematic.Specifi-cally,if ratings of creative performance do not specify or simply do not correspond to the same time period in which participants’recalled mood states were actually in force,then support for an affect-creativity relationship is on shaky ground.Finally,with experimental approaches,the effects of positive moods can be compared against different referent moods or baselines,that is, neutral moods and negative moods.Advantages of experimental evidence notwithstanding,few existing studies in this area are non-experimental.Although results can prove unstable with small samples,meta-analytic procedures can be applied to as few as two studies.With that proviso in mind, correlational evidence for the mood–creativity relationship is incorporated as an extension of the results section.MethodRecently,Lyubomirsky et al.(2005)conducted a meta-analytic review of the benefits of frequent positive affect.A portion of that review examined the impact of induced mood states on creativity and problem solving.Lyubomirksy et al.interpreted the mean ef-fect size(r)of.25as small,but consistent with evidence that pleas-ant moods promote original thinking.The authors reported a virtually equivalent effect size for non-experimental studies of po-sitive affect and creativity(mean r=.26).Although the review helps inform our understanding of mood–creativity relations,it does not effectively answer the question addressed by the present meta-analysis.First,effect sizes for experimental studies were computed from a mix of positive versus neutral and positive versus negative mood contrasts.That is,the estimate combined the effects of positive mood from two different referent moods or baselines: neutral and negative.Second,the analysis combined studies of cre-ativity and complex problem solving.Further,the analysis of non-experimental studies included constructs associated with,though not identical to,positive affect(e.g.,optimism,life satisfaction, and mypomania).Third,no unpublished studies involving mood and creativity were included in the analysis.Fourth,the analysis did not correct for measurement error.Finally,the review focused on the benefits of frequent positive affect,effectively disregarding the distinction between state and trait affect.Criteria for inclusionThree decision rules determined the studies included in the meta-analysis of experimental research.First,mood state had to be a manipulated variable and an intentional feature of the re-search design.Hence,the study had to describe or mention the mood induction procedure employed.Studies that employed measures of naturally occurring(i.e.,non-manipulated)mood states or trait affect as the independent variable were excluded. Second,one or more of the dependent variables had to be an out-come or rating of performance on a creativity-related task.Exam-ples of creative tasks include unusual uses,word association, categorization,remote associates,and insight problem-solving tasks.Ratings of creative performance had to reflect judgments of creativity by an independent observer or judge.Studies involv-ing self-perceptions of creativity were excluded.Third,the study had to report sample sizes and the requisite outcome statistics (e.g.,r,univariate F,t,v2)for calculating effect sizes.With one exception,the inclusion rules for the meta-analysis of non-exper-imental studies were identical to the criteria for experimental studies.In the case of non-experimental research,studies that employed measures of naturally occurring mood states were in-cluded.That is,mood state was a measured,but non-manipulated independent variable.Search for primary dataA multi-stage search was conducted tofind published and unpublished studies that met these inclusion criteria.First,a com-puterized bibliographic search in the ABI/INFORM,PsycINFO,Web of Science and Dissertation Abstracts databases was conducted. Keywords for the search included mood,emotion,and affect.These results werefiltered using the terms creativity and innovation to construct a preliminary set of relevant studies.Second,additional studies were found using references cited in the studies obtained from the database search.Third,original studies were identified through a manual search of recent issues(2006–2007)from jour-nals likely to publish studies on emotion and/or creativity:Acad-emy of Management Journal,Administrative Science Quarterly, Creativity Research Journal,Journal of Applied Psychology,Journal of Applied Social Psychology,Journal of Creative Behavior,Journal of Management,Journal of Personality and Social Psychology,and Orga-nizational Behavior and Human Decision Processes.Finally,recog-nized mood-affect researchers were contacted to solicit additional unpublished studies.The overall sample used in the meta-analysis of the experimen-tal literature consisted of62empirical studies(48published and 14unpublished).When multiple estimates of effect sizes were available from a single study,the results were averaged unless sta-tistical values were based on independent samples.This process yielded48effect sizes(n=2794)involving the contrast of positive and neutral affective states,41effect sizes(n=3409)representing the comparison of positive and negative mood states,and24effect sizes(n=1204)comprising the contrast between negative and neutral affective states.Ten published studies yielded10effect sizes(n=1566)used in the meta-analysis of the non-experimental literature.M.A.Davis/Organizational Behavior and Human Decision Processes108(2009)25–3829。