SONJALYUBOMIRSKY,CHRISTKACHandM.ROBINDIMATTEOWHATARETHEDIFFERENCESBETWEENHAPPINESSANDSELF-ESTEEM?(Accepted24June2005)ABSTRACT. Thepresent studyinvestigatedtheoreticallyandempiricallyderivedsimilaritiesanddierencesbetweentheconstructsof enduringhappinessandself-esteem. Participants (N=621), retired employees ages 5195, completed stan-dardized measures of aect, personality, psychosocial characteristics, physicalhealth, anddemographics. Therelationsbetweeneachof thetwotarget variables(happiness andself-esteem) andthe full set of remainingvariables were assessedthroughaseriesofsuccessivestatistical analyses: (1)simplePearsonscorrelations,(2) partial correlations, and (3) hierarchical regression analyses. The results revealedthat happiness and self-esteem, while highly correlated (r=0.58), presented uniquepatterns of relations with the other measured variables. The best predictors ofhappiness were the following: mood and temperamental traits (i.e., extraversion andneuroticism), social relationships (lackof loneliness andsatisfactionwithfriend-ships),purposeinlife,andgloballifesatisfaction.Bycontrast,self-esteemwasbestpredictedbydispositionsrelatedtoagencyandmotivation(i.e.,optimismandlackof hopelessness). Implications for the understanding of happiness and self-esteem arediscussed.KEY WORDS: aect, agency, happiness, nomological net, personality, self-esteem,subjectivewell-beingINTRODUCTIONHappinessandself-esteemappear,onthesurface,tobeinextricablylinked. Intheireverydayexperience, happyindividualstendtofeelgoodabout themselves, andpeople wholackself-worthandself-respect are generally unhappy. Empirical evidence supports thisintuition,revealingmoderatetohighcorrelationsbetweenmeasuresofhappinessandself-esteem(rangingfrom0.36to0.58)(Andrews,1991; A. Campbell et al., 1976; A. Campbell, 1981; Diener andDiener, 1995; Fordyce, 1988; Kozma and Stones, 1978; LyubomirskyandLepper, 1999; Schimmacket al., 2004). TheseassociationsareSocialIndicatorsResearch(2006)78:363404 Springer2005DOI10.1007/s11205-005-0213-yneither perfect nor consistent, however, leaving much of the varianceunexplainedandpromptingtheneedtofurther examinethe simi-laritiesanddierencesbetweenthesetwoconstructs.Althoughfewresearcherswouldopenlyarguethathappinessandself-esteemaresynonymous,self-esteemisoftenusedasanindexofglobal happiness or psychological well-being (e.g., Baruch andBarnett, 1986;Ry,1989;Whitley,1983). Morecommon istheviewthat happiness andself-esteemare sointimately relatedthat it isdicult, if not impossible, toseparate themconceptually. Indeed,happiness may not be possible or realizable without a healthy dose ofself-condenceandself-acceptance.WilliamJames(1910),forexam-ple, spoke of well-being and self-esteemin the same breath(Epstein, 1973, p. 405), positing that all three aspects of the self thematerial, the social, and the spiritual are capable of evoking feelingsofhappiness.Nearlyacenturylater,inareviewoftheensuingper-spectives on happiness, Ry (1989) concluded that the most recurrentcriterion for positive well-being has been the individuals sense of self-acceptanceorself-esteem(seealsoTaylorandBrown,1988;Myers,1992; Diener, 1996). Manywhoaresocializedinindividualisticcul-turesmaynotevenmakeadistinctionbetweenhowhappytheyarewiththeir lives andhowsatisedtheyarewiththemselves (Lucaset al., 1996).An alternate view holds that happiness and self-esteem are distinctand discriminable constructs. Although self-esteem may seem crucialand adaptive for happiness, it does not provide an adequatedescriptionofhappinessandmaybeunrelatedtomanyofourmosthappyorunhappyexperiences(Parducci, 1995). Just asagoodin-come,agoodjob,oragoodmarriagedoesnotguaranteehappiness(seeDiener,1984;Dieneretal.,1999),highself-esteemisnotasuf-cient conditionfor happiness. This perspective mayhelpexplainwhytherelationshipbetweenself-esteemandlifesatisfaction(akeycomponent of happiness) varies inindividualist versus collectivistcultures(DienerandDiener, 1995). Wherethegroupandthecom-munityarevaluedmorehighlythantheself,self-esteemmaysimplynotbeascriticalasourceofhappiness.In the only study to date to analyze systematically the relationshipbetweenself-esteemandwell-being,Lucasandhiscolleagues(1996)usedmultitrait-multimethodmatrixanalysestoshowthatlifesatis-faction is empirically distinguishable fromself-esteem(see alsoSONJALYUBOMIRSKYETAL. 364Diener and Diener, 1995). A number of important questions have yetto be explored, however: Are global happiness and self-esteem indeeduniqueconstructs and, if so, what is thenature of thedierencesbetween them? Which characteristics discriminate between happy andunhappyindividualsandwhichonesdiscriminatebetweenhighself-esteem (HSE) and low self-esteem (LSE) individuals? These questionsarethefocusofthecurrentstudy.DeningHappinessandSelf-EsteemHappinessHoward Mumford Jones once said that happinessbelongs to thatcategoryofwords, themeaningofwhicheverybodyknowsbutthedenitionof whichnobody cangive (citedinFreedman, 1978).Although happiness may have dierent meanings for dierent people,most agree that it is aglow word(Parducci, 1995) that is, apervasive andlasting sense that life is fullling, meaningful, andpleasant (Myers, 1992). Tostudythis sometimes elusiveconstruct,researchershaveachievedamodestamountofagreementonhowitshould be measured and dened. The most widely accepted denitionis that of Diener andhis colleagues, whoprefer touse the labelsubjectivewell-being,deningitasacombinationoflifesatisfaction(acognitivejudgment)andthebalanceofthefrequencyofpositiveandnegativeaect (i.e., hedonictone) (Larsenet al., 1985, Dieneretal.,1991).Makingtheassumptionthatmostpeopleknowwhethertheyarehappy, anumberof researchershaveallowedtheindividualsbeingmeasured to dene happiness for themselves (e.g., Gurin et al., 1960;Kozma and Stones, 1980; Lyubomirsky and Lepper, 1999). Thisassumptionisalsomadehere.Mostpeoplearecapableofreportingontheirownglobal happiness, andthisjudgmentisnotnecessarilyequivalenttoasimpleaggregateof theirrecentlevelsof aectandratings of life satisfaction (or, reports of their resources). Forexample, one may conceivably consider oneself a very happy person,despiteleadingonlyasomewhathappylifeornottypicallyexperi-encingmorepositivethannegativeemotions.Thus,forourmeasureof happiness in the present study, we use global, subjectiveassessments of whether one is a happy or unhappy individual(LyubomirskyandLepper,1999;Lyubomirsky,2001).HAPPINESSANDSELF-ESTEEM 365Self-EsteemGreaterconsensus existsregarding thedenitionof self-esteem.Self-esteem has been dened as a global feeling of self-worth or adequacyas a person, or generalized feelings of self-acceptance, goodness, andself-respect (Coopersmith, 1967; Crocker and Major, 1989;Rosenberg, 1965; Wylie, 1979). This global, personal judgment ofworthinessischaracterizedastheevaluativecomponent of theself(e.g.,J.D.Campbell,1990),andasdistinctfromcollectiveorracialself-esteem (Crocker and Major, 1989). According to Epstein (1973),peoplehaveabasicneedfor self-esteem, and, at least inWesterncultures, they use numerous strategies to maintain it (Dunning et al.,1995; Taylor andBrown, 1988; cf. Markus andKitayama, 1991;DienerandDiener, 1995). Self-esteemformsearlyinthecourseofdevelopment, remains fairly constant over time, and is relativelyimmunetochange(J.D.Campbell,1990).CorrelatesandPotentialSourcesofHappinessandSelf-EsteemThemore-or-less accepteddenitions of happiness andself-esteemsuggestimportantdierencesbetweenthesetwoconstructs, nottheleastofwhichisthathappinessisabroaderandmoreoverarchingconcept, whereasself-esteemappearstobemorespecicandmorecognitiveinnature.Thus,itiscriticaltodelvedeeperintotheliter-ature and explore the particular constructs that have been empiricallyrelatedtothetwoconstructsnamely, whatattributescharacterizepeople who are happy (vs. unhappy) and what attributes characterizepeople who are high (vs. low) in self-esteem. Whether these attributeshappen to be quite comparable or quite distinct might oer clues intothedierencesandsimilaritiesbetweenhappinessandself-esteem.CorrelatesofHappinessTraitsanddispositions.Typically,personalitytraitsaccountforasmuch as 40% to 50% of the proportion of the variability in well-being(Diener et al., 1999); thus, traits anddispositions appear tobe critical tohappiness. According to Myers and Diener (1995), four traits consis-tently characterize happy people: self-esteem(e.g., Kozma and Stones,1978; Fordyce, 1988; Diener and Diener, 1995), optimism (A. Camp-bell, 1981; Carver and Gaines, 1987), extraversion (Costa and McCrae,1980; Costaet al., 1981; Emmons andDiener, 1985; HeadeyandSONJALYUBOMIRSKYETAL. 366Wearing, 1989; Pavot et al., 1990; Brebner et al., 1995), and a sense ofpersonal mastery or control (Ry, 1989; Csikszentmihalyi and Wong,1991; Grob et al., 1999). Thus, in general, happy individuals have so-cial, outgoingpersonalities, aswell aspositivefeelingsaboutthem-selves, their sense of mastery, and the future. They are also more likelyto be active and energetic and less likely to be neurotic (DeNeve andCooper, 1998). Theseattitudes canbeself-fullling, leadinghappypeople to experience more positive events (e.g., Headey and Wearing,1989; Magnus et al., 1993) andmorefulllingsocial relationships,which can further enhance well-being.McCraeandCosta(1991) proposedtwoways that traits mightinuence well-being temperamental andinstrumental. The tem-peramentalsequenceisillustratedbyatraitleadingtoamood,andthen,inturn,bothtraitandmoodbearingonhappiness(CostaandMcCrae, 1980; McCraeandCosta, 1991). Similarly, Rusting(1998)positedthat traits predispose people tointerpret events ina dis-positionallycongruent manner. Thus, happypeoples positiveatti-tudesandjudgmentsmayleadthemtoperceivelifeexperiencesinaway that sustains their positive moods (e.g., Lyubomirsky andTucker, 1998; cf. Lyubomirsky, 2001) forexample, byperceivingcontrol intheir actionsorbyconstruingvalueindailylifeevents.Notably, experimental evidence supports the temperamentalsequenceforbothextraversionandneuroticism. Specically, resultsshowthatextravertedindividualsseemtoexperiencepositiveeventsas more intenselypositive thandointroverts, whereas individualshighinneuroticismseemtoexperiencenegativeeventsmorenega-tively than do their more emotionally stable peers (Larsen andKetelaar, 1989).Incontrast,theinstrumentalcausalsequenceisillustratedbythenotion that particular traits compel people towards certain situations,which lead to later happiness or unhappiness. The empirical evidencesupportsthissequenceaswell. Forexample, inarecent study, therelationbetweenextraversionandhappinesswaspartiallymediatedby social aliation and active leisure activities (Tkach andLyubomirsky, 2005).Evenmoreconvincing isastudyofpersonalityanddailyactivitiesthat foundthat extravertschosetospendmoretime in social situations than did introverts, and that extravertsexperiencedgreater positive aect during chosensocial situationsthanduringimposedsolitaryones(Emmonsetal.,1986).HAPPINESSANDSELF-ESTEEM 367Social aliation. One of the most important sources of happiness ispersonal relationships (Argyle, 1987; Diener, 1984; Ry, 1989; Myers,1992; Kahana et al., 1995; Myers and Diener, 1995). Numerous studiessupport a link between happiness and friendship, marriage, intimacy,and social support (e.g., see Lyubomirsky et al., in press, for a review).Forexample,onestudyshowedthatthosewhonamedveormorefriends with whomthey discussed important matters in the last6months were 60%more likely to report being very happy (cited inMyers, 1992; see alsoHendersonet al., 1981). Indeed, people arehappiest when with friends (Csikszentmihalyi and Hunter, 2003). Also,happy people are more likely to have friends who encourage and sup-port them(Myers, 1992). Anumber of studies suggest that closefriendships can help buer stress (Reis, 1984) and avert distress due toloneliness, anxiety, boredom, andloss of self-esteem(PeplauandPerlman, 1982; seeArgyle, 1987). Notsurprisingly, lonelinessisin-verselycorrelatedwithhappiness, especiallyinolderadults(LeeandIshii-Kuntz, 1987), and positively correlated with depression (Weeks etal., 1980; Peplau and Perlman, 1982; Seligman, 1991).Positivemoods. Theexperienceof happiness is markedbymorefrequent positive aective states than negative ones (Bradburn, 1969;Diener et al., 1985b; Diener et al., 1991). Indeed, the possibility existsthat because people may rely on their moods at the time of judgmenttosimplifythecomplextaskofappraisingtheirhappiness(Schwarzand Clore, 1983; Schwarz and Bohner, 1996), such appraisals may beaectedbytransient moods. However, althoughhappiness reportsaresomewhatdependentonmood,theyareinfactrelativelystable,with good long-termreliability (see Diener, 1994, for a review).Furthermore, dailymoodisnotconsistentlycorrelatedwithhappi-ness (Lucas et al., 1996; Lyubomirsky and Tucker, 1998; Diener et al.,1999), doesnotpredictoutcomesover andabove levelsof happiness(e.g., Seidlitz and Diener, 1993; Lyubomirsky and Ross, 1997;Lyubomirsky and Tucker, 1998), and peer reports of well-beingpredict current well-beingbetter thandoes transient mood(PavotandDiener, 1993). Thus, the evidence suggests that bothcurrentmoodandlong-termaect are reectedinmeasures of happiness(Diener, 1984). Inaddition, whenpeopleareaskedtodescribehowthey decide whether they are happy, the decisions are predominantlybasedonaect(Rossetal.,1986).SONJALYUBOMIRSKYETAL. 368Satisfactionwithlife. Sincesubjectivewell-beingiscommonlyde-ned as an aggregate of life satisfaction and the balance of aect, it isnotsurprisingthathappyindividualsdemonstratebothglobalsatis-factionwiththeirlives(seeDiener, 1984; Argyle, 1987; MyersandDiener, 1995, for reviews) and satisfaction within specic life domains,such as work, recreation, friendship, marriage, health, and the self (A.Campbell, 1981; Argyle, 1987; Eysenck, 1990; Lepper, 1996; Diener etal., 1999). An inuential top-down view is that happiness inuencesonesoutlook,whichcolorsonesperceptionsofspecicdomains(Stones and Kozma, 1986; Feist et al., 1995; Veenhoven, 1997). How-ever, evidence alsosuggests that happiness results inpart fromasummation of various domains of satisfaction (Feist et al., 1995).Satisfactionof psychological needs. Althoughalargenumber ofpsychological needs havebeenproposed, recentlyresearchers havebegun to focus on three universal needs competence, autonomy, andrelatedness (Deci and Ryan, 1985). Regardless of the number of needsput forward, however, theorists agree that the satisfaction of needs isrelated to well-being (e.g., Rogers, 1961; Maslow, 1970; Omodei andWearing, 1990), and research supports this relationship. For example,adiarystudydemonstratedthatpeoplesself-reportedgooddaysweredaysinwhichtheywerefeelingabovetheirbaselineincompe-tencyandautonomythat is, theyfelt capableandself-motivated(Sheldonetal.,1996).Researchprovidesadditionalsupportforthislinkage at the daily level (Reis et al., 2000), as well as for longer-termperiods (e.g., Omodei and Wearing, 1990; Sheldon and Elliot, 1999).Demographics and life events. Finally, the literature on happiness ismarkedbythe robust ndingthat external blessings namely,demographic variables and life circumstances are less important thana happy disposition. For example, smaller-than-expected correlationshavebeenfoundbetweenwell-beingandobjectivevariablessuchasincome, age, gender, race, occupation, education, religion, children,and life events in both younger and older adults (see Diener et al., 1999;Lyubomirsky, 2005, for reviews; see also George, 1978).CorrelatesofSelf-EsteemMost investigations of self-esteemhave been concerned with thethoughts,moods,andactionsoftenobservedincollege studentsinthe laboratory that are associated with high versus low levels of theconstruct(e.g., CrockerandMajor, 1989). Forexample, relativetoHAPPINESSANDSELF-ESTEEM 369peoplewithlowself-esteem, thosewithhighself-esteemhavebeenfoundtopossessclearerself-concepts(J.D. Campbell andLavallee,1993); to be less vulnerable to depression (e.g., Harter, 1993; TennenandAeck,1993;seealsoKernisetal.,1993)andanxiety(Flemingand Courtney, 1984); to be more resilient to self-image threats(Spenceretal., 1993); andtobemorelikelytosavorpositiveaect(Woodet al., 2003), topersist inthefaceof failure(DiPaulaandJ. D. Campbell, 2002),andtoperceivenegativefeedbackasachal-lengerather thanathreat (Seeryet al., 2004). It shouldbenoted,however, thatlittleempirical researchonself-esteemhasbeencon-ductedwiththeelderly,whowerethesubjectofourstudy.Agencyandcompetence.Althoughlessresearchhasaddressedtherelationshipbetweenself-esteemandstable,dispositionalconstructs,some clues regarding the source of feelings of self-worth are providedin theoretical accounts of self-esteem. For example, according to onetheoreticalperspective,self-esteemisgainedthroughecaciousandsuccessful navigationofonesenvironment, wherebyoneacquiresasenseof control, competence, andability(Bandura, 1977; CrockerandMajor, 1989; cf. VanTuinenandRamanaiah, 1979). Further-more, according to a developmental perspective, self-worth is derivedfrom having a sense of competence in domains that are valued by theindividualandimportantsignicantothers(Harter,1993).Thus, self-esteemwouldbeexpectedtobecloselylinkedwithasense of agency or mastery and control of ones environment. Indeed,thenotionthatasenseofpersonalcontroliscriticaltoself-conceptandself-esteemhas beenendorsedbyanumber of theorists (e.g.,Fenichel, 1945; Heider, 1958; White, 1959; seeTaylor andBrown,1988,forareview).Positive expectations. Self-esteemis alsohighly correlatedwithoptimismandlackofhopelessnessincollegestudents(Scheieretal.,1994; Lucaset al., 1996; seealsoTennenandAeck, 1993). Opti-mists anticipate bright futures andexpect favorable outcomes fortheiractions.Thus,onemightexpectoptimiststopersistlongerandharder through lifes tasks and challenges, creating self-fulllingprophecies,and,consequently,bolsteringtheirself-regard.Insum, it wouldappear that self-esteemis stronglyrelatedtomotivationalconstructssuchasoptimism,mastery,andcompetencethatis,feelingsthatoneisacompetentagent,capableofsuccess.SONJALYUBOMIRSKYETAL. 370ThePresentStudyOurexaminationoftherelevantliteraturespointstomarkeddier-ences betweentheconstructs that areempiricallyandtheoreticallylinkedtohappiness, comparedwiththoselinkedtoself-esteem. In-deed,happinessand self-esteemseemtosharehighcorrelationsonlywithasenseofmastery, optimism, andhopefulness. Inthepresentstudy, weexamineacomprehensivesetofvariablestoassessprevi-ouslyunexploredrelationshipsinvolvinghappinessandself-esteem.The primaryapproachundertakeninthe current studywas toexamine the nomological networks of bothhappiness andself-esteemthat is, todeterminethosefactorsthat aremost stronglyassociatedwitheachconstruct(Lucasetal.,1996).Establishingthathappiness andself-esteemeachhave aunique nomological net ofrelationshipssupportstheirdiscriminantvalidityandprovidescluestothedierencesandsimilaritiesbetweenthem.Unlike much of the literature reviewed here, the present study usedasample of retiredadults, ages 5195. This sample oers severaladvantages,includingtheopportunitytobolsterthegeneralizabilityof previous ndings. In contrast to a college sample, an older samplemaymoreaccuratelyreectthethoughts,feelings,andbehaviorsofthe general populationineverydaylife (Sears, 1986). Inaddition,unlike undergraduates, older people arguably have relatively strongerself-denitionsandmorecertainandsolidiedattitudes. Theyhaveconceivablyenjoyedalonglifetoreectonandjudgetheirperson-ality, feelings, andbeliefs e.g., whether theyhaveachievedtheircherishedgoals,howhappyoroptimisticorneurotictheytrulyare,or how much they genuinely accept and like themselves, their friends,ortheiractivities.Much more empirical research using elderly samples has focused onwell-being thanonself-esteem. Self-esteemtends tobe viewedasimportant during the formative years and not the retirement years, andresearchers interest in it appears to wane with the age of participants(e.g., Dietz, 1996). By contrast, the experience of happiness has been anenduringconcernofgerontologists,inpartbecausesomeofthepre-dictors of happiness are correlated with age and because happiness isfrequentlytakentoindicatesuccessful adaptationtoaging(Schulz,1985; Whitbourne, 1985; Baker et al., 1992). According to Freedman(1978), older people who are happy are more likely to feel acceptanceHAPPINESSANDSELF-ESTEEM 371andsatisfaction withtheirlivesthanyounger folks,aremorecertainthat their lives have meaning and direction, and are more condent intheir guiding values. Paradoxically, they may also be more optimisticthan younger people because most of their important decisions i.e.,regardingtheiraccomplishments,relationships,andloveshaveal-ready been made. Indeed, older individuals have been found to reporthigher levels of happiness than younger ones (Roberts and Chapman,2000; Sheldon and Kasser, 2001).Incontrast, inconsistent results have beenfoundregardingagedierencesinsourcesof self-esteem(e.g., Wylie, 1979; Goveet al.,1989;Ry,1989;Colemanetal.,1993;Atchley,1994;Brandtstadterand Greve, 1994; Dietz, 1996). This pattern may be a consequence oftwo opposing processes (Tennen and Aeck, 1993), which could leadtodeclinesinthestabilityofself-esteemduringmiddleandoldage(Trzesniewski et al., 2003). The usual suspects of self-esteemenhancement (e.g., inWesternculture power andstatus, careersuccess, beauty, athletic ability, etc.) are less available inoldage.Furthermore, ones peers and mentors, who probably have played animportantroleinboostingonesself-esteem,mayhavepassedaway,andones healthhas beguntodecline. Onthe other hand, withmaturity, people are able to gain condence in their abilities and altertheir goals adaptively andappropriately for example, by relin-quishingsomeoftheirunattainabledreamsandacceptingtheirfail-ingsaswellastheirgifts.Followingourearlier discussionof previous work, wehypothe-sized that self-reported happiness, assessed with the SubjectiveHappinessScale(LyubomirskyandLepper,1999),wouldberelatedtoself-reportedpositive aect, needsatisfaction, satisfactionwithoneslifeasawholeaswell aswithavarietyoflifedomains, tem-peramental traits (anextraverteddemeanor andalackof neuroti-cism), an optimistic outlook, a sense of mastery, fullling socialrelationships(e.g., satisfactionwithonesfamilyandfriends, socialsupport,lackofloneliness),andasenseofpurposeinlife.Objectivecharacteristics and circumstances such as sex, age, and life events werenotexpectedtocorrelatestronglywithhappiness.With respect to self-esteem(assessed with Rosenbergs (1965)widelyusedscale),previoustheoreticalandempiricalworkledustoexpect a signicant association with agentic dispositions such asoptimism, asenseofmasteryorcontrol, andaninverseassociationSONJALYUBOMIRSKYETAL. 372with hopelessness. However, due to a paucity of research, no specicpredictions were made about the relationship between self-esteem andother factors. Finally, we examined links between happiness and self-esteem, respectively, andaset of additional potentiallyimportantconstructs. For example, givenour older sample, several variablestappinghealthwereincludednamely, perceptionsofoneshealth,physical symptoms, physical functioning, severity of pain, and energylevel. Researchhasdemonstratedaconsistent relationshipbetweenpoorhealthstatusandunhappinessanddistress, especiallyinolderpeople (Okun et al., 1984; Lubin et al., 1988; for reviews, seePressmanandCohen, inpress). However, otherstudieshavefoundthat whenother factors are takenintoaccount, subjective healthstatus has little eect onhappiness levels (see Diener, 1994). Pre-sumably, however, the happiness of older folks may be more stronglyrelatedtotheirhealththanthatofyoungerones.METHODSampleParticipants were retired employees of a large utility company servingmuchofSouthernCalifornia.Asystematic,computer-drivenproce-durewasusedbythecompanytoselectarandomsubsampleofthenamesandaddressesofindividualsfromthecompanyspopulationof retirees. The sample was comprised of 621 individuals, agesranging from 51 to 95 (M=70), who were 80% male, 80% married,and 96% Caucasian. They were typically well-educated (56% with atleastsomecollegeeducation)andhadbeenretiredforanaverageof10years. Only 10% of the participants were widowed (n=65), withtheaveragetimesincewidowhoodequaling7.5years. Inaddition,fully 85% of this sample reported sharing a residence with at least oneother.ProceduresandMeasuresSelf-reportquestionnairesweremailedtothehomeaddressesoftheparticipants, whowereaskedtocompletethemattheirleisureandwhile alone. Respondents returned the completed questionnaires in aHAPPINESSANDSELF-ESTEEM 373postage-paid envelope and were not compensated for their time. Withtheexceptionof demographics, all measuresusedinthisstudyaredescribedinTableI. Henceforth, thevariableshappiness1andself-esteem will be referred to as the target variables, whereas all othervariablesofinterestwillbereferredtoasthemeasuredvariables.RESULTSOverviewofStatisticalAnalysesOurprimaryaimwastoexplorethesimilaritiesanddierencesbe-tween the two target variables, happiness and self-esteem. To this end,weperformedaconrmatoryfactoranalysisonourtwotargetvari-ables; computedsimplezero-orderandpartial correlationsbetweenthe target variables andthe measuredvariables; and, nally, per-formed hierarchical regression analyses on a subset of the variables.First, aconrmatory factor analysis (CFA) wasconducted to testwhetheraone-factororatwo-factorsolutiontthedatabest.Thatis, we set out to test whether happiness and self-esteem share so muchconstruct overlap that they are indistinguishable in the population, orwhethertheyarerelatedbutdistinguishableconstructs.Second, zero-order Pearsons correlations were computedsepa-rately for happiness and self-esteem with each measured variable. Theaim of these analyses was to examine the similarity and uniqueness ofassociationsamongthetwotargetvariableswiththemeasuredvari-ables.Furthermore,wecomputeddependentsampleZ-scorestotestthemagnitudeofthedierencesbetweenthecorrelationsforhappi-ness andthosefor self-esteemacross eachmeasuredvariable. Theeect size index described by Cohen (1977) was used to determine therelativesizeofthezero-ordercorrelations.Wefocusedprimarilyonthose measured variables that demonstrated a large relation to thetarget variables (i.e., r 0.50) and had signicant Z-scores.Third,partialcorrelationswerecomputedtotesttheindependentrelationshipsbetweeneachofthetargetvariablesandthemeasuredvariables (cf. BlockandKremen, 1996). Thesepartial correlationsprovidedinformationaboutthestrengthoftheassociationbetweenone target variable with each measured variable, while controlling forthe inuenceof the other target variable. Here, our focus was onSONJALYUBOMIRSKYETAL. 374TABLEIDescriptivestatisticsofthemeasuredvariablesMeasureItemsDescriptionMSDaHappiness4LyubomirskyandLepper(1999):Levelsofglobalandcomparativesubjectivehappiness(7-point).5.670.930.86Self-esteem10Rosenberg(1965):Globalfeelingsofself-worth(4-point).2.360.420.87Domainsatisfactionsa10Michalos(1985):Includessatisfactionwithhealth,familyrelations,friendships,housing,residentialarea,recreation,self-esteem,transportation,education,andmaritalrelationship(7-point).5.480.78EnergyLevel4Stewartetal.(1992):Fatigueandenergyinthelastfourweeks(6-category).3.180.960.91Extraversion10EysenckandEysencks(1985)shortform:Dispositionalextraversion(Yes/No).0.650.240.71Healthperceptions4Stewartetal.(1992):Ratingsofcurrentstateofhealth(5-category).2.790.720.75Hopelessness11Becketal.(1974):Hopefulfeelingsaboutlifeandexpectationsforthefuture(4-point).0.700.480.93Lifesatisfaction5Dieneretal.(1985):Ratingsofgenerallifesatisfaction(7-point).4.155.360.89Loneliness7HaysandDiMatteo(1987):Feelingsoflonelinessandisolationinthelast4weeks(4-point)0.720.670.89Moodstate1Ratingofpresentmood(7-point).5.861.05HAPPINESSANDSELF-ESTEEM 375TABLEIContinuedMeasureItemsDescriptionMSDaNeedsatisfaction15OmodeiandWearing(1990):BasedonMurraysneeds;measureslevelsofneedfulllmentacross15separateneeds(7-point).5.290.840.93Negativeaect5Bradburn(1969):Levelsofnegativeaectduringthepastfewweeks(Yes/No).0.150.230.65Neuroticism10EysenckandEysencks(1985)shortform:Dispositionalneuroticism(Yes/No).0.140.190.68Optimism8ScheierandCarvers(1985)LifeOrientationTest:Dispositionaloptimism(4-point).2.140.410.85Painseverity2Sherbourne(1992):Interferenceindailylifecausedbypain(5-category)andoveralllevelofpainseverity(6-category).1.301.040.82Physicalsymptoms12Sherbourneetal.(1992):Physicalsymptomsexperiencedinthelast4weeks(6-category).0.740.54Positiveaect5Bradburn(1969):Levelsofpositiveaectduringthepastfewweeks(Yes/No).0.780.260.69Purposeinlife7Crumbaughs(1968)shortform:Senseofpurposeandmeaninginlife(4-point).5.410.710.83Recentexperiences41Hendersonetal.(1981):Listoflifeeventsduringthepreviousyear.Dividedintonegativeandpositiveeventsviathreeexpertraters(interraterreliability=0.91).NegativeEvents(31items)2.682.16PositiveEvents(10items)0.080.08SONJALYUBOMIRSKYETAL. 376TABLEIContinuedSenseofmastery7Pearlinetal.(1981):Perceptionthatlifechangesareunderonescontrol(4-category).2.020.350.73Socialsupport12SherbourneandStewart(1991):Levelsofperceivedsocialsupportsavailableduringthepreviousmonth(4-category).3.140.960.95Note:Meanvalue(M)representsthemeanscalescore.Highervaluesrepresentgreaterlevelsofthevariable.SD,standarddeviation;a,Cronbachsalpha.denotesthatcomputationofalphareliabilitywasnotappropriate.aDescriptivestatisticsareprovidedforthecompositeofdomainsatisfactionsandnotseparatelyacrosstheindividualdomains.HAPPINESSANDSELF-ESTEEM 377those variables that demonstrated a medium partial relation to thetargetvariables(i.e.,rp0.30).Finally, we focused on those variables that appeared to bedominantpredictorsofthetargetvariablesthatis,thosevariablesthatemergedassignicantpredictorsevenaftercontrollingfortheeects of other measuredvariables. Tothat end, aseries of hier-archical regression analyses were performed, such that happinessand self-esteemwere regressed on the measured variables. Theadvantageofthisstudy, withitslargesetof measuredvariables, isthat it providedtheopportunitytodeterminetherelativestrengthof the associationbetweeneachtarget variable andthe measuredvariables, while taking into account other measured variables.Concernsregardingtheinter-relatednessofthenumerouspredictorvariables (i.e., multicollinearity) were addressed by reducing thetotal number of measured variables into smaller, theoreticallyimportant subsets.In sum,thesuccessivestepsof ouranalyticstrategy allowedustonarrowdown the full set of measured variables to a more parsimonioussetofconstructsthatwerebothstronglyanddierentiallyrelatedtohappiness and self-esteem. These nal variables would have jumped allthe hurdles we had placed in their paths. In the nal hour, these vari-ables would have met the following four criteria: First, the measuredvariables zero-order correlation with the target variable was large.Second, the measured variables zero-order correlation with the targetvariable was signicantly larger than that with the comparison targetvariable. Third, themeasuredvariablespartial correlationwiththetarget variable, controlling for the comparison target variable,remainedhigh. Finally, the measuredvariable was asignicantpredictor of the target variable in a hierarchical multiple regression.RelationsBetweentheTargetVariablesandDemographicsCharacteristicsHappinessAlthough education level has been consistently, albeit slightly, relatedto happiness in national probability samples (cf. Diener et al., 1999), itwas not here. Likewise, the consistent relationship of marital status tohappinessorwell-being(e.g.,Mastekaasa,1994)wasnotreplicated.Most participants in our sample were married, however, precluding aSONJALYUBOMIRSKYETAL. 378meaningful analysis of this association. Happiness was also unrelatedto age, living arrangement (alone or not), and time since widowhood.Theonlydemographicvariableinthissampletoholdasignicantrelationship with happiness was gender, with females reporting slightlyhigher levels of happiness than males, t(611)=2.15; p