building positive resources: effects of positive … · building positive resources: effects of...

BUILDING POSITIVE RESOURCES: EFFECTS OF POSITIVE EVENTSAND POSITIVE REFLECTION ON WORK STRESS AND HEALTH

JOYCE E. BONOUniversity of Florida

THERESA M. GLOMBUniversity of Minnesota

WINNY SHENUniversity of South Florida

EUGENE KIMGeorgia Institute of Technology

AMANDA J. KOCHHuman Resources Research Organization

This three-week longitudinal field study with an experimental intervention examinesthe association between daily events and employee stress and health, with a specificfocus on positive events. Results suggest that both naturally occurring positive workevents and a positive reflection intervention are associated with reduced stress andimproved health, though effects vary across momentary, lagged, daily, and day-to-evening spillover analyses. Findings are consistent with theory-based predictions:positive events, negative events, and family-to-work conflict independently contribute toperceived stress, blood pressure, physical symptoms, mental health, and work detach-ment, suggesting that organizations should focus not only on reducing negative events, butalso on increasing positive events. These findings show that a brief, end-of-workdaypositive reflection led to decreased stress and improved health in the evening.

Research has established that work stress physi-cally and psychologically damages workers and ec-onomically burdens organizations and societies

(Pfeffer, 2010; Schnall, Dobson, & Rosskam, 2009).Numerous studies have addressed sources of workstress (e.g., Kamarck et al., 2002), ways to eliminatethem from work environments (e.g., Israel, Baker,Goldenhar, Heaney, & Schurman, 1996), and ways tomitigate their negative effects (e.g., Rau, Georgiades,Fredrikson, Lemne, & de Faire, 2001; Van der Doef &Maes, 1998). Overwhelmingly, this line of researchhas focused on negative aspects of work (e.g., longhours, time pressure, role ambiguity; see Crawford,LePine, & Rich, 2010), with positive aspects of workplaying primarily a buffering role. Moreover, researchhas tended to treat positive work resources (e.g., au-tonomy, support) as relatively stable characteristics ofan environment, despite the fact that theory suggestsa more ongoing, dynamic, and continuous depletionand replenishment of resources.

Over the past decade, a contrasting line of re-search has emerged that focuses explicitly on pos-itive events (e.g., positive psychology [Seligman &Csikszentmihalyi, 2000]; positive organizational

The first two authors contributed equally to this re-search. Data were collected when all the authors were atthe University of Minnesota. Funding was provided bythe University of Minnesota Institute for AdvancedStudy, the Carlson School of Management Small GrantsProgram, and a sabbatical supplement awarded to thefirst author by the University of Minnesota. We thankDevasheesh Bhave, Jonathan Booth, and Matthew Borne-man for helping pilot our protocol. We are grateful toJoanna Mark, Erin Spoerri, Nafisa Teague, and ChantelleShoff for assistance in data collection and to participantsof the Carlson School OB Research Group, Department ofWork and Organizations Workshop, and 2009 CREWEmotions Conference for their input and suggestions. Wealso offer a special thanks to our organizational partici-pants for their willingness to wear blood pressure mon-itors, respond to beeping devices, and take eveningphone calls, all while carrying on with already full lives.

1601

� Academy of Management Journal2013, Vol. 56, No. 6, 1601–1627.http://dx.doi.org/10.5465/amj.2011.0272

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download, or email articles for individual use only.

scholarship [Cameron, Dutton, & Quinn, 2003]) andways to capitalize on the beneficial effects of theseevents (e.g., Fredrickson & Joiner, 2002; Gable,Reis, Impett, & Asher, 2004; Ilies, Keeney, & Scott,2011). To date, there has been little integration ofthese lines of research, in part because they repre-sent fundamentally different points of view. Workstress and occupational health research has focusedon understanding how negative work events affectemployees, with the ultimate goal of reducing neg-ative events and their concomitant psychological,physical, and economic costs. In contrast, positivepsychology and positive organizational scholarshipresearch has highlighted the experience and ampli-fication of positive experiences in promotinghealth and well-being. One assumption that under-lies positive organizational scholarship has beenthat the factors that lead to stress are not the sameas those that lead to thriving (Dutton, Glynn, &Spreitzer, 2006). Another assumption has been theimportance of focusing on positive events to coun-teract the natural human tendency toward focusingon negative events (Baumeister, Bratslavsky, Finke-nauer, & Vohs, 2001; Taylor, 1991). A bias towardfocusing on negative antecedents has also been sug-gested in the management and psychology litera-tures. Seligman and Csikszentmihalyi wrote, “Itseems an unspoken (or more precisely, unwritten)and underexamined assumption of the literaturethat positive events, if and when they do occur,have relatively limited impact on individual well-being and interpersonal life” (2000: 10). Spreitzer,Sutcliffe, Dutton, Sonenshein, and Grant suggestedthat much more is known about combating “diseaseand infirmity” than about supporting “positivehealth, wellness, and positive functioning” at work(2005: 537).

Despite their differences in perspective, modelsof work stress and positive psychology theoriesshare similarities; in both, resources are viewed asessential to optimal human functioning. Our goal isto integrate and extend theory from positive psy-chology into literature on work stress and occupa-tional health (broaden and build theory [Fredrick-son, 1998, 2001], the conservation of resourcesmodel [Hobfoll, 1989], and the job demands-resources model [Demerouti, Bakker, Nachreiner, &Schaufeli, 2001]). We do this by first examiningdynamic variation in a broad variety of workevents, including those that build resources (posi-tive events), those that deplete resources (negativeevents), and those that may deplete resources inone domain while building them in another (family-

to-work conflict). Second, we examine the effects ofan assigned positive reflection intervention de-signed to help workers capitalize on positiveevents. Third, following the conservation of re-sources and job demands-resources traditions, weexamine the effects of work events and the inter-vention on stress, but we also expand the scope ofour study to include other outcomes relevant toemployees, including blood pressure, health com-plaints, and work detachment, a measure of thriv-ing derived from the positive psychology tradition.Work detachment is defined as an “individual’ssense of being away from the work situation”(Etzion, Eden, & Lapidot, 1998: 579), and it repre-sents a psychological detachment, in that employ-ees are able to switch off thoughts of work whenthey leave (Sonnentag & Bayer, 2005).

We contribute to theory by providing a morecomplete test of conservation of resources theory(via dynamic examination of a wide variety ofresource-building and -depleting events), by testinga potential boundary condition of popular positivepsychology interventions, and by extending therange of outcomes affected by these interventions.Our study recognizes the interplay of positive ex-periences, stress, and health as a dynamic processincluding both immediate reactions to work eventsand cumulative effects that build over the course ofa workday and spill over into evening to bettercapture the depleting/restorative properties ofevents and resources. Following Thoresen, Kaplan,Barsky, Warren, and de Chermont (2003), we alsochallenge the widely held notion in psychologythat positively and negatively valenced constructshave different nomological networks, whereby pos-itive outcomes (e.g., job satisfaction, organizationalcommitment) are associated with positive events,traits, and mood states, and negative outcomes(e.g., stress, poor health) are associated with nega-tive events, traits, and mood states. We also seek todetermine whether assigning an average worker toan intervention designed to help her/him capitalizeon positive events will have the same beneficialeffects as such interventions do when they are vol-untarily chosen (often by mildly depressed individ-uals; e.g., Seligman, Steen, Park, & Peterson, 2005),or when they explicitly involve sharing positiveexperiences with others (e.g., Ilies et al., 2011).Moreover, we contribute to the practice of manage-ment by examining the health benefits of smallwork events that can be initiated by managers (e.g.,positive feedback) as well as more formal exercises(e.g., positive reflection). Understanding the rela-

1602 DecemberAcademy of Management Journal

tionship between positive work events and healthis valuable to employers, given the financial impli-cations associated with poor employee health (e.g.,health insurance costs, illness-related productivitylosses).

We tested our hypotheses in a sample of healthcare workers who wore ambulatory blood pressuremonitors at work and at home for three weeks andreported four times daily on work events and stressas they occurred. We also interviewed them nightlyabout evening stress, work detachment, and health.At the study midpoint, we implemented a positivereflection intervention at the end of each workday,which we link to stress, health, and work detach-ment in the evening.

THEORY AND HYPOTHESES

Hobfoll’s (1989) conservation of resources modelmade an important theoretical contribution to un-derstanding of stress by focusing explicitly on therole of resources, including positive resources. Thecentral notion of the model is that humans strive toprotect and enhance the self through the acquisi-tion and maintenance of resources. When environ-mental conditions deplete or threaten resources,people suffer increased stress, and when environ-mental conditions provide or build resources, peo-ple enjoy better health and reduced stress. Theconservation of resources model recognizes the im-portance of stable characteristics of individuals(e.g., personal characteristics) and work environ-ment (e.g., job characteristics) but also explicitlyposits ongoing depletion and replenishment of re-sources over time. Yet most empirical studies treatresources as stable characteristics of individuals ortheir environments. For example, Neveu (2007)used one-time assessments of typical levels of workresources, such as skill utilization and coworkersupport. In keeping with conservation of resourcestheory, we treat stress as a dynamic function ofevents and experiences. This notion is also consis-tent with the propositions of affective events theory(Weiss & Cropanzano, 1996), which posits workevents as the proximal causes of employees’ affec-tive experiences. We suggest that, in addition af-fecting mood, day-to-day workplace events andexperiences serve as mechanisms for depleting re-sources and fostering their replenishment.

An important issue to consider when examiningthe effects of positive (and negative) events is theextent to which they have immediate effects orbuild over time. For example, in the stress litera-

ture, it has been clearly established that stress re-actions to events are immediate (e.g., Dickerson &Kemeny, 2004) but can also build and have a last-ing effect (Block & Zautra, 1981). In the healthliterature, the effects of events, especially on bloodpressure, are more complicated. It has been wellestablished that both negative and positive eventscan create immediate spikes in blood pressure,which seems more sensitive in a given moment toarousal than to event valence (see Schwartz, Pick-ering, & Landsbergis, 1996). However, in the longrun, evidence suggests that positive events may beassociated with lower blood pressure (Uchino,2006). For this reason, we consider the effects ofpositive workplace events on health and well-beingoutcomes in the moment, throughout workdays,and as they spill over into evening.

Positive Workplace Events

Meta-analytic work by Thoresen and colleagues(2003) challenged the notion of affect symmetry,whereby positive traits and moods predict positiveoutcomes and negative traits and moods predictnegative outcomes. For positive attitudes, such asjob satisfaction, they showed that both positive andnegative affect contribute equally, but for negativeworkplace outcomes (e.g., burnout), the resultswere less clear. A usefulness analysis showed pri-marily symmetrical links (negative traits and moodstates predicted negative outcomes), but someasymmetrical effects (positive traits and moodstates predicted negative outcomes) were found,even though studies linking positive affect withnegative outcomes are uncommon.

Much of the research on positive events and ex-periences has emerged from positive psychologyand its foundational broaden and build theory(Fredrickson, 1998, 2001), which explicitly ad-dresses positive experiences and emotions as theyenhance health and flourishing; this research hasfollowed the tradition of affect symmetry (positiveevents ¡ positive emotions ¡ positive outcomes).Taking an evolutionary approach, Fredrickson ar-gued that in contrast to negative emotions, whichnarrow a person’s thought and action repertoires inpreparation for quick action (e.g., to fight off animminent threat), positive emotions broaden peo-ple’s thought and action repertoires, allowing themto consider more expansive ideas, actions, or solu-tions. Positive emotions are associated with play,creativity (cf. George & Zhou, 2002), exploration,and the ability to absorb new information; they

2013 1603Bono, Glomb, Shen, Kim, and Koch

inspire people to savor and share experiences. In aline of reasoning congruent with the conservationof resources model, Fredrickson suggested that pos-itive emotions provide immediate and enduringbenefits because they create physical, intellectual,and psychological resources that individuals canuse on the spot or hold in reserve to manage futurethreats. Application of broaden and build theory tothe work setting suggests that when employees ex-perience positive events at work, they will generatenew ideas, additional resources, better social rela-tionships, and improved problem solving via rec-ognition of a wider scope of possible solutions,thereby reducing stress. The links between workevents, emotions, and mood states proposed bybroaden and build theory are consistent with re-sults from a number of studies testing affectiveevents theory. For example, when employees expe-rience positive work, supervisor, and coworkerevents, they report more positive momentary mood(Miner, Glomb, & Hulin, 2005), and on days whenemployees receive more than they expected (i.e.,promises are exceeded), they experience more pos-itive emotions (Conway & Briner, 2002). Althoughthe positive psychology tradition tends to focus onpositive events and positive outcomes (but seeSeligman et al. [2005] for an exception), our linkageof positive psychology perspectives to the conser-vation of resource theory leads us to expect thatpositive work events, via their resource-buildingcapacity, will also affect negative work outcomes,such as stress and health complaints.

Positive events may directly build psychologicalresources by fulfilling basic human needs, includ-ing belongingness and autonomy (Baumeister &Leary, 1995; Ryan & Deci, 2000). In an effort todefine positive human health as something morethan the absence of illness or disease, Ryff andSinger (1998) identified four core, universal featuresof well-being: mastery, a purposeful life, quality in-terpersonal connections, and positive self-regard. Ac-cording to the conservation of resources model, work-place events that build or enhance these featuresshould enhance core resources and combat stress. Forexample, accomplishing a task might enhance a senseof mastery, and fun interactions with coworkersmight highlight social resources.

Empirical research has suggested links betweenwork events and core resources. Research testingthe job demands-resources model in flight atten-dants (Xanthopoulou, Bakker, Heuven, Demerouti,& Schaufeli, 2008) linked daily fluctuations incoworker support to daily fluctuations in self-

efficacy, which affected achievement of work-related goals. Spreitzer et al. (2005) suggested thatworking in a climate of trust and respect can buildrelational resources via more positive day-to-dayinteractions, which in turn lead to behaviors thatsupport self-determination and autonomy. In anexperimental study, Grant and Gino (2010) foundthat positive events (e.g., helpers being thanked fortheir efforts) increased self-efficacy and social self-worth. Research outside the work context also hasshown that positive events can build core resources.Results of a series of experience sampling studiessuggested that daily positive events are associatedwith increased self-esteem and perceived control(Nezlek & Plesko, 2001; Reis & Gable, 2003).

Emerging research also suggests that positive workevents may have salutary effects on immune func-tioning (Heaphy & Dutton, 2008). Research directlylinking positive experiences and health is far lesscommon than research linking negative experiencesand health, but human and animal studies are sug-gestive. For example, Panksepp (1993) found thatplay in animals increased dopamine use in theirbrains, and numerous studies (see Ryff and Singer[1998] for a review) have linked social interactions toreduced cortisol levels; both cortisol and dopamineare implicated in healthy immune functioning andstress. Additionally, a set of empirical studies follow-ing employees over several days showed that positiveevents lowered heart rates during work hours andevenings (Evans & Steptoe, 2001; Rau et al., 2001).

Hypothesis 1. Positive workplace events are neg-atively associated with (a) stress, (b) blood pres-sure, and (c) health complaints and positivelyassociated with (d) work detachment.

Positive Reflection Intervention

The positive psychology literature includes in-tervention techniques that reduce physical com-plaints, increase life satisfaction, and decrease de-pressive symptoms (Emmons & McCullough, 2003;Seligman et al., 2005; see Lyubomirsky [2008] for areview). These techniques include interventionssuch as meditating (Fredrickson, Cohn, Coffey, Pek,& Finkel, 2008), helping others (Steger, Kashdan, &Oishi, 2008), focusing on blessings (Emmons & Mc-Cullough, 2003) or positive events (Seligman,Rashid, & Parks, 2006), and others (Lyubomirsky,2008). Many of these interventions have not beenexamined in work settings. More importantly, theyare often chosen by those seeking improvement in


well-being (Seligman et al., 2005, 2006). We chosean intervention that promotes reflection on positiveevents by asking study participants to write aboutthree good things that happened during their day(Seligman et al., 2005, 2006). The “three-good-things intervention” is designed to improve well-being by causing people to reflect on and “savor”positive events (Bryant, 1989). One unique featureof this intervention, especially as an assigned exer-cise, is that it does not rely on social elements (e.g.,capitalizing on positive events by sharing themwith others [see Gable et al., 2004; Ilies et al.,2011]). Rather, it relies on internal processeswhereby participants ruminate on positive events,counteracting their natural human tendencies toruminate on negative events.

In addition to the mechanisms noted for positiveevents, there are four additional theoretical expla-nations for the efficacy of a positive reflection in-tervention. First, as mentioned above, the interven-tion works against the human tendency to payspecial attention to negative events (Baumeister etal., 2001; Taylor, 1991). “Human beings are natu-rally biased toward remembering the negative, at-tending to the negative, and expecting the worst”(Seligman et al., 2006: 783). Positive reflection at-tempts to break this pattern by moving the focus ofattention from negative events to positive ones,leading people to experience more positive emo-tions and concomitant positive effects (Fredrick-son, 2001). Second, the three-good-things interven-tion may counteract hedonic adaptation, whichcauses people to grow accustomed to their circum-stances (Diener & Diener, 1996). Focusing explic-itly on good things may decrease the likelihood ofbecoming accustomed to and thus failing to noticeor benefit from positive events and resources (Fri-jda, 1988). Third, writing about a positive eventprovides an opportunity to relive the positiveevent, making the event and the resources associ-ated with it more accessible in memory and makingthe event more likely to be shared with others(Gable et al., 2004). Fourth, a cognitive adaptationexplanation suggests identifying the cause ofevents allows individuals to make sense of eventsand incorporate them into their understanding ofthe world and of themselves (Janoff-Bulman, 1992).Identifying and writing about why three goodthings happened may highlight available resourcesin individuals and their environments.

Empirical results for the efficacy of this interven-tion among the general population are impressive.Seligman et al. (2006) reported an uncontrolled

study in which participants’ pre- and post-testscores on a depression inventory showed a dra-matic decrease in depressive symptoms following athree-good-things exercise. Seligman et al. (2005)reported similar results in a randomized controlledstudy, with effects persisting six months later. Al-though research evidence is impressive for the ben-eficial effects of positive refocusing on stress andwell-being (Seligman et al., 2006), existing researchis limited in two important ways. First, the efficacyof the three-good-things intervention has been testedprimarily among those seeking improvements inwell-being and those who reported being mildly de-pressed at baseline. Second, most research has exam-ined the effects of the intervention on depression orhappiness. Seligman et al. noted that more behavior-based assessment is needed; they “welcome the daywhen objective productivity and health measuressupplement subjective happiness measures” (2005:420). We build on existing research by linking anassigned positive reflection intervention at work toa broader variety of outcomes.

Hypothesis 2. Reflecting on positive events atthe end of a workday is negatively associatedwith (a) stress, (b) blood pressure, and (c)health complaints and positively associatedwith (d) work detachment.

Negative Workplace Events

Our primary focus is on the resource-buildingeffects of positive experiences and events, but wewould be remiss if we failed to consider the effectsof negative events. Existing research tends to linkpositive events to positive outcomes (e.g., positivecommunication and life satisfaction [Gable et al.,2004]) and negative events to negative outcomes(but see Crawford et al. [2010] for an exception).Stress, health, and well-being are expected to beaffected by the net resources resulting from re-source-building positive experiences and resource-depleting negative experiences, yet studies havechiefly examined either positive or, more often,negative experiences, rarely examining both posi-tive and negative experiences as they occur andco-occur at work (e.g., Miner et al., 2005). Thequestion of whether both positive and negativework events independently influence health has notbeen thoroughly examined, though theory suggestsboth are important. Because our interest is in theoverall effects of both resource-building andresource-depleting workplace events, the inclusion


of negative events in this research is important.But, given the fairly deep literature—theoreticaland empirical—linking negative work events tostress, health, and work detachment, we are notbreaking new ground in our examination of nega-tive events. Rather, it serves as a way for us to testthe notion that positive and negative events haveindependent effects.

Occupational health psychology has long linkednegative workplace characteristics with employeestress and health (e.g., Cohen & Wills, 1985; Hob-foll, 1998; see Van der Doef and Maes [1999] for asummary). For the most part, this research has alsotaken a resource-based approach, whereby negativeworkplace events diminish or threaten resources inseveral ways, via negative mood states, deleteriouseffects on physiological functioning, and threats tobasic needs such as belongingness and autonomy.Numerous studies have linked negative job-relatedcharacteristics such as long work hours, low con-trol at work, and high work demands to cardiovas-cular disease as well as other psychological andphysical outcomes (e.g., Ilies, Dimotakis, & De Pa-ter, 2010; Sonnentag & Bayer, 2005; Teuchmann,Totterdell, & Parker, 1999; Totterdell, Wood, &Wall, 2006). Other studies (e.g., DeLongis, Coyne,Dakof, Folkman, & Lazarus, 1982) have linked dailywork hassles (e.g., disliking work duties) to somaticsymptoms such as headaches, chest pain, and backtrouble. Research has also shown that negative workevents influence health by increasing negative moodstates (Bolger, DeLongis, Kessler, & Schilling, 1989;Evans, Johansson, & Rydstedt, 1999).

In addition to their effects on mood, negativework events may deplete resources via physiologi-cal means. In a study on immune responses result-ing from exposure to a cold virus, Cohen, Tyrrell,and Smith (1993) showed that negative events hadeffects on illness that were independent from per-ceived stress or negative emotions. Stone and col-leagues (1993) also found that among study partic-ipants infected with a virus, those with morenegative life events exhibited more symptoms,even though mood and stress did not differ forthose who did and did not develop colds. Addi-tionally, Kamarck et al. (1998) linked psychosocialdemands at work with ambulatory blood pressure,and Ilies et al. (2010) found that work overload re-sulted in higher blood pressure. In keeping with ex-isting research, we expect negative work events tonegatively affect employee health and well-being.

Hypothesis 3. Negative workplace events arepositively associated with (a) stress, (b) bloodpressure, and (c) health complaints and nega-tively associated with (d) work detachment.

Family-to-Work Conflict

Up to this point, we have focused primarily onpositive and negative events, without consider-ation of the nature of these events. However, withinthe organizational sciences, there is an entire liter-ature focused on a specific type of workplaceevent—work-family conflict. Given our aim to ex-amine the independent effects of a broad variety ofresource-building or -depleting workplace events,we incorporate work-family conflict into this re-search. A number of studies have examined thehealth consequences of work-to-family and family-to-work conflict, which are defined as interroleconflict due to mutually incompatible role pres-sures from work and family domains (Greenhaus &Beutell, 1985). Because our focus is explicitly onevents that occur at work, we narrow our attentionto family-to-work conflict or the extent to whichfamily demands interfere with employees’ work.Although most of work-family research posits con-flict between work and family as a negative stres-sor, some research suggests that work-family con-flict’s effects are not uniformly negative. Forexample, work-family enrichment is defined as“the extent to which experiences in one role im-prove the quality of life in the other role” (Green-haus & Powell, 2006: 72). Surface examination ofwork-to-family conflict may suggest resource-de-pleting effects. But more careful considerationleads us to argue that family-to-work conflict mayhave both resource-building and resource-depletingeffects, but in different domains and during differ-ent time periods. When family tasks interfere withwork, work-related resources may diminish, butfamily-related resources may grow. For example,complications related to transportation of childrenfrom school to activities may creep into the work-day and have negative effects at work, but dealingwith such issues at work may make evening timewith family more enjoyable. Thus, family-to-workconflict may both build and deplete resources. Fur-ther, family-to-work conflict may operate differ-ently in the moment and over time. Thoughts offamily at work may interfere with performance inthe moment, but such thoughts aggregated acrossthe day may have positive effects on general well-being, because family interactions can build re-


sources and are good for one’s health (Gable et al.,2004; Heaphy & Dutton, 2008).

Despite the potential for family-to-work conflictto build and deplete resources, several reviewshave suggested that it generally damages psycho-logical and physical health (Allen, Herst, Bruck, &Sutton, 2000; Greenhaus, Allen, & Spector, 2006;Kelly et al., 2008). Allen and colleagues’ meta-analytic work (2000) linked work-family conflict topsychological strain (r � .29), somatic/physicalsymptoms (r � .29), and work-related stress (r �.41). Other meta-analytic results confirm negativeassociations between work-family conflict andhealth (Mesmer-Magnus & Viswesvaran, 2005).These assessments, however, often rely on narrowhealth indicators (e.g., anxiety or alcohol use) oroverall health perceptions. Allen et al. (2000) sug-gested a need for research that examines the effectsof work-family conflict on a broader range of healthindexes, including objective measures. Currentwork-family conflict research is also limited by itspredominant use of a between-participants ap-proach, which allows for the possibility of spuriouscorrelations between work-family conflict andhealth outcomes due to something about an indi-vidual (e.g., personality) that causes both work-family conflict and its outcomes.

Because existing research has built a strong em-pirical case for the resource-depleting effects offamily-to-work conflict, we hypothesize only neg-ative effects. However, the possibility that this typeof negative event may be unique leads us to sep-arate family-to-work conflict from more generalwork-related negative events, so that we can in-dependently assess the effects of family-to-workconflict on well-being during the workday, whenits resource depleting effects should be the stron-gest, and in the evening at home, when employ-ees would be most likely to experience resource-building effects via stronger family relationships.

Hypothesis 4. Family-to-work conflict is posi-tively associated with (a) stress, (b) blood pres-sure, and (c) health complaints and negativelyassociated with (d) work detachment.

Buffering and Enhancing Effects ofPositive Events and Intervention

Conservation of resources theory and broadenand build theory both suggest that positive eventsmay cumulate over time to build enduring re-sources that increase resiliency in the face of re-

source loss (Fredrickson, 2001). Hence, our posi-tive reflection intervention and naturally occurringpositive events might not only have direct effects,but they might also serve as a buffer, moderatingthe effects of negative events and family-to-workconflict on stress and health. The empirical litera-ture, however, provides limited and inconsistentempirical support for buffering effects (e.g., Van derDoef & Maes, 1998, 1999).

Typically, the stress literature has treated positiveresources as a buffer against the effects of negativeevents, but recent research suggests that interven-tions such as ours might also amplify the beneficialeffects of naturally occurring positive workplaceexperiences. Fredrickson’s (1998, 2001) broadenand build theory suggests this possibility, as doesempirical research showing that when positiveevents are savored or shared, their effects are en-hanced (Gable et al., 2004; Ilies et al., 2011; Langs-ton, 1994). Because our positive reflection is anexercise that involves savoring but not sharing, itwas not clear whether it would magnify the effectsof positive events.

Given inconsistent support for the buffering ef-fects of positive events and limited theoretical andempirical rationale for hypothesizing enhancing ef-fects for our positive reflection exercise, we offer noformal moderation hypotheses. However, we testedfor effects of interactions between work events andour positive reflection intervention on stress andhealth, whereby the intervention buffered employ-ees from the effects of negative events and family-to-work conflict and enhanced the effects of positiveevents. We also tested whether naturally occurringpositive events buffered the effects of negative eventsand family-to-work conflict.

METHOD

Procedures

We collected data over three weeks (Monday–Friday, 15 days total) from three sources: ambula-tory blood pressure monitors, personal digitalassistant (PDA) surveys, and evening phone in-terviews. Each day, participants (a) reported workevents and stress through four PDA surveys atwork, (b) wore ambulatory blood pressure monitorsthat automatically measured their blood pressureevery 30 minutes for two hours in the morning(between two morning PDA signals), two hours inthe afternoon (between two afternoon PDA signals),and two hours in the evening (between a phone


reminder and an evening interview), and (c) an-swered questions about their current stress andhealth during the evening phone interview. At themidpoint of the three-week study, we introducedan additional task: at the end of each workday(study days 8–15), following the last PDA signal,participants performed a daily positive reflectionexercise. Figure 1 shows the elements and timing ofthe data collection.

Participants were paid $250 for any level of par-ticipation and $100 more if they completed 80 per-cent of activities. Over the three weeks, partici-pants completed 86 percent of daily PDA surveys,92 percent of daily blood pressure measurements,92 percent of evening interviews, 84 percent ofevening blood pressure measurements, and 82 per-cent of positive reflections.

Participants

Participants worked in nine outpatient familypractice clinics operated by two health care organ-izations in a large metropolitan area. All these clin-ics were quite similar in that they were traditionalphysicians’ offices providing nonemergency care.The organizations invited employees via e-mail toan information session. After providing study in-formation, we directed interested attendees to ourstudy website, where they completed a background

survey (e.g., health, work schedule). Following en-rollment, participants were trained to use theequipment.

Eighty-two employees expressed interest in ourstudy, but 12 did not complete the backgroundsurvey, 8 were scheduled to be away from workduring the data collection period, and 1 did notmeet requirements (i.e., no known hypertension,good general health). We dropped a single malefrom the study. Our final sample consisted of 61women who worked at least 32 hours a week andwhose jobs involved some direct contact with pa-tients. Job titles included nurse, receptionist,medical assistant, and lab technician. Althoughresponsibilities varied across jobs (e.g., recep-tionists helping with scheduling and billing is-sues, nurses and medical assistants helping withclinical procedures), in this clinic setting, all par-ticipants were part of the direct patient care team.The average wage was $15.93 per hour (s.d. �$2.68/hour; range $11.75–$26.50). The averageage of participants was 33.65 years (s.d. � 10.33)and most were white/Caucasian (69%; 18%,Asian/Pacific Islander; 10% black/African-American;3% Hispanic). Most (79%) had a two-year collegedegree; 18 percent had completed high school;and 3 percent had a bachelor’s degree. Averagetenure at the clinic and in the current job wasfour years.

FIGURE 1Daily Data Collection over the 15-Day Period

Blood Pressure (every

30 min.)

Evening Survey

PDA Survey

Reminder to Turnon BP

Positive Reflection

(Days 8–15)

PDA Survey

PDA Survey

PDA Survey


30 min.)


30 min.)


30 min.)

Morning (2 hours)

Afternoon (2 hours)

Evening (2 hours)


30 min.)


30 min.)


30 min.)


30 min.)


30 min.)


30 min.)

Blood Pressure

(every 30 min.)


30 min.)

Phone


Measures

Before beginning the three-week daily survey pe-riod, participants completed an online survey re-porting demographic characteristics (e.g., age, race,gender), work schedules, and trait positive affectiv-ity (PA) and negative affectivity (NA). PA and NAwere assessed with the ten PA (e.g., “enthusiastic,”“active”) and ten NA (e.g., “upset,” “distressed”)items from the Positive Affect Negative AffectSchedule (PANAS; Watson, Clark, & Tellegen,1988). Participants indicated the extent to whichthey “generally feel this way,” using a Likert-typescale (1 � “very slightly or not at all,” 5 � “verymuch”).

Daily PDA survey. PDAs were scheduled to sig-nal participants two hours after they arrived atwork and then approximately every two hoursthereafter (i.e., two signals in the morning and twosignals in the afternoon). Daily surveys took ap-proximately two minutes to complete and askedabout work events and stress.

Work events. In keeping with other event sam-pling studies (e.g., Ilies et al., 2011; Wang, Liao,Zhan, & Shi, 2011), we developed a work eventscale by drawing on existing literature (e.g., Bono &Campana, 2007; Hart, Wearing, & Headey, 1994;Oishi, Diener, Choi, Kim-Prieto, & Choi, 2007;Stone & Neale, 1982). We developed 13 items thatwere broad enough to apply to many jobs and alsoencompassed the wide array of work events foundin the literature. Items fell into three broad catego-ries: positive events, negative events, and family-to-work conflict.

Because discrete work events such as those as-sessed tend not to occur frequently during smalltime segments (e.g., two-hour periods), the use ofdichotomous items is common in event studies(Ilies et al., 2011; Kim & Yoon, 2012; Tsai & Huang,2002; Wang et al., 2011). Skarlicki, van Jaarsveld,and Walker (2008) provided evidence that—at leastfor negative events directed toward customers—most of the variance in work events can be capturedwith a dichotomous (yes/no) scale. However, be-cause we assessed a broad array of work events,including family-to-work conflict, which is notcommon in event studies, we asked participants toreport whether each work event had occurred sincethe last signal (or since arriving at work for the firstsignal of the day), using a scale (0 � “not at all,” to3 � “a lot”) that allowed us to verify the appropri-ateness of a dichotomous (yes/no) scaling for theevents we studied. As with past research, we found

nonnormal distributions for the work event items;96 percent of responses (across items) fell into tworesponse categories: “not at all” or “a little.” There-fore, we dichotomized the work event items (0 �“did not occur,” 1 � “occurred”; “a little,” “a mod-erate amount,” and “a lot” were all coded as 1). Wethen summed the number of events experienced toform scales for positive events (four items: “accom-plish what you hoped to,” “have fun and socialize,”“receive information that positively affected yourwork schedule, duties, or pay,” and “receive posi-tive feedback or praise”), negative events (fiveitems: “receive information that negatively affectedyour work schedule, duties, or pay,” “receivenegative feedback or criticism/complaints,” “gettreated disrespectfully,” “work with difficult peo-ple,” and “have work-related conflict”), and family-to-work conflict (two items: “have personal tasksinterfere with your work” and “have thoughts offamily interfere with your work”).1

We created two versions of each event scale. Thefirst represents events experienced during eachtwo-hour measurement period (momentary scales).The second represents events experienced eachday, averaged across the four PDA signals (workdayscales). There is redundancy between the momen-tary and workday scales, but both are important, asmomentary scales allowed us to assess immediateresponses to workplace events, and workday scalesallowed us to assess the effects of these events asthey cumulate over the course of a workday, whichis especially important for our spillover analyses.Coefficient alpha is not an appropriate measure ofreliability for these scales because they include dis-tinct events that may not co-occur and are notmeant to represent a latent psychological construct(e.g., Liu,Wang, Zhan, & Shi, 2009; Wang et al.,2011). In Figure 2, we present the average fre-quency of the 11 events.

1 Two items were dropped after data collection. Wedropped “regret” from the negative events scale becauseit occurred in less than 1 percent of our responses. Re-sults were the same with the inclusion or exclusion ofthis item. We also dropped a family interference itemthat had ambiguous wording (“talk or email with familyand friends”), because we did not ask if the activityinterfered with work; it is possible that this activity oc-curred during a break. Also, talking with family might bepositive (sharing a positive work event with family) ornegative (getting a call from a family member whoneeds help).


Stress. Participants indicated whether their dayhad been stressful since the last signal or since thestart of their workday (1 � “strongly disagree,” 5 �“strongly agree”).

Systolic blood pressure. Participants woreA&D™ TM-2430 ambulatory blood pressure moni-tors for two hours in the morning, afternoon, andevening. Each monitor included a small computerworn at the waist, with a tube running under orover their clothing to a cuff worn on the upper arm.The computer was programmed to measure bloodpressure every 30 minutes, on the hour and halfhour. Participants activated their monitor on thefirst morning and first afternoon PDA signal anddeactivated it on the second morning and secondafternoon signal. In the evening, two hours beforetheir scheduled phone interview, participants re-ceived an automated telephone reminder to acti-vate their monitor, which they removed after theevening interview. Although the monitors recordboth systolic and diastolic pressure, systolic pres-sure is most responsive to work events (Theorell,Knox, Svensson, & Waller, 1985), is the best pre-dictor of cardiovascular disease (Benetos, Thomas,Bean, Gautier, Smulyan, & Guize, 2002), and hasbeen the focus of work-related blood pressure re-search (Ilies et al., 2010; Kamarck et al., 2002).

Evening interview. Participants responded to afive-minute structured phone interview, at a timearranged the prior evening and as close to theirbedtime as possible. We asked questions abouthealth complaints, stress, and work detachment.

Evening health complaints. We drew interviewquestions about health from Goldberg (1972). Weasked about both physical and mental health com-plaints, prefacing all items with “Since you leftwork today, to what extent did you experience thefollowing?” “Upset stomach,” “neck or back pain,”“headaches,” “painful or tense muscles” comprisedthe physical scale, and “felt tired or fatigued,” “dif-ficulty concentrating,” and “difficulty making de-cisions” comprised the mental scale. Participantsanswered interviewer questions using a five-pointscale (1 � “not at all,” 2 � “slight,” 3 � “moderate,”4 � “a great deal,” 5 � “severely”). Items wereaveraged to form scales for physical and mentalcomplaints (� � .72 and .63, respectively).

Evening stress. We used a single interview ques-tion for stress. Participants reported the level ofstress they had experienced since leaving work thatday, using the same responses as for the healthcomplaint items.

Evening work detachment. Drawing from theliterature (e.g., Sonnentag & Fritz, 2007), we as-

FIGURE 2Frequency of Events in a Day


sessed work detachment in three ways. First, weasked participants to agree or disagree with thestatement, “I was able to forget about my work thisevening” (1 � “strongly disagree,” 2 � “disagree,”3 � “neither agree nor disagree,” 4 � “agree,” or5 � “strongly agree”; mean � 3.87, s.d. � .72). Laterin the interview, we asked, “To what extent did youexperience difficulty switching off your mind afterwork?” using the same response options as forhealth complaints (mean � 1.26, s.d. � .61). Fi-nally, we asked them the number of hours or min-utes they had spent “thinking about work, prepar-ing for work, or doing work activities” that evening(range � 0–3 hours, mean � 8.5 minutes, s.d. �22.10). The second and third items were reverse-scored, then items were standardized and averagedto form a work detachment scale (� � .74).


On study days 8 through 15, participants loggedinto our website at a work computer after theycompleted their final PDA survey of the day. Par-ticipants recorded three good things that had hap-pened that day (personal or work-related) andexplained why they thought these events had oc-curred (see the Appendix), an intervention basedon those used in prior research (Seligman et al.,2005, 2006). On study day 7, we visited each par-ticipant’s clinic with instructions for accessing thesurvey. To determine whether participants took ourintervention seriously, we examined the goodthings they recorded. Three of the authors indepen-dently coded participants’ responses, assigning 1 ifa participant did not appear to take the interventionseriously (e.g., [good thing] “It’s Friday”; [why] “Itcomes after Thursday”) or 2 if they completed theintervention in a thoughtful, reflective manner(e.g., [good thing] “A doctor gave me a complimenttoday” [why] “Because I knew exactly what to do inan emergency situation, and I helped a patient whowas having a seizure”). Initial agreement betweenthe raters was 96 percent; the research team dis-cussed and resolved 20 disputed ratings. Becauseonly 10 incidents appeared to lack some effort, alldata were included in our analyses.

RESULTS

Data Analysis

We used HLM 6 (Raudenbush, Bryk, Cheong, &Congdon, 2004) with group mean centering. In

group-mean-centered analyses, a positive relation-ship between negative work events and stressmeans that on days when participants experiencedmore negative work events than was typical, theyalso experienced elevated stress levels. We focusedon within-person effects because they remove theeffects of individual characteristics (e.g., “trait”positive or negative affectivity), allowing us to di-rectly examine the association between events andoutcomes.2

We modeled the effects of work events occurringover three time periods/levels, to examine whetherthe association between work events and healthbuilt or diminished over time. For the momentarylevel, work events and outcomes were assessedduring the same time period (e.g., during a morningat work). For the lagged momentary level, welinked work events from the morning to health inthe afternoon, controlling for events in the after-noon and health in the morning. For the workdaylevel, work events and outcomes were aggregatedover the course of an entire workday. For the eve-ning, or spillover analyses, work events during anentire workday were linked to outcomes that eve-ning. For completeness, Table 1 presents associa-tions between variables both within- and between-persons, but these correlations do not appropriatelymodel the nested nature of our data.

Partitioning of Variance

Before hypothesis testing, we examined varianceat the within- and between-individual levels forour level 1 variables. At the momentary level, 33percent of the variance in stress and 51 percent ofthe variance in systolic blood pressure was within-individual. At the workday level, 48 percent of thevariance in stress and 65 percent of variance insystolic blood pressure was within-individual. Alarge portion of the variance for evening outcomeswas within-individual: evening stress (40%), sys-tolic blood pressure (37%), physical complaints(49%), mental health complaints (60%), and workdetachment (34%). Results suggested that multi-level modeling was appropriate. We also examined

2 Within-person centering effectively controls for theeffects of all individual difference variables, as each per-son’s data are centered around her own mean. Nonethe-less, because there is a debate in the literature aboutwhether PA and NA should be controlled for in stressand health research, we reran all our analyses controllingfor trait PA and NA. Results were unchanged.


the effects of clinic, organization, and weekday onthe outcomes and found them to be nonsignificant;therefore, we did not model them in subsequentanalyses.

Work Events, Health, and Stress

Our first set of analyses examined the associationbetween work events, stress, and health, using onlydata from study days 1 through 7, as our goal was toexamine the association between work events onstress and health as a baseline, prior to our positivereflection intervention.

Momentary results. We present the results of ourmomentary analyses, in which we measured workevents and outcomes concurrently, in columns twoand three of Table 2. Because our goal was to ex-amine the incremental effects of each type of workevent, all three types of events were entered intothe model simultaneously. Results supported anegative association between positive events atwork and stress (Hypothesis 1a: � � �.14, p � .01);when a person experienced more positive eventsthan normal, she also reported less stress. A similarpattern of findings emerged for negative events, butin the opposite direction; negative events were as-

sociated with increased stress (Hypothesis 3a: � �.29, p � .01). When family interfered with work,participants also experienced increased stress (Hy-pothesis 4a: � � .16, p � .05). We found no signif-icant associations between work events and bloodpressure in our momentary analysis.

Lagged momentary results. Because the mo-mentary results involved analyses in which events,stress, and blood pressure were measured concur-rently, it is possible that the results we report couldbe spurious because of the effects of another factorpresent during that two-hour time period, such astransient mood, which might affect reactions to(and reporting of) work events and perceptions ofstress. For this reason, we conducted lagged mo-mentary analyses in which we examined the asso-ciation between work events in the morning andstress and blood pressure in the afternoon, control-ling for both morning stress levels and subsequent(afternoon) work events. Our results in columnsfour and five in Table 2 reveal a significant,negative association between morning positiveevents and stress (� � �.10, p � .05); when aperson experienced more positive events than nor-mal in the morning, she also reported less stress

TABLE 1Means, Standard Deviations, and Within- and Between-Person Correlationsa

Variable Mean s.d. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Momentary1. Positive events 1.89 0.55 .00 .03 �.10 �.012. Negative events 0.89 0.79 .02 .32 .61 .193. Family conflict 0.41 0.41 .11 .01 .33 .234. Stress 3.10 0.67 �.14 .28 .07 .175. Blood pressure 124 10 .05 .05 .05 .14

Workday6. Positive events 7.40 2.20 �.01 .06 �.05 �.04 �.02 �.17 �.09 �.11 .08 .10 .257. Negative events 3.42 2.94 .00 .32 .59 .14 .35 .08 .32 .31 �.37 �.02 .418. Family conflict 1.66 1.66 .09 �.03 .40 .23 .28 �.06 .37 .33 �.20 �.14 .249. Stress 3.04 0.64 �.20 .26 .12 .17 .54 .04 .46 .42 �.26 �.06 .28

10. Blood pressure 123 10 .00 .05 .03 .11 .14 .77 .10 .22 �.02 .09 �.21Evening11. Stress 1.66 0.61 �.06 .12 .06 .24 .00 �.05 .78 .78 �.68 .08 .2512. Blood pressure 125 9.42 �.05 �.01 .02 .00 .12 .10 �.20 �.03 .19 .08 �.1813. Mental complaints 1.48 0.46 �.03 .05 .06 .10 �.02 .29 .02 .82 �.68 �.05 .2214. Physical complaints 1.38 0.44 �.03 .12 �.02 .09 �.03 .19 .07 .27 �.67 .00 .2115. Work detachment 0.00 0.50 .08 �.17 .02 �.19 �.07 �.22 �.11 �.13 �.15 .04 �.39Person-level16. Positive affect 3.68 0.63 �.2317. Negative affect 1.68 0.41

a The upper-left block represents momentary correlations; the lower-right block represents workday and spillover correlations. Withineach block, within-person correlations (computed for each person and averaged across people [Miner et al., 2005]) are below the diagonal,and between-person correlations are above the diagonal. Within-person n’s range from 14 through 30 for momentary events and from 4through 15 for workday and spillover. n � 61 for between-person analyses.


TA

BL

E2

Res

ult

sof

Mom

enta

ryan

dW

ork

day

Reg

ress

ion

An

alys

esfo

rth

eE

ffec

tsof

Wor

kE

ven

tson

Str

ess

and

Hea

lth

,S

tud

yD

ays

1–7a

Var

iabl

es

Mom

enta

ryA

nal

yses

Lag

ged

Mom

enta

ryA

nal

yses

Wor

kd

ayA

nal

yses

Str

ess

Sys

toli

cB

lood

Pre

ssu

reS

tres

sS

ysto

lic

Blo

odP

ress

ure

Str

ess

Sys

toli

cB

lood

Pre

ssu

re

Inte

rcep

t3.

15**

(.07

,43

.14)

123.

31**

(1.3

5,91

.32)

3.19

**(.

10,

32.3

5)12

3.78

**(1

.44,

86.2

0)3.

07**

(.07

,45

.69)

123.

27**

(1.3

2,93

.12)

Pos

itiv

eev

ents

�.1

4**

(.06

,�

2.48

).3

3(.

50,

0.66

)�

.10*

(.06

,�

1.87

).8

9(.

73,

1.21

)�

.06*

*(.

02,

�3.

68)

.23

(.19

,1.

16)

Neg

ativ

eev

ents

.29*

*(.

04,

6.47

).5

2(.

37,

1.40

)�

.03

(.06

,�

0.51

).2

9(.

46,

0.62

).0

9**

(.01

,6.

32)

.35*

(.21

,1.

66)

Fam

ily-

to-w

ork

con

flic

t.1

6*(.

07,

2.35

).8

2(.

57,

1.43

).1

2(.

10,

1.18

).2

2(1

.09,

0.20

).0

8**

(.02

,4.

64)

.57*

(.30

,1.

88)

aV

alu

esar

eu

nst

and

ard

ized

HL

Mco

effi

cien

ts(s

.e.,

t).

For

mom

enta

ryan

alys

es,

vari

able

sw

ere

mea

sure

dco

ncu

rren

tly

(n�

738–

741

obse

rvat

ion

s,d

f�

734–

737)

.M

omen

tary

lagg

edan

alys

esp

red

ict

afte

rnoo

nou

tcom

esfr

omm

orn

ing

even

ts(n

�35

1–35

3ob

serv

atio

ns,

df

�34

3–34

5),w

ith

mor

nin

gst

ress

and

afte

rnoo

nev

ents

con

trol

led

for.

For

wor

kday

anal

yses

,va

riab

les

wer

eag

greg

ated

acro

ssa

sin

gle

wor

kday

(n�

332

obse

rvat

ion

s;d

f�

328)

.*

p�

.05

**p

�.0

1

in the afternoon. There were no lagged effects fornegative events, family-to-work conflict, or bloodpressure.

Workday results. In the last two columns of Ta-ble 2, we present the results of our workday anal-yses, which examined the association betweenwork events aggregated over an entire day and av-erage stress and blood pressure for that workday.The pattern of associations between work eventsand stress at the workday level was similar to thatat the momentary level. Workers were significantlyless stressed on the days they experienced morepositive events throughout the workday (� � �.06,p � .01). Negative events were significantly andpositively associated with stress (� � .09, p � .01),as was family interference with work (� � .08,p � .01).

In the workday analysis, we also found signifi-cant associations between negative work eventsand blood pressure (� � .35, p � .05) and family-to-work conflict and blood pressure (� � .57, p �.05). On days when participants experienced morenegative events than normal, and on days whentheir family interfered with work more than nor-mal, their blood pressure was slightly and signifi-cantly higher. Positive events were not signifi-cantly associated with blood pressure.

Spillover results. Next, we examined the linkbetween workday events and evening health andwell-being. Table 3 presents results. Comparing Ta-bles 2 and 3 clearly shows that work events areassociated with stress and health differently overtime. Although positive events were not associatedwith lower blood pressure during workdays (per-haps because positive events can be exciting andraise blood pressure in the moment [Pressman &Cohen, 2005]), we found a significant spillover ef-

fect. On days when participants had more thantheir normal number of positive events, blood pres-sure was lower in the evening (� � �.55, p � .05).Positive workplace events were also associatedwith greater work detachment in the evening (� �.03, p � .05).

Examining the spillover effects of negativeevents, we found that experiencing more negativeevents during a workday was associated with in-creased evening stress (� � .06, p � .01) and re-duced work detachment (� � �.06, p � .01). Nospillover effects were found for family-to-workconflict.


Main effects. One of the central purposes of ourstudy was to determine whether an assigned (notvoluntarily chosen), brief, positive reflection per-formed at the end of the workday would improveemployee stress and health (Hypothesis 2). We con-ducted all tests of the intervention effects within-persons, with participants serving as their owncontrols. That is, we compared evening stress andhealth before and after the positive reflection inter-vention. We examined intervention effects in twoways. First, we compared evening health and stressin the preintervention period (study days 1–7) tohealth and well-being during the intervention(study days 8–15). However, some participantsmissed a day of the study during the interventionperiod, so we also compared health and well-beingon days when participants completed the positivereflection intervention and days when they did not,regardless of where they were in the study (noparticipants completed the intervention before day8, but not all completed the intervention on each of

TABLE 3Results of Spillover Regression Analyses for the Effects of Work Events on Evening Stress and Health, Study Days 1–7a

Variables StressSystolic

Blood PressurePhysical

ComplaintsMental

ComplaintsWork

Detachment

Intercept 1.76** (.08, 21.67) 124.99** (1.32, 94.36) 1.42** (.05, 27.57) 1.52** (.05, 32.01) �.03 (.07, �0.40)Intrapersonal effects

Positive events �.02 (.03, �0.70) �.55* (.26, �2.11) .00 (.01, 0.16) .02 (.01, 1.11) .03* (.02, 1.75)Negative events .06** (.02, �2.25) .12 (.27, 0.45) .02 (.01, 1.25) .00 (.01, 0.07) �.06** (.02, �3.04)Family-to-work

conflict�.01 (.04, �0.27) .67 (.44, 1.50) .01 (.02, 0.72) .02 (.02, 1.23) .01 (.04, 0.14)

a Values are unstandardized HLM coefficients (s.e., t). n � 327–329 observations. Spillover analyses link events aggregated acrossworkday with evening outcomes. df � 323–325.

* p � .05** p � .01


days from 8 through 15). Our results in Table 4clearly demonstrate the efficacy of a positive reflec-tion intervention. Comparing the preinterventionperiod with the intervention period, we found thatparticipants experienced reduced stress (� � �.16,p � .01) and reported fewer health complaints(physical: � � �.07, p � .01; mental: � � �.07, p �.05). Comparing outcomes on days when partici-pants did not complete the intervention with thoseon days when they did complete it, we also foundsignificant effects for evening stress (� � �.07, p �.01) and health complaints (physical: � � �.11;mental: � � �.04, p � .01). Additionally, partici-pants experienced greater detachment from work(� � .06, p � .05) on days when they completed theintervention. Effect sizes were generally small(physical complaints: d � �.16; mental com-plaints: d � �.15; inability to switch off work: d ��.09; stress: d � �.23). The only outcome ourintervention failed to affect significantly (p � .05)was evening blood pressure. In Figure 3, we depictresults comparing mean levels of each outcome forall participants before and during the interventionperiod.3 Overall, the results shown in Table 4 andFigure 3 suggest that our intervention meaningfullyaffected participants’ stress, health, and well-being,providing support for Hypothesis 2.

Buffering effects. In addition to examining themain effects of our intervention, we were interestedin whether the positive reflection or naturally oc-curring positive events might buffer the effects ofnegative events and family-to-work conflict andwhether the positive reflection intervention might

help employee capitalize on the beneficial effectsof positive events.

Intervention. For positive and negative events,we found no enhancing or buffering effects of theintervention; Table 5 shows results. Although ourintervention had main effects (reducing healthcomplaints and stress), it was not helpful in pro-tecting employees from the effects of negative workevents, nor was it helpful in amplifying the benefitsfrom positive events employees experienced. Whenpeople experienced negative events, they felt morestressed, suffered more health complaints, and haddifficulty detaching from work in the evening, andthese effects were not mitigated when they re-flected on good things in their lives.

We did find buffering effects for family-to-workconflict. Results in Table 5, which are depicted inFigure 4, suggest that the positive reflection in-tervention reduced the negative effects of family-to-work conflict on blood pressure and mentalcomplaints. The main effect of family-to-workconflict on blood pressure was not significant(i.e., simple slope � .67, n.s.; see Table 3). None-theless, our moderation test indicated a signifi-cant reduction in the effects of family-to-workconflict on blood pressure during the interven-tion period (� � �1.56, p � .05). The same pat-tern of effects was found for health complaints;the positive reflection exercise reduced the neg-ative effects of family-to-work conflict on eve-ning mental health (� � �.07, p � .05).

To better understand why we found bufferingeffects for the intervention on family-to-work con-flict and not for other types of negative events, weconducted a post hoc examination of the content ofparticipants’ positive reflections. Specifically, wecoded for whether the positive reflections refer-enced family members or family events. For allparticipants and days (n � 397), participants refer-

3 Work detachment is standardized (mean � 0, s.d. �1), and the average value before the intervention was lessthan zero, and the average value after the interventionwas greater than zero, indicating an increase in workdetachment during the intervention period.

TABLE 4Effects of the Positive Reflection Intervention on Stress and Healtha

VariablesEveningStress

Evening SystolicBlood Pressure

PhysicalComplaints

MentalComplaints

WorkDetachment

Intervention period �.16** (.08, 21.60) 0.99 (.81, 1.23) �.07** (.03, 2.46) �.07* (.03, �2.26) .06 (.07, 0.89)Intervention completed �.07** (.03, �2.43) .36 (.42, 0.86) �.11** (.02, �4.61) �.04** (.01, �2.75) .06* (.03, 2.22)

a Values are unstandardized HLM coefficients (s.e., t). n � 61 participants; 773–838 observations, df � 60, 771–836. “Interventionperiod” is the comparison of study days 1–7 (preintervention) to days 8–15 (intervention period). “Intervention completed” is thecomparison of days when individuals actually completed the intervention to days when they did not.

* p � .05** p � .01


enced family in almost half of their daily reflectionexercises (44%). Only five individuals (8% of sam-ple) failed to reference family in their positive re-flections at some point. Thus, it appears that manyparticipants used the positive reflection to reflecton family members, family events, and family is-sues in their lives, which possibly explains why wefound buffering effects of the intervention for fam-ily-to-work conflict but not for negative eventsmore generally. As an additional post hoc analysis,we created a variable for each participant that in-dicated the percentage of her entries that includedfamily content. For example, if a person wrote threegood things a day for eight days and talked aboutfamily only five times, the score was .21 (5/24); fora person who talked about family in two entriesevery day for eight days, the score was .67 (16/24).A cross-level analysis tested whether the bufferingeffects of the intervention for family-to-work con-flict were stronger for participants who wrote aboutfamily more; results of this analysis were not sta-tistically significant (p � .05).

Positive workplace events. We also testedwhether positive workplace events foster the con-

servation or building of resources by buffering em-ployees from the effects of negative events andfamily-to-work conflict. In contrast to our interven-tion results, positive events were found to have nobuffering effects on family-to-work conflict (Ta-ble 5). However, we did find significant bufferingeffects for negative events; specifically, the associ-ation between negative events and stress (� � �.01,p � .05) and physical health (� � �.004, p � .05)was smaller on days when employees experiencedmore positive events (Figure 4).

DISCUSSION AND CONCLUSION

Studs Terkel opened his classic book Workingwith these lines: “This book, being about work, is,by its very nature about violence—to the spirit aswell as to the body. It is about ulcers �as wellas accidents, about shouting matches as well asfistfights, about nervous breakdowns as well askicking the dog around. It is, above all (or beneathall), about daily humiliations” (1972: xiii). Thispunishing view of work serves as the basis formuch existing research on employee stress, but our

FIGURE 3Effects of Positive Intervention on Evening Stress and Health

No = days on which participants did not complete the positive reflection intervention

Panel 1

Panel 2

0

0.5

1

1.5

2

Mean Levels of Stress, Mental Complaints, and Physical Complaints

Work Detachmentas Standardized Score

(deviation from themean of 0)

No

Yes

−0.1

−0.05

0

0.05

0.1

No

Yes

Yes = days on which participants completed the positive reflection intervention

Stress Mental Complaints

Physical Complaints

Work Detachment


TA

BL

E5

En

han

cin

gan

dB

uff

erin

gE

ffec

tsof

Pos

itiv

eE

ven

tsan

dR

efle

ctio

na

Var

iabl

esE

ven

ing

Str

ess

Eve

nin

gS

ysto

lic

Blo

odP

ress

ure

Ph

ysic

alC

omp

lain

tsM

enta

lC

omp

lain

tsW

ork

Det

ach

men

t

En

han

cin

gef

fect

sP

osit

ive

even

ts�

inte

rven

tion

.01

(.06

,�

0.19

).7

5(.

90,

�0.

83)

�.0

1(.

01,

�0.

87)

�.0

4(.

03,

�1.

14)

�.0

4(.

04,

�0.

94)

Bu

ffer

ing

effe

cts:

Inte

rven

tion

Neg

ativ

eev

ents

�in

terv

enti

on�

.08

(.07

,�

1.12

).2

4(.

65,

�0.

37)

.01

(.01

,�

0.63

).0

2(.

03,

�0.

52)

�.0

0(.

07,

�0.

00)

Fam

ily-

to-w

ork

even

ts�

inte

rven

tion

�.0

2(.

07,

�0.

23)

�1.

56*

(.68

,�

2.31

)�

.00

(.01

,�

0.11

)�

.07*

(.03

,�

2.31

)�

.02

(.02

,�

0.04

)B

uff

erin

gef

fect

s:P

osit

ive

even

tsN

egat

ive

even

ts�

pos

itiv

eev

ents

�.0

1*(.

00,

�1.

87)

�.0

3(.

05,

�0.

56)

�.0

04**

(.00

,�

2.55

)�

.00

(.00

,�

0.32

)�

.00

(.00

,�

0.95

)F

amil

y-to

-wor

kev

ents

�p

osit

ive

even

ts.0

1(.

00,

�1.

61)

.09

(.11

,�

0.81

)�

.00

(.00

,�

0.18

).0

0(.

00,

�0.

11)

.00

(.01

,�

0.46

)

aV

alu

esar

eu

nst

and

ard

ized

HL

Mco

effi

cien

ts(s

.e.,

t).

Inte

ract

ion

sw

ere

mod

eled

sim

ult

aneo

usl

yw

ith

mai

nef

fect

s(n

otp

rese

nte

dh

ere;

resu

lts

inT

able

s3

and

4).

n�

61p

arti

cip

ants

;63

1–70

1ob

serv

atio

ns;

df

�60

,62

7–69

7.*

p�

.05

**p

�.0

1

research suggests that daily work comprises morethan pain and suffering. In support of Kanov, Mait-lis, Worline, Dutton, Frost, and Lilius, who claimedthat organizations “are also places of healing”(2004: 809), our results suggest that—even if onlyin small ways—positive daily experiences at work,such as socializing, positive feedback, and goal ac-complishment, relate directly to reduced stress andimproved health. Spreitzer et al. noted that simplyreducing unsafe working conditions or job insecu-rity “does not mean that an individual will thrive”(2005: 539). Our results converge with existing re-search in documenting links between negativework experiences, stress, and health, but they alsoclearly suggest that incorporating positive work ex-periences has salutary benefits. In Working, Terkelalso suggested the promise of positive experiences:“It [work] is about a search, too, for daily meaningas well as daily bread, for recognition as well ascash, for astonishment rather than torpor; in short,for a sort of life rather than a Monday throughFriday sort of dying” (1972: xiii).

Our research explicitly incorporates positiveevents—both naturally occurring and induced viaintervention—as a key contribution to the study of

work stress. Partly because the positive psychologyliterature emerged in reaction to psychology’s his-torical focus on negative events, the work stressliterature has failed to soundly integrate the prin-ciples and practices of positive psychology. More-over, although popular stress theories (e.g., conser-vation of resources, job demands-resources)include positive aspects of the work environment(e.g., job control), researchers have treated themmore as stable buffers than as direct and dynamicinfluencers of work stress. Few researchers haveexamined the direct effects of discrete positivework events as they affect stress, blood pressure, orpsychosomatic symptoms. Given the history of in-consistent findings when researchers treat positiveevents as buffers, our research makes an importantcontribution by showing that stress is reduced bothwhen employees experience fewer negative eventsand when they have more positive experiences.Naturally occurring positive work events were as-sociated (negatively) with perceived stress, in themoment and over a workday. Moreover, our laggedanalyses are consistent with a causal interpretationof the results. Specifically, positive morning eventswere associated with reduced afternoon stress,

FIGURE 4Buffering Effects of the Positive Reflection Intervention and Positive Events


even when we controlled for afternoon events andmorning stress levels. We also found associationsbetween workday positive events and reducedblood pressure and enhanced ability to detach inthe evening. Our results also highlight the potentialof positive intervention experiences. Participantsreported lowered stress, decreased physical andmental complaints, and improved detachment fromwork in the evening. The intervention failed, how-ever, to reduce blood pressure. Although our effectswere generally small, they are impressive, givenour “rather minimal intervention” (Emmons & Mc-Cullough, 2003: 386), and they support the efficacyof positive reflection exercises at work.

Implications for Theory and Research

One way we contribute to theory is by challeng-ing assumptions of symmetry held in basic psy-chology that are reflected in both theory and em-pirical research in the organizational sciences(Thoresen et al., 2003). We do not dispute the ex-istence of two distinct affect systems (i.e., positiveand negative affect), nor do we dispute that positiveand negative events may have their strongest effectson positive and negative outcomes, respectively.Rather, what this research does is highlight theimportance of considering asymmetrical effectswhereby positive events are not only associated withpositive health (lower blood pressure), but also—perhaps via their resource-building capacities—negatively associated with stress. Indeed, ourmomentary lagged analysis clearly suggests the im-portance of positive events for stress reduction, asthe negative association between positive eventsand stress persisted throughout workdays, evenwhen we controlled for subsequent positive andnegative events. This research clearly illustrates theimportance of including asymmetrical effects intheories explicitly and is consistent with emergingstudies in the positive psychology tradition (e.g.,Seligman et al., 2005) in documenting asymmet-rical effects. Decades of work stress research hasfocused on removing stressors (e.g., negativeevents) from work environments. Our study sup-ports this approach but also suggests that, takentogether, positive events stand as an untappedpotential waiting to be leveraged in theories ofstress and in practice.

Although our study highlights the importance ofnaturally occurring positive events, effect sizes fornegative events were stronger—a pattern consistentwith existing literature on the asymmetry of nega-

tive and positive stimuli (Baumeister et al., 2001;Taylor, 1991). However, it is noteworthy that in ourmomentary lagged analysis, it was only the associ-ation between positive events and stress that re-mained over the course of workdays. Our resultssuggest that although the effects of bad may bestronger, the effects of good may be longer, at leastwith respect to employee perceptions of stress. Ourspillover results for blood pressure also suggest thatthere may be either a cumulative effect or a delayedeffect for positive events. The pattern of results inour data also underscores the rationale behind theintervention: people typically do not pay as muchattention to positive events as they do to negativeevents. Although this tendency persists for numer-ous reasons grounded in the evolutionary advan-tages of paying attention to negative stimuli, thesimple fact is that to fully exploit the power ofpositive events, specific positive refocusing or re-flection exercises, such as the one used in thisstudy, may be needed.

We do not intend to downplay the importance ofnegative events in the daily lives of workers. Inkeeping with the literature, our results show thatnegative events have damaging effects on perceivedstress and blood pressure at work and spill over toinfluence evening stress, physical complaints, andability to detach from work. However, our resultsalso suggest that positive events, which are oftenoverlooked in studies of negative workplace out-comes such as stress or health complaints, deservemore attention. It is also important to note thatalthough we did not find lagged effects for negativeevents on family-to-work conflict, this does notmean these events are short lived. Rather, theireffects might be fully mediated via stress and sub-sequent negative events in a cycle, whereby nega-tive events in the morning cause stress (immedi-ately), and that stress causes more negative eventsand more stress in the afternoon, in a repeatingcycle. If this were the case, significant lagged ef-fects might or might not be found, because theeffects of morning work events on afternoon stressmight be fully mediated by morning stress andsubsequent afternoon events.

This need for a shift in approaches to stress canalso be illustrated in the work-family conflict liter-ature. Our work-family results are intriguing in thatthey demonstrate consistent negative associationsbetween family-to-work conflict and stress andhealth during workdays, but our intervention actedas a buffer, preventing these effects from spillingover into evenings. Our post hoc analysis suggests


that the positive reflection may have acted as abuffer, because family was a common topic for thepositive reflection. Work-family research has beenoverwhelmingly focused on the negative untilfairly recently, when researchers began to examinetopics such as enrichment and positive spillover.Our research extends work-family literature by sug-gesting that explicitly bringing thoughts of familyinto their workplace via positive reflections maybenefit, rather than harm, employees and theirfamilies.

Our study is also unique in that we examine bothbuffering effects of daily positive events and buff-ering and enhancing effects of an intervention. Wedid not find enhancing effects for the positive re-flection intervention, in that it did not increase theassociation between positive events and outcomes,but we did find that naturally occurring positiveevents tended to reduce the association betweennegative events and outcomes, further reinforcingthe importance of including positive events instress theories.

Implications for Practice

Emphasis on the role of positive events and ex-periences has important implications for organiza-tions, managers, and employees. Considering thatorganizations, much like individuals, often focus onthe negative—what went wrong, what problems needsolving, what threats are present—organizationsmight benefit from giving positive events moreprominence. Some companies are beginning to fo-cus on strengths rather than deficiencies (e.g.,Buckingham & Clifton, 2001) and on performancemanagement and positive feedback rather thanperformance appraisals and corrective action(Williams, 1997). Such efforts should foster morepositive work events. Companies such as South-west Airlines are using celebrations, encouragingemployees to have fun, and generating positiveevents for customers and employees. Case studiesoften showcase organizations that focus on positiveactions, such as Southwest, suggesting that atten-tion to positive activities is still innovative and thatorganizations could do more to foster positive ex-periences. Imagine a manager modifying the three-good-things intervention, starting weekly staffmeetings by asking employees to share good eventsfrom the past week. Focusing on accomplishments,sharing positive experiences, celebrating success,and expressing gratitude would be a change formost organizations and their employees. Our re-

sults suggest that such a focus could have impor-tant effects on employee stress and health, in keep-ing with the focus of many corporate wellnessprograms (Hollander & Lengermann, 1988).

Limitations and Future Directions

This research has a number of importantstrengths, including collecting data on stress andwork experiences as they occurred, objectivelymeasured health indicators on the job, and nightlyinterviews about a broad range of health indicators,along with a field experiment. Nonetheless, it alsohas important limitations. One prominent limita-tion is our all-female sample. On the one hand, thispresents minimal concern for generalizability, be-cause the sex balance in our sample is representa-tive of health care workers; 80 percent of employ-ees in the health care sector are women (UnitedStates Bureau of Labor Statistics, 2009). On theother hand, the absence of men and the health caresetting of our sample mean that caution should betaken when generalizing results to male workers orother job settings, especially predominantly malesettings or work with fewer interpersonal interac-tions. Existing research suggests that men andwomen differ in physiological reactivity (e.g.,blood pressure) to events (Stoney, Matthews,McDonald, & Johnson, 1988), underscoring the im-portance of future research with male samples andother occupations.

A strength of our study is its assessment of abroad range of health and well-being indexes, in-cluding ambulatory blood pressure. However, thisalso presents challenges, because blood pressure issensitive to many factors. We did not control forfactors such as caffeine use or physical activityprior to each blood pressure measurement, leavingus with a great deal of measurement error, therebylimiting our power to find effects that might exist.Moreover, the association between events andblood pressure is complex, because both positive(exciting) and negative (stressful) events can in-crease blood pressure in the moment. This mayexplain why our aggregated blood pressure effectswere stronger than momentary effects. Our designmay also have interfered with our ability to findblood pressure effects. Our positive interventionwas intended to lower blood pressure, but someparticipants may have viewed it as one more de-manding study activity at the end of the workday.Thus, the nature and timing of our manipulationmay have mitigated its salutary effects. Further-


more, although we assessed ambulatory blood pres-sure, which predicts cardiovascular disease betterthan clinical measurements, we captured onlythree 2-hour segments each day. Emerging researchsuggests that capturing circadian blood pressurerhythms is superior to episodic ambulatory mea-surement. For this reason, our initial design in-cluded 24-hour blood pressure measurement, butpilot testing suggested such an approach wouldeliminate most of our participants because bloodpressure measurement made it difficult to simultane-ously eat, sleep, or drive. The current literature offersno clear guidance—theoretically or empirically—asto what types of events show immediate effects andwhich have cumulative effects over time. Thus,24-hour monitoring might aid in understanding therole of time in links between work events, stress,and health, as would jointly examining a diverseset of physiological measures (e.g., both blood pres-sure and cortisol).

Our goal of comprehensively sampling workevents also has benefits and drawbacks. Despitecarefully developing an event scale drawn from theexisting literature and discussing it with organiza-tional personnel to determine its comprehensive-ness, we acknowledge that by creating a scale thatis both short (few items) and general (broad items),we may have achieved comprehensiveness at therisk of specificity. For example, we ask only if anemployee has received positive information, butthere are many types of positive information thatmight have been received, and they may have dif-fering effects. Another measurement drawback wasour exclusive focus on whether certain positive ornegative events occurred (“yes”/“no”) and not onwhich events, how often they occurred during atime period, or their intensity. Our underlying as-sumption is that the number of events matters morethan which events occur (e.g., disrespect vs. con-flict) or how bad they are. It is a challenge in expe-rience sampling research to collect fine-tuned data,but using rough measures (dichotomous scales andsimple counts of event types), as we did in thisstudy, results in considerable loss of information.Other measures, such as asking about best andworst events (and how bad and how good theywere), are critical in future research if we are togain a full understanding of the effects of workevents. Perhaps even more important is the devel-opment of a comprehensive theory about when andwhy frequency and intensity matter, and whichtypes of events are likely to have varying effects

according to their intensity, to aid researchers inmore precisely capturing the effects of work events.

Another drawback of our study was that some ofour analyses used data collected concurrently. It ispossible that our momentary results were inflatedby factors such as transient affect, which mightinfluence both perception and reporting of workevents and perceptions and reporting of stress.However, given significant results in both ourlagged analyses (lagged momentary and day-to-evening spillover) and significant spillover resultsfor evening blood pressure, which we measuredobjectively, common method variance is not a plau-sible explanation for our overall pattern of results,and especially not for the associations we found forpositive events.

Perhaps the most important limitation of our designwas our decision not to include a no-intervention con-trol group, which would have provided stronger ev-idence of internal validity and greater confidencethat it was our intervention that caused our results.Common threats to validity with this type of design(Cook & Campbell, 1979) are mitigated by our shorttime interval (three weeks) and the fact that wecollected data over ten months in nine differentclinics, which reduces threats associated with his-tory and maturation. The biggest threat to validityis the possibility that our participants may haveexperienced less stress and better health during theintervention stage because they knew the study wascoming to an end. This seems unlikely, because theintervention actually created additional work forparticipants in the second half of the study. Thisconcern is also mitigated by the fact that timetrends such as these have not been found in otherexperience-sampling studies with similar depen-dent variables but without interventions (e.g., Beal& Ghandour, 2011; Miner et al., 2005). Nonetheless,research replicating our results with a no-interven-tion control group is needed.

Although our study demonstrated the efficacy ofa positive reflection, we cannot speak to the effi-cacy of other positive reflection interventions. Priorresearch has shown our intervention to have bene-ficial effects, although it was not designed for busyworkplaces. We asked participants to complete theintervention at the end of their workday, when therush to finish daily tasks and go home may havediminished their opportunity to reflect deeply andsavor positive events. Given these challenges, oureffects—though small and nonsignificant for bloodpressure—are impressive, and future research canbuild on them. For example, future research might


consider whether limiting reflection to positiveevents that occurred in participants’ work settingcould improve work-related outcomes such as jobsatisfaction and organizational commitment. Thetiming of intervention might also be considered,especially for workers with less discrete boundar-ies between work and life. Our participants workedin health care, an industry that represents morethan 10 percent of the US workforce and continuesto grow, and they earned the 2010 median wage fora US worker (United States Bureau of Labor Statis-tics, 2012). In this industry, shift work is common,but for a growing number of managerial, profes-sional, and technical jobs, work-life boundaries arebecoming blurred due in part to increasing use oftechnology. For workers in these jobs, the natureand timing of positive reflection interventions maydiffer.

Future research might also consider whether apositive reflection would be more effective forsome employees than for others. There is clear ev-idence that high-negative-affect individuals reactmore to negative events (Gray, 1981; Larsen & Kete-laar, 1991). Others may respond more strongly topositive events or reflections. Another area for fu-ture research would be to examine the effects of apositive reflection intervention on work recoveryactivities. A growing body of research examines theefficacy of various activities for work recovery(Fritz & Sonnentag, 2005; Trougakos, Beal, Green, &Weiss, 2008). Positive reflection exercises at theend of a workday might increase proclivity for re-covery activities such as socialization or exercise.Indeed, this may be the mechanism by which ourintervention had its effects. Interventions focusedon dealing with negative events might also betested, though much of the stress literature is al-ready focused on coping with negative events.

In this research, we posit mechanisms by whichpositive work events build resources and improveworker well-being but do not explicitly examinethese mediating mechanisms. Future work mightdirectly test the pathways we suggest, wherein pos-itive events build core psychological resources. Pe-rusing the good things responses, we found abun-dant evidence that the intervention promptedreflection on resources identified as universal fea-tures of well-being (Ryff & Singer, 1998), includingmastery ([good thing] “Several coworkers have ex-pressed their faith in me” [why] “Today was myfirst day as ‘lead’ and others wanted to let me knowthat they knew I could do it”), a purposeful life([good thing] “Received a thank you from our clinic

administrator for doing a great job” [why] “I tryvery hard to do my best and I take a lot of pride inthe work I do!”), quality interpersonal connections([good thing] “My husband texted me to let meknow he loves me and that he hopes I have a goodday at work” [why] “Because my husband lovesme, supports me at work, and cares about how myday goes”), and positive self-regard ([good thing] “Imanaged to keep up with Dr. and maintained com-plete control of the flow and tasks needed to bedone” [why] “Because I am good at what I do, I dowell with a busy pace, and am great at prioritizingmy tasks”). These quotations provide compellingexamples of mechanisms suggested in the litera-ture, but research explicitly examining mediatingmechanisms is needed.

Although we had no explicit theory of time andduration, we found that work events have some-what different effects at the momentary, daily, andday-to-evening spillover levels. Our results shouldbe considered preliminary, but they seem to sug-gest that negative events and family-to-work con-flict elicit more immediate reactions, whereas pos-itive workday events have stronger effects whenthey accumulate throughout the course of a work-day. The lack of guiding theory and inconsistentempirical results in the literature suggest that moreempirical and theoretical work is needed to betterunderstand the role of time in the effects of workevents on stress. This study is a first step.

Conclusion

Considered as a whole, our results highlight thepowerful role that discrete work events play inemployees’ health and well-being. Although theliterature is replete with studies focusing on nega-tive work events, our results suggest that positiveevents are important because they are indepen-dently associated with stress and health. Our re-sults showing the potential benefits of an easy toimplement, positive refocusing intervention in theworkplace are also promising.

REFERENCES

Allen, T. D., Herst, D. E., Bruck, C. S., & Sutton, M. 2000.Consequences associated with work-to-family con-flict: A review and agenda for future research. Jour-nal of Occupational Health Psychology, 5: 278–308.

Baumeister, R. F., Bratslavsky, E., Finkenauer, C., &


Vohs, K. 2001. Bad is stronger than good. Review ofGeneral Psychology, 5: 323–370.

Baumeister, R. F., & Leary, M. R. 1995. The need tobelong: Desire for interpersonal attachments as afundamental human motivation. Psychological Bul-letin, 117: 497–529.

Beal, D. J., & Ghandour, L. 2011. Stability, change, andthe stability of change in daily workplace affect.Journal of Organizational Behavior, 32: 526–546.

Benetos, A., Thomas, F., Bean, K., Gautier, S., Smulyan,H., & Guize, L. 2002. Prognostic value of systolic anddiastolic blood pressure in treated hypertensivemen. Archives of Internal Medicine, 162: 577–581.

Block, M., & Zautra, A. 1981. Satisfaction and distress ina community: A test of the effects of life events.American Journal of Community Psychology, 9:165–180.

Bolger, N., DeLongis, A., Kessler, R. C., & Schilling, E. A.1989. Effects of daily stress on negative mood. Jour-nal of Personality and Social Psychology, 57: 808–818.

Bono, J. E., & Campana, K. 2007. Linking negative workevents and employee health: Does social supportmatter? Symposium Presentation at the 22nd An-nual Conference for the Society of Industrial andOrganizational Psychology, New York.

Bryant, F. B. 1989. A four-factor model of perceivedcontrol: Avoiding, coping, obtaining, and savoring.Journal of Personality, 57: 773–797.

Buckingham, M., & Clifton, D. O. 2001. Now, discoveryour strengths. New York: Free Press.

Cameron, K. S., Dutton, J. E., & Quinn, R. E. 2003. Posi-tive organizational scholarship: Foundations of anew discipline. San Francisco: Berrett-Koehler.

Cohen, S., Tyrrell, D. A., & Smith, A. P. 1993. Negativelife events, perceived stress, negative affect, and sus-ceptibility to the common cold. Journal of Person-ality and Social Psychology, 64: 131–140.

Cohen, S., & Wills, T. A. 1985. Stress, social support, andthe buffering hypothesis. Psychological Bulletin, 98:310–357.

Conway, N., & Briner, R. 2002. A daily diary study ofaffective responses to psychological contract breachand exceeded promises. Journal of OrganizationalBehavior, 23: 287–302.

Cook, T. D., & Campbell, D. T. 1979. Quasi-experimen-tation: Design & analysis issues for field settings.Chicago: Rand McNally.

Crawford, E. R., LePine, J. A., & Rich, B. L. 2010. Linkingjob demands and resources to employee engagementand burnout: A theoretical extension and meta-ana-

lytic test. Journal of Applied Psychology, 95: 834–848.

DeLongis, A., Coyne, J. C., Dakof, G., Folkman, S., &Lazarus, R. S. 1982. Relationship of daily hassles,uplifts, and major life events to health status. HealthPsychology, 1: 119–136.

Demerouti, E., Bakker, A. B., Nachreiner, F., & Schaufeli,W. B. 2001. The job demands-resources model ofburnout. Journal of Applied Psychology, 86: 499–512.

Dickerson, S. S., & Kemeny, M. E. 2004. Acute stressorsand cortisol responses: A theoretical integration andsynthesis of laboratory research. Psychological Bul-letin, 130: 355–391.

Diener, E., & Diener, C. 1996. Most people are happy.Psychological Science, 7: 181–185.

Dutton, J. E., Glynn, M. A., & Spreitzer, G. 2006. Positiveorganizational scholarship. In J. Greenhaus & G. Cal-lanan (Eds.), Encyclopedia of career development:641–644. Thousand Oaks, CA: Sage.

Emmons, R. A., & McCullough, M. E. 2003. Countingblessings versus burdens: An experimental investi-gation of gratitude and subjective well-being in dailylife. Journal of Personality and Social Psychology,84: 377–389.

Etzion, D., Eden, D., & Lapidot, Y. 1998. Relief from jobstressors and burnout: Reserve service as a respite.Journal of Applied Psychology, 83: 577–585.

Evans, G. W., Johansson, G., & Rydstedt, L. 1999. Hassleson the job: A study of a job intervention with urbanbus drivers. Journal of Organizational Behavior,20: 199–208.

Evans, O., & Steptoe, A. 2001. Social support at work,heart rate, and cortisol: A self-monitoring study.Journal of Occupational Health Psychology, 6:361–370.

Fredrickson, B. L. 1998. What good are positive emo-tions? Review of General Psychology, 2: 300–319.

Fredrickson, B. L. 2001. The role of positive emotions inpositive psychology: The broaden-and-build theoryof positive emotions. American Psychologist, 56:218–226.

Fredrickson, B. L., Cohn, M. A., Coffey, K. A., Pek, J., &Finkel, S. M. 2008. Open hearts build lives: Positiveemotions, induced through loving-kindness medita-tion, build consequential personal resources. Jour-nal of Personality and Social Psychology, 95:1045–1062.

Fredrickson, B. L., & Joiner, T. 2002. Positive emotionstrigger upward spirals toward emotional well-being.Psychological Science, 13: 172–175.


Frijda, N. H. 1988. The laws of emotion. American Psy-chologist, 43: 349–358.

Fritz, C., & Sonnentag, S. 2005. Recovery, health, and jobperformance: Effects of weekend experiences. Jour-nal of Occupational Health Psychology, 10: 187–199.

Gable, S. L., Reis, H. T., Impett, E. A., & Asher, E. R. 2004.What do you do when things go right? The intraper-sonal and interpersonal benefits of sharing positiveevents. Journal of Personality and Social Psychol-ogy, 87: 228–245.

George, J. M., & Zhou, J. 2002. Understanding when badmoods foster creativity and good ones don’t: The roleof context and clarity of feelings. Journal of AppliedPsychology, 87: 687–697.

Goldberg, D. P. 1972. The detection of psychiatric ill-ness by questionnaire. London: Oxford UniversityPress.

Grant, A. M., & Gino, F. 2010. A little thanks goes a longway: Explaining why gratitude expressions motivateprosocial behavior. Journal of Personality and So-cial Psychology, 98: 946–955.

Gray, J. A. 1981. A critique of Eysenck’s theory of per-sonality. In H. J. Eysenck (Ed.), A model for person-ality: 246–276. New York: Springer.

Greenhaus, J. H., Allen, T. D., & Spector, P. E. 2006.Health consequences of work-family conflict: Thedark side of the work-family interface. In P. L. Per-rewé & D. C. Ganster (Eds.), Research in occupa-tional stress and well-being, vol. 5: 61–98. Amster-dam: JAI/Elsevier.

Greenhaus, J. H., & Beutell, N. J. 1985. Sources of conflictbetween work and family roles. Academy of Man-agement Review, 10: 76–88.

Greenhaus, J. H., & Powell, G. N. 2006. When work andfamily are allies: A theory of work-family enrich-ment. Academy of Management Review, 31: 72–92.

Hart, P. M., Wearing, A. J., & Headey, B. 1994. Perceivedquality of life, personality and work experiences:Construct validation of the police daily hassles anduplifts scales. Criminal Justice and Behavior, 21:283–311.

Heaphy, E. D., & Dutton, J. E. 2008. Positive social inter-actions and the human body at work: Linking organ-izations and physiology. Academy of ManagementReview, 33: 137–162.

Hobfoll, S. E. 1989. Conservation of resources: A newattempt at conceptualizing stress. American Psy-chologist, 44: 513–524.

Hobfoll, S. E. 1998. Stress, culture, and community: Thepsychology and philosophy of stress. New York:Plenum.

Hollander, R. B., & Lengermann, J. J. 1988. Corporatecharacteristics and worksite health promotion pro-grams: Survey findings from Fortune 500 companies.Social Science and Medicine, 26: 491–501.

Ilies, R., Dimotakis, N., & De Pater, I. 2010. Psychologicaland physiological reactions to high workloads: Im-plications for well-being. Personnel Psychology, 63:407–436.

Ilies, R., Keeney, J., & Scott, B. A. 2011. Work-familyinterpersonal capitalization: Sharing positive workevents at home. Organizational Behavior and Hu-man Decision Processes, 114: 115–126.

Israel, B. A., Baker, E. A., Goldenhar, L. M., Heaney,C. A., & Schurman, S. J. 1996. Occupational stress,safety, and health: Conceptual framework and prin-ciples for effective prevention interventions. Journalof Occupational Health Psychology, 1: 261–286.

Janoff-Bulman, R. 1992. Shattered assumptions: To-wards a new psychology of trauma. New York: FreePress.

Kamarck, T. W., Janicki, D. L., Shiffman, S., Polk, D. E.,Muldoon, M. F., Liebenauer, L. L., & Schwartz, J. E.2002. Psychosocial demands and ambulatory bloodpressure: A field assessment approach. Physiology &Behavior, 77: 699–704.

Kamarck, T. W., Shiffman, S. M., Smithline, L., Goodie,J. L., Paty, J. A., Gnys, M., & Jong, J. Y. 1998. Effectsof task strain, social conflict, and emotional activa-tion on ambulatory cardiovascular activity: Daily lifeconsequences of recurring stress in a multiethnicadult sample. Health Psychology, 17: 17–29.

Kanov, J. M., Maitlis, S., Worline, M. C., Dutton, J. E.,Frost, P. J., & Lilius, J. 2004. Compassion in organi-zational life. American Behavioral Scientist, 47:808–827.

Kelly, E. L., Kossek, E. E., Hammer, L. B., Durham, M.,Bray, J., Chermack, K., Murphy, L. A., & Kaskubar, D.2008. Getting there from here: Research on the ef-fects of work-family initiatives on work-family con-flict and business outcomes. In J. P. Walsh & A. P.Brief (Eds.), Academy of Management annals, vol.2: 305–349. Essex, UK: Routledge.

Kim, E., & Yoon, D. J. 2012. Why does service with asmile make employees happy? A social interactionmodel. Journal of Applied Psychology, 97: 1059–1067.

Langston, C. A. 1994. Capitalizing on and coping withdaily-life events: Expressive responses to positiveevents. Journal of Personality and Social Psychol-ogy, 67: 1112–1125.

Larsen, R. J., & Ketelaar, T. 1991. Personality and suscep-tibility to positive and negative emotional states.


Journal of Personality and Social Psychology, 61:132–140.

Liu, S., Wang, M., Zhan, Y., & Shi, J. 2009. Daily workstress and alcohol use: Testing the cross-level mod-eration effects of neuroticism and job involvement.Personnel Psychology, 62: 575–597.

Lyubomirsky, S. 2008. The how of happiness: A scien-tific approach to getting the life you want. NewYork: Penguin.

Mesmer-Magnus, J. R., & Viswesvaran, C. 2005. Conver-gence between measures of work-to-family and fam-ily-to-work conflict: A meta-analytic examination.Journal of Vocational Behavior, 67: 215–232.

Miner, A. G., Glomb, T. M., & Hulin, C. L. 2005. Experi-ence sampling mood and its correlates at work. Jour-nal of Organizational and Occupational Psychol-ogy, 78: 171–193.

Neveu, J. 2007. Jailed resources: Conservation of re-sources theory as applied to burnout among prisonguards. Journal of Organizational Behavior, 28:21–42.

Nezlek, J. B., & Plesko, R. M. 2001. Day-to-day relation-ships among self-concept clarity, self-esteem, dailyevents, and mood. Personality and Social Psychol-ogy Bulletin, 27: 201–211.

Oishi, S., Diener, E., Choi, D. W., Kim-Prieto, C., & Choi,I. 2007. The dynamics of daily events and well-beingacross cultures: When less is more. Journal of Per-sonality and Social Psychology, 93: 685–698.

Panksepp, J. 1993. Neurochemical control of moods andemotions: Amino acids to neuropeptides. In M.Lewis & J. M. Haviland (Eds.), Handbook of emo-tions: 87–107. New York: Guilford.

Pfeffer, J. 2010. Building sustainable organizations: Thehuman factor. Academy of Management Perspec-tives, 24(1): 34–45.

Pressman, S. D., & Cohen, S. 2005. Does positive affectinfluence health? Psychological Bulletin, 131: 925–971.

Rau, R., Georgiades, A., Fredrikson, M., Lemne, C., & deFaire, U. 2001. Psychosocial work characteristicsand perceived control in relation to cardiovascularrewind at night. Journal of Occupational HealthPsychology, 6: 171–181.

Raudenbush, S., Bryk, A., Cheong, Y. F., & Congdon, R.2004. HLM6: Hierarchical linear and nonlinearmodeling. Lincolnwood, IL: Scientific SoftwareInternational.

Reis, H. T., & Gable, S. L. 2003. Toward a positive psy-chology of relationships. In C. L. Keyes & J. Haidt(Eds.), Flourishing: The positive person and thegood life: 129–159. Washington: American Psycho-logical Association.

Ryan, R. M., & Deci, E. L. 2000. Self-determination theoryand the facilitation of intrinsic motivation, socialdevelopment, and well-being. American Psycholo-gist, 55: 68–78.

Ryff, C. D., & Singer, B. 1998. The contours of positivehuman health. Psychological Inquiry, 9: 1–28.

Schnall, P. L., Dobson, M., & Rosskam, E. (Eds.). 2009.Unhealthy work: Causes, consequences and cures.Amityville, NY: Baywood Press.

Schwartz, J. E., Pickering, T. G., & Landsbergis, P. A.1996. Work-related stress and blood pressure: Cur-rent theoretical models and considerations from abehavioral medicine perspective. Journal of Occu-pational Health Psychology, 1: 287–310.

Seligman, M. E., & Csikszentmihalyi, M. 2000. Positivepsychology: An introduction. American Psycholo-gist, 55: 5–14.

Seligman, M. E., Rashid, T., & Parks, A. C. 2006. Positivepsychotherapy. American Psychologist, 61: 774–788.

Seligman, M. E., Steen, T. A., Park, N., & Peterson, C.2005. Positive psychology progress: Empirical vali-dation of interventions. American Psychologist, 60:410–421.

Skarlicki, D. P., van Jaarsveld, D. D., & Walker, D. D.2008. Getting even for customer mistreatment: Therole of moral identity in the relationship betweencustomer interpersonal injustice and employee sab-otage. Journal of Applied Psychology, 93: 1335–1347.

Sonnentag, S., & Bayer, U. V. 2005. Switching off men-tally: Predictors and consequences of psychologicaldetachment from work during off-job time. Journalof Occupational Health Psychology, 10: 393–414.

Sonnentag, S., & Fritz, C. 2007. The recovery experiencequestionnaire: Development and validation of a mea-sure for assessing recuperation and unwinding fromwork. Journal of Occupational Health Psychology,12: 204–221.

Spreitzer, G., Sutcliffe, K., Dutton, J. E., Sonenshein, S., &Grant, A. M. 2005. A socially embedded model ofthriving at work. Organization Science, 16: 537–549.

Steger, M. F., Kashdan, T. B., & Oishi, S. 2008. Beinggood by doing good: Daily eudaimonic activity andwell-being. Journal of Research in Personality, 42:22–42.

Stone, A. A., Bovbjerg, D. H., Neale, J. M., Napoli, A.,Valdimarsdottir, H., Cox, D., Hayden, F. G., & Gwalt-ney, J. M. 1993. Development of common cold symp-toms following experimental rhinovirus infection isrelated to prior stressful life events. BehavioralMedicine, 8: 115–120.


Stone, A. A., & Neale, J. M. 1982. Development of amethodology for assessing daily experiences. In A.Baum & J. Singer (Eds.), Advances in environmentalpsychology: Environment and health, vol. 4: 49–83. Hillsdale, NJ: Erlbaum.

Stoney, C. M., Matthews, K. A., McDonald, R. H., &Johnson, C. A. 1988. Sex differences in lipid, lipo-protein, cardiovascular, and neuroendocrine re-sponses to acute stress. Psychophysiology, 25: 645–656.

Taylor, S. E. 1991. Asymmetrical effects of positive andnegative events: The mobilization-minimization hy-pothesis. Psychological Bulletin, 110: 67–85.

Terkel, S. 1972. Working. New York: Pantheon.

Teuchmann, K., Totterdell, P., & Parker, S. K. 1999.Rushed, unhappy, and drained: An experience sam-pling study of relations between time pressure, per-ceived control, mood, and emotional exhaustion in agroup of accountants. Journal of OccupationalHealth Psychology, 4: 37–54.

Theorell, T., Knox, S., Svensson, J., & Waller, D. 1985.Blood pressure variations during a working day atage 28: Effects of different types of work and bloodpressure level at age 18. Journal of Human Stress,11: 36–41.

Thoresen, C. J., Kaplan, S. A., Barsky, A. P., Warren,C. R., & de Chermont, K. 2003. The affective under-pinnings of job perceptions and attitudes: A meta-analytic review and integration. Psychological Bul-letin, 129: 914–945.

Totterdell, P., Wood, S., & Wall, T. 2006. An intra-individual test of the demands-control model: Aweekly diary study of psychological strain in portfo-lio workers. Journal of Occupational and Organi-zational Psychology, 79: 63–84.

Trougakos, J. P., Beal, D. J., Green, S. G., & Weiss, H. M.2008. Making the break count: An episodic exami-nation of recovery activities, emotional experiences,and affective delivery. Academy of ManagementJournal, 51: 131–146.

Tsai, W. C., & Huang, Y. M. 2002. Mechanisms linkingemployee affective delivery and customer behavioralintentions. Journal of Applied Psychology, 87:1001–1008.

Uchino, B. N. 2006. Social support and health: A reviewof physiological processes potentially underlyinglinks to disease outcomes. Journal of BehavioralMedicine, 29: 377–387.

United States Bureau of Labor Statistics. 2009. Women inthe labor force: A databook. http://www.bls.gov/cps/wlftable14.htm. Accessed February 21, 2011.

United States Bureau of Labor Statistics. 2012. Employ-

ment projections: 2010–20. http://bls.gov/news.release/ecopro.nr0.htm. Accessed July 30, 2013.

Van der Doef, M., & Maes, S. 1998. The job demand-control (-support) model and physical health out-comes: A review of the strain and buffer hypotheses.Psychology & Health, 13: 909–936.

Van der Doef, M., & Maes, S. 1999. The job demand-control (-support) model and psychological well-being: A review of 20 years of empirical research.Work and Stress, 13: 87–114.

Wang, M., Liao, H., Zhan, Y., & Shi, J. 2011. Daily cus-tomer mistreatment and employee sabotage againstcustomers: Examining emotion and resource per-spectives. Academy of Management Journal, 54:312–334.

Watson, D., Clark, L. A., & Tellegen, A. 1988. Develop-ment and validation of brief measures of positiveand negative affect: The PANAS scales. Journal ofPersonality and Social Psychology, 54: 1063–1070.

Weiss, H. M., & Cropanzano, R. 1996. Affective eventstheory: A theoretical discussion of the structure,causes, and consequences of affective experiences atwork. In B. M. Staw & L. Cummings (Eds.), Researchin organizational behavior, vol. 18: 1–74. Green-wich, CT: JAI.

Williams, M. J. 1997. Performance appraisal is dead.Long live performance management. Harvard Man-agement Update, no. U9702A: 1–6.

Xanthopoulou, D., Bakker, A. B., Heuven, E., Demerouti,E., & Schaufeli, W. B. 2008. Working in the sky: Adiary study on work engagement among flight atten-dants. Journal of Occupational Health Psychology,13: 345–356.

APPENDIX


We think too much about what goes wrong and notenough about what goes right in our lives and work. Ofcourse, sometimes it makes sense for us to analyze badevents so that we can learn from them and avoid them inthe future. However, people tend to spend more timethinking about what is bad in life or work than is helpful.Worse, this tendency to focus on bad events sets us up foranxiety and depression. One way to keep this from hap-pening is to develop our ability to think about the good inwork and in life.

Your assignment is as follows:Every day for the next 8 days, you will be writing for

5–10 minutes about three things that went really well onthat day and why they went well. At the last signal ofeach day, you will log into this website and write yourthoughts for the day. Remember, your employer has said


http://www.bls.gov/cps/wlftable14.htm

http://www.bls.gov/cps/wlftable14.htm

http://bls.gov/news.release/ecopro.nr0.htm

http://bls.gov/news.release/ecopro.nr0.htm

it is OK for you to take 10 minutes away from work tocomplete this exercise.

The three things you list each day can be relativelysmall in importance (“My husband picked up my favoriteice cream for dessert on the way home from work today”or “I was able to make someone feel better at work to-day”) or relatively large in importance (“I got a raisetoday” or “My sister just gave birth to a healthy babyboy”). What you write about can be related to good thingsthat happened that day at work, but they do not need tobe work related.

Next to each positive event in your list, answer thequestion, “Why did this good thing happen?” For exam-ple, someone might write that her husband picked up icecream “because my husband is really thoughtful some-times” or “because I remembered to call him from workand remind him to stop by the grocery store.” Whenasked why you got a raise today, someone might writethat “God was looking out for me” or “I worked hard anddid well on my job.”

Writing about “why” the positive events in your lifehappened may seem awkward at first, but as you do thiseach day for the next 8 days, it will get easier.

It is critical to our research that you take the time tocarefully complete this exercise for the next eight days.Some days you will have only small “good things” towrite about, but even these small things are important torecord. When the study is complete, we will link your IDnumber to your name so that we can email you a recordof your “Good Things” journal.

Joyce E. Bono ([email protected]) is a professor of man-agement at the Warrington College of Business, Univer-sity of Florida. She received her Ph.D. in organizationalbehavior from the University of Iowa. Her primary researchinterest is in employees’ quality of work life, which in-cludes the effects of managers, work events, and personalityon employee attitudes and motivation. Other research in-

terests include personality and factors that affect the ad-vancement of women in leadership positions.

Theresa M. Glomb ([email protected]) is the McFarlandProfessor of Organizational Behavior in the CarlsonSchool of Management at the University of Minnesota.She received her Ph.D. in social, organizational, andindividual differences psychology from the University ofIllinois. Her research focuses on issues of workplacewell-being including emotions and mood in organiza-tions, job attitudes and behaviors, emotional labor, andworkplace victimization, including incivility, aggres-sion, and sexual harassment.

Winny Shen ([email protected]) is an assistant professor ofindustrial/organizational psychology in the Departmentof Psychology at the University of South Florida. Shereceived her Ph.D. in industrial/organizational psychol-ogy from the University of Minnesota. Her current re-search interests center around diversity and inclusion inorganizational and educational settings, leadership, per-sonality, and the intersection of these domains.

Eugene Kim ([email protected]) is an as-sistant professor of organizational behavior in theScheller College of Business at the Georgia Institute ofTechnology. He received his Ph.D. from the University ofMinnesota. His current areas of research include work-place aggression and victimization, moral cognition andbehavior, emotion and well-being at work, individualdifferences (e.g., intelligence and personality), and hu-man and social capital.

Amanda J. Koch ([email protected]) is a research scien-tist at Human Resources Research Organization. She re-ceived her Ph.D. in industrial/organizational psychologyfrom the University of Minnesota. Her research interestsinclude topics related to personnel selection, testing andassessment, and diversity.


building positive resources: effects of positive … · building positive resources: effects of...

Documents