The Costs and Benefits of Writing, Talking, and Thinking About Life's Triumphs and Defeats

Journal of Personality and Social Psychology Copyright 2006 by the American Psychological Association
2006, Vol. 90, No. 4, 692–708 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.90.4.692
The Costs and Benefits of Writing, Talking, and Thinking About Life’s
Triumphs and Defeats
Sonja Lyubomirsky, Lorie Sousa, and Rene Dickerhoof
University of California, Riverside
Three studies considered the consequences of writing, talking, and thinking about significant events. In
Studies 1 and 2, students wrote, talked into a tape recorder, or thought privately about their worst (N ⫽
96) or happiest experience (N ⫽ 111) for 15 min each during 3 consecutive days. In Study 3 (N ⫽ 112),
students wrote or thought about their happiest day; half systematically analyzed, and half repetitively
replayed this day. Well-being and health measures were administered before each study’s manipulation
and 4 weeks after. As predicted, in Study 1, participants who processed a negative experience through
writing or talking reported improved life satisfaction and enhanced mental and physical health relative
to those who thought about it. The reverse effect for life satisfaction was observed in Study 2, which
focused on positive experiences. Study 3 examined possible mechanisms underlying these effects.
Students who wrote about their happiest moments— especially when analyzing them— experienced
reduced well-being and physical health relative to those who replayed these moments. Results are
discussed in light of current understanding of the effects of processing life events.
Keywords: positive experience, health, well-being, rumination, expressive writing
Significant life events can be processed in a number of ways. whether the outcome for mental and physical health is favorable or
Some people write about their experiences in a journal; others talk unfavorable (Greenberg, Wortman, & Stone, 1996; Lyubomirsky
about their concerns with friends or family; still others prefer to & Nolen-Hoeksema, 1995; Murray & Segal, 1994; Pennebaker &
think about their situation privately or not do anything at all. Seagal, 1999).
Whether the significant experience is negative (e.g., an injury, In contrast to the literature on negative life events, work on
death of a friend, loss of salary, divorce) or positive (e.g., mar- processing positive experiences is relatively sparse. Research has
riage, birth of a child, promotion, graduation), the way that one tended to focus on negative life circumstances and the ways in
responds to the experience may differentially affect the outcome which they are processed and managed (Seligman & Csikszent-
for one’s well-being and health. The primary question motivating mihalyi, 2000). Demonstrating the utility of processing positive
the present research is, Which ways of processing negative and life experiences in specific ways—for example, by journaling,
positive experiences are most beneficial and why? talking, or thinking privately about them—would represent a much
Experiencing unpleasant or traumatic circumstances can affect needed advance in research.
an individual’s sense of meaning and order in the world. Conse-
quently, seeking to restore meaning and order is a common and
adaptive way of coping with negative events. However, it appears THEORY AND RESEARCH ON NEGATIVE
that not all methods of seeking meaning are created equal. Previ- EXPERIENCES
ous research suggests that the ways in which people process their
thoughts under adverse circumstances—namely, whether they en- Effects of Writing and Talking
gage in thinking privately versus writing or talking— can influence
A large and growing literature supports the premise that writing
about past traumatic experiences has beneficial consequences for
well-being and health (for reviews, see Frattaroli, in press; Smyth,
Sonja Lyubomirsky, Lorie Sousa, and Rene Dickerhoof, Department of
1998), such as less distress, negative affect (NA), and depression
Psychology, University of California, Riverside.
Lorie Sousa is now at Sanford Systems, Wildomar, California. over time (Dominguez et al., 1995; Greenberg & Stone, 1992;
This research was supported in part by an intramural grant from the Murray & Segal, 1994). Furthermore, those who simply talk about
Academic Senate of the University of California, Riverside, and a disser- a traumatic experience into a tape recorder (as though they are
tation grant from the Graduate Division. We thank Sanford Systems for talking to a friend) are just as likely to benefit as those who write
graciously hosting our study Web site. We are also grateful to Katie about the experience (Murray & Segal, 1994). Discussing a trau-
Frieboes, Antoinette Lee, James Chin, Andrea Gutierrez, Mia Lyon, matic event during brief psychotherapeutic sessions has also been
Meaghan Burroughs, and Wayne Roemhil for assisting with data collec- found to increase well-being (Donnelly & Murray, 1991). Finally,
tion; to Robin DiMatteo, Sunil Erevelles, Joanne Frattaroli, and Charlotte
the literature on social support is relevant, as levels of social
Castro-Markey for valuable comments on this research; and to Bob
Rosenthal and Dan Ozer for their statistical advice. contact have been found to be related to better mental and physical
Correspondence concerning this article should be addressed to Sonja health (DeLongis, Folkman, & Lazarus, 1988; Holahan & Moos,
Lyubomirsky, Department of Psychology, University of California, River- 1985; Vinokur, Schul, & Caplan, 1987; see also Gable, Reis,
side, CA 92521. E-mail: [email protected] Impett, & Asher, 2004).
692
COSTS AND BENEFITS OF WRITING, TALKING, AND THINKING 693
In a meta-analysis of 146 studies, Frattaroli (in press) found a By contrast, thinking is inherently disorganized and even cha-
positive and significant benefit of both writing and talking about otic and includes not only words but also images, intense emo-
negative life events. Thus, on the basis of existing research, we tions, and memories. Indeed, the disorganized and unintegrated
suspect that writing and talking will have similar outcomes for nature of traumatic memories (Foa & Riggs, 1993; van der Kolk &
people’s well-being and health. By contrast, thinking about an van der Hart, 1991) may keep them cycling in one’s mind until
unhappy life event is expected to produce relatively less favorable they are organized, integrated, and recorded. In addition, as dis-
results. cussed above, private thought appears to be inclined to degrade
into repetitive dwelling. Perhaps writing and talking provide a
Effects of Thinking channel leading to the release of painful experiences, whereas
thinking provides a channel leading to the reexperience and main-
Contrary to studies on the effects of writing and talking, re- tenance of painful experiences in memory (cf. Martin & Tesser,
search overwhelmingly suggests that thinking about traumatic life 1989).
events does not result in beneficial outcomes. The search for Of course, not all private thought results in rumination and
meaning and understanding that typically follows the experience dwelling. A distinction can be made between the relatively less
of a traumatic event—although deemed necessary and beneficial— adaptive types of repetitive thought, which include rumination and
unfortunately has the potential to degenerate into a series of intrusive thinking, and the relatively more adaptive types, which
repetitive, negative, and intrusive thoughts—namely, “rumination” include working through (Horowitz, Field, & Classen, 1993),
(Hixon & Swann, 1993; Nolen-Hoeksema, McBride, & Larson, reflection (Hixon & Swann, 1993; Trapnell & Campbell, 1999),
1997; Nolen-Hoeksema & Morrow, 1991). Numerous studies have and emotional processing (Stanton, Kirk, Cameron, & Danoff-
documented that self-focused rumination produces a host of ad- Burg, 2000). A pattern emerges, such that the less adaptive types
verse outcomes, including prolonging dysphoric mood, enhancing of thought are typically repetitive, less controllable, and associated
negatively biased and pessimistic thoughts and memories, and with negative mood, whereas the more adaptive types are rela-
interfering with problem solving, motivation, and concentration tively more deliberate, systematic, searching, and analytical in
(for reviews, see Lyubomirsky & Tkach, 2004; Nolen-Hoeksema, nature, akin to problem solving and curiosity (see Segerstrom,
2003). Stanton, Alden, & Shortridge, 2003).
In sum, thinking about traumatic or stressful life experiences can Finally, writing and talking involve recording or documenting
often be self-perpetuating and repetitive and can result in a host of one’s thoughts externally— either on a piece of paper or into a tape
negative consequences. Thus, in Study 1, we hypothesized that recorder or other source—whereas thinking does not. Perhaps
thinking, unlike writing and talking, would be associated with people tend to rehearse, reexperience, and elaborate significant
detrimental outcomes when processing negative experiences. events to record them in their memories. Writing or speaking about
one’s thoughts out loud may be a means to unburden oneself by
Mechanisms Underlying the Different Outcomes of chronicling them externally, thus allowing oneself to move past
Writing, Talking, and Thinking one’s troubles.
Why might different ways of processing traumatic events result

in such disparate outcomes? A review of the literature suggests THEORY AND RESEARCH ON POSITIVE
that writing, talking, and thinking might result in differing levels of EXPERIENCES
integration and synthesis. Thus, we can speculate about the ways
in which writing and talking are distinct from thinking. Unfortunately, the processing of positive experiences has not
Writing and talking tend to involve organizing, integrating, and received comparable attention from researchers. Researchers may
analyzing one’s problems with a focus on solution generation or at assume that processing positive events is unimportant because it
least acceptance. The highly structured nature of language and does not involve suffering and its alleviation. However, one could
syntax invites organization and analysis that occur in the process argue that learning what processes maintain and stimulate positive
of creating a narrative, which often leads to searching for meaning, emotions, which undoubtedly accompany the reexperience and
enhanced understanding, and identity formation (Singer, 2004; processing of happy events, does help advance understanding of
Smyth, True, & Souto, 2001). Once the structure and meaning of how to relieve suffering. For example, researchers have suggested
an experience is understood, the individual gains a sense of reso- that the cultivation of positive emotions may stimulate well-being,
lution and control and is better able to manage his or her emotions health, and general functioning (Fredrickson, 2001; Salovey, Roth-
about the experience (Pennebaker & Graybeal, 2001). Further- man, Detweiler, & Steward, 2000). According to Fredrickson’s
more, writing or talking about traumatic experiences helps a per- (2001) broaden-and-build theory, positive emotions broaden one’s
son label his or her emotions, which may allow him or her to cognitions and actions and foster growth and coping skills. As a
understand them and to let go (Esterling, L’Abate, Murray, & result, positive emotions build durable physical, cognitive, and
Pennebaker, 1999; Swinkels & Giuliano, 1995). Other studies have social resources—for example, physical play, intellectual explora-
shown that improvements in health and other outcomes are asso- tion, and sharing with others (e.g., Fredrickson & Joiner, 2002;
ciated with the use of causal and insight words (presumably related Fredrickson, Tugade, Waugh, & Larkin, 2003). These resources
to integration) during the writing process (Esterling et al., 1999; can promote multiple successful outcomes—in the domains of
Pennebaker, Mayne, & Francis, 1997); with organization, accep- social relationships, work, and physical and mental health (see
tance, and the formation of a narrative (Pennebaker & Seagal, Lyubomirsky, King, & Diener, 2005).
1999); and with integration of memories in self-understanding Positive emotions may also serve as a resource in the prevention
(Blagov & Singer, 2004). and treatment of psychological problems stemming from negative
694 LYUBOMIRSKY, SOUSA, AND DICKERHOOF
emotions, such as anxiety, depression, and stress-related health 2005). Their findings corroborate our prediction that systematic
problems (Fredrickson, 2000). For example, the emotions of joy analysis of positive events is counterproductive. According to
and contentment have been found to have an undoing effect on Wilson and Gilbert (2003), analyzing a positive event and trying to
negative emotions (Fredrickson & Levenson, 1998; Fredrickson, make sense of it serves to “ordinize” it. That is, “meaning making”
Mancuso, Branigan, & Tugade, 2000) and on physiological arousal through systematic analysis removes any uncertainty or mystery
(Hughes, Uhlmann, & Pennebaker, 1994; Salovey et al., 2000). surrounding positive experiences, transforming them from some-
Indeed, writing about the positive (as opposed to negative) emo- thing surprising, thrilling, and extraordinary into something mun-
tions associated with a traumatic experience is associated with dane, ordinary, and less emotionally intense.
lowered heart rate and skin conductance levels (Hughes et al., In contrast to the arguments regarding analysis, we submit that
1994). Inductions of positive affect (PA) have also been found to thinking repetitively (e.g., ruminating) about positive events could
enhance good health and immune function, including a return to be beneficial for the individual. Repetitive replaying and rehearsal
baseline for levels of negative emotional arousal and cardiovascu- of a particular experience tends to maintain its surrounding emo-
lar activation (Fredrickson & Levenson, 1998), the release of tions—a process that is undesirable when the experience is nega-
antibodies responsible for fighting off the cold virus (Stone, Cox, tive but valuable when it is positive. Thus, thinking about a happy
Valdimarsdottir, Jandorf, & Neale, 1987), and fewer illnesses and moment or wonderful experience—passively, repetitively, without
physician visits (Goldman, Kraemer, & Salovey, 1996). analysis—may help sustain PA because it allows the individual to
Another path by which PA and positive experiences may be savor, rehearse, and reexperience it.
associated with health is by providing the psychological resilience
needed to face health threats. Salovey et al. (2000) suggested that Relevant Prior Research
individuals who experience positive emotions may be more likely
to use health-promoting behaviors and to respond to health threats. Unfortunately, to our knowledge, no studies have directly com-
For example, studies have shown that individuals who experience pared different ways of responding to happy experiences, and only
more PA may be better equipped to manage and cope with health a handful of investigations to date, reviewed below, have exam-
issues when they arise (e.g., Aspinwall & Brunhart, 1996; Irving, ined people’s reactions to such experiences.
Snyder, & Crowson, 1998). Research on “capitalization” suggests that, unlike negative
events, positive events can provide opportunities to capitalize on
Comparing Writing, Talking, and Thinking good fortune. For example, expressive responses to positive life
events (e.g., telling others or throwing a party) have been associ-
In sum, the research described above suggests that by helping an ated with PA above and beyond the valence of the positive events
individual to cultivate physical, intellectual, and social resources, themselves (Langston, 1994). Communicating positive events to
the experience and maintenance of positive emotions can be in- others has also been related to increased PA and well-being,
strumental in promoting well-being, boosting health, and aiding especially when this communication is responded to in active and
coping with life’s disappointments and failures. Thus, it would be constructive ways (e.g., with enthusiasm rather than criticism;
important to determine which particular ways of responding to Gable et al., 2004). However, capitalizing responses to happy
happy life experiences best maintain PA. Because writing, talking, experiences cannot be equated with either talking or writing per se,
and thinking about negative life events appear to be associated as they can include a diverse set of behaviors and need not involve
with different outcomes, we suggest that processing positive life a narrative or analysis of the event in question. Furthermore, the
events by using these three methods may also trigger and maintain effects of capitalization on well-being are likely due in part to the
different levels of PA and thus be associated with differential benefits of communication and social support.
benefits to happiness and health. Several studies by King and her colleagues have investigated
Recall that writing and talking tend to invoke organization, whether writing about positive topics promotes successful self-
integration, analysis, and a methodical approach. This quality regulation and, consequently, produces benefits for health and
appears to be beneficial in the context of NA (i.e., when processing well-being. The results of one study supported the hypothesis that
negative events), yet it might interfere with the experience of PA writing about the positive aspects of a traumatic life event pro-
(i.e., when processing positive events). For example, processing a motes self-regulatory processes that, in turn, lead to some of the
joy or triumph analytically, step by step (as one may do when same health benefits found when writing about trauma alone (King
writing or talking about it), may lead to the questioning of one’s & Miner, 2000). Two other studies, which addressed positive
good fortune (e.g., “Maybe I don’t deserve this”), to consideration events more directly, obtained evidence of health benefits for
of the possible counterfactuals (e.g., “What if I had not been at the participants who wrote about their best possible future selves
right place at the right time?”), or to the deliberation of possible (King, 2001) or about peak experiences (Burton & King, 2004).
down sides (e.g., “My friends will be jealous now and might snub These findings appear to support our intuitions that replaying and
me”). Consequently, systematic analysis of positive events through rehearsing happy moments is beneficial, as King’s instructions
writing or talking could be detrimental, as it may serve to break directed her participants to write in a manner that replays rather
such events down into their constituent parts, to reduce the plea- than analyzes the experience in question—that is, “write about the
sure associated with them, and even to evoke negative emotions, experience in as much detail as possible trying to include the
such as guilt or worry. feelings, thoughts, and emotions that were present at the time . . .
Theory and research by Wilson and his colleagues bolster our try your best to reexperience the emotions involved” (Burton &
arguments. These researchers have shown that uncertainty follow- King, 2004, p. 155).
ing a positive event prolongs the pleasure it yields, whereas cer- Researchers have also found benefits for well-being through a
tainty reduces pleasure (Wilson, Centerbar, Kermer, & Gilbert, writing exercise—specifically, the practice of writing down on a
daily or weekly basis the five things for which one is grateful or described above, we anticipated that when it comes to the best
thankful (Emmons & McCullough, 2003; Lyubomirsky, Sheldon, experiences in life—an individual’s joys and victories—it is re-
& Schkade, 2005). However, because the writing involves a simple petitive replaying and rehearsal of the experience that is likely to
list (e.g., “my mom, my new girlfriend, last week’s A in math”), it maintain the positive emotions surrounding it and to enhance
cannot be equated with the more complex writing examined in the well-being generally. In contrast, step-by-step analysis of a posi-
present studies. tive event by breaking it down into its component parts is likely to
Finally, work on “savoring” has shown that people’s perceptions diminish pleasure and well-being. Accordingly, the purpose of
of their ability to savor positive experiences—that is, relishing Study 3 was to explore an important possible mechanism (i.e.,
positive moments, anticipating future positive events, and remi- replay vs. analysis) that underlies the relationship between differ-
niscing about pleasant past experiences—are robustly related to ent modes of processing positive events and outcomes for positive
reports of greater well-being (Bryant, 1989, 2003). Furthermore, emotions and health. To this end, participants were instructed to
researchers have identified individual differences in the ways that either write or think about one of their happiest days for 8 min each
people respond to positive experiences—that is, whether they tend day for 3 consecutive days. Half were induced to systematically
to savor or dampen their PA (Wood, Heimpel, & Michela, 2003). analyze their positive experience while writing or thinking, and
These findings suggest that savoring is a valuable and beneficial half were induced to repetitively replay it.
exercise, which, we submit, comes more easily when an individual Our primary hypothesis for this study was that thinking about
replays a positive event than when he or she systematically ana- positive experiences— especially when the thinking involves re-
lyzes it. petitive replaying and rehearsal—would be the most beneficial
In summary, researchers are only beginning to explore the ways method of processing to a person’s long-term PA, general well-
that people process and respond to happy life events. The aim of being, and health. In contrast, writing about a happy life experi-
our Studies 2 and 3 was to compare directly the impact of different ence—particularly when the writing involves step-by-step analy-
modes of processing of positive experiences. sis—was expected to be the most detrimental method.
Accordingly, we predicted that writing while analyzing would be
PRESENT RESEARCH especially detrimental when compared with thinking while
replaying.
Study 1
Previous research suggests that processing traumatic experi- STUDY 1
ences by writing or talking about them results in beneficial out- Method
comes. To our knowledge, no studies have compared writing and
talking about a traumatic experience with thinking about it, al- Participants
though a separate literature on private thought suggests that the
Ninety-six students (70 female and 26 male) were recruited from the
latter may be detrimental to functioning. To remedy this gap, our
undergraduate psychology student participant pool at the University of
first study was aimed at determining which ways of processing California, Riverside. Three additional participants failed either to follow
traumatic experiences are most beneficial to students’ well-being directions or to complete the entire study and were dropped from the
and health. Participants generated a traumatic or negative event analyses. Although data regarding ethnicity were not collected, the partic-
and either wrote, talked aloud, or thought privately about that ipant pool typically consists of approximately 40% Asians, 20% Lati-
event for 15 min during each of 3 consecutive days. Our primary no(a)s, 20% Caucasians, 10% African Americans, and 10% “other.” Par-
hypothesis for this study was that thinking about significant neg- ticipants’ ages ranged from 17 to 38 years (M ⫽ 19.88 years, SD ⫽ 3.35
ative life experiences would be associated with less favorable years). Students were matched across conditions for gender. On the basis
outcomes 4 weeks later than would writing or talking about them. of their responses to a mass-administered questionnaire, those who scored
above 16 on the Beck Depression Inventory (Beck, 1967) were excluded
from participation in the study.
Study 2
Our second and third studies focused on pleasant experiences. Procedure
The procedure of Study 2 was identical to that of Study 1, except Participants completed a packet of outcome measures administered
that students considered their happiest life event. We hypothesized during their introductory psychology class. From this initial mass admin-
that thinking about significant positive life experiences would be istration, 36 students (who did not participate in any manipulation) were
associated with more favorable outcomes, such as enhanced life randomly selected to serve as the comparison group, and the remaining 60
satisfaction, than would writing or talking about them. This hy- were randomly assigned to one of the experimental groups—that is, writ-
pothesis, which was essentially the inverse of that for Study 1, was ing, talking, or thinking. These three experimental groups were asked to
based on the assumption that the organized, analytical, and exter- sign up for 3 consecutive days of participation followed by an e-mail
nalizing nature of writing and talking may be inherently incom- contact to take place 4 weeks subsequent to their 3rd day of initial
participation. The experimental participants were tested individually. First,
patible with the maintenance of PA, resulting in unfavorable
they were asked to read an information form that briefly described the tasks
outcomes for these two groups.
involved in the study and informed them of their rights as participants.
Then they received a packet of outcome measures.
Study 3
The aim of our third study was to manipulate directly the Outcome Measures
amount of systematic analysis or repetitive replaying during the The following measures were administered to participants two times
exercise of writing or thinking about a positive life event. As during the course of the study—at Time 1 (i.e., during the 1st week of their
psychology class for the comparison group and on their 1st day of partic- Following recommendations (Stewart et al., 1988), we standardized
ipation for the three experimental groups) and at Time 2 (i.e., 4 weeks overall scores on the Health Survey by converting them to a 100-point
subsequent to administration of the Time 1 measures via an e-mail message scale. Overall, the Health Survey has good internal consistency, with
for all groups). alphas ranging between .81 and .87. Consistent results have been found
across age, gender, diagnostic, and educational groups. In addition, the
convergent and discriminant validity of this scale has been evaluated by
Demographics Questionnaire using multitrait analysis with good results (Stewart et al., 1988).
The first questionnaire requested information about each participant’s
date of birth, gender, e-mail address, and college major. Symptom Checklist
The Symptom Checklist (Sherbourne, Allen, Kamberg, & Wells, 1992)
Life Satisfaction is intended to tap the experience of several common health problems (e.g.,
backaches, upset stomach, cold, allergies; ␣s ⫽ .75 and .74 at Time 1 and
The next measure administered was the Satisfaction With Life Scale Time 2, respectively). Throughout the questionnaire, participants are asked
(SWLS; Diener, Emmons, Larsen, & Griffin, 1985), which assesses re- to respond by indicating the frequency of their experience of each of 13
spondents’ current satisfaction with their life in general. It consists of five symptoms over the previous month (1 ⫽ never, 2 ⫽ a few times in 4 weeks,
questions (e.g., “In most ways my life is close to my ideal”), which are 3 ⫽ more than once a week, 4 ⫽ 3 or 4 times a week, 5 ⫽ nearly every day,
rated on 7-point Likert-type scales (1 ⫽ strongly disagree, 7 strongly and 6 ⫽ everyday). This scale has been shown to have an alpha of .75
agree; ␣s ⫽ .88 and .89 for Time 1 and Time 2, respectively). Validation (Sherbourne et al., 1992).
studies have shown that the SWLS comprises a single factor and possesses
high internal consistency (␣ ⫽ .87) and high test–retest reliability (r ⫽ .82;
Diener et al., 1985).
Experimental Manipulation
Transient Affect Preliminary Measures

The following measures were administered to the three experimental
Participants then completed the Positive and Negative Affect Schedule
groups on their 1st day of participation. The comparison group did not
(PANAS; Watson, Clark, & Tellegen, 1988), which lists 20 adjectives. Ten
complete these questionnaires.
adjectives reflect PA (e.g., proud, interested, alert), and 10 reflect NA (e.g.,
Worst Experience Questionnaire. This questionnaire asked partici-
nervous, irritated, afraid). The instructions ask respondents to rate their
pants to list the three worst or most traumatic experiences of their life and
current experience of each item (i.e., how they feel right now) by using
to rate how upsetting each experience was on a 10-point Likert-type scale
5-point scales (1 ⫽ very slightly or not at all, 5 ⫽ extremely). Composites
(1 ⫽ not upsetting, 10 ⫽ extremely upsetting). Then, they were asked to
for PA and NA were created by averaging separately the scores of the
highlight the one experience that was most upsetting to them.
adjectives denoting PA and those denoting NA (␣ ⫽ .70 for Time 1 PA,
Life Experience Questionnaire. Next, students completed a set of
␣ ⫽ .77 for Time 1 NA, ␣ ⫽ .91 for Time 2 PA, and ␣ ⫽ .90 for Time 2
ratings regarding their very worst life experience. First, they were asked,
NA). In previous studies, the PANAS has been shown to have alphas
“How recent was this experience?” (responses ranged from within the last
ranging between .76 and .85 (Watson et al., 1988).
week to 9 years ago to more than 9 years ago). The remaining items were
rated on 10-point Likert-type scales: “How significant is this experience in
Health Survey your life?” (1 ⫽ not at all significant, 10 ⫽ very significant); “How much
time have you spent writing [talking aloud to others] [thinking (privately)]
The Medical Outcomes Study Short-Form–20 Health Survey (Stewart, about this experience?” (1 ⫽ none, 10 ⫽ a lot). The final question asked,
Hays, & Ware, 1988) is an inventory intended to tap an individual’s current “Is this experience resolvable (meaning, do you think the experience can be
general functioning and physical health (Time 1, ␣ ⫽ .70; Time 2, ␣ ⫽ easily resolved by you, or is it out of your control)?” (1 ⫽ not easily
.86). The measure consists of six subscales, assessing health perceptions (5 resolvable, 10 ⫽ easily resolvable).
items); mental health (5 items); pain (1 item); and physical (6 items), role
(2 items), and social functioning (1 item).
The Health Perceptions subscale asks participants to rate their beliefs
Experimental Conditions
about their health (e.g., “I am as healthy as anybody I know”), by using Participants then experienced one of three manipulations. Participants in
5-point Likert-type scales (1 ⫽ definitely false, 5 ⫽ definitely true). The the writing group were asked to write about their experience. Those in the
Mental Health subscale asks participants to rate their mental health status talking group were instructed to talk about their experience into an audio-
(e.g., “During the past month, how much of the time have you . . . felt calm tape recorder. Finally, those in the thinking group were asked to think
and peaceful?”; 1 ⫽ all of the time, 6 ⫽ none of the time). The subscale privately about their experience. Participants in these three experimental
measuring physical functioning requests that participants assess their phys- groups were exposed to the experimental conditions for exactly 15 min
ical ability to perform various physical tasks (e.g., “For how long [if at all] each day of their 3 days of participation.
has your health limited you in the kinds or amounts of vigorous activities Writing condition. In this condition, students were asked to write about
you can do, like lifting heavy objects, running, or participating in strenuous their worst life experience on each day of participation. Participants were
sports?”; 1 ⫽ not at all limited, 3 ⫽ limited for more than 3 months). The given several sheets of blank paper and then were read the following
Role Functioning subscale consists of two questions (e.g., “Does your instructions, adapted from Pennebaker and Francis (1996):
health keep you from working at a job, doing work around the house, or
going to school?”; 1 ⫽ no, 2 ⫽ limited for 3 months or less, 3 ⫽ yes, for For the next three days, I would like for you to write about your
more than 3 months). The Social Functioning subscale consists of the item, deepest thoughts and feelings regarding the significant life experience
“How much of the time, during the past month, has your health limited you highlighted on the questionnaire. In your writing, I’d like you to
your social activities (like visiting with friends or close relatives)?” (1 ⫽ really let go and explore your deepest emotions and thoughts. You
all of the time, 6 ⫽ none of the time). Finally, the Pain subscale simply might tie your topic to your relationships with others including par-
asks, “How much bodily pain have you had in the past 4 weeks?” (1 ⫽ ents, significant others, friends, or relatives, to your past, your present,
none, 6 ⫽ very severe). or your future, or to who you have been, who you would like to be,
or who you are now. You may write about the same general issues or On the basis of previous research, the writing and talking groups
experiences on all days of writing or on different things each day. . . . were expected to experience more favorable outcomes overall than
Don’t worry about using complete sentences or being logical. Just write the thinking group. To test this primary hypothesis, we performed
whatever comes to your mind about this experience.
planned contrasts (with Time 1 as a covariate) by comparing the
When the experimenter returned after 15 min, the pages were collected and thinking group with the writing and talking group, respectively, as
the final questionnaire of the session was administered. well as with the two groups combined.2 Because these contrast
Talking condition. In this condition, participants were asked to talk analyses tested specific directional hypotheses, all reported p val-
about their worst life experience on each day of participation. An audio ues are one-tailed. Finally, we explored whether thinking about
recorder containing a 90-min tape labeled with the participant number and one’s worst life event produces more adverse outcomes than not
day of participation (Day 1, Day 2, or Day 3) was placed on the desk in considering or processing a negative event at all, by comparing the
front of the student. A small microphone was clipped to the participant’s
thinking group with the comparison group (i.e., a sample from the
shirt. The instructions students heard were identical to those in the writing
condition except that participants were told to think out loud as if they were general population of participants).
talking to themselves. They were further told that, “although talking aloud Examination of the demographic data revealed little variabil-
may seem awkward at first, we find that most people get comfortable with ity in age and major (most students were psychology majors),
it after a few minutes.” When the 15 min were up, the experimenter and analyses of the primary dependent variables by gender, age,
reentered the room, turned off the audio recorder and microphone, and set and major revealed no significant main effects or interactions.
the equipment aside. The experimenter then administered the final ques-
Thus, all analyses were collapsed across these variables. There
tionnaire of the session.
Thinking condition. In this condition, students were asked to think
were 20 participants in the writing group, 19 in the talking
privately about their worst life experience on each day of participation. group, 21 in the thinking group, and 36 in the comparison
Thus, the participants’ task involved no writing or talking. The instructions group.
for this condition were identical to those provided in the writing condition
except that participants were asked to think about their significant life
experience rather than write about the experience. The experimenter reen- Ratings of Respondents’ Worst Experiences
tered after 15 min and administered the final questionnaire of the session.
Comparison condition. Students in this group did not consider or Post hoc analyses comparing each of the three experimental
process any life event. They were simply contacted via e-mail 4 weeks conditions were conducted to determine whether there were any
subsequent to the initial administration of the outcome measures and asked initial significant differences among groups in their ratings of
to complete them once again.
their most traumatic experiences. Death of a family member
(e.g., mother, grandmother, uncle, brother), death of a pet, car
Final Questionnaire accident, breaking up with a significant other, and domestic
violence were some of the experiences mentioned. Overall,
After experimental participants completed 15 min of the manipulation
each day, they were asked this final question, which we used as an students rated their worst life event as extremely upsetting
approximate manipulation check: “How many of the last 15 minutes did (M ⫽ 9.01 on a 10-point scale) and very significant (M ⫽ 8.35
you spend focusing on the task you were given?” (0 –5 min, 5–10 min, on a 10-point scale). Participants’ ratings of the recency of their
10 –15 min, all 15 min). Ratings for all 3 days were averaged to obtain a self-generated negative life experiences revealed that, overall,
mean attention score. this event occurred fairly recently— on average, between 1 and
3 years prior to the study. Participants also perceived their
experience as being somewhat unresolvable (M ⫽ 4.24 on a
Follow-Up 10-point scale).
Analyses of ratings of how much time participants had spent
Four weeks following their participation, all participants were contacted
via e-mail and asked to complete on the Web the final set of outcome writing, talking, or thinking about the negative experience before
measures, which were identical to the first set of measures. Finally, the study revealed that they spent very little time writing (M ⫽
participants were sent a thorough debriefing e-mail detailing the purpose of 3.50 on a 10-point scale), a moderate amount of time talking (M ⫽
the study and were asked to return an end consent form giving us permis- 5.52 on a 10-point scale), and a lot of time thinking (M ⫽ 7.13 on
sion to use their data. a 10-point scale) about the experience.
No initial differences were found between the talking and
thinking groups on any of these variables. However, despite
Results random assignment, several significant differences emerged
Overview of Statistical Analyses

In this study, levels of students’ life satisfaction, transient affect,
and health were assessed before and after they were induced to 1
In all three studies, the results obtained by using covariance analyses
process their worst life experience—that is, at Time 1 (premanipu- (i.e., regressed change scores) were virtually identical to those obtained by
lation) and at Time 2 (postmanipulation). The extent to which using difference scores.
levels of these variables changed over time was examined by 2
The use of both pairwise and combined contrast codes in our three
analyzing Time 2 data with Time 1 as a covariate (i.e., by using studies was deemed appropriate and desirable, after consultation with an
regressed change scores).1 Contrast analyses were then performed expert on contrast analysis (R. Rosenthal, personal communication, May
by using the adjusted scores. 14, 2004).
Table 1
Marginal Means (Adjusted for Time 1) and Standard Errors for Processing Negative Events by Condition (Study 1)
Condition
Writing Talking Thinking Comparison

(n ⫽ 20) (n ⫽ 19) (n ⫽ 21) (n ⫽ 36)
Measure M SE M SE M SE M SE
Satisfaction With Life 4.81a 0.19 4.76a 0.20 4.25b 0.19 4.67a 0.15
Positive affect 3.08a 0.18 2.88a 0.18 3.16a 0.17 3.15a 0.13
Negative affect 1.60a 0.15 1.59a 0.15 1.66a 0.15 1.60a 0.11
Overall health 87.28a 1.55 87.72a 1.57 83.62b 1.50 84.45a,b 1.17
Physical Functioning 97.23a 1.87 97.17a 1.91 97.72a 1.82 96.54a 1.39
Pain 87.55a 3.19 82.56a 3.30 80.74a 3.11 77.35a 2.38
Role Functioning 99.63a 2.02 99.70a 2.06 99.90a 1.96 95.79a 1.50
Social Functioning 99.83a 2.21 97.59a,b 2.27 94.17b 2.15 96.80a,b 1.65
Mental Health 75.98a 2.37 79.19a 2.43 70.67b 2.30 75.04a,b 1.79
Health Perceptions 67.70a 4.03 67.89a 4.12 62.11a 3.92 65.32a 2.99
Physical Symptoms 1.31a 0.08 1.36b 0.09 1.54b,c 0.08 1.62c 0.06
Note. High scores indicate increasing functioning, except for negative affect and physical symptoms. Means that share subscripts are not significantly
different from one another. The means for the thinking group are significantly different from those of the talking and writing groups (combined) for (a)
satisfaction with life, (b) overall health, (c) social functioning, (d) mental health, and (e) physical symptoms.
between the writing group and the thinking group.3 Fortunately, by the SWLS, relative to those who thought about it. Supporting
despite these unexpected group differences, covariance analy- this hypothesis, planned pairwise contrasts (with Time 1 as a
ses, using each of these variables (e.g., significance, recency) as covariate) revealed that participants in the thinking condition re-
separate covariates, revealed that these initial differences did ported lower life satisfaction relative to those in the writing con-
not alter our results. Indeed, the results were stronger in the dition, F(1, 91) ⫽ 4.26, p ⫽ .02, ␩ ⫽ .22, as well as relative to
predicted direction with covariation. those in the talking condition, F(1, 91) ⫽ 3.42, p ⫽ .03, ␩ ⫽ .20.
Furthermore, planned contrasts indicated that the thinking group
Manipulation Check reported lower life satisfaction relative to both the talking and
writing groups, F(1, 91) ⫽ 5.21, p ⫽ .02, ␩ ⫽ .22, as well as
Immediately after participants had completed each of the 3 days relative to the comparison group, F(1, 91) ⫽ 3.05, p ⫽ .04, ␩ ⫽
of the experimental manipulation, they were asked how much of .17 (Table 1).
the 15 min they spent processing their worst life event as instructed
(i.e., by writing, talking, or thinking about it). Overall, students
Transient Affect
reported that they spent approximately 10 –15 min (M ⫽ 2.60 on
a 4-point scale) focusing on their self-generated negative life The thinking group was hypothesized to show reduced transient
event. It is interesting to note that analyses revealed significant PA and increased NA relative to the other groups. However,
group differences, F(3, 92) ⫽ 17.88, p ⬍ .0001, ␩ ⫽ .40. That is, contrary to predictions, no significant group differences were
the writing group reported spending more time on the task (M ⫽ found for either PA or NA, as measured by the short-term PANAS
3.38 on a 4-point scale; 10 –15 min) than did the talking group (all Fs ⬍ 2, ns).
(M ⫽ 2.44; 5–10 min) or the thinking group (M ⫽ 2.02; 5–10 min),
possibly because writing was a relatively more compelling and Health Survey
engaging exercise and because hand writing an idea takes longer
than thinking or saying it. Once again, however, despite differ- Overall health. Supporting our hypothesis, a planned contrast,
ences among the experimental groups on this variable, covariance with Time 1 as a covariate, indicated that students who simply
analyses with attention as a covariate revealed that these differ-
ences did not alter the results. In fact, group differences in our 3
outcome measures were stronger in the predicted direction follow- Specifically, Tukey’s honestly significant difference (HSD) post hoc
analyses revealed that compared with the thinking group, the writing group
ing covariation.
reported their experiences to be significantly more upsetting (Ms ⫽ 9.10
vs. 8.71 on a 10-point scale; HSD ⫽ 0.74, p ⬍ .04), more significant (Ms ⫽
Outcome Measures 9.10 vs. 7.67 on a 10-point scale; HSD ⫽ 1.43, p ⬍ .03), and, finally, less
recent (Ms ⫽ 5.00 vs. 6.83—i.e., ⬍1 year ago vs. 3– 6 years ago; HSD ⫽
The estimated standardized marginal means (adjusted for Time ⫺1.81, p ⬍ .003). Additionally, the writing and thinking groups differed on
1) and standard errors for all our primary outcome measures are how much time they reportedly spent writing (Ms ⫽ 4.35 vs. 2.52 on a
presented in Table 1. 10-point scale; HSD ⫽ 1.82, p ⬍ .05) and thinking (Ms ⫽ 8.30 vs. 6.29 on
a 10-point scale; HSD ⫽ 2.01, p ⬍ .01) about their significant life
Life Satisfaction experience. Finally, the writing and talking groups were found to differ on
the recency of the experience (Ms ⫽ 5.00 vs. 7.05—i.e., ⬍1 year ago vs.
Participants who wrote or talked about their worst life experi- 3– 6 years ago; HSD ⫽ ⫺2.05, p ⬍ .001). Gain scores were used in these
ence were predicted to show improved life satisfaction, as assessed analyses.
thought about their negative life experience reported diminished Contrary to expectations, the experimental groups did not differ
overall health, as revealed in total Health Survey scores (Stewart et in their experience of short-term PA and NA 4 weeks after the
al., 1988), relative to those who wrote about it, F(1, 91) ⫽ 3.51, induction. In retrospect, this finding should not be unexpected, as
p ⫽ .03, ␩ ⫽ .20, and those who talked aloud about it, F(1, 91) ⫽ affect is a state variable and, as such, is transient in nature. It is
3.61, p ⫽ .03, ␩ ⫽ .20, as well as relative to both those who wrote possible that affect is too transitory to expect an effect that endures
and talked combined, F(1, 91) ⫽ 4.81, p ⫽ .02, ␩ ⫽ .22 (see Table 4 weeks after the induction. Further research using a broader range
1). However, a contrast of the thinking group and the comparison of instruments to measure affect may be needed to test this con-
group did not reach statistical significance (F ⬍ 1). jecture. To this end, a long-term assessment of affect was used in
Because the Health Survey comprises several subscales, further our last study (Study 3).
analyses could examine the role of various aspects of health. Unlike the consequences of managing negative experiences, the
Analyses of two of the subscales—Mental Health and Social effects of processing positive life experiences in different ways
Functioning—revealed parallel findings to those for overall health. have received little attention in the literature and remain something
Mental health. The thinking group reported significantly di- of a mystery. The goal of Study 2 was to investigate this issue.
minished mental health relative to the writing group, F(1, 91) ⫽
3.38, p ⫽ .03, ␩ ⫽ .19, and the talking group, F(1, 91) ⫽ 6.38, p ⬍ STUDY 2
.01, ␩ ⫽ .26, as well as relative to both groups combined, F(1,
91) ⫽ 5.89, p ⫽ .01, ␩ ⫽ .24. Thinkers did not differ significantly, Method
however, from those in our comparison group (Fs ⬍ 3).
Social functioning. The thinking group reported reduced so- Participants
cial functioning relative to the writing and talking groups com-
One hundred eleven participants (82 female and 29 male) were recruited
bined, F(1, 91) ⫽ 3.11, p ⫽ .04, ␩ ⫽ .17, as well as to the writing from the undergraduate psychology student participant pool at the Univer-
group only, F(1, 91) ⫽ 3.78, p ⫽ .03, ␩ ⫽ .20. However, the sity of California, Riverside. Again, the participant pool is typically com-
thinking group did not differ significantly from the talking group posed of roughly 40% Asians, 20% Latino(a)s, 20% Caucasians, 10%
(F ⬍ 2) or the comparison group (F ⬍ 1) on this variable. African Americans, and 10% “other.” Participants ranged in age from 17 to
38 years (M ⫽ 19.43 years, SD ⫽ 2.59 years) and, as in Study 1, were
matched across conditions for gender.
Physical Symptoms
Analyses of Symptom Checklist (Sherbourne et al., 1992) scores Procedure and Materials
(i.e., frequency of backaches, colds) corroborated in part the find- Our second study aimed to explore the relative costs and benefits of
ings for overall physical health. Students who thought privately writing, talking, and thinking about positive experiences—namely, the
about their worst experience reported significantly more physical “best” experience of one’s life. To this end, as in Study 1, participants were
health symptoms than those who wrote or talked about it (com- exposed to one of the three experimental inductions—writing, talking, or
bined), F(1, 91) ⫽ 4.43, p ⫽ .02, ␩ ⫽ .22, as well as to those who thinking about the significant life experience. The design was identical to
wrote about it (individually), F(1, 91) ⫽ 4.08, p ⫽ .02, ␩ ⫽ .20, that of Study 1, except for the point at which all participants, except for
but not compared with those who talked about it (F ⬍ 3) or did not those in the comparison group, were asked to generate a significant life
experience. In Study 2, experimental participants were asked the following
consider an event at all (F ⬍ 1; see Table 1).
question: “Please list the three (3) BEST or HAPPIEST experiences of
your life.” Next, they were asked to rate how joyful or happy each
experience was on a 10-point Likert-type scale (1 ⫽ not happy, 10 ⫽
Discussion
extremely happy). Finally, they were instructed to highlight the one expe-
Our findings generally supported our predictions. Overall, par- rience that was the most joyful or happy for them.
Cronbach’s alphas for each of the primary outcome variables (Time 1
ticipants who were instructed to write or talk into a tape recorder
and Time 2, respectively) were as follows: ␣s ⫽ .74 and .93 for transient
about their very worst life experience for 15 min on each of 3 PA; ␣s ⫽ .82 and .86 for transient NA; ␣s ⫽ .88 and .91 for the
consecutive days reported improved well-being and health relative Satisfaction With Life Scale; ␣s ⫽ .74 and .81 for the Health Survey; and
to participants instructed to think privately to themselves. When ␣s ⫽ .78 and .86 for the Symptom Checklist.
assessed 4 weeks subsequent to their initial participation, students
who wrote or talked reported increased life satisfaction, enhanced
Results
mental health, improved overall health, improved social function-
ing, and fewer physical health symptoms relative to students who Overview of Statistical Analyses
thought about their negative experience. In addition, students who
thought privately about their unpleasant life event reported dimin- As in Study 1, levels of participants’ life satisfaction, transient
ished life satisfaction relative to no-treatment controls as well as a affect, and health were assessed before (Time 1) and after (Time 2)
(nonsignificant) trend toward inferior mental health and greater they processed their happiest life experience. Again, the extent to
physical symptoms. It is worth noting, however, that the expected which levels of these variables changed over time was determined
pattern of results with respect to physical health could be a reflec- by analyzing Time 2 data with Time 1 as a covariate (i.e., by using
tion of a general positivity bias—that is, a happy, satisfied person regressed change scores). Finally, the set of planned contrasts
is likely to rate his or her health and general functioning more conducted to test our primary hypothesis was identical to that in
positively. Future research could address this issue by relying on Study 1.
relatively more objective health reports from doctors and close Once again, analyses of all the primary outcome variables by
others. gender and age revealed no significant main effects or interactions.
Thus, presentation of the results collapsed across these variables. 107) ⫽ 9.89, p ⬍ .001, ␩ ⫽ .29. As in Study 1, the writing group
There were 36 participants in the comparison group (also used in reported spending the most time on the task on average (M ⫽ 3.17
Study 1), 24 in the writing group, 25 in the talking group, and 26 on a 4-point scale; 10 –15 min), followed by the talking group
in the thinking group. (M ⫽ 2.51; 5–10 min) and the thinking group (M ⫽ 2.21; 5–10
min), respectively.
Ratings of Respondents’ Best Experiences Covariance analyses were conducted on all the primary outcome
variables by using attention as a covariate. Despite significant
As in Study 1, post hoc analyses comparing each of the three group differences on this variable, these analyses revealed that the
experimental conditions were conducted to ascertain any initial attention differences did not alter the critical results.
significant group differences on participants’ ratings of their hap-
piest life experiences. Gaining admission to the college of their Outcome Measures
choice, meeting their significant other, receiving a new car from
their parents, becoming engaged, experiencing the birth of a sib- Life Satisfaction
ling, going to a concert with friends, and graduating from high According to our primary hypothesis, participants who thought
school were some of the experiences mentioned by participants. about their happiest experience were expected to report enhanced
Overall, students rated their best life event as extremely happy life satisfaction 4 weeks later relative to those who wrote or talked
(M ⫽ 9.81 on a 10-point scale), very significant (M ⫽ 9.09 on a about it. Planned pairwise comparisons revealed that the thinking
10-point scale), and as occurring between 6 months and 1 year group reported improved life satisfaction relative to the writing
previous to their participation in the study. Students also rated their group, F(1, 107) ⫽ 2.26, p ⫽ .07, ␩ ⫽ .14, and relative to the
self-generated positive event as somewhat within their own control talking group, F(1, 107) ⫽ 7.02, p ⬍ .01, ␩ ⫽ .25, as well relative
(M ⫽ 6.51 on a 10-point scale). to the writing and talking groups combined, F(1, 107) ⫽ 5.79, p ⫽
Furthermore, analyses revealed that participants spent a moder- .009, ␩ ⫽ .23. Indeed, this pattern of results, shown in Table 2, is
ate amount of time writing (M ⫽ 5.57 on a 10-point scale), a lot of even more notable when compared with that found for the same
time talking (M ⫽ 7.57 on a 10-point scale), and a lot of time variable in Study 1, in which students processed negative life
thinking (M ⫽ 8.09 on a 10-point scale) about their positive events (shown in Table 1). Students in the thinking group, how-
experience prior to the study. ever, did not differ significantly from those in the comparison
Finally, Tukey’s HSD post hoc analyses revealed no significant group (F ⬍ 2).
differences among the three experimental groups on ratings of the
happiness, significance, resolvability– controllability, and recency Transient Affect and Health
of their most pleasant life experience as well as on how much time
As evident from Table 2, our primary hypothesis was not
they had spent writing, talking, and thinking about this experience
supported for either transient PA (both Fs ⬍ 2), transient NA (both
before the study.
Fs ⬍ 1), overall and subscale Health Survey scores (all Fs ⬍ 2),
or Symptom Checklist scores (both Fs ⬍ 1).
Manipulation Check
Analyses of how much time students spent focusing on the task Discussion
during the study (i.e., writing, talking, or thinking about their Our hypothesis followed from the argument that writing and
happiest life event) showed significant group differences, F(3, talking may promote a type of analysis incompatible with positive
Table 2
Marginal Means (Adjusted for Time 1) and Standard Errors for Processing Positive Events by Condition (Study 2)
Condition
Writing Talking Thinking Comparison

(n ⫽ 24) (n ⫽ 25) (n ⫽ 26) (n ⫽ 36)
Satisfaction With Life 4.33a 0.31 3.86b 0.30 4.97c 0.30 4.67a,c 0.15
Positive affect 3.08a 0.16 3.10a 0.15 2.99a 0.15 3.15a 0.13
Negative affect 1.56a 0.11 1.46a 0.11 1.53a 0.11 1.60a 0.11
Overall health 84.88a 1.36 85.72a 1.35 85.20a 1.30 84.45a 1.17
Physical Functioning 97.34a 2.04 98.05a 2.03 94.66a 1.96 96.54a 1.39
Pain 80.05a 3.46 80.99a 3.36 78.85a 3.31 77.35a 2.38
Social Functioning 95.67a 1.72 98.21a 1.68 97.32a 1.65 96.80a 1.65
Mental Health 69.89a 1.81 69.19a 1.78 68.98a 1.74 75.04a 1.79
Health Perceptions 69.88a 3.17 76.09a 3.10 78.97a 3.05 65.32a 2.99
Physical Symptoms 1.66a 0.09 1.57a 0.09 1.66a 0.09 1.62a 0.06
Note. High scores indicate increasing functioning, except for negative affect and physical symptoms. Means that share subscripts are not significantly
different from one another. The means for the thinking group for satisfaction with life are significantly different from those of the talking and writing groups
(combined).
emotions. On the basis of this assumption, we expected those who replaying (n ⫽ 30), thinking while analyzing (n ⫽ 29), writing while
thought about their happiest experience to report more favorable replaying (n ⫽ 27), and writing while analyzing (n ⫽ 26). A no-treatment
outcomes than those who analyzed it through writing or talking. control group was not used in this study, as it was not necessary to test our
Support for this hypothesis was lacking for health, but it was found central hypotheses.
for one important outcome variable—namely, life satisfaction. As in the first two studies, participants were required to complete four
That is, private thought about a positive life event was associated separate sessions—three sessions on consecutive days (i.e., Day 1, Day 2,
with higher satisfaction with life than writing or talking about that and Day 3) and the fourth session (i.e., Day 4) approximately 4 weeks later.
event. Furthermore, although inspection of the means suggests that Furthermore, a packet of outcome measures was administered on Day 1
(the 1st day of the study and prior to assignment to one of the four
comparison group participants reported higher life satisfaction
manipulations) and Day 4 (4 weeks subsequent to completing the first
than did writers and talkers but lower life satisfaction than did
packet).
thinkers, these group differences did not reach statistical signifi-
cance. Thus, we cannot conclude whether the effects on life
satisfaction were due to the thinking condition producing a favor-
able outcome (relative to no-treatment controls) versus the talking
and writing conditions producing a relatively unfavorable 4
In Studies 1 and 2, we attempted a preliminary examination of the
outcome. mechanisms underlying the observed effects through an analysis of the
As in Study 1, thinking did not differentially affect transient written and spoken words generated over the course of the 3 days. To this
mood. Although this null finding appears to challenge our argu- end, judges coded the percentage of writing or speech for (a) the amount
ment that thinking privately about happy moments maintains PA, of systematic, step-by-step analysis of the positive or negative experience
we speculate that the 4-week interval between our assessments of and (b) the amount of replaying or recounting of that experience. Interrater
transitory mood limited the possibility of finding significant dif- reliability was modest to good for both types of ratings in both studies (␣s
ferences. However, our findings with regard to life satisfaction—a from .50 to .79). Furthermore, in each study, we ran a proxy sample of
construct that incorporates PA and positive self-judgments—are participants who experienced our thinking condition (albeit, for 1 day only)
indeed consistent with the maintenance-of-PA argument. and then estimated the percentage of their allotted time that they spent
replaying versus analyzing the event.
Because the organized and systematic nature of both writing and talking
lends itself to greater step-by-step analysis and reduced repetitive, chaotic
STUDY 3 replaying, we expected participants who were induced to write or talk to
engage in increasingly more analysis and less replaying of the experience
Our next step was to investigate the mechanisms underlying the over the course of the 3-day induction. Supporting this argument, repeated
effects of processing positive events—that is, repetitive replay measures analyses revealed that the amount of time spent analyzing in-
versus systematic analysis. To this end, in Study 3, we examined creased over the course of the 3 days for the writing group in both
the role of replaying versus analysis in producing differential studies—Study 1, F(1, 18) ⫽ 34.91, p ⬍ .0001, ␩ ⫽ .81; Study 2, F(1,
outcomes for long-term PA, aspects of psychological well-being, 14) ⫽ 14.86, p ⬍ .001, ␩ ⫽ .72—whereas replaying decreased—Study 1,
and physical health by using an experimental controlled design.4 F(1, 18) ⫽ 34.17, p ⬍ .0001, ␩ ⫽ .81; Study 2, F(1, 14) ⫽ 14.86, p ⬍ .001,
On the basis of previous research, we hypothesized that those who ␩ ⫽ .72. Students who talked out loud showed similar—though not all
wrote about a positive life experience while systematically ana- significant—trends, such that replaying decreased over the 3-day period—
lyzing it would show the least favorable outcomes of all our Study 1, F(1, 10) ⫽ 4.00, p ⫽ .07, ␩ ⫽ .53; Study 2, F(1, 8) ⫽ 21.31, p ⬍
.001, ␩ ⫽ .85—whereas analyzing generally increased—Study 1, F ⬍ 2;
groups, whereas those who thought about a positive life experience
Study 2, F(1, 8) ⫽ 13.83, p ⬍ .01, ␩ ⫽ .80.
while repetitively replaying it would show the most favorable When asked what percentage of the designated 15 min they spent
outcomes. replaying versus analyzing their happy or unhappy life event, proxy (1-day)
participants reported in both studies that they had spent a little more than
half the time (M ⫽ 53.6% in Study 1; M ⫽ 54.1% in Study 2) replaying and
Method a little less than half the time analyzing (M ⫽ 46.2% in Study 1; M ⫽
44.5% in Study 2).
Participants Finally, we tested whether the more analysis students engaged in during
writing and talking, the more beneficial outcomes they accrued 4 weeks
One hundred twelve (65 female, 47 male) undergraduate students at the later. In Study 1, regression analyses using data from the writing group
University of California, Riverside, participated in this study. Over half of revealed that greater percentages of time students were coded as system-
the participants were of Asian (32%) or Latino(a) (31%) descent, 21% were atically analyzing their negative experience (and the less time they spent
Caucasian, 6% were African American, and 10% identified themselves as replaying) were associated with higher transient PA, F(1, 18) ⫽ 5.32, p ⫽
“other.” Students ranged in age from 17 to 29 years (M ⫽ 19.23 years, .03, ␩ ⫽ .48; lower NA, F(1, 18) ⫽ 8.44, p ⬍ .009, ␩ ⫽ .56; better overall
SD ⫽ 2.03 years). Participants were again matched across conditions for mental health, F(1, 18) ⫽ 7.65, p ⫽ .01, ␩ ⫽ .55; better overall health, F(1,
gender. Four additional participants failed to complete all the sessions of 18) ⫽ 9.29, p ⬍ .01, ␩ ⫽ .58; and better health perceptions, F(1, 18) ⫽
the study and were not included in the analyses. 9.61, p ⬍ .01, ␩ ⫽ .59. These findings support the premise that system-
atically analyzing (i.e., writing about) a negative life experience is more
Procedure beneficial than replaying or repetitively processing the experience (as when
thinking about it). Unfortunately, possibly because of technical and coding
In this study, half of the participants were randomly assigned to think difficulties, the same pattern of results was not observed in codings of the
about one of their happiest days, and half were assigned to write about it. speech samples generated in the talking condition. In Study 2, neither
Furthermore, half were induced to repetitively replay the day while think- replaying nor analysis was found to be related to the outcome variables
ing or writing, and half were induced to systematically analyze it. Conse- when participants processed a happy experience by writing or talking about
quently, four experimental conditions were created—that is, thinking while it (all Fs ⬍ 2, ns).
Outcome Measures own leisure and to skip over any questions that you feel do not
particularly pertain to your experience.
Long-Term PA
Once participants confirmed that they understood these instructions, they
In place of the SWLS, we used a measure of long-term, or dispositional,
were left to contemplate the following statement and questions, which were
PA. Long-term PA is assessed with the PANAS (Watson et al., 1988), but
printed on five separate pages:
instead of rating their current experience, participants were instructed to
rate the extent they had “felt this way during the last 3 months.” The
Remember exactly what happened in as much detail as you can.
reliabilities for this scale were very good (␣s ⫽ .84 and .89 for long-term
Think about: What exactly did you do or say?
PA, and ␣s ⫽ .88 and .89 for long-term NA). In previous studies, long-term
Think about: If another person (or people) were involved, what
PA has correlated strongly with the SWLS, with rs ranging from the .30s
exactly did they do or say?
to the .60s (e.g., Tkach, 2005).
Think about: How did you feel about this experience at the time it
occurred?
Psychological Well-Being Think about: How did you feel about yourself following this
experience?
For a multidimensional assessment of well-being, we implemented the
Psychological Well-Being Scale (PWB; Ryff, 1989), an 84-item assess- Participants returned to these instructions on Day 2 and Day 3.
ment of positive functioning (␣s ⫽ .96 and .98 for Time 1 and Time 2, Think–analyze condition. In this condition, participants were asked to
respectively). The PWB consists of six major subscales—Autonomy, En- think about their positive life experience by systematically analyzing their
vironmental Mastery, Personal Growth, Positive Relations With Others, thoughts. The instructional packet they received was identical to that used
Purpose in Life, and Self-Acceptance. All PWB items are rated on the same in the think–replay condition, except for the first few sentences, which read
6-point Likert-type scale (1 ⫽ strongly disagree, 6 ⫽ strongly agree). as follows:
The Autonomy subscale of the PWB consists of such items as “My
decisions are not usually influenced by what everyone else is doing.” An Please spend the next 8 minutes privately thinking about the day you
example of the Mastery subscale is “I am quite good at managing the many described on the previous page. Please try to analyze your thoughts by
responsibilities of my daily life.” The Personal Growth subscale asks such doing a step-by-step breakdown of the events of the day from the
questions as “For me, life has been a continuous process of learning, beginning to the end. Ask yourself such questions as “why” or “how”
changing, and growth.” An example of the Positive Relations With Others did this experience happen to me. Use your time thinking about the
subscale is “I feel like I get a lot out of my friendships.” The Purpose in experience in an effort to understand it and to make sense of why this
Life subscale contains such items as “I have a sense of direction and experience might have happened to you.
purpose in life.” Finally, the Self-Acceptance subscale includes such state-
ments as “In general, I feel confident and positive about myself.” In the
Next, participants in this condition were left to contemplate the following
past, the PWB has displayed good psychometric properties, with internal
statement and questions, printed on five separate pages:
consistency for all six subscales ranging between .86 and .93 (Ryff, 1989).
Analyze the events of the day in a step-by-step manner from begin-
Physical Health ning to end, with an emphasis on “why” and “how” this experience
happened to you.
With the exception of the Symptoms Checklist (which was excluded), Think about: In what ways could this happy event be viewed
the same measures of health (␣s ⫽ .88 [Time 1] and .89 [Time 2] for the differently?
Health Survey) used in the first two studies were administered in Study 3. Think about: What could you have done differently to change the
course of the event?
Experimental Manipulation Think about: What would you do or say differently if you could
reexperience this event again?
Happiest Day Form Think about: Do you think others would feel the same way if they
experienced the same thing you did? Explain your response.
After identifying one of the happiest days they had ever experienced,
participants completed the same Life Experience Questionnaire used in the Write–replay condition. In this condition, participants were asked to
previous two studies. write about their positive life experience by replaying the experience over
and over again as though rewinding a tape and replaying it. Students were
Experimental Conditions given a standard blue book to record their writing and were asked to return
to this blue book on each of the 3 consecutive days of their participation.
Next, participants were exposed to one of four experimental manipula- In addition to receiving this “journal,” participants were provided an
tions for exactly 8 min a day over a 3-day period. instructional packet that was identical to that of the think–replay condition
Think–replay condition. In this condition, students were asked to sim- save the one novel instruction to write instead of think.
ply think about their positive life experience strictly by replaying their Write–analyze condition. In this condition, participants were asked to
thoughts over and over again. Prior to beginning their first session, partic- write about their positive life experience by systematically analyzing the
ipants were given an instructional packet containing the following event. Again, students were given blue books to record their writing and
instructions: were referred to an instructional packet identical to that of the think–
analyze condition except for the one novel instruction to write instead of
Please spend the next 8 minutes privately thinking about the day you think.
described on the previous page. Please replay these thoughts as
though you were rewinding a videotape and playing it back. Think
about the events of the day with an emphasis on what happened, how Follow-Up
you were feeling at the time, and how you behaved. The following
pages will contain questions/instructions that should help you through Four weeks subsequent to their participation in one of the four condi-
this thought process. . . . Feel free to move from page to page at your tions, all participants returned to the laboratory to complete their final day
of participation in the study. Upon arrival, participants were asked to the most time focused on the task (M ⫽ 7.38 min), followed by the
complete the same packet of outcome measures they received on Day 1. write–replay group (M ⫽ 7.33 min), the think–analyze group (M ⫽
5.55 min), and then the think–replay group (M ⫽ 5.47 min). The
Results act of writing—a systematic and organized process—may have
kept students more focused on the task than did the act of think-
Overview of Statistical Analyses ing—a loose and somewhat chaotic process that can lead to mind
wandering.
In this study, participants’ long-term affect, psychological well-
With regard to reported difficulty staying on task, participants
being, and health were assessed before and after they processed
on average indicated that it was fairly easy to concentrate on the
one of their happiest moments. Similar to the first two studies, the
task (M ⫽ 2.83 on a 7-point scale). Post hoc analyses again
extent to which participants’ scores on these variables changed as
revealed significant group differences on this variable, F(3, 108) ⫽
a result of exposure to one of our four conditions was examined by
7.67, p ⬍ .001, ␩ ⫽ .26. That is, the think–analyze group had the
analyzing Time 2 data with Time 1 as a covariate.
most difficulty staying on task (M ⫽ 3.66), followed by the
The reader will recall that, out of all our groups, students in the
think–replay group (M ⫽ 3.30), the write–replay group (M ⫽
think–replay group were expected to reap the greatest benefits for
2.30), and then the write–analyze group (M ⫽ 1.92). This finding
their happiness and health, whereas those in the write–analyze
may again be due to the more focused nature of writing as
group were expected to show the poorest results. Although, theo-
compared with thinking.
retically, replaying should produce relatively favorable outcomes,
In light of these differences in each group’s ability to maintain
and analysis should produce unfavorable outcomes, no matter what
focus, we conducted covariance analyses by using both time spent
the mode of processing (i.e., writing, talking, thinking), we could
on task and difficulty staying on task each day as a covariate. The
not make clear predictions regarding the outcomes of the two
results of those analyses were virtually identical to our original
mixed conditions—namely, write–replay and think–analyze—as
results.
each involved two manipulations predicted to have opposite ef-
fects on happiness and health.
To test our primary hypotheses, we performed planned contrasts Condition Compliance
(with Time 1 variables as a covariate) by comparing the write– To verify that our participants had complied with our instruc-
analyze group and the think–replay group, respectively, to the tions to either replay or analyze their happiest day, three judges
remaining three conditions. These two hypotheses combined led to subsequently coded the narratives generated by participants in the
the prediction that writers who analyzed would benefit signifi- two writing conditions. For obvious reasons, participants’ thoughts
cantly less from our study than thinkers who replayed. This hy- in the two thinking conditions could not be similarly coded.
pothesis was tested by performing a planned pairwise comparison In sum, participants’ written essays were coded to determine the
directly between the write–analyze group and the think–replay percentage of repetitive replay versus systematic analysis that
group. actually appeared in the writing (␣ ⫽ .84; see Footnote 4 for
coders’ instructions). The average of judges’ ratings indicated that
Ratings of Positive Life Experiences participants in the replay condition engaged in replaying 69% of
the time and engaged in analysis 31% of the time, whereas those
Examples of happiest days named by participants included a day in the analyze condition engaged in analysis 57% of the time and
spent with the object of one’s love, the day of one’s high school in replaying 43% of the time. This group difference was statisti-
graduation, and a thrilling vacation day. On average, students cally significant, t(51) ⫽ 3.83, p ⬍ .001, ␩ ⫽ .26.
indicated that their happy day had occurred somewhere between 3
months and 6 months ago (M ⫽ 4.47 on a 10-point scale) and that
Outcome Measures
the event had been highly significant to them (M ⫽ 8.93 on a
9-point scale). Furthermore, participants indicated that they had Again, the relevant statistics for all outcome variables are pre-
previously spent a fair amount of time thinking about the positive sented in a single table (see Table 3).
experience (M ⫽ 6.69 on a 10-point scale) and relatively little time
writing about it (M ⫽ 3.28 on a 10-point scale). As expected, a Long-Term PA
series of Tukey’s HSD post hoc analyses revealed no significant
differences among the four experimental groups for any of these A planned contrast revealed that the think–replay group had the
variables prior to manipulation. highest long-term PA relative to all the other groups (Ms ⫽ 3.63
vs. 3.41), F(1, 107) ⫽ 2.22, p ⫽ .07, ␩ ⫽ .14. This finding is
consistent with our hypothesis that repetitively replaying one’s
Manipulation Checks
happiest life experience while privately thinking about it would
Staying on Task maintain the positive emotions that surround it, thus producing the
greatest benefits for an individual’s well-being. Other planned
Analyses were also conducted to establish group differences in comparisons with the affect variable failed to reach conventional
participants’ ability to stay on task during the timed manipulation. levels of statistical significance.
Overall, students reported that they spent over 6 of the 8 min
allotted to them each day focused on the task at hand (M ⫽ 6.38
Psychological Well-Being
min out of 8). Post hoc analyses did reveal significant differences
among the four experimental groups on this variable, F(3, 108) ⫽ Participants who wrote while analyzing their happiest day were
21.97, p ⬍ .001, ␩ ⫽ .41. That is, the write–analyze group spent expected to report decreased psychological well-being, whereas
Table 3
Marginal Means (Adjusted for Time 1) and Standard Errors for Processing Positive Experiences by Condition (Study 3)
Condition
Think–replay Think–analyze Write–replay Write–analyze

(n ⫽ 30) (n ⫽ 29) (n ⫽ 27) (n ⫽ 26)
Long-term PA 3.63a 0.12 3.43b 0.12 3.39b 0.12 3.42b 0.12

Long-term NA 2.32a 0.13 2.26a 0.13 2.30a 0.13 2.18a 0.13
Overall PWB 4.52a 0.07 4.61a 0.07 4.59a 0.07 4.47a 0.07
Autonomy 4.23a 0.08 4.36a 0.08 4.35a 0.09 4.37a 0.09
Positive Relations With Others 4.55a 0.09 4.64a 0.09 4.66a 0.09 4.61a 0.09
Purpose in Life 4.71a 0.10 4.78a 0.10 4.72a 0.10 4.65a 0.10
Personal Growth 5.03a 0.09 5.02a 0.09 4.94a 0.10 4.76b 0.10
Self-Acceptance 4.36a 0.10 4.54a 0.10 4.55a 0.11 4.22b 0.11
Environmental Mastery 4.11a 0.10 4.40b 0.10 4.30b 0.11 4.30b 0.11
Overall health 82.03a 1.34 82.57a 1.36 80.82a 1.41 79.06b 1.44
Physical Functioning 97.47a 1.34 97.38a 1.36 96.65a 1.41 93.95b 1.43
Pain 76.90a 3.54 78.06a 3.54 71.07a 3.67 64.90b 3.75
Social Functioning 88.71a 4.60 92.42a 4.66 90.38a 4.84 86.86a 4.94
Mental Health 64.95a 3.17 66.18a 3.20 64.80a 3.32 64.87a 3.39
Health Perception 74.45a 2.40 74.35a 2.42 74.91a 2.51 72.03a 2.56
Note. Except for NA, high scores indicate increasing functioning. Means for three groups that share subscripts (combined) are significantly different from
the mean of the fourth group (with different subscript). The means of the think–replay group and the write–analyze group are significantly different for
(a) personal growth, (b) overall health, (c) physical functioning, and (d) pain. PA ⫽ positive affect; NA ⫽ negative affect; PWB ⫽ Psychological
Well-Being Scale.
those who thought while replaying this day were expected to show analyzing it, and the best general health was expected for those
increased well-being. However, these predictions were not sup- who thought privately about their happiest day while replaying it.
ported for our overall measure of psychological well-being, and Supporting this hypothesis, a planned contrast, with Time 1 health
analyses involving the six subscales proved to have mixed results. as a covariate, as measured by the overall Health Survey, revealed
That is, two of the subscales for PWB (Personal Growth and significantly lower health at Time 2 for the write–analyze group
Self-Acceptance) did provide support for our propositions, relative to the think–replay, think–analyze, and write–replay
whereas one subscale (Environmental Mastery) revealed an unex- groups combined (Ms ⫽ 79.06 vs. 81.83), F(1, 107) ⫽ 2.90, p ⫽
pected finding. No group differences were found for Autonomy, .05, ␩ ⫽ .16. Yet no significant differences in health were found
Purpose in Life, and Positive Relations With Others. when comparing the think–replay group with the remaining three
Personal growth. As shown in Table 3, a planned contrast, conditions (F ⬍ 1, ns). However, when compared directly with one
with Time 1 personal growth as a covariate, indicated that the another, the write–analyze group showed marginally significant
write–analyze group reported less personal growth relative to the decline in health relative to the think–replay group, F(1, 107) ⫽
three other groups combined (Ms ⫽ 4.76 vs. 5.00), F(1, 107) ⫽ 2.32, p ⫽ .07, ␩ ⫽ .15.
4.59, p ⫽ .02, ␩ ⫽ .20. No significant group differences were Because the measure of health used in this study comprised six
found when the think–replay condition was compared with the subscales, further analyses could examine more specific aspects of
remaining three conditions, F(1, 107) ⫽ 1.42, ns. However, writers general health. These analyses revealed that several of our hypoth-
who analyzed reported decreased personal growth relative to eses were supported for two of these subscales—Physical Func-
thinkers who replayed, F(1, 107) ⫽ 4.22, p ⫽ .02, ␩ ⫽ .19.
tioning and Pain. No significant group differences were found for
Self-acceptance. A planned contrast, with Time 1 self-
the remaining four subscales—Health Perceptions, Role Function-
acceptance as a covariate, showed that participants in the write–
ing, Social Functioning, and Mental Health (see Table 3).
analyze group reported less self-acceptance relative to the remain-
Physical functioning. Planned contrasts with Time 1 physical
ing groups (Ms ⫽ 4.22 vs. 4.48), F(1, 107) ⫽ 4.76, p ⫽ .02, ␩ ⫽
functioning as a covariate revealed that the write–analyze group
.21. No other significant group differences emerged (Fs ⬍ 1).
reported decreased physical functioning relative to the other
Environmental mastery. Contrary to our predictions, the
groups combined (Ms ⫽ 93.95 vs. 97.18), F(1, 107) ⫽ 3.90, p ⫽
think–replay group reported marginally significant decreases in
.03, ␩ ⫽ .19, but the think–replay group did not report increased
environmental mastery relative to the other groups (Ms ⫽ 4.11 vs.
4.33), F(1, 107) ⫽ ⫺3.12, p ⫽ .08, ␩ ⫽ ⫺.17. physical functioning relative to the other groups (F ⬍ 1, ns).
Finally, participants who wrote while analyzing reported dimin-
ished physical functioning relative to those who thought while
Health Survey
replaying, F(1, 107) ⫽ 3.26, p ⫽ .04, ␩ ⫽ .17.
Overall health. The poorest general health at Time 2 was Pain. In addition to finding some support for our hypotheses
expected for participants who wrote about their happiest day while regarding reported physical functioning, we found that the write–
analyze group also reported more pain relative to the think–replay, sense of why and how a fortunate circumstance has befallen them.
think–analyze, and write–replay groups (Ms ⫽ 64.90 vs. 75.30), Instead, perhaps people should simply be content to relive and
F(1, 107) ⫽ 5.90, p ⬍ .01, ␩ ⫽ .23, and the think–replay group savor these experiences without trying to determine their causes or
reported less pain relative to the other three groups (Ms ⫽ 76.90 meanings (Bryant, 2003; Wilson et al., 2005). Additional work is
vs. 71.34), F(1, 107) ⫽ 1.85, p ⫽ .09, ␩ ⫽ .13, although this needed to further bear out and clarify this interesting phenomenon.
difference was nonsignificant. Finally, writers who analyzed their
happiest moments reported greater pain than thinkers who re- GENERAL DISCUSSION
played those moments, F(1, 107) ⫽ 5.42, p ⫽ .01, ␩ ⫽ .22.
Study 1: Negative Life Experiences
Discussion We hypothesized that students who wrote or talked into a tape
recorder about their worst life experience would report improved
We expected individuals who wrote analytically about one of life satisfaction and health relative to participants who thought
their happiest life moments to report less favorable outcomes (e.g., privately about the experience. This hypothesis was generally
reduced affect, well-being, and health) relative to those who supported. The only outcome variable that did not show this
thought repetitively about such experiences. Modest support for pattern of results was transient affect, which may fluctuate too
our specific predictions was found for several of the outcome much to expect stable changes 4 weeks after the induction.
variables, including long-term PA, personal growth, self- Previous research on the relationship between cognition and
acceptance, overall health, physical functioning, and pain. health has provided evidence for substantial benefits of writing and
First, supporting our hypothesis, relative to the other conditions, talking about traumatic events (DeLongis et al., 1988; Murray &
our results showed that thinking while replaying one’s happiest Segal, 1994; Pennebaker, 1993). This body of work has generated
day yielded significantly enhanced long-term PA and, to some speculation concerning the mechanisms by which writing and
extent, decreased pain. This finding is consistent with the notion talking about one’s experiences lead to positive outcomes. For
that repetitive, cyclical thoughts about happy experiences maintain example, Pennebaker has suggested that writing and talking re-
positive emotions. Second, as predicted, writing while analyzing quire organization and structure, which allow individuals to let go
led to reduced personal growth, self-acceptance, general health, and move past their problems, resulting in improved mental and
and physical functioning and to increased reported pain. Finally, physical health (Pennebaker, 1993; Pennebaker & Francis, 1996).
thinking while replaying one’s happiest moment was associated Our first study not only provided evidence for this assumption, but
with superior personal growth, general health, and physical func- went beyond earlier research by directly comparing writing and
tioning—and with lower pain—than writing while analyzing that talking (which presumably involve analytical processing of a trau-
moment. matic experience) with private thought (which we assume to
It is interesting to note one finding that was unexpected and involve repetition or replaying). Thinking was predicted to be
contrary to our predictions: Students induced to think while re- detrimental to well-being and health because it can quickly and
playing their happiest moments actually reported decreased feel- easily degrade into negative repetitive cognitions that are relatively
ings of environmental mastery relative to the other groups. This more difficult to integrate, condemning the person to the reexpe-
finding is particularly difficult to interpret given that thinking rience and maintenance of painful memories. The findings of
while replaying simultaneously produced increased long-term PA Study 1 were consistent with this position, supporting the argu-
relative to these same groups. Perhaps repetitively replaying happy ment that the inherently organized nature of writing and talking
memories in one’s mind helps to maintain or increase the positive about one’s thoughts about a traumatic life experience may pro-
feelings associated with those events, but also impairs one’s ability vide a way for participants to accept their experiences and record
to self-regulate. As previously discussed, self-regulation is one of them in an external fashion, thus allowing them to move beyond
the proposed mechanisms thought to account for the benefits of their troubles and ultimately resulting in favorable outcomes.
trauma writing. Conceivably, then, it is this lack of self-regulatory
processing— occurring when people are thinking and especially
when they are thinking while repetitively replaying—that may Positive Life Experiences
have impaired rather than improved at least one aspect of well-
Study 2: Writing and Talking Versus Thinking
being (i.e., environmental mastery). Additionally, because having
a sense of mastery often involves analysis of one’s performance In Study 2, we hypothesized that the organized and systematic
and potential for improvement, this sense may be inhibited by nature of writing and talking may be somewhat incompatible with
unsystematic replaying. Further research is needed to test these the maintenance of PA and might therefore interfere with the
arguments empirically. potential benefits associated with the experience of positive emo-
In conclusion, the results of Study 3 provide some evidence that tion (Fredrickson, 2001). These hypotheses were supported in this
writing while systematically analyzing positive experiences is study for satisfaction with life. Participants who thought privately
more deleterious to well-being and health than thinking while about their happiest experience reported greater life satisfaction
repetitively replaying these experiences. These findings suggest than participants who wrote or talked about such an experience.
that breaking down one’s happiest moments through writing—a It may be too early in this research endeavor to understand
process that seems to create a detailed narrative analysis—should completely why thinking about a positive event would improve life
perhaps be avoided, even though the very same process appears to satisfaction but not show similar outcomes for physical health.
be useful when applied to negative life events. Furthermore, when However, a consideration of the construct of life satisfaction may
it comes to happy life experiences, our modest evidence indicates provide some clues. The assessment of satisfaction with life in-
that perhaps people ought not to overanalyze or attempt to make volves measures that tap beliefs and perceptions of one’s own life
and one’s emotions and feelings. Writing and talking (analytical ably serve to increase error variance and thus would tend to
processes) about a positive experience may lead one down a path obscure differences that might be seen more clearly with more
of evaluation and seeking meaning that is incompatible with sus- discriminating measures. However, future research would be
taining positive emotion and, therefore, incompatible with positive strengthened by using other means of assessment, including phys-
beliefs about oneself and one’s circumstances. That is, because life iological measures (e.g., heart rate, blood pressure, skin conduc-
satisfaction comprises one’s beliefs and feelings, this measure may tance, white blood cell count, and antibody and cortisol levels) and
be more sensitive to any changes one might experience from behavioral measures (e.g., number of doctor visits and direct
processing experiences in systematic and analytical ways. In con- observations of activity level) as well as reports from knowledge-
trast, the measures of physical health and health symptoms are able informants.
relatively more likely to tap behavior and physiology and, thus, are Another concern is about the generalizability of our results
perhaps less sensitive to experimental manipulation when writing, beyond a student sample. The average age of our participants in
talking, and thinking about positive experiences are involved. The these three studies was approximately 19 years. The relatively
lack of specific instructions about exactly how participants should modest life experience of this age group may have influenced the
write, talk, and think might contribute to this relative lack of strength of the effects, making it relatively difficult to predict the
sensitivity. Future research using more finely tuned physiological impact of the different ways of event processing on older adults
and behavioral measures is needed to further clarify these findings. who may have endured greater hardships (e.g., poor health, di-
vorce, or financial problems) or greater joys (e.g., wedding, birth
Study 3: Analysis Versus Replaying of a child, or career success). However, experiences and the
emotions surrounding them are relative. One’s experience of an
Our first two studies raised the question of mechanisms, and event is anchored by one’s prior experiences. It could be argued
Study 3 aimed to address this question. Specifically, Study 3 that a first love could be just as thrilling and significant as getting
manipulated the precise ways that people write or think about married and having a child might be to someone else with more
significant events. We found modest support for the hypothesis maturity and resources.
that individuals induced to write about a happy life event in an The present studies have begun to establish some differential
analytic fashion would report poorer well-being and physical benefits and drawbacks of processing positive and negative
health relative to those induced to process the same event by events—life’s victories and defeats. However, an investigation that
simply rehearsing and replaying it in their thoughts. These findings compares the processing of positive and negative experiences in a
suggest that when considering positive experiences, it is the inher- single study would be needed to answer a variety of questions. For
ently organized, narrative-based, and analytic nature of writing example, is writing analytically about a trauma a superior practice
that may be counterproductive to well-being and health. In con- than thinking repetitively about a triumph? Is analyzing a positive
trast, at least with respect to the maintenance of PA, the repetitive, event more harmful to one’s well-being and functioning than
circular, inherently unorganized nature of thought is beneficial. replaying a negative one? Answers to these questions may provide
Recalling a wonderful moment—a triumph, a creative break- practical suggestions concerning the healing effects of managing
through, a first declaration of love— by reliving it through re- traumatic versus joyful events.
hearsal and replaying may allow the individual to savor (Bryant,
1989, 2003) and capitalize on (Gable et al., 2004; Langston, 1994) Concluding Remarks
the experience. Furthermore, reviewing a happy event—without
sense making or analysis— helps the person retain a sense of Our research suggests that systematic step-by-step analysis
mystery and thrill about the event (Wilson et al., 2005; Wilson & (which presumably tends to occur while writing or speaking) is
Gilbert, 2003). In sum, we believe these processes act to maintain worthwhile and beneficial when directed at unhappy, stressful, or
and bolster positive emotions, which contribute to a person’s traumatic life events, but may be harmful when applied to happy
overall physical and emotional well-being (Fredrickson, 2001). times. In contrast, repetitive, circular replaying (which presumably
Achieving a sense of meaning, resolution, and control over is inclined to occur during private thought) is somewhat advanta-
significant past life experiences has long been held to have enor- geous when the target is one’s highest moment, but may be
mous value. Our research introduces a potential caveat to this damaging when the target is one’s lowest ebb. Prior research has
recommendation—that is, that it may apply only to those experi- almost exclusively focused on how people process negative expe-
ences that are negative. As G. K. Chesterton (1905) asserted, riences and has generated a remarkably consistent pattern of re-
“Happiness is a mystery like religion, and should never be ratio- sults. By contrast, our findings highlight how a positive psycho-
nalized” (p. 103). Future studies that replicate and advance our logical approach can reveal previously undetected mechanisms
findings would be much desired. underlying the existing evidence and, we hope, galvanize further
research.
Limitations and Future Questions Both negative and positive life experiences undoubtedly greatly
impact people’s happiness, mental health, and physical well-being.
Several limitations related to design and data collection should However, this impact is rarely long-term, in large part due to
be mentioned. Because all of our outcomes in our three studies hedonic adaptation (Frederick & Loewenstein, 1999), but also due
were measured via self-report, we must be vigilant of social to people’s remarkable capacity to cope with and process the
desirability and response biases. However, we believe that the events and circumstances of their lives. A greater understanding of
participants themselves are our best resources for information how best to process both one’s worst hours and one’s most
about their own emotions and cognitions. It is also worth noting wondrous moments, how to optimize their impact on well-being
that any weakness or unreliability in our measures would presum- and health, and how to be better for having experienced them
would be a valuable contribution. The present research represents Fredrickson, B. L., Mancuso, R. A., Branigan, C., & Tugade, M. M.
a first step in this direction. (2000). The undoing effect of positive emotions. Motivation and Emo-
tion, 24, 237–258.
References Fredrickson, B. L., Tugade, M. M., Waugh, C. E., & Larkin, G. R. (2003).
What good are positive emotions in crisis? A prospective study of
Aspinwall, L. G., & Brunhart, S. M. (1996). Distinguishing optimism from resilience and emotions following the terrorist attacks on the United
denial: Optimistic beliefs predict attention to health threats. Personality States on September 11th, 2001. Journal of Personality and Social
and Social Psychology Bulletin, 22, 993–1003. Psychology, 84, 365–376.
Beck, A. T. (1967). Depression: Clinical, experimental, and theoretical Gable, S. L., Reis, H. T., Impett, E. A., & Asher, E. R. (2004). What do you
aspects. New York: Harper & Row. do when things go right? The intrapersonal and interpersonal benefits of
Blagov, P. S., & Singer, J. A. (2004). Four dimensions of self-defining sharing positive events. Journal of Personality and Social Psychology,
memories (specificity, meaning, content, and affect) and their relation-
87, 228 –245.
ships to self-restraint, distress, and repressive defensiveness. Journal of
Goldman, S. L., Kraemer, D. T., & Salovey, P. (1996). Beliefs about mood
Personality, 72, 481–511.
moderate the relationship of stress to illness and symptom reporting.
Bryant, F. B. (1989). A four-factor model of perceived control: Avoiding,
Journal of Psychosomatic Research, 41, 115–128.
coping, obtaining, and savoring. Journal of Personality, 57, 773–797.
Greenberg, M. A., & Stone, A. A. (1992). Emotional disclosure about
Bryant, F. B. (2003). Savoring Beliefs Inventory (SBI): A scale for mea-
traumas and its relation to health: Effects of previous disclosure and
suring beliefs about savoring. Journal of Mental Health, 12, 175–196.
trauma severity. Journal of Personality and Social Psychology, 63,
Burton, C. M., & King, L. A. (2004). The health benefits of writing about
intensely positive experiences. Journal of Research in Personality, 38, 75– 84.
150 –163. Greenberg, M. A., Wortman, C. B., & Stone, A. A. (1996). Emotional
Chesterton, G. K. (1905). Heretics. London: Butler and Tanner. expression and physical health: Revising traumatic memories or foster-
DeLongis, A., Folkman, S., & Lazarus, R. S. (1988). The impact of daily ing self-regulation? Journal of Personality and Social Psychology, 71,
stress on health and mood: Psychological and social resources as medi- 588 – 602.
ators. Journal of Personality and Social Psychology, 54, 486 – 495. Hixon, J. G., & Swann, W. B. (1993). When does introspection bear fruit?
Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The Self-reflection, self-insight, and interpersonal choices. Journal of Per-
Satisfaction With Life Scale. Journal of Personality Assessment, 49, sonality and Social Psychology, 64, 34 – 43.
71–75. Holahan, C. J., & Moos, R. H. (1985). Life stress and health: Personality,
Dominguez, B., Valderrama, P., Meza, M. A., Perea, S. L., Silva, A., coping, and family support in stress resistance. Journal of Personality
Martinez, G., et al. (1995). The roles of emotional reversal and disclo- and Social Psychology, 49, 739 –747.
sure in clinical practice. In J. W. Pennebaker (Ed.), Emotion, disclosure, Horowitz, M. J., Field, N. P., & Classen, C. C. (1993). Stress response
and health (pp. 255–270). Washington, DC: American Psychological syndromes and their treatment. In L. Goldberger & S. Breznitz (Eds.),
Association. Handbook of stress (2nd ed., pp. 757–773). New York: Free Press.
Donnelly, D. A., & Murray, E. J. (1991). Cognitive and emotional changes Hughes, C. F., Uhlmann, C., & Pennebaker, J. W. (1994). The body’s
in written essays and therapy interviews. Journal of Social and Clinical response to processing emotional trauma: Linking verbal text with
Psychology, 10, 334 –350. autonomic activity. Journal of Personality, 62, 565–585.
Emmons, R. A., & McCullough, M. E. (2003). Counting blessings versus Irving, L. M., Snyder, C. R., & Crowson, J. J. (1998). Hope and coping
burdens: An experimental investigation of gratitude and subjective well- with cancer by college women. Journal of Personality, 66, 195–214.
being in daily life. Journal of Personality and Social Psychology, 84, King, L. A. (2001). The health benefits of writing about life goals.
377–389. Personality and Social Psychology Bulletin, 27, 798 – 807.
Esterling, B. A., L’Abate, L., Murray, E. J., & Pennebaker, J. W. (1999). King, L. A., & Miner, K. N. (2000). Writing about the perceived benefits
Empirical foundations for writing in prevention and psychotherapy: of traumatic events: Implications for physical health. Personality and
Mental and physical health outcomes. Clinical Psychology Review, 19, Social Psychology Bulletin, 26, 220 –230.
79 –96. Langston, C. A. (1994). Capitalizing on and coping with daily-life events:
Foa, E., & Riggs, D. (1993). Post-traumatic stress disorder in rape victims. Expressive responses to positive events. Journal of Personality and
In J. Oldham, M. B. Riba, & A. Tasman (Eds.), American Psychiatric
Social Psychology, 67, 1112–1125.
Press review of psychiatry (Vol. 12, pp. 273–303). Washington, DC:
Lyubomirsky, S., King, L. A., & Diener, E. (2005). The benefits of
American Psychiatric Press.
frequent positive affect: Does happiness lead to success? Psychological
Frattaroli, J. (in press). Experimental disclosure and its moderators: A
Bulletin, 131, 803– 855.
meta-analysis. Psychological Bulletin.
Lyubomirsky, S., & Nolen-Hoeksema, S. (1995). Effects of self-focused
Fredrick, S., & Loewenstein, G. (1999). Hedonic adaptation. In D. Kah-
rumination on negative thinking and interpersonal problem solving.
neman, E. Diener, & N. Schwarz (Eds.), Well-being: The foundations of
hedonic psychology (pp. 302–329). New York: Russell Sage Foundation. Journal of Personality and Social Psychology, 69, 176 –190.
Fredrickson, B. L. (2000, March 7). Cultivating positive emotions to Lyubomirsky, S., Sheldon, K. M., & Schkade, D. (2005). Pursuing happi-
optimize health and well-being. Prevention & Treatment, 3, Article ness: The architecture of sustainable change. Review of General Psy-
0001a. Retrieved March 18, 2002, from http://journals.apa.org/ chology, 9, 111–131.
prevention/volume3/pre0030001a.html Lyubomirsky, S., & Tkach, C. (2004). The consequences of dysphoric
Fredrickson, B. L. (2001). The role of positive emotions in positive rumination. In C. Papageorgiou & A. Wells (Eds.), Rumination: Nature,
psychology: The broaden-and-build theory of positive emotions. Amer- theory, and treatment of negative thinking in depression (pp. 21– 41).
ican Psychologist, 56, 218 –226. Chichester, England: Wiley.
Fredrickson, B. L., & Joiner, T. (2002). Positive emotions trigger upward Martin, L. L., & Tesser, A. (1989). Toward a motivational and structural
spirals toward emotional well-being. Psychological Science, 13, 172– theory of ruminative thought. In J. S. Uleman & J. A. Bargh (Eds.),
175. Unintended thought (pp. 306 –326). New York: Guilford Press.
Fredrickson, B. L., & Levenson, R. W. (1998). Positive emotions speed Murray, E. J., & Segal, D. L. (1994). Emotional processing in vocal and
recovery from the cardiovascular sequelae of negative emotions. Cog- written expression of feelings about traumatic experiences. Journal of
nition & Emotion, 12, 191–220. Traumatic Stress, 7, 391– 405.
Nolen-Hoeksema, S. (2003). Women who think too much: How to break traumatic experiences: The necessity for narrative structuring. Journal of
free of overthinking and reclaim your life. New York: Henry Holt. Social and Clinical Psychology, 20, 161–172.
Nolen-Hoeksema, S., McBride, A., & Larson, J. (1997). Rumination and Stanton, A. L., Kirk, S. B., Cameron, C. L., & Danoff-Burg, S. (2000).
psychological distress among bereaved partners. Journal of Personality Coping through emotional approach: Scale construction and validation.
and Social Psychology, 72, 855– 862. Journal of Personality and Social Psychology, 78, 1150 –1169.
Nolen-Hoeksema, S., & Morrow, J. (1991). A prospective study of depres- Stewart, A. L., Hays, R. D., & Ware, J. E. (1988). The MOS short-form
sion and posttraumatic stress symptoms after a natural disaster: The general health survey: Reliability and validity in a patient population.
1989 Loma Prieta earthquake. Journal of Personality and Social Psy- Medical Care, 26, 724 –735.
chology, 61, 115–121. Stone, A. A., Cox, D. S., Valdimarsdottir, H., Jandorf, L., & Neale, J. M.
Pennebaker, J. W. (1993). Social mechanisms of constraint. In D. M. (1987). Evidence that secretory IgA antibody is associated with daily
Wegner & J. W. Pennebaker (Eds.), Handbook of mental control (pp. mood. Journal of Personality and Social Psychology, 52, 988 –993.
200 –219). Englewood Cliffs, NJ: Prentice Hall. Swinkels, A., & Giuliano, T. A. (1995). The measurement and conceptu-
Pennebaker, J. W., & Francis, M. E. (1996). Cognitive, emotional, and alization of mood awareness: Monitoring and labeling one’s mood
language processes in disclosure. Cognition & Emotion, 10, 601– 626. states. Personality and Social Psychology Bulletin, 21, 934 –949.
Pennebaker, J. W., & Graybeal, A. (2001). Patterns of natural language Tkach, C. (2005). Unlocking the treasury of human kindness: Enduring
use: Disclosure, personality, and social integration. Current Directions improvements in mood, happiness, and self-evaluations. Unpublished
in Psychological Science, 10, 90 –93. doctoral dissertation, Department of Psychology, University of Califor-
Pennebaker, J. W., Mayne, T. J., & Francis, M. E. (1997). Linguistic nia, Riverside.
predictors of adaptive bereavement. Journal of Personality and Social Trapnell, P. D., & Campbell, J. D. (1999). Private self-consciousness and
Psychology, 72, 863– 871. the five-factor model of personality: Distinguishing rumination from
Pennebaker, J. W., & Seagal, J. (1999). Forming a story: The health reflection. Journal of Personality and Social Psychology, 76, 284 –304.
benefits of narrative. Journal of Clinical Psychology, 55, 1243–1254. van der Kolk, B., & van der Hart, O. (1991). The intrusive past: The
Ryff, C. D. (1989). Happiness is everything, or is it? Explorations on the flexibility of memory and the engraving of trauma. American Imago, 48,
meaning of psychological well-being. Journal of Personality and Social 424 – 454.
Psychology, 57, 1069 –1081. Vinokur, A., Schul, Y., & Caplan, R. D. (1987). Determinants of perceived
Salovey, P., Rothman, A. J., Detweiler, J. B., & Steward, W. T. (2000). social support: Interpersonal transactions, personal outlook, and tran-
Emotional states and physical health. American Psychologist, 55, 110 – sient affective states. Journal of Personality and Social Psychology, 53,
121. 1137–1145.
Segerstrom, S. C., Stanton, A. L., Alden, L. E., & Shortridge, B. E. (2003). Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and vali-
A multidimensional structure for repetitive thought: What’s on your dation of brief measures of positive and negative affect: The PANAS
mind, and how, and how much? Journal of Personality and Social scales. Journal of Personality and Social Psychology, 54, 1063–1070.
Psychology, 85, 909 –921. Wilson, T. D., Centerbar, D. B., Kermer, D. A., & Gilbert, D. T. (2005).
Seligman, M. E. P., & Csikszentmihalyi, M. (2000). Positive psychology: The pleasures of uncertainty: Prolonging positive moods in ways people
An introduction. American Psychologist, 55, 5–14. do not anticipate. Journal of Personality and Social Psychology, 88,
Sherbourne, C. D., Allen, H. M., Kamberg, C. J., & Wells, K. B. (1992). 5–21.
Physical psychophysiologic symptoms measure. In A. L. Stewart & J. E. Wilson, T. D., & Gilbert, D. T. (2003). Affective forecasting. Advances in
Ware (Eds.), Measuring functioning and well-being (pp. 260 –272). Experimental Social Psychology, 35, 345– 411.
Durham, NC: Duke University Press. Wood, J. V., Heimpel, S. A., & Michela, J. L. (2003). Savoring versus
Singer, J. A. (2004). Narrative identity and meaning making across the dampening: Self-esteem differences in regulating positive affect. Jour-
adult lifespan: An introduction. Journal of Personality, 72, 437– 459. nal of Personality and Social Psychology, 85, 566 –580.
Smyth, J. M. (1998). Written emotional expression: Effect sizes, outcome
types, and moderating variables. Journal of Consulting and Clinical Received August 8, 2003
Psychology, 66, 174 –184. Revision received November 2, 2005
Smyth, J. M., True, N., & Souto, J. (2001). Effects of writing about Accepted November 4, 2005 䡲

The Costs and Benefits of Writing, Talking, and Thinking About Life's Triumphs and Defeats

Uploaded by

Copyright:

Available Formats

The Costs and Benefits of Writing, Talking, and Thinking About Life's Triumphs and Defeats

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Costs and Benefits of Writing, Talking, and Thinking About Life's Triumphs and Defeats

Uploaded by

Copyright:

Available Formats

Journal of Personality and Social Psychology Copyright 2006 by the American Psychological Association

2006, Vol. 90, No. 4, 692–708 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.90.4.692

Keywords: positive experience, health, well-being, rumination, expressive writing

Why might different ways of processing traumatic events result

Transient Affect Preliminary Measures

Overview of Statistical Analyses

Writing Talking Thinking Comparison

Writing Talking Thinking Comparison

Think–replay Think–analyze Write–replay Write–analyze

Long-term PA 3.63a 0.12 3.43b 0.12 3.39b 0.12 3.42b 0.12

You might also like