You are on page 1of 19

The Journal of Positive Psychology

Dedicated to furthering research and promoting good practice

ISSN: 1743-9760 (Print) 1743-9779 (Online) Journal homepage: http://www.tandfonline.com/loi/rpos20

Four experiments on the relational dynamics and


prosocial consequences of gratitude

Jo-Ann Tsang & Stephen R. Martin

To cite this article: Jo-Ann Tsang & Stephen R. Martin (2017): Four experiments on the relational
dynamics and prosocial consequences of gratitude, The Journal of Positive Psychology, DOI:
10.1080/17439760.2017.1388435

To link to this article: http://dx.doi.org/10.1080/17439760.2017.1388435

Published online: 12 Oct 2017.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=rpos20

Download by: [UNIVERSITY OF ADELAIDE LIBRARIES] Date: 13 October 2017, At: 05:48
The Journal of Positive Psychology, 2017
https://doi.org/10.1080/17439760.2017.1388435

Four experiments on the relational dynamics and prosocial consequences of


gratitude
Jo-Ann Tsang and Stephen R. Martin 
Department of Psychology and Neuroscience, Baylor University, Waco, TX, USA

ABSTRACT ARTICLE HISTORY


Four experiments utilized experimental inductions of gratitude and behavioral measures of Received 30 March 2017
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

prosociality to explore the effects of 4 variables on gratitude and grateful prosocial outcomes: Accepted 20 September 2017
benefactor similarity (Study 1), intention (Study 2), future benefits (Study 3), and anonymity (Study
KEYWORDS
4). We consistently found that the receipt of a favor increased prosocial behavior, and this effect was Gratitude; prosocial
mediated by gratitude. Benefactor similarity did not meaningfully influence either prosocial behavior behavior; altruism
or gratitude. Although benefactor intention did meaningfully affect gratitude, the effect was too
small to influence the mediational effect of gratitude on prosocial behavior. Neither anonymity nor
the possibility of future procurement decreased the role of gratitude in distribution; instead, both
of these variables may have enhanced the role of gratitude. These data support gratitude as an
important component of prosocial behavior and suggest that gratitude may contain an altruistic
component, consistent with its relational function.

As noted by Aristotle (1905), humans are social animals the welfare of the benefactor, or are they motivated by less
and require society to reach their full potential. Not only communal concerns such as self-presentation and procur-
do we need the cooperation of others for physical survival, ing future benefits?
but also for mental health: loneliness and ostracism are
among the most debilitating human experiences. But liv-
The relationship-strengthening function of
ing in the company of others is not easy. Gratitude is one
gratitude
social lubricant that can encourage cooperation, maintain
social harmony, and build the relationships we need to Early research focused primarily on the role of gratitude
survive (Algoe, 2012; Algoe, Haidt, & Gable, 2008; Simmel, in social exchange. For example, Trivers (1971) theorized
1950; Trivers, 1971). that gratitude evolved to increase the likelihood that
Gratitude is a positive emotion that can occur in a prosocial behaviors would be reciprocated, thus perpet-
recipient when a benefactor intentionally gives a valued uating reciprocal altruism and its associated fitness bene-
gift to them (e.g. McCullough, Kimeldorf, & Cohen, 2006; fits. McCullough and colleagues theorized that gratitude
Tsang, 2006b). Although much is known about the indi- served as a moral or prosocial barometer that alerted
vidual benefits of gratitude (Barrett-Cheetham, Williams, & individuals to the fact that a benefactor intentionally
Bednall, 2016; Emmons & McCullough, 2003; Wood, Froh, expended costly effort in order to provide them with a
& Geraghty, 2010), less is known about the relationship valued benefit, and this realization then motivated the
dynamics and prosocial motivations associated with grat- recipient to behave prosocially toward the benefactor
itude. The current set of four experiments utilize experi- in return (e.g. McCullough, Kilpatrick, Emmons, & Larson,
mental inductions of gratitude and behavioral measures 2001; McCullough et al., 2006). Research has supported
of grateful prosociality within a distribution paradigm in that individuals are more likely to experience gratitude
order to explore two types of questions about the rela- when the favor is intentional, valuable (Lane & Anderson,
tional function of gratitude. First, how might aspects of the 1976; Tesser, Gatewood, & Driver, 1968), and costly (Tesser
relationship-worthiness of the benefactor, such as benefac- et al., 1968). Thus, gratitude may promote and maintain an
tor similarity and benevolent intention, elicit gratitude and economy of exchangeable prosocial actions.
grateful behavior? Second, are the prosocial behaviors trig- In contrast to an economic view of gratitude, Algoe
gered in the grateful recipient motivated toward increasing and colleagues (e.g. Algoe, 2012; Algoe et al., 2008; see

CONTACT  Jo-Ann Tsang  JoAnn_Tsang@Baylor.edu


© 2017 Informa UK Limited, trading as Taylor & Francis Group
2   J.-A. TSANG AND S. R. MARTIN

also Buck, 2004) theorize that in addition to encouraging In Study 1, we expand upon existing literature by exam-
reciprocity, gratitude also functions to create and maintain ining another benefactor characteristic: similarity. Previous
close relationships. They suggest that gratitude evolved research has shown that individuals are more attracted to
to alert individuals to high-quality relationship partners similar than dissimilar targets (Montoya & Horton, 2013;
who care about their well-being. Research has been accu- Montoya, Horton, & Kirchner, 2008). Because individuals
mulating to support this relational theory of gratitude. In should find similar others attractive, and thus be more
research between strangers, grateful recipients prefer willing to initiate a close relationship with them, they
the company of the benefactor (Bartlett, Condon, Cruz, should experience more gratitude in response to similar
Baumann, & Desteno, 2012), are more willing to provide compared to dissimilar benefactors.
the benefactor with their contact information (Williams
& Bartlett, 2015), and report more communal feelings
Benefactor intention
toward their benefactor (Simão & Seibt, 2015). Looking at
close relationships, correlational ( Gordon, Impett, Kogan, Another benefactor characteristic that should be impor-
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

Oveis, & Keltner, 2012; Gordon, Arnette, & Smith, 2011; Sun, tant in a relational account of gratitude is benefactor
Jiang, Chu, & Qian, 2014), experimental (Puente-Diaz & intention. Benefactors who provide intentional benefits
Cavazos-Arroyo, 2016), and longitudinal (Algoe, Gable, & and who appear responsive to the recipient’s needs have
Maisel, 2010; Algoe & Zhaoyang, 2016; O’Connell, O’Shea, & the potential to be high-quality relationship partners, and
Gallagher, 2016) research has revealed links between grat- should elicit more gratitude (Algoe, 2012).
itude and relationship satisfaction. Correlational studies Research has conceptualized benefactor intention
have also uncovered links between gratitude and taking in different ways. The most common way has been
the partner’s perspective (Gordon & Chen, 2013), as well as to use scenarios to portray benefactors as motivated
relationship closeness (Algoe & Haidt, 2009) and commit- either benevolently or selfishly. For example, Tesser et
ment (Joel, Gordon, Impett, MacDonald, & Keltner, 2013). al. (1968) presented participants with scenarios that
Longitudinal research has shown links between partner varied whether the benefactor was providing a favor
responsiveness, gratitude and appreciation toward the due to completely altruistic reasons, partially selfish
partner, and relationship maintenance behavior and sta- reasons, or completely selfish reasons. They found that
bility (Algoe & Way, 2014; A. Gordon et al., 2012; Joel et gratitude was higher when the benefactor was altruis-
al., 2013; Kubacka, Finkenauer, Rusbult, & Keijsers, 2011). tic (see also Pelser et al., 2015; Tsang, 2006a; Watkins,
Expressions of gratitude have also been shown to increase Scheer, Ovnicek, & Kolts, 2006; Wood, Maltby, Stewart,
perceptions of the communal strength of a relationship Linley, & Joseph, 2008). Similarly, Weinstein et al. (2010)
(Lambert, Clark, Durtschi, Fincham, & Graham, 2010). found that when participants read vignettes that por-
trayed benefactors as acting of their own volition and
from their own values, they stated they would feel more
Benefactor relationship qualities
grateful, compared to those reading about favors from a
If gratitude is most centrally about relationships, then indi- benefactor who helped because of societal expectations
viduals might experience more gratitude when the ben- or to avoid guilt (see also Graham, 1988). This supports
efactor has higher relationship potential. In other words, the relational function of gratitude because it is more
not only should characteristics of the gift, such as cost and likely that benefactors who help out of their own volition
value, determine gratitude, but characteristics of the giver actually care about the recipients personally, compared
should also influence gratitude. Research utilizing video- to benefactors who help for more controlled, external
tapes of hypothetical interactions between strangers have reasons.
found correlations between gratitude and liking as well Beliefs about free will are also relevant to intention and
as attraction (Hendrickson & Goei, 2009). Other research the relational function of gratitude. MacKenzie, Vohs, and
utilizing vignettes (Rotkirch, Lyons, David-Barrett, & Jokela, Baumeister (2014) found that experimentally-induced
2014; Weinstein, DeHaan, & Ryan, 2010) and recalled favors reductions in the belief in free will caused lower levels of
(Gruszecka, 2014) has found correlations between close- gratitude for scenarios, recalled favors, and actual favors
ness to the benefactor and gratitude. Close relationship in the lab. This occurred because reduced belief in free will
research has found that couples experiencing higher caused recipients to infer that benefactors did not have
marital satisfaction report more gratitude toward each sincere motivations for helping them, providing further
other (Gordon et al., 2011), and perceptions of benefac- support for the role of intention in gratitude. Together,
tor thoughtfulness are predictive of gratitude, above and these results support Algoe’s (2012) assertion that bene-
beyond the cost and the value ascribed to the gift (Algoe factors who appear responsive toward the recipient elicit
et al., 2008). more gratitude.
THE JOURNAL OF POSITIVE PSYCHOLOGY   3

With the exception of MacKenzie et al. (2014), the which are regions associated with moral judgments and
research on intention and gratitude has examined hypo- theory of mind. This indirectly suggests that gratitude
thetical scenarios and recalled favors. To expand upon may be related to altruistic motivation (Batson, 2011).
existing literature, we investigate whether induced grati- Mathews and Green (2010) found that the self-focused
tude is affected by perceived benefactor intention in Study construct of social anxiety was negatively related to trait
2. In addition, rather than relying on self-report measures gratitude, and that inducing high-self focus increased
of gratitude, we provide an additional measure of grati- feelings of indebtedness in participants recalling a ben-
tude in the form of prosocial behavior. efit, although the induction did not have any effect on
gratitude. This suggests that gratitude might instead
be related to other-focused motivations such as altru-
Grateful motivation
ism, although the evidence is again indirect. Algoe and
In addition to influencing the experience of gratitude, rela- Haidt (2009) found that when individuals recalled expe-
tional concerns may also affect the behaviors that follow riences of gratitude, they reported being motivated to
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

from gratitude (Algoe, 2012). The grateful recipient may enhance the reputation of the benefactor, acknowledge
be motivated to continue the upward spiral of gratitude their actions, repay them in some way, and develop a
by acting prosocially–not simply due to the norm of reci- closer relationship. However, because of social desira-
procity, but out of concern for the benefactor. bility concerns (Jones & Pittman, 1982), and due to the
Recent research supports the link between gratitude fact that individuals are often unaware of their cognitive
and subsequent prosocial behavior. In vignette research, processes (Nisbett & Wilson, 1977), it is not entirely clear
participants report that they would behave prosocially whether gratitude is motivated by benefactor enhancing
toward someone to whom they felt gratitude (Graham, goals, or whether individuals are simply presenting their
1988; Naito & Sakata, 2010; Watkins et al., 2006). Trait and motivations as such.
state gratitude are related to self-reports (Dijker, Nelissen, We address the dearth of research on the motivations
& Stijnen, 2013; Froh, Bono, & Emmons, 2010; Emmons underlying gratitude by pitting a more communal, altruistic
& McCullough, 2003; Li & Chow, 2015; McCullough, gratitude motivation against two possible egoistic motiva-
Emmons, & Tsang, 2002; Sikkel & Schoenmakers, 2012) tions for gratitude: the procuring of future benefits (Study 3)
and peer-reports of helping others (Chang, Lin, & Chen, and the presentation of the self as a grateful person (Study 4).
2012; McCullough et al., 2002; Michie, 2009). Research
in the laboratory has shown a positive relationship
Methodology in gratitude research
between gratitude and actual prosocial behavior
(Bartlett & DeSteno, 2006; Bartlett et al., 2012; DeSteno, Another contribution we hope to make to the literature
Bartlett, Baumann, Williams, & Dickens, 2010; Froh et al., is by providing experimental tests of gratitude using
2014; Goei, Roberto, Meyer, & Carlyle, 2007; Ma, Tunney, real-time gratitude inductions and behavioral measures
& Ferguson, 2014; Tsang, 2006b; 2007; Tsang, Schulwitz, of gratitude and its correlates. Much of the current grati-
& Carlisle, 2012). tude literature makes use of scenarios and/or self-reports,
However, just because a person behaves prosocially which can be problematic. For instance, although indi-
does not mean that their ultimate goal is to increase the viduals may have access to the outcomes of cognitive
person’s welfare (e.g. Batson, 2011). Egoistically motivated processes, they are not consciously aware of the specific
grateful behaviors might include returning the favor in variables that have influenced those cognitive processes
order to procure future favors (Trivers, 1971) or thanking (e.g. Nisbett & Wilson, 1977), introducing limitations to
the benefactor in order to present the self as an apprecia- scenario and even some recall methods of studying grati-
tive, moral person (Jones & Pittman, 1982). In contrast, if tude. Furthermore, the use of self-report measures of grat-
gratitude serves a relational function and triggers com- itude is problematic due to the socially desirable nature
munal norms, one would expect that grateful prosociality of gratitude (Tsang, 2006b). Therefore, it is important to
would be directed not only egoistically, but altruistically supplement self-reports and scenarios with methods that
toward the needs of the benefactor. induce gratitude in the laboratory and measure its behav-
Little research directly speaks to the question of ioral outcomes.
whether grateful prosociality is fueled by altruistic as In the current set of four experiments, we induce grat-
well as egoistic motivations. Fox, Kaplan, Damasio, and itude in the laboratory, in real time, reducing potential
Damasio (2015) found that individuals who read stories cognitive biases that might result from recalling past epi-
about gratitude showed increased activity in the medial sodes of gratitude and from reading gratitude scenarios.
prefrontal cortex and the anterior cingulate cortex, We also provide a behavioral measure of gratitude that
4   J.-A. TSANG AND S. R. MARTIN

may be less affected by self-presentation concerns than Grateful self-presentation


are self-report measures. If grateful individuals are oriented toward egoistic
self-presentation motives, they should behave more
prosocially when their benefactor will know that they are
Predictions
reciprocating. In contrast, if gratitude motivates individu-
We make the following predictions in the context of the als more altruistically, then grateful individuals should act
relational function of gratitude. prosocially toward the benefactor even if the benefactor
would not know that the prosocial behavior originates
from them.
Gratitude vs. positive mood
Consistent with previous research (Tsang, 2006b; 2007;
Study 1
Tsang et al., 2012), we predict that participants who receive
a favor from a benefactor will report more gratitude and In Study 1, we tested the first two predictions – that a
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

demonstrate more grateful prosocial behavior compared favor would lead to more gratitude than a positive out-
to participants who receive the same positive outcome come by chance, and that individuals would experience
by chance. more gratitude when their benefactor was more similar
to them. Participants were told that they and a participant
in another room were engaging in a number of resource
Relationship-relevant characteristics of the
distributions. We portrayed some partners as attitudinally
benefactor
similar to participants via answers on a questionnaire. For
We investigate the effects of the following characteristics the gratitude manipulation, either the partner or ‘random
of the benefactor on gratitude and grateful behavior. chance’ provided the participant with resources. The par-
ticipant rated their felt gratitude and grateful motivation
Benefactor similarity (self-report measures of gratitude), and were provided
If gratitude helps individuals find good relationship part- with their own opportunity to distribute resources (behav-
ners (Algoe, 2012), then individuals who are more attrac- ioral measure of gratitude).
tive should elicit more gratitude. Benefactors who have
more similar attitudes, and are therefore more attractive,
Method
should elicit more gratitude and grateful behavior com-
pared to less similar benefactors. Participants
Participants were 179 female undergraduate psychol-
Benefactor intention ogy students at a small, southwestern private religious
Gratitude should also vary depending on the intentions university. Because the experimenters for this study
of the benefactor. The presence of benevolent motivation were female, only female participants were recruited
might indicate increased relationship potential and the to eliminate cross-gender self-presentation concerns
presence of communal norms. Benevolent partners should (Jones & Pittman, 1982). Demographic information
therefore elicit more gratitude and grateful behaviors than was not collected, but given the undergraduate demo-
should partners who communicate expectations of reci- graphic at this university, the sample consisted primarily
procity (Watkins et al., 2006). of European Americans aged 18–22 We decided a priori
to have at least 20 participants per cell, and collected
data until the end of the semester when this minimum
Prosocial motivations of the grateful recipient
number of participants was reached. Seventeen indi-
We investigate the possibility that grateful prosociality viduals were excluded due to suspicion, leaving a total
may be altruistically motivated, and contrast that moti- N = 162.
vation with the following possible egoistic motivations.
Procedure
Procuring future benefits Participants were run singly in a small laboratory room.
If grateful individuals are oriented toward social exchange When participants first arrived, they were asked to fill out
concerns, then they should engage in increased prosocial a 7-item attitude questionnaire while the experimenter
behaviors when they know that they will interact with the waited for a ‘second participant’ to arrive. After partici-
benefactor in the future. In contrast, if gratitude activates pants completed the questionnaire, they were told that
communal norms and altruistic motivation, then individu- they and another same-gender psychology student would
als should behave prosocially regardless of the possibility engage in four rounds of a resource distribution task. They
of future interaction. would not interact with their partner, but would receive
THE JOURNAL OF POSITIVE PSYCHOLOGY   5

the chance to communicate by written note during certain of the study. When the participant had completed their
rounds. They were also informed that in order to study the review of their partner’s questionnaire, they were given
effect of first impressions on distributions, one participant an impression questionnaire to fill out about their partner,
in the study would get to see the other’s attitude question- which included a number of personality ratings along with
naire. In reality, only three distribution rounds were car- manipulation check items assessing the participant’s per-
ried out, the other participant was fictitious, and the other ceived similarity with their partner (‘Under normal condi-
participant’s answers were filled out by the experimenter. tions, how enjoyable do you think it would be to work with
Participants were told that during each round, 10 raf- this person?’ ‘How similar to you do you think this person
fle tickets good for a $50 gift card would be distributed is?’ and, ‘How likely is it that you and this person could
between the two participants. In some rounds, they or become friends?’ For all items 1 = not at all, 7 = extremely)
the other participant would be assigned to distribute the The experimenter then started the first round of distri-
resources, whereas in other rounds resources would be butions. All participants received three raffle tickets distrib-
distributed by chance. It was emphasized to participants uted by chance, and were told that their partner was given
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

that they and their distribution partner were not in compe- seven. Participants were then randomly assigned to Favor
tition with each other, but would each have opportunities conditions. On Round 2, participants in the Favor: Present
to receive resources. All distribution decisions were made condition were told that their partner had given them nine
using paper-and-pencil forms, and resources for each tickets, keeping only one ticket to herself. A handwritten
round took the form of 10 blue raffle tickets. note, seemingly from their partner, accompanied the dis-
tribution outcome and read, ‘I saw that you didn’t get a
Conditions.  Participants were randomly assigned to lot in the last round – that must have been a bummer’.
Similar and Dissimilar conditions. These conditions were This note served to communicate benevolent intention.
manipulated based on participants’ earlier answers to Participants in the Favor: Absent control conditions were
the attitude questionnaire. The questionnaire consisted told that they received nine raffle tickets by chance, and
of seven forced-choice items: (1) What are your views their partner had received one. There was no note from
on abortion? (check one: Prolife/Prochoice), (2) What are the other participant in this condition. Thus, participants
your views on fraternities and sororities? (check one: I in the Favor: Absent condition received the same positive
am against/in favor of fraternities and sororities as they outcome as participants in the Favor: Present condition, but
usually function), (3) Where would you prefer to live in the absence of a benefactor. Experimenters remained
after college? (check one: urban area/rural area), (4) How blind to the Favor: Present/Absent conditions until the end
do you feel about the city of Waco’s ban on smoking in of the study.
restaurants? (check one: I am in favor/against the ban
on smoking in restaurants), (5) Which type of movies do Dependent measures.  In Round 3, participants in all
you prefer? (check one: Comedies/Dramas), (6) Which conditions received the opportunity to distribute 10
of the following TV shows would you prefer to watch? raffle tickets. This constituted our behavioral measure of
(check one: The Simpsons/American Idol), (7) Which of prosociality. After they made their distribution decision,
the following better describes your political affiliation? participants were given a questionnaire that contained
(check one: Democrat/Republican). While the participant manipulation-check items, along with items that asked
was reading further information about the study, the them the reasons behind their decision. Participants
experimenter surreptitiously filled out a questionnaire were asked the extent to which they were motivated by
supposedly from the other participant. Participants were the following concerns: to get money, be fair, help the other
randomly assigned to have a partner that was either participant, express appreciation, establish justice, fulfill an
similar or dissimilar to them. In the Similar condition, the obligation, reciprocate a favor and act morally. Participants
experimenter marked items #1, 2, 4, 6, and 7 the same rated these motivations on a 7-item Likert-type scale
as the participant’s; whereas in the Dissimilar condition, (1 = Not at all, 7 = Totally). The item express appreciation
only items # 3 and 5 were marked the same as the constituted a measure of grateful motivation, and the
participant’s. Although the experimenters were not blind item fulfill an obligation was a measure of indebted
to the participants’ similarity condition, they were kept motivation. In addition, after each of the rounds
naïve to the study hypotheses. the experimenter gave participants a questionnaire
After the experimenter created the partner’s ques- asking them to rate the emotions they were currently
tionnaire, she informed participants that they had been experiencing toward their partner on a 7-point scale,
randomly assigned to read their partner’s questionnaire, including the adjectives pleased, grateful, indebted, happy,
and they were given a few minutes to do so. The part- resentful, thankful, annoyed, appreciative, obligated, upset,
ner’s questionnaire was left in the room for the duration sympathetic, and angry. A composite measure of grateful
6   J.-A. TSANG AND S. R. MARTIN

emotion was created by combining the adjectives 95% CI [−0.76, 0.23], BF01 = 6.97. Similarity also failed to
grateful, thankful, and appreciative reported after Round 2 moderate the relationship between receiving resources
(α = 0.97), and a composite measure of indebted emotion and experiencing gratitude, t(158) = 1.31, p = 0.19, 95% CI
was created by combining the adjectives indebted and [−1.65, 0.34], BF01 = 7.18. Those who received resources
obligated reported after Round 2 (α = 0.81). There were from a benefactor reported a greater motivation to express
also additional personality questionnaires not analyzed gratitude (M = 6.21, SD = 0.92) than those who received
in this study. resources by chance (M = 2.67, SD = 1.95), t(154) = 14.17,
After the questionnaires were completed, the experi- p < 0.001, 95% CI [3.05, 4.03], d = 2.28. However, similarity
ment was terminated. During the postexperimental inter- failed to exhibit a main effect on grateful motivation (t(154)
view, participants were carefully probed for suspicion, and = 0.80, p = 0.43, 95% CI [−0.29, 0.69], BF01 = 4.38) or to
the experimenter fully explained the experimental proce- moderate the relationship between receiving resources
dure to participants, including the reasons for deception, and grateful motivation, t(154) = 0.08, p  =  0.94, 95% CI
using guidelines from Aronson, Ellsworth, Carlsmith, and [−0.47, 0.51], BF01 = 19.76. Those who received resources
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

Gonzales (1990). No participant expressed any distress from a benefactor gave more resources (M  =  7.10,
about the experiment, and all seemed to understand SD = 1.26) than those who received resources by chance
the reasons for deception. All participants were given 20 (M = 4.66, SD = 1.98), t(158) = 9.30, p < 0.001, 95% CI [1.95,
entries into a $50 raffle. A raffle was held at the end of each 3.00], d = 1.48. Similarity neither moderated [t(158) = 0.92,
semester in which the study was conducted, and winners p  =  0.19, BF01  =  9.65] the effect nor exhibited a main
were awarded $50. effect on distribution of resources, t(158) = 1.06, p = 0.29,
To summarize, this experiment tested the effect of ben- BF01 = 5.67. Therefore, receiving a favor seemed to cause
efactor similarity on gratitude by using a laboratory induc- an increase in both grateful emotion and prosocial behav-
tion and behavioral measure of gratitude. Participants ior, but these effects were unaffected by the similarity of
were told that they were distributing resources with the benefactor.
another student. In the second round, some participants
were told that they received nine raffle tickets from the Explanatory models
‘other participant’, the gratitude induction, whereas other A serial mediational model was estimated in which receiv-
participants received nine raffle tickets through ‘chance’, ing tickets from a benefactor predicts grateful emotions,
which led to a positive outcome without gratitude. All which in turn predicts the motivation to express gratitude,
participants were then provided the opportunity to act and finally the number of tickets the participant distrib-
prosocially by distributing raffle tickets. Some participants uted to the benefactor. Our hypothesis states that grateful
believed that the other participant held similar attitudes emotions should be greater when the benefactor is similar
to them on important issues such as abortion and polit- to the participant. To model this, similarity moderates the
ical affiliation, whereas other participants believed their relationship between receiving tickets from a benefactor
partner held dissimilar beliefs. and grateful emotions. Mediation can thus be assessed
for the dissimilar and similar conditions. Bootstrapping
was used to estimate standard errors. Continuous varia-
Results
bles were standardized to ease interpretation. Finally, the
For all analyses, factors were effect coded. mediational model partials out the relationships between
grateful and indebted emotions, and between gratitude
Experimental effects expression and obligation motivations. The overall medi-
A manipulation check showed that the rating of benefac- ational effect was significant, abc = 0.26, z = 2.45, p = 0.01,
tor similarity correlated highly with the condition assign- 95% CI [0.07, 0.48]. The mediational effect in the similar
ment of similarity, r = 0.84. Likewise, individuals accurately condition, abc  =  0.24, did not significantly differ from
recalled whether they received tickets by chance or from the effect in the dissimilar condition, abc = 0.28, z = 0.90,
the other participant, r = 0.96. p  =  0.37, 95% CI [−0.04, 0.17]. The mediational effect
We hypothesized that those who received resources through grateful emotions and motivation to express grat-
from another should especially feel grateful if the benefac- itude accounted for 21.66% of the direct effect of receiving
tor was similar to themselves. Those who received resources resources from a benefactor on distributing, all controlling
from a benefactor did report feeling more grateful on for indebted emotions and obligation. Therefore, the rela-
Round 2 (M = 6.54, SD = 0.68) than those who received tionship between receiving a favor and prosocial behavior
resources by chance (M = 2.90, SD = 2.12), t(158) = 14.25, was mediated by grateful emotions and the self-reported
p < 0.001, 95% CI [3.09, 4.08], d = 2.26. Similarity exhib- motivation to express gratitude. Similarity was not a sig-
ited no main effect on gratitude, t(158) = −1.06, p = 0.29, nificant moderator of this relationship.
THE JOURNAL OF POSITIVE PSYCHOLOGY   7

Study 1 discussion to eliminate cross-gender self-presentation concerns


with our female experimenters (Jones & Pittman, 1982).
In Study 1, we replicated previous research that showed
Demographic information was not collected, but the
that gratitude affects prosocial behavior above and beyond
population demographic consisted primarily of European
positive mood (Bartlett & DeSteno, 2006; Tsang, 2006a;
Americans ages 18–22. We decided a priori to have at least
2007; Tsang et al., 2012). The prosocial effects of gratitude
20 participants per cell, and collected data until the end of
did not appear to differ depending on the similarity of
the semester when this minimum number of participants
the benefactor to the grateful recipient, suggesting that
was reached. Nine participants were excluded due to sus-
grateful prosociality generalizes beyond attitudinally sim-
picion, leaving a total N = 116.
ilar others.
Procedure
Study 2 As in Study 1, participants were told that they would
participate in distribution rounds with another, unseen
In Study 2, we examine how another relationship-relevant
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

(fictitious) participant. In this study, there was no pre-dis-


aspect of the benefactor – benefactor intention – might tribution attitude questionnaire. After receiving only
affect the experience of gratitude. Previous research has 3 raffle tickets on the first round distributed by chance,
demonstrated that benefactor intention is a determinant participants were randomly assigned to Favor conditions
of whether one experiences gratitude or indebtedness in Round 2. Participants in the Favor: Present condition
(Tsang, 2006b; Watkins et al., 2006). However, most research were further divided into note conditions. Participants in
investigating intention and gratitude has looked at scenar- the Benevolent Note condition received a note with the
ios or self-reported gratitude (e.g. Graham, 1988; Lane & nine tickets that said ‘I saw that you didn’t get a lot in the
Anderson, 1976; Tesser et al., 1968; Tsang, 2006a; Watkins last round–maybe I can help you out’. Participants in the
et al., 2006; Weinstein et al.,2010). In Study 2, we examine Egoistic Note condition received a note that said ‘Now you
the effects of benefactor intention on gratitude by inducing owe me’. Participants in the Favor: Absent condition were
actual gratitude in the laboratory and assessing gratitude told their 9 tickets were distributed by chance, and did
both via self-report and through prosocial behaviors. not receive a note.
All participants then had a chance in Round 3 to distrib-
Predictions ute raffle tickets. Dependent variables included the distri-
bution as a behavioral measure of prosociality, along with
We made the following predictions. the same self-report measures of grateful and indebted
motivations for the distribution, and ratings of grateful and
Gratitude versus positive mood indebted emotions after Round 2.
We aimed to replicate basic gratitude findings from Study After the questionnaires were completed, the experi-
1, predicting that the receipt of a favor would induce grate- ment was terminated. The postexperimental interview was
ful emotion and increase prosocial behavior toward the completed similarly to Study 1. No participant expressed
benefactor. any distress, and all seemed to understand the reasons for
deception. All participants were given 20 entries into a $50
Benevolent vs. selfish intention raffle. A raffle was held at the end of each semester of the
We also examined whether benefactor intention affects study, and winners were awarded $50.
grateful emotion and behavior. We predicted that when To summarize, we tested the effect of benevolent vs.
the benefactor communicated benevolent intentions, egoistic benefactor intentions on gratitude. Some partic-
both grateful emotion and prosocial behavior motivated ipants were told that their partner gave them raffle tick-
by gratitude would increase. In contrast, when the ben- ets out of sympathy, whereas others were told that their
efactor communicated selfish motivations, gratitude partner helped out of a selfish motivation to make them
and prosocial behavior would fail to increase or would reciprocate the favor.
decrease compared to a baseline.

Results
Method
For all analyses, factors were effect-coded.
Participants
Participants were 125 female undergraduate psychol- Experimental effects
ogy students at the same private religious university. As We performed two manipulation checks. Participants accu-
before, only female participants were recruited in order rately recalled whether they received tickets by chance or
8   J.-A. TSANG AND S. R. MARTIN

from the partner, r = 0.98, and 88% of participants accu- a base model, the pathway through grateful emotions and
rately recalled the type of note received. desire to express gratitude accounted for 25.49% of the
We hypothesized that the selfish note would decrease direct effect of receiving resources on distributing. For the
gratitude. Those receiving the selfish note (M  =  7.39, sake of contrast, a similar model was fit wherein receiving
SD = 1.98) indicated marginally less gratitude (α = 0.98) resources predicts indebtedness, which predicts obliga-
than those receiving the benevolent note (M  =  7.99, tion, and in turn, distribution. Likewise, the mediation
SD = 1.86), t(113) = −1.85, p = 0.07, 95% CI [−1.66, 0.09], controls for relationships between grateful emotions and
d = −0.41. We contrasted this by examining whether the indebtedness, and desire to express gratitude and obliga-
note affected post-favor indebtedness, defined as a com- tion. Mediation, in this case, was nonsignificant, abc = 0.09,
posite of reported indebtedness and obligation after z = 1.41, p = 0.16, 95% CI [−0.02, 0.24]. Whereas obliga-
Round 2, α = 0.86. Unexpectedly, post-favor indebtedness tion and indebtedness, controlling for gratitude, did not
was not significantly different between note conditions, significantly mediate the relationship between receiving
t(114) = 0.70, p = 0.49, 95% CI [−0.75, 1.24], BF01 = 7.63. and giving, expression of gratitude and gratefulness did
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

However, the selfish note increased endorsement of mediate the relationship.


obligation as the reason for distributing tickets (t(109) =
2.57, p = 0.01, 95% CI [0.26, 1.96], d = 0.61) and decreased
Study 2 discussion
endorsement of expressing gratitude as the underlying
motivation, t(109) = −2.36, p = 0.02, 95% CI [−1.49, 0.131], Similarly to Study 1, we found that receipt of a favor caused
d  =  −0.56. This suggests that benefactor intention may participants to react prosocially. Although stated benefac-
alter one’s own reason for reciprocating, though one may tor intention did not alter prosocial behavior, it did affect
experience both indebted and grateful emotions upon participants’ experience of gratitude: Participants faced
receiving resources. with a benevolent benefactor experienced more grati-
As expected, those who received resources from the tude compared to those faced with a selfish benefactor.
benefactor distributed more tickets (M = 7.47, SD = 1.47) Participants were equally prosocial toward a benevolent
than those who received resources by chance (M = 5.23, and a selfish benefactor, and feelings of gratitude mediated
SD = 2.08), t(113) = 6.70, p < 0.001, 95% CI = [1.570, 2.880], their prosocial behavior. One limitation of Study 2, pointed
d = 1.31. However, no significant difference was detected out by an anonymous reviewer, is that benefactor intention
between those who received the benevolent note was confounded with benefactor rudeness – participants
(M = 7.08, SD = 1.40), and those who received the selfish may have perceived the statement, ‘Now you owe me’ as
note (M = 7.71, SD = 1.55), t(113) = 1.31, p = 0.19, 95% CI being rude.
= [−0.27, 1.28], d = 0.30, BF01 = 2.71. Although benefactor similarity in Study 1 did not seem
to affect gratitude or grateful behavior, benefactor inten-
Explanatory model tion in Study 2 seemed to have affected gratitude (but
To better understand the extent to which gratitude was not grateful behavior). It appears that some aspects of the
responsible for the relationship between receiving and benefactor, namely benefactor motivation, may affect the
distributing resources, and whether such a pathway was experience of gratitude.
modified by the intention of the benefactor, a serial medi-
ational model was fit using bootstrapped estimates of
Study 3
standard error. In this model, we examined the effect of
receiving resources and the effect of note type on grate- We next move from examining the motivations of the
ful emotions after controlling for indebted emotions. We benefactor to the motivations of the grateful recipient.
expected grateful emotions to in turn predict the motiva- We investigate whether grateful prosocial behaviors are
tion to express gratitude, after controlling for obligation motivated by egoistic, self-oriented concerns, or altru-
as a self-reported motivation. Desire to express gratitude istic, other-oriented concerns. Given that expressions
then predicts distribution of resources. This model rep- of gratitude can serve a prosocial reinforcer function
resents a serial moderated mediation that controls for (McCullough et al., 2006), in Study 3 we looked at the
the effects of indebted emotions and obligation. The egoistic motivation of securing future benefits from the
overall mediational model was significant, abc  =  0.270, benefactor.
z = 2.15, p = 0.03, 95% CI [0.04, 0.53]. No significant dif-
ference in mediation was detected between individuals
Predictions
who received a selfish note, abc  =  0.24, and those who
received a benevolent note, abc = 0.30, z = −1.66, p = 0.10, We again predicted that the receipt of a favor would
95% CI [−0.129, 0.001]. Comparing the mediated effect to increase grateful emotions and behavior compared to
THE JOURNAL OF POSITIVE PSYCHOLOGY   9

the receipt of a non-social positive outcome. Additionally, Present condition were told that the other participant
we tested an egoistic motivation for grateful prosociality. would again be the distributor in Round 4, whereas par-
If gratitude in this context is motivated by the desire to ticipants in the Future Benefits: Absent condition were told
secure future benefits, then grateful recipients should act that the tickets in Round 4 would be distributed by chance.
more prosocially when they know that their benefactor The Future Benefits conditions thus varied whether par-
will have another opportunity to benefit them. In con- ticipants had the potential to benefit from their partner in
trast, if gratitude is associated in this context with a more future rounds if they helped them in Round 3.
altruistic motivation, then grateful recipients should act After participants reviewed the distribution schedule,
prosocially even if they do not anticipate receiving future they began the first round of distributions, where they
benefits. received 3 raffle tickets distributed by chance. On Round
2, participants in the Favor: Present condition were told that
their partner had given them nine tickets, while keeping
Method
only one ticket to him/herself. A handwritten note, seem-
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

Participants ingly from their partner, read, ‘I saw that you didn’t get a
Participants were 78 undergraduate psychology students lot in the last round – that must have been a bummer’.
(53 women and 25 men, ages 18–22 years) at the same uni- Participants in the Favor: Absent control conditions were
versity with predominantly European American students. told that they received nine raffle tickets by chance, and
More women were recruited than men because we had that their partner had received one. There was no note in
more female experimenters. We decided a priori to have this condition.
at least 20 participants per cell, and collected data until Participants in all conditions then had a chance in
the end of the semester in which this minimum number Round 3 to distribute raffle tickets. Dependent variables
of participants was reached. Four suspicious participants included the distribution as a behavioral measure of proso-
were excluded, leaving N = 74. ciality, along with the same self-report measures of grate-
ful and indebted intentions for the distribution, and ratings
Procedure of grateful and indebted emotions after Round 2.
Participants were again told that they were distributing After the questionnaires were completed, the experi-
resources with another participant. In this study, partici- ment was terminated without a fourth round. During the
pants were informed that they had been assigned to see postexperimental interview (Aronson et al., 1990), no par-
the distribution schedule ahead of time. Therefore, they ticipant expressed any distress, and all seemed to under-
would know who the distributor and receiver would be on stand the reasons for deception. All participants were
each round from the beginning of the study, but their part- given 20 entries into a $50 raffle. A raffle was held at the
ner would not know until right before the round started. end of each study semester, and winners were awarded
To this end, participants received a schedule listing four $50.
rounds, and detailing who would be the distributor, who To summarize, this study tested the effect of future ben-
would be the receiver, or whether the round would be efit potential on gratitude in a laboratory context. Some
distributed by chance. This schedule was provided on a participants believed that the other participant would
piece of paper, and participants were encouraged to refer again have the opportunity to distribute raffle tickets,
to it throughout the study. Participants were randomly providing participants with an egoistic motivation to be
assigned to be in one of four conditions that varied based prosocial, whereas others believed that the other partici-
on the two variables of (1) Favor and (2) Future Benefits. pant would not have another chance to distribute, remov-
Participants in all conditions were told that the first round ing this egoistic motivation.
would be distributed by chance. Participants in the Favor:
Present conditions were told that in Round 2, they would
Results
be the receiver, and their partner would be the distributor,
whereas participants in the Favor: Absent conditions were For all analyses, factors were effect-coded.
told that Round 2 would also be distributed by chance.
The favor condition served as a replication of Studies 1 Experimental effects
& 2, testing the effects of gratitude compared to positive Manipulation checks demonstrated that participants accu-
mood. All participants were told that they would be the rately recalled whether they were given tickets by chance
distributor for Round 3, with the other participant as the or from the benefactor, r = 0.89, and correctly identified
receiver. who the distributor would be on the fourth round, r = 0.96.
Participants were also randomly assigned to a Future Participants who received tickets from the benefactor
Benefits conditions. Participants in the Future Benefits: indicated more gratitude (M = 7.46, SD = 1.93) than those
10   J.-A. TSANG AND S. R. MARTIN

who received them by chance (M = 3.28, SD = 3.22), t(70) = Study 3 discussion


6.79, p < 0.001, 95% CI [2.93, 5.37], d = 1.59. Consistent with
As with Studies 1 and 2, participants in Study 3 who
an altruistic motivation explanation, the future interaction
received a favor expressed more gratitude and behaved
condition was unrelated to grateful emotion (t(70) = −0.32,
more prosocially toward their benefactor, compared to
p = 0.75, 95% CI [−1.41, 1.03]) and did not moderate the
participants who received a similar positive outcome by
effect of receiving tickets from another participant, t(70) =
chance. Neither grateful emotion/motivation nor grate-
1.13, p = 0.26, 95% CI [−1.06, 3.82], BF01 = 7.25. Likewise,
ful prosocial behaviors were affected by the possibility of
receiving tickets from a benefactor produced greater
receiving future benefits from the benefactor, suggesting
desire to express gratitude as a reason for giving tickets
that this particular egoistic motivation was not responsi-
(t(70) = 8.19, p < 0.001, 95% CI [2.58, 4.24], d = 1.92), but
ble for participants’ grateful behaviors. This does not nec-
the future interaction condition had neither a main nor
essarily mean that grateful behaviors were motivated by
interactive effect, |t|(70) < 0.84, ps > 0.40, BF01 = 9.54.
altruism, however. The possibility of future benefits is not
Participants who received tickets from the benefactor
the only egoistic motivation that may have been triggered
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

gave more tickets (M = 6.80, SD = 1.80) than those who


in this situation. Grateful recipients might also have been
received them by chance (M  =  4.68, SD  =  2.78), t(70) =
motivated to act prosocially in order to increase their posi-
4.30, p  <  0.001, 95% CI [1.17, 3.19], d  =  1.01. The future
tive reputation and receive praise from the benefactor. This
interaction manipulation failed to moderate the relation-
self-presentation motivation may have been present even
ship between benefactor condition and distribution, t(70)
if participants did not expect that the benefactor could
= 1.93, p = 0.06, BF01 = 3.01. The main effect of the future
help them in the future, because despite the inability of the
interaction manipulation was nonsignificant, d  =  −0.04,
benefactor to distribute in a future round, they would still
t(70) = −0.19, p = 0.85; counter to the egoistic interpre-
have known whether or not the participant had returned
tation of gratitude, those who received tickets from the
the favor.
benefactor tended to give more tickets when they did not
expect a future interaction (M = 7.39, SD = 1.15) than when
they did (M = 6.32, SD = 2.10), t(38) = 1.94, p = 0.06, 95% Study 4
CI [−2.190, 0.049], d = 0.50. Study 4 was designed to address the additional egoistic
motivation of self-presentation in gratitude by manipulat-
Explanatory model ing whether or not the benefactor would know that the
We again estimated a serial mediational model in which prosocial behavior came from the participant. A gener-
receiving tickets from a benefactor predicts gratitude, ous distribution with the benefactor’s knowledge can be
which in turn predicts the desire to express gratitude motivated by self-presentation, whereas an anonymous
and the number of tickets given. As before, we estimated generous distribution is less likely to be motivated by
these effects after controlling for indebted emotions and self-presentation, and is possibly altruistic.
obligation as a motivation. We theorized that, if grate-
ful prosociality is motivated by the egoistic motivation
Predictions
to procure more benefits, then those who expected a
future interaction would distribute more tickets. To Similar to the studies above, we predicted that grateful
model this, we assume that the relationship between recipients would experience more gratitude and act more
the desire to express gratitude and the number of tickets prosocially compared to recipients of a nonsocial benefit.
given is moderated by whether the participant expects Furthermore, if gratitude is associated with an egoistic
a future interaction, such that the variance in distribu- motivation to self-present to the benefactor, then grateful
tion explained by the desire to express gratitude is lower recipients should act more prosocially when their benefac-
when egoism is activated. tor will know that they behaved prosocially, compared to
The overall mediational effect was significant, when the benefactor does not know. In contrast, if gratitude
abc  =  0.25, z  =  2.78, p  <  0.01, 95% CI [0.06, 0.44]. The is associated with a more altruistic motivation, then grateful
mediational effect when expecting a future interac- recipients should act prosocially even if their benefactor will
tion (abc = 0.38) was not significantly different from the remain ignorant to the source of the prosocial outcome.
effect when not expecting a future interaction (abc, 0.13),
z = 1.44, p = 0.15, 95% CI [−0.69, 0.01]. The overall medi-
Method
ational effect explained 28.76% of the direct effect of
receiving tickets from a benefactor on ticket distribution, Participants
after controlling for indebted emotions and obligation Participants were 105 undergraduate psychology students
as a motivation. (98 women and 7 men, ages 18–22  years) at the same
THE JOURNAL OF POSITIVE PSYCHOLOGY   11

university with predominantly European American stu- Results


dents. We decided a priori to have at least 20 participants
For all analyses, factors were effect-coded.
per cell, and collected data until the end of the semester
when this minimum number of participants was reached.
Experimental effects
Manipulation checks demonstrated that participants accu-
Procedure
rately recalled whether they themselves received tickets
As with the previous studies, participants were told that
by chance or from the benefactor, r = 0.96, and who the
they would participate in several distribution rounds with
benefactor was told the distributor was, r = 0.93.
an unseen participant. After receiving three raffle tick-
As before, those who received tickets from the bene-
ets distributed by chance, participants were randomly
factor reported greater post-round gratitude (M  =  7.80,
assigned to Favor: Present versus Absent conditions in
SD  =  2.03) than those who received tickets by chance
Round 2. Additionally, during Round 3 participants were
(M = 3.48, SD = 2.99), t(101) = 8.69, p < 0.001, 95% CI [3.35,
further assigned to knowledge conditions. Participants in
5.34], d = 1.70. We did not compute main effects of the
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

the Knowledge: Absent condition received additional writ-


knowledge manipulation on grateful emotions, as the
ten instructions with their Round 3 distribution:
grateful emotion measure appeared before the knowledge
There is additional information about your experimental
manipulation. Looking at grateful motivation, those who
condition about which you should know.
received tickets from a benefactor endorsed the expres-
On Round 3 you are assigned to be the Distributor.
sion of gratitude as a reason to give tickets more than
However, the other participant will be told that this con-
dition will be decided by Chance. This is the only round those who received tickets by chance, t(101) = 10.30, 95%
where these special conditions apply. CI [2.64, 3.91], d = 2.02. The knowledge manipulation nei-
The other participant is not given this information, and ther exhibited a main nor interactive effect on this meas-
does not know that you are receiving extra information ure, |t|(100) < 1.12, ps > 0.27.
about this round. Whatever you decide to distribute to For prosocial behavior, those who received tickets from
him/her, he/she will think it is a result of Chance. the benefactor gave back more tickets (M = 6.53, SD = 1.71)
Participants were then asked to make their distribu- than those who received tickets by chance (M  =  4.65,
tion decision as in previous studies. Participants in the SD  =  2.22), t(101) = 4.86, p  <  0.001, 95% CI [1.10, 2.61],
Knowledge: Present control condition did not receive addi- d = 0.95. Those who distributed anonymously gave fewer
tional instructions, and believed, as in Studies 1–3, that tickets (M = 5.09, SD = 2.51) than those who distributed
their partner would know that they were the one who with benefactor knowledge (M = 6.04, SD = 1.70), t(101)
distributed on the third round. = 2.34, p = 0.02, 95% CI [0.14, 1.65], d = 0.46, BF10 = 2.36.
The dependent variables included the ticket distribu- However, the interaction was not significant, t(101) = 0.17,
tion as a behavioral measure of prosociality, along with p = 0.86, ΔBF01 = 3.71.
the same self-report measures of grateful and indebted
motivations for the distribution, and ratings of grateful and Explanatory model
indebted emotions after Round 2. As conducted previously, a serial mediational model
After the questionnaires were completed, the exper- that controls for indebted emotions and obligation was
iment was concluded and the participants were given a estimated. In this model, receiving tickets from a bene-
postexperimental interview. No participant expressed factor predicts grateful emotions, which then predicts
any distress, and all seemed to understand the reasons for one’s motivation to express gratitude, and finally predicts
deception. All participants were given 20 entries into a $50 ticket distribution. We investigated whether anonymous
raffle. A raffle was held at the end of each study semester, distribution affected the relationship between gratitude
and winners were awarded $50. and the distribution of tickets in a manner consistent
To summarize, this study tested the relationship with egoistic motivation. To examine this hypothesis, we
between egoistic self-presentation concerns and grati- tested a model where the relationship between desire to
tude. Some participants were told that their partner would express gratitude and ticket distribution is moderated
think the raffle tickets they were given were actually dis- by knowledge condition. Although the overall media-
tributed to them by chance, minimizing self-presentation tion was significant, abc = 0.26, z = 1.89, p = 0.059, 95%
motivations for distributing the tickets. If participants still CI [0.003, 0.26], the knowledge condition moderated the
provided their benefactor with tickets even if they would mediation through grateful emotions and expression
not know that the tickets were coming from them, this of gratitude (the indirect effect differed by knowledge
would suggest that gratitude can lead to altruistic behav- condition), z  =  2.11, p  =  0.04, 95% CI [0.07, 0.61]. When
ior on the part of the grateful person. distributing anonymously, the mediational effect was
12   J.-A. TSANG AND S. R. MARTIN

significant, abc = 0.26, % of total effect = 30.39%, z = 2.30, to be generated. Because the modeler is uncertain about
p = 0.02, 95% CI [0.07, 0.53]. However, when distributing the generating parameters, one expresses uncertainty in
non-anonymously, the mediational pathway was nonsig- such parameters by selecting a prior distribution for each
nificant, abc = −0.03, z = −0.34, p = 0.692, 95% CI [−0.18, parameter. Bayesian modeling thus is defining a joint prob-
0.09]. In other words, in Study 4, our mediational model ability distribution in the form of a likelihood and prior
was replicated only when participants were told that their uncertainty in the parameters. This approach is not only
benefactor would not know that the resources were com- flexible (modelers choose assumptions and can incorpo-
ing from them; contrary to Studies 1–3 the model failed rate prior beliefs), but coherent in interpretation: The pos-
to replicate when the benefactor knew from whom the terior marginal distributions of each parameter represent
favor came. Indeed, the motivation to express gratitude the combination of prior uncertainty and data to yield a
was highly related to ticket distribution when distributing new probability distribution defining the ‘updated’ plau-
anonymously (BETA = 0.52), but was unrelated when dis- sibility of each parameter, conditioned on the observed
tributing non-anonymously (BETA = 0.05). After controlling data.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

for grateful motivation and its interaction with benefactor Hierarchical Bayesian models are Bayesian models for
knowledge, receiving tickets had little additional effect on which some unknown parameters themselves depend
ticket distribution (Mdiff = 0.14) when distributing anon- on higher-order unknown parameters, called hyperpa-
ymously, but did maintain a predictive effect when dis- rameters. Again, the modeler does not know the hyper-
tributing non-anonymously (Mdiff  =  2.07), suggesting parameters, and thus hyperparameters are assigned prior
egoistic motives. However, see the analyses below for a distributions called hyperpriors. Hierarchical Bayesian
further possibility. models are especially useful when one has observed sev-
eral sets of observations whose parameters are presumed
dependent on some higher-order parameter-generating
Partially pooled hierarchical Bayesian models
process. Here, we have four studies that attempt to esti-
One strength of this research is the use of similar meth- mate similar parameters, and thus a hierarchical model is
ods to replicate the behavioral effects of gratitude. Across appropriate. Partial pooling is a consequence of hierarchical
four experiments, we examined the effect of (1) receiv- models, and allows the parameter estimates of each study
ing resources from another person (as opposed to from to inform the parameter estimates of each other study; this
chance) and (2) some contextual factor that was hypoth- is in contrast to complete pooling (treating four datasets
esized to undermine one’s experience of gratitude or as one data-set) and no pooling (treating four datasets as
provide an egoistic reason for giving resources. Because estimating separate parameters). This procedure is anal-
all studies are similarly constructed and often attempt to ogous to random effects modeling or meta-analysis, for
estimate similar parameters (e.g. the effect of receiving those more familiar with such procedures.
resources from a person rather than chance on distribu-
tion), we constructed two hierarchical Bayesian models
Partially pooled effects of conditions
that partially pool estimates for parameters shared across
studies. The first model examines the effect of the groups All four studies employed an ANOVA testing for the com-
on distribution and the desire to express gratitude, and bined effects of receiving tickets from a participant and
the second model is a partially pooled variant of the serial some aspect of the benefactor or context. A partially
moderated mediational models estimated per study. These pooled ‘ANOVA’ (i.e. effect coded linear model with an
models will allow us to, in a meta-analytic sense, use all interaction term) can provide an overall estimate of the
four samples to estimate shared parameters of interest and main effect of receiving resources from a person, the over-
to borrow information from each sample to better inform all effect of the undermining contexts on distribution, and
estimates of study-specific parameters. We briefly describe the overall effect of their interactions. This model assumes
what is meant by a partially pooled hierarchical Bayesian that the parameter estimates of interest computed in each
model and proceed to describe the models in detail. sample are estimating either the same parameter or are
Bayesian modeling enables researchers to build prob- generated from a common distribution of parameters. For
abilistic models for estimating parameter estimates of example, all four studies estimated the effect of receiving
interest in a flexible, coherent manner. Modelers assume resources from a benefactor on distribution; pooled esti-
that a set of parameters underlie the data generating mates provide a stronger estimate for such an effect. In
process used to generate one’s observed data. Modelers contrast, all four studies featured different contexts which
then assert how the data are connected to the param- should affect the role of gratitude in resource distribution;
eters by specifying a distribution from which the data, although the contexts differ, all four context effects are
conditioned on the generating parameters, are assumed presumed to be from a population of contexts that affect
THE JOURNAL OF POSITIVE PSYCHOLOGY   13

the role of gratitude, and thus a pooled estimate provides A similar story is suggested by the effect of the experi-
the expected effect of such contexts on the dependent mental conditions on the motivation to express gratitude.
variable. We examined this pooled ANOVA model on two The partially pooled effect of receiving tickets on the
outcome variables: Number of tickets distributed and the motivation to express gratitude was large and consistent,
motivation to express gratitude. Groups are effect-coded M = 1.46 [1.19, 1.67]. However, the partially pooled effect
to be one unit apart for ease of interpretation. To aid of the context was negligible, M = 0.07 [−0.27, 0.40]. The
interpretation and to ease prior selection, both outcome shrunken effects of similarity, future interaction expecta-
variables are standardized within sample. All effect esti- tion, and anonymity, respectively, were M = 0.07 [−0.11,
mates were assigned a prior of ~ Normal(0, 1), implying 0.27], M = 0.015 [−0.27, 0.25], and M = 0.111 [−0.10, 0.35].
that we have no a priori belief about directionality, and Again, Study 2 was excluded from this because the manip-
the 95% prior credible interval is approximately [−1.96, ulation is nested within the receiving resources condition.
1.96]. All residuals are assigned an a priori distribution The context did not appear to exert an effect on grateful
of ~ Cauchy(0,1)+, following the recommendations of motivations. The partially pooled interactive effect of con-
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

Gelman (2006). Estimation was performed using Stan, a text and receiving resources on the motivation to express
Hamiltonian Monte Carlo sampler for probabilistic mode- gratitude was, likewise, negligible, M = −0.180 [−0.54, 0.22].
ling (Carpenter et al., 2017). The shrunken estimates for the four studies were as fol-
The posterior partially pooled effect [and 95% poste- lows: M = −0.100 [−0.42, 0.27], M = −0.27 [−0.56, −0.01],
rior credible intervals] of receiving resources from a bene- M = −0.226 [−0.66, 0.15], and M = −0.138 [−0.48, 0.28]. Of
factor, rather than by chance, on distribution of resources these, only benefactor intention predictably altered the
to the participant was consistent and large, M = 0.99 [0.60, effect of receiving resources on grateful motivations. In a
1.29]. However, the posterior expected partially pooled supplementary analysis, the same pattern of results were
effect of context on distribution was negligible with con- observed for the effect of receiving resources on experi-
siderable variability, M = 0.11 [−0.48, 0.65]. The shrunken enced gratitude toward the participant.
estimates implicate that the context in Study 4 (anon-
ymous distribution) decreased distribution, M  =  −0.30
Partially pooled moderated mediational path model
[−0.65, 0.00]. However, the effect of context in Study 3
(knowing the participant will distribute later) was neg- Each study above included a serial moderated media-
ligible, M = 0.002 [−0.36, 0.35], as was the effect of simi- tional analysis, described as follows: Receiving resources
larity, M = 0.006 [−0.20, 0.30] (Study 2 was not included, from another participant predicts grateful emotions (after
as the effect of benefactor intention is nested within controlling for indebted emotions), which in turn predicts
the effect of receiving resources from the participant). the motivation to express gratitude (after controlling for
Turning to the interactive effect of context and receiv- perceived obligation), which finally predicts the number
ing resources from a participant, the overall interactive of tickets given to the participant. Each study included a
effect between the contexts and receiving resources from different moderator, all expected to alter this mediational
the person was, again, negligible but expected to vary, pathway. In Study 1, dissimilarity was expected to alter
M  =  −0.01 [−0.68, 0.44]. The moderating effects of the the relationship between receiving resources and grateful
contexts in Studies 1, 2, and 4 were negligible, M = −0.10 emotions. In Study 2, messages that suggested the recip-
[−0.56, 0.32], M  =  0.10 [−0.25, 0.48], M  =  −0.01 [−0.52, ient now owes them were expected to alter the relation-
0.50], respectively. As noted in the singular analysis above, ship between receiving resources and grateful emotions.
Study 3 may suggest an unexpected interaction between In Study 3, expecting a future interaction was predicted
receiving resources from a participant and knowing the to alter the relationship between grateful motivations
participant will be distributing in the future. In contrast and distributions due to an assumed increase in social
to the egoistic hypothesis, those who received resources exchange concerns. Finally, in Study 4, distributing anon-
from the participant provided fewer tickets when the ymously was predicted to alter the relationship between
participant was expected to distribute later. However, grateful motivations and distributions due to a decreased
after pooling information from other studies, the effect role of self-presentation. Thus, all of these contexts should
was less probable than even the marginally significant affect the mediational role of gratitude.
estimate before, M  =  −0.40 [−1.15, 0.19]. Therefore, we These models were nearly identical across studies and
suggest there is little evidence that these contexts, over- attempted to estimate many of the same parameters. We
all, had a marginal main effect or interactive effect on once again employed a Bayesian hierarchical partially
distribution. Nevertheless, distributing anonymously reli- pooled model to obtain better estimates of each shared
ably predicted lesser distribution of tickets, but does not path coefficient, the shrunken mediational effects, the
interact with the effect of benefaction. overall mediational effect, and the moderating roles of
14   J.-A. TSANG AND S. R. MARTIN

these undermining contexts on the mediational effect. the mediated effect (see Figure 2). The moderating effect
All variables were standardized, the interaction variables of benefactor intention was also negligible, m  =  −0.26
were created from the standardized data, and conditions [−0.56, 0.03]. Even with a 95.96% probability of reducing
were coded to be effects-coded to be one unit apart. See the effect of benefaction on grateful emotions, the 95%
Figure 1 for the model and posterior estimates. most probable effects were not sufficient in nullifying the
Let a, b, c, and d represent the pathways between ben- effects of benefaction on grateful emotions, and bene-
efaction and emotional gratitude, emotional gratitude factor intention did not meaningfully modify the indirect
and motivation to express gratitude, gratitude expres- effect through gratitude (see Figure 2).
sion and distribution, and benefaction and distribution, The effect of future interaction on the relationship
respectively. Pathways a, b, c, and d had a 99% posterior between grateful expression and distribution was incon-
probability of being positive; this was true for all shrunken sistent, m = 0.398 [−0.15, 0.95]. Future interaction only had
estimates and the hierarchical path estimates. Indebted a 7.05% probability of decreasing the effect of expression
emotions and obligation both had a 99% posterior proba- on distribution, and of the negative effects, none were
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

bility of positively predicting grateful emotion and grateful strong enough to negate the relationship. This was incon-
expression motivation, respectively. The pooled, marginal sistent with an egoistic interpretation of grateful prosocial-
indirect effect of benefaction on resources given, through ity. If anything, the future interaction condition increased
grateful emotions and motivations, had a 98.7% posterior the indirect effect of gratitude relative to the no future
probability of being positive, with 95% credible intervals interaction condition, mno-future − mfuture  =  −0.23 [−0.58,
[0.05, 0.40], after controlling for indebted emotions and 0.08]. Finally, the effect of anonymity on the relationship
obligation. Likewise, the shrunken estimates of the indirect between grateful expression and distribution was notably
effect for each study all had a 99% posterior probability of positive, m = 0.497 [0.05, 0.94]. With a 98.55% probability
being positive. of increasing the relationship between grateful expression
The interactional effects of the manipulation onto the and distribution, and a 94.38% probability of decreasing
hypothesized pathways were not partially pooled, because the direct effect of benefaction on distribution, the ano-
they were not presumed to estimate similar parameters. nymity manipulation in Study 4 increased the mediational
Nevertheless, the interaction parameters were presumed role of gratitude. One may interpret this pattern of results
to modify shrunken estimates of path parameters, and as suggesting that, when anonymous, gratitude is a strong
thus indirectly used the partial pooling methodology. The mediator of the effect of benefaction on giving, but when
effect of dissimilarity on the relationship between bene- not anonymous, other motivations may be activated. That
faction was negligible, m = −0.21 [−0.62, 0.17]. Even with is, although individuals on average distribute fewer tickets
an 85% probability of decreasing the relationship, the 95% when anonymous, those who do, do so through grateful
most probable effects were not enough to notably change emotions and a desire to express gratitude. Indeed, the

Figure 1 Partially pooled hierarchical Bayesian path model.


Notes: Dashed lines represent the posterior distributions of the shrunken path coefficient for each study. The study is labeled above each dashed posterior. The
dotted line is the prior for the path. The solid lines represent the posterior distributions of the pooled estimates.
THE JOURNAL OF POSITIVE PSYCHOLOGY   15

Figure 2. (Left) The posterior distribution(s) for the indirect effect of benefaction through grateful emotions and motivation to express
gratitude. (Right) The posterior distributions for the difference of indirect effect estimates between conditions: Similar – Dissimilar, No
Owe – Owe, No future interaction – Future interaction, and No knowledge – Knowledge.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

difference in mediational effects between the two con- meaningfully affect grateful emotions, the effect was too
ditions was notable, mno-knowledge − mknowledge = 0.28 [0.03, small to influence the mediational effect of gratitude on
0.59]. resource distribution. Next, we investigated whether cer-
However, this may not necessarily be the case. The tain characteristics of the gratitude situation might make
resource distributions in Studies 1–3 were all similar to the egoism or altruism a stronger motivation in grateful proso-
non-anonymity condition of Study 4, because the partic- ciality. We predicted that, if grateful prosociality was moti-
ipants knew the recipient could see who distributed to vated by egoism, then expecting a future interaction and
them. With this in mind, we compared the mediational distributing non-anonymously to the benefactor would
effect in the anonymity condition in Study 4 to the media- increase the number of resources distributed. However, if
tional effects from Studies 1–3. The indirect effect for those grateful prosociality is in part motivated by altruism, then
in the anonymous condition in Study 4 was not different prosocial behaviors should not decrease when future inter-
from the indirect effects for those in Studies 1–3, M = 0.09 actions and anonymity are altered. Although anonymous
[−0.13, 0.31], M = 0.05 [−0.17, 0.28], M = 0.05 [−0.17, 0.29], distribution indeed predicted decreased prosocial behav-
respectively. However, the indirect effect in the non-anon- ior, anonymity and the expectation of a future interaction,
ymous condition in Study 4 differed consistently from if anything, increased the mediating role of gratitude; this
the indirect effects observed in Studies 1–3, M  =  −0.19 latter effect, however, is less dependable compared to the
[−0.42, 0.00], M = −0.23 [−0.43, −0.04], M = −0.23 [−0.44, other findings.
−0.03]. This is puzzling, given that these conditions should These experiments contained a number of strengths.
be equivalent. Because we observed significant indirect The laboratory setting allowed us to design four experi-
effects in Studies 1–3, but not in the non-anonymous con- ments with similar gratitude inductions and thus provide
dition of Study 4, we suggest that the lack of an indirect multiple replications of the effect of a favor on gratitude
effect in the non-anonymous condition of Study 4 may and grateful prosociality. Another strength is that the
be due to sampling error, and thus the relatively larger experimental design allowed us to test the causality
indirect effect in the anonymous condition may in actuality between variables of interest and gratitude. The use of a
be due to a chance underestimate of the indirect effect of behavioral measure of grateful prosociality helped address
the non-anonymous condition. social desirability concerns associated with gratitude and
aided in the investigation of potential prosocial outcomes
associated with gratitude.
General discussion
Along with strengths come associated limitations. For
Four experiments were conducted to assess the impact of one, these experiments were conducted in a laboratory
benefactor (similarity, intention) and situational character- setting. Although this helped with controlling extraneous
istics (future interaction, anonymity) on responses within a variables, thereby increasing internal validity, it created an
resource distribution paradigm. We consistently found that artificial environment. We hoped that the elaborate cover
the receipt of a favor increased prosocial behavior, and this stories and the existence of a monetary prize helped to
effect was mediated by increases in grateful emotion and offset the artificiality with increased psychological realism.
grateful motivation, even controlling for feelings of indebt- In a related limitation, laboratory-induced gratitude can be
edness and self-reported obligated motivation. Benefactor less intense than grateful reactions to more dramatic real-
similarity did not meaningfully influence either distri- life events. However, many everyday instances of gratitude
bution or gratitude. Although benefactor intention did are in response to small favors, so it is important to study
16   J.-A. TSANG AND S. R. MARTIN

this level of gratitude in addition to more intense reac- self. Gratitude may lead us to help benefactors even when
tions. Another thing to note is that benefactors in these they lack the potential to help us again in the future, or
experiments were unseen strangers. Although this could when they will never know the source of our help. In this
provide a strong test of the ‘find’ component in Algoe’s way, gratitude might contribute to unsung acts of car-
(2012) ‘find-remind-bind’ theory of gratitude, it remains ing that can help form and solidify our important social
to be seen whether the results of these studies apply to relationships.
gratitude in existing relationships. Additionally, we did not
recruit enough male participants in this study to explore
Disclosure statement
gender differences in gratitude and grateful prosociality.
Previous research has documented that women tend to No potential conflict of interest was reported by the authors.
rate themselves as higher on gratitude than do men (Karris
& Craighead, 2012) and that men may feel uncomfortable ORCID
expressing and experiencing gratitude (Kashdan, Mishra,
Stephen R. Martin   http://orcid.org/0000-0001-8085-2390
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

Breen, & Froh, 2009). Further research is needed on the


potential for gender to moderate the relationship between
gratitude and prosocial behavior. References
Despite these limitations, this research paves the way
Algoe, S. B. & Haidt, J. (2009). Witnessing excellence in action:
for future studies in gratitude. Although our null effects of The 'other-praising' emotions of elevation, gratitude, and
benefactor similarity may contradict the relational theory admiration. The Journal of Positive Psychology, 4, 105–127.
of gratitude – specifically, that gratitude helps people find doi:10.1080/17439760802650519.
good relationship partners (Algoe, 2012), we are not the Algoe, S. B. (2012). Find, remind, and bind: The functions of
first to find a nonsignificant relationship between benefac- gratitude in everyday relationships. Social and Personality
Psychology Compass, 6, 455–469.
tor similarity and gratitude. Weinstein et al. (2010) did not
Algoe, S. B., & Way, B. M. (2014). Evidence for a role of the
find a significant relationship between self-reported grat- oxytocin system, indexed by genetic variation in CD38, in the
itude and perceptions of benefactor similarity in response social bonding effects of expressed gratitude. Social Cognitive
to a hypothetical helping scenario. Although previous and Affective Neuroscience, 9, 1855–1861. doi:10.1093/scan/
research has found links between attitude similarity and nst182
attraction (e.g. Bond, Byrne, & Diamond, 1968; Byrne & Algoe, S. B., & Zhaoyang, R. (2016). Positive psychology in context:
Effects of expressing gratitude in ongoing relationships
Blaylock, 1963), it may be that attitude similarity between depend on perceptions of enactor responsiveness. The
unseen strangers in a laboratory is not a strong enough Journal of Positive Psychology, 11, 399–415. doi:10.1080/174
indicator of high benefactor relationship potential. Future 39760.2015.1117131
research could explore whether other variables related to Algoe, S. B., Haidt, J., & Gable, S. L. (2008). Beyond reciprocity:
attraction, such as physical attractiveness (e.g. Eastwick, Gratitude and relationships in everyday life. Emotion, 8, 425–
429.
Eagly, Finkel, & Johnson, 2011), reciprocal attraction (Aron,
Algoe, S. B., Gable, S. L., & Maisel, N. C. (2010). It’s the little
Dutton, Aron, & Iverson, 1989), or mere exposure (Saegert, things: Everyday gratitude as a booster shot for romantic
Swap, & Zajonc, 1973) would lead to differences in grati- relationships. Personal Relationships, 17, 217–233.
tude in response to a favor. doi:10.1111/j.1475-6811.2010.01273.x
Our findings also hint at the altruistic potential of Aristotle (1905). Politics. Oxford: Oxford University Press.
grateful prosocial behavior. The ability to receive further Aronson, E., Ellsworth, P. C., Carlsmith, J. M., & Gonzales, M. H.
(1990). Methods of research in social psychology (2nd ed.). New
benefits from a benefactor affected neither felt grati- York, NY: McGraw-Hill.
tude nor prosocial behavior, suggesting that this egoistic Aron, A., Dutton, D. G., Aron, E. N., & Iverson, A. (1989).
motivation was not a motivator for grateful prosociality in Experiences of falling in love. Journal of Social and Personal
this context. Making the prosocial behavior anonymous Relationships, 6, 243–257.
decreased prosociality, but the prosocial behavior that did Barrett-Cheetham, E., Williams, L. A., & Bednall, T. C. (2016).
A differentiated approach to the link between positive
occur in the anonymous context appeared to be motivated
emotion, motivation, and eudaimonic well-being. The
by gratitude, again suggesting an altruistic component to Journal of Positive Psychology, 11, 595–608. doi:10.1080/174
gratitude, at least in some situations. Future research could 39760.2016.1152502
explore whether the manipulation of altruism-relevant var- Bartlett, M. Y., & DeSteno, D. (2006). Gratitude and prosocial
iables such as empathy or relationship closeness (Batson, behavior: Helping when it costs you. Psychological Science,
2011) increases the altruistic potential of gratitude. 17, 319–325.
Bartlett, M. Y., Condon, P., Cruz, J., Baumann, J., & Desteno,
In support of its relational function, gratitude may D. (2012). Gratitude: Prompting behaviours that build
thus emphasize motivations centered on the welfare of relationships. Cognition and Emotion, 26, 2–13. doi:10.1080/
others, rather than an exclusive focus on benefits for the 02699931.2011.561297
THE JOURNAL OF POSITIVE PSYCHOLOGY   17

Batson, C. D. (2011). Altruism in humans. New York, NY: Oxford Gordon, A. M., Impett, E. A., Kogan, A., Oveis, C., & Keltner, D.
University Press. (2012). To have and to hold: Gratitude promotes relationship
Bond, M., Byrne, D., & Diamond, M. J. (1968). Effect of maintenance in intimate bonds. Journal of Personality and
occupational prestige and attitude similarity on attraction Social Psychology, 103, 257–274. doi:10.1037/a0028723
as a function of assumed similarity of attitude. Psychological Graham, S. (1988). Children’s developing understanding of the
Reports, 23, 1167–1172. motivational role of affect: An attributional analysis. Cognitive
Buck, R. (2004). The gratitude of exchange and the gratitude Development, 3, 71–88.
of caring: A developmental interactionist perspective of Gruszecka, E. (2014). The gratitude in close and less close
moral emotion. In R. A. Emmons & M. E. McCullough (Eds.), relationships. Przegląd Psychologiczny, 57, 447–461.
The Psychology of Gratitude (pp. 100–122). Oxford: Oxford Hendrickson, B., & Goei, R. (2009). Reciprocity and dating:
University Press. Explaining the effects of favor and status on compliance
Byrne, D., & Blaylock, B. (1963). Similarity and assumed similarity with a date request. Communication Research, 36, 585–608.
of attitudes among husbands and wives. Journal of Abnormal doi:10.1177/0093650209333036
and Social Psychology, 67, 636–640. Joel, S., Gordon, A. M., Impett, E. A., MacDonald, G., & Keltner, D.
Carpenter, B., Gelman, A., Hoffman, M. D., Lee, D., Goodrich, B., (2013). The things you do for me: Perceptions of a romantic
Betancourt, M., … Riddell, A. (2017). Stan: A probabilistic partner’s investments promote gratitude and commitment.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

programming language. Journal of Statistical Software, 76(1), Personality and Social Psychology Bulletin, 39, 1333–1345.
doi:10.18637/jss.v076.i01 Jones, E. E., & Pittman, T. S. (1982). Toward a general theory
Chang, Y., Lin, Y., & Chen, L. (2012). Pay it forward: Gratitude in of strategic self-presentation. In J. Suls (Ed.), Psychological
social networks. Journal of Happiness Studies, 13, 761–781. perspectives on the self, 1, 231–262. Hillsdale: Lawrence
doi:10.1007/s10902-011-9289-z Erlbaum Associates.
DeSteno, D., Bartlett, M. Y., Baumann, J., Williams, L. A., & Dickens, Karris, M. A., & Craighead, W. E. (2012). Differences in character
L. (2010). Gratitude as moral sentiment: Emotion-guided among U.S. college students. Individual Differences Research,
cooperation in economic exchange. Emotion, 10, 289–293. 10, 69–80.
doi:10.1037/a0017883 Kashdan, T. B., Mishra, A., Breen, W. E., & Froh, J. J. (2009).
Dijker, A. J. M., Nelissen, R. M. A., & Stijnen, M. M. N. (2013). Gender differences in gratitude: Examining appraisals,
Framing posthumous organ donation in terms of reciprocity: narratives, the willingness to express emotions, and changes
What are the emotional consequences? Basic and Applied in psychological needs. Journal of Personality, 77, 691–730.
Social Psychology, 35, 256–264. doi:10.1080/01973533.2013. doi:10.1111/j.1467-6494.2009.00562.x
785401 Kubacka, K. E., Finkenauer, C., Rusbult, C. E., & Keijsers, L. (2011).
Eastwick, P. W., Eagly, A. H., Finkel, E. J., & Johnson, S. E. (2011). Maintaining close relationships: Gratitude as a motivator and
Implicit and explicit preferences for physical attractiveness in a detector of maintenance behavior. Personality and Social
a romantic partner: A double dissociation in predictive Psychology Bulletin, 37, 1362–1375.
validity. Journal of Personality and Social Psychology, 101, Lambert, N. M., Clark, M. S., Durtschi, J., Fincham, F. D., &
993–1011. Graham, S. M. (2010). Benefits of expressing gratitude:
Emmons, R. A., & McCullough, M. E. (2003). Counting blessings Expressing gratitude to a partner changes one’s view
versus burdens: An experimental investigation of gratitude of the relationship. Psychological Science, 21, 574–580.
and subjective well-being in daily life. Journal of Personality doi:10.1177/0956797610364003
and Social Psychology, 84, 377–389. Lane, J., & Anderson, N. H. (1976). Integration of intention and
Fox, G. R., Kaplan, J., Damasio, H., & Damasio, A. (2015). Neural outcome in moral judgment. Memory and Cognition, 4, 1–5.
correlates of gratitude. Frontiers in Psychology, 6, 1491. Li, K., & Chow, W. (2015). Religiosity/spirituality and prosocial
Froh, J. J., Bono, G., & Emmons, R. (2010). Being grateful is beyond behaviors among Chinese Christian adolescents: The
good manners: Gratitude and motivation to contribute to mediating role of values and gratitude. Psychology of Religion
society among early adolescents. Motivation and Emotion, 34, and Spirituality, 7, 150–161. doi:10.1037/a0038294
144–157. doi:10.1007/s11031-010-9163-z Ma, L. K., Tunney, R. J., & Ferguson, E. (2014). Gratefully
Froh, J. J., Bono, G., Fan, J., Emmons, R. A., Henderson, K., Harris, received, gratefully repaid: The role of perceived fairness in
C., & Wood, A. M. (2014). Nice thinking! An educational cooperative interactions. PLoS One, 9(12), 15. doi:10.1371/
intervention that teaches children to think gratefully. School journal.pone.0114976
Psychology Review, 43, 132–152. MacKenzie, M. J., Vohs, K. D., & Baumeister, R. F. (2014). You
Gelman, A. (2006). Prior distributions for variance parameters in didn’t have to do that: Belief in free will promotes gratitude.
hierarchical models. Bayesian Analysis, 1, 515–533. Personality and Social Psychology Bulletin, 40, 1423–1434.
Goei, R., Roberto, A., Meyer, G., & Carlyle, K. (2007). The effects of doi:10.1177/0146167214549322
favor and apology on compliance. Communication Research, Mathews, M. A., & Green, J. D. (2010). Looking at me, appreciating
34, 575–595. doi:10.1177/0093650207307896 you: Self-focused attention distinguishes between gratitude
Gordon, A. M., & Chen, S. (2013). Does power help or hurt? and indebtedness. Cognition and Emotion, 24, 710–718.
The moderating role of self–other focus on power and doi:10.1080/02699930802650796
perspective-taking in romantic relationships. Personality and McCullough, M. E., Emmons, R. A., & Tsang, J. (2002). The grateful
Social Psychology Bulletin, 39, 1097–1110. disposition: A conceptual and empirical topography. Journal
Gordon, C. L., Arnette, R. A. M., & Smith, R. E. (2011). Have you of Personality and Social Psychology, 82, 112–127.
thanked your spouse today?: Felt and expressed gratitude McCullough, M. E., Kilpatrick, S. D., Emmons, R. A., & Larson, D. B.
among married couples. Personality and Individual Differences, (2001). Is gratitude a moral affect? Psychological Bulletin, 127,
50, 339–343. doi:10.1016/j.paid.2010.10.012 249–266.
18   J.-A. TSANG AND S. R. MARTIN

McCullough, M. E., Kimeldorf, M. B., & Cohen, A. D. (2006). An Simão, C., & Seibt, B. (2015). Friendly touch increases gratitude
adaptation for altruism? The social causes, social effects, by inducing communal feelings. Frontiers in Psychology, 6,
and social evolution of gratitude. Current Directions in 815.
Psychological Science, 17, 281–285. Simmel, G. (1950). The sociology of Georg Simmel. (K. H. Wolff,
Michie, S. (2009). Pride and gratitude: How positive emotions Trans., Ed.). Glencoe, IL: The Free Press.
influence the prosocial behaviors of organizational leaders. Sun, P., Jiang, H., Chu, M., & Qian, F. (2014). Gratitude and
Journal of Leadership & Organizational Studies, 15, 393–403. school well-being among Chinese university students:
doi:10.1177/1548051809333338 Interpersonal relationships and social support as mediators.
Montoya, R. M., & Horton, R. S. (2013). A meta-analytic investigation Social Behavior and Personality, 42, 1689–1698. doi:10.2224/
of the processes underlying the similarity-attraction effect. sbp.2014.42.10.1689
Journal of Social and Personal Relationships, 30, 64–94. Tesser, A., Gatewood, R., & Driver, M. (1968). Some determinants
Montoya, R. M., Horton, R. S., & Kirchner, J. (2008). Is actual of gratitude. Journal of Personality and Social Psychology, 9,
similarity necessary for attraction? A meta-analysis of actual 233–236.
and perceived similarity. Journal of Social and Personal Trivers, R. L. (1971). The evolution of reciprocal altruism.
Relationships, 25, 889–922. Quarterly Review of Biology, 46, 35–57.
Naito, T., & Sakata, Y. (2010). Gratitude, indebtedness, and Tsang, J. (2006a). The effects of helper intention on
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 05:48 13 October 2017

regret on receiving a friend’s favor in Japan. Psychologia: An gratitude and indebtedness. Motivation and Emotion, 30,
International Journal of Psychological Sciences, 53, 179–194. 198–204.
doi:10.2117/psysoc.2010.179 Tsang, J. (2006b). Gratitude and prosocial behavior: An
Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can experimental test of gratitude. Cognition and Emotion, 20,
know: Verbal reports on mental processes. Psychological 138–148.
Review, 84, 231–259. Tsang, J.-A. (2007). Gratitude for small and large favors: A
O'Connell, B. H., O'Shea, D., & Gallagher, S. (2016). Enhancing behavioral test. The Journal of Positive Psychology, 2(3) 157–
social relationships through positive psychology activities: A 167.
randomised controlled trial. The Journal of Positive Psychology, Tsang, J., Schulwitz, A., & Carlisle, R. D. (2012). An experimental
11, 149–162. doi:10.1080/17439760.2015.1037860 test of the relationship between religion and gratitude.
Pelser, J., de Ruyter, K., Wetzels, M., Grewal, D., Cox, D., & van Psychology of Religion and Spirituality, 4, 40–55.
Beuningen, J. (2015). B2B channel partner programs: Watkins, P. C., Scheer, J., Ovnicek, M., & Kolts, R. (2006). The
Disentangling indebtedness from gratitude. Journal of debt of gratitude: Dissociating gratitude and indebtedness.
Retailing, 91, 660–678. doi:10.1016/j.jretai.2015.05.006 Cognition and Emotion, 20, 217–241.
Puente-Diaz, R., & Cavazos-Arroyo, J. (2016). How remembering less Weinstein, N., DeHaan, C. R., & Ryan, R. M. (2010). Attributing
acts of gratitude can make one feel more grateful and satisfied autonomous versus introjected motivation to helpers and
with close relationships: The role of ease of recall. European the recipient experience: Effects on gratitude, attitudes,
Journal of Social Psychology. Online first posting. Retrieved from and well-being. Motivation and Emotion, 34, 418–431.
https://doi.org.ezproxy.baylor.edu/10.1002/ejsp.2177 doi:10.1007/s11031-010-9183-8
Rotkirch, A., Lyons, M., David-Barrett, T., & Jokela, M. (2014). Williams, L. A., & Bartlett, M. Y. (2015). Warm thanks: Gratitude
Gratitude for help among adult friends and siblings. expression facilitates social affiliation in new relationships
Evolutionary Psychology, 12, 673–686. via perceived warmth. Emotion, 15, 1–5. doi:10.1037/
Saegert, S., Swap, W., & Zajonc, R. B. (1973). Exposure, context, emo0000017
and interpersonal attraction. Journal of Personality and Social Wood, A. M., Maltby, J., Stewart, N., Linley, P. A., & Joseph, S. (2008).
Psychology, 25, 234–242. A social-cognitive model of trait and state levels of gratitude.
Sikkel, D., & Schoenmakers, E. (2012). Bequests to health-related Emotion, 8, 281–290. doi:10.1037/1528-3542.8.2.281
charitable organisations: A structural model. International Wood, A. M., Froh, J. J., & Geraghty, A. W. A. (2010). Gratitude
Journal of Nonprofit and Voluntary Sector Marketing, 17, 183– and well-being: A review and theoretical integration. Clinical
197. doi:10.1002/nvsm.1421 Psychology Review, 30, 890–905.

You might also like