Ernan Haruvy 1
June, 2018
1
Professor of Marketing, The University of Texas at Dallas, Jindal School of Management, SM
32, 800 W. Campbell Rd., Richardson, TX 75080. Phone: 972-883-4865; Fax: 972-883-6727;
E-mail: eharuvy@utdallas.edu
2
Eric Geddes Professor of Business, School of Business, University of Alberta, Edmonton, AB,
Canada, T6G 2R6; Phone: (780) 492-1866; E-mail: ppopkows@ualberta.ca.
1
Charities increasingly utilize social media tools in fundraising. This research studies three
mechanisms by which Facebook “likes” affect charitable behavior: (1) signaling, (2)
commitment escalation, and (3) social contagion. The first study manipulates the number of
Facebook likes on a page and finds support for a signaling effect of the aggregate Facebook likes
on charitable intent. The second study manipulates whether the visitor clicks a Facebook like or
not. Results show that the act of “liking” a fundraising page in and of itself escalates one’s
charitable intent. An auction field study sells artwork created by under-privileged youth, and
utilizes Facebook and Facebook likes to disseminate word of mouth about these auctions. Results
show that social media increases willingness to pay (WTP) in charity auctions. The number of
Facebook likes has a direct signaling effect and an indirect social contagion effect on WTP, as
Facebook friends pass on information, through their likes. Consistent with the commitment
Social media platforms are effective in disseminating information (Toubia and Stephen
2013), particularly for social causes (Madianou 2013; Saxton and Wang 2014) and non-profits
(Novak 2012; Valverde et al. 2013). With recent increased competition for charity dollars, non-
profits are increasingly interested in using social media, such as Facebook and Twitter, to
promote their causes. A recent survey (Ganim Barnes, 2014) studying the usage of social media
among US charities indicates that 98% of the largest charities use at least one form of social
media. This increased ubiquity of social media usage by non-profits has resulted in a sizable
literature on ways that non-profits can implement social media as part of their strategy (e.g.,
Briones et al. 2011; Campbell, Lambright, and Wells 2014; Waters et al. 2009; Young 2017).
Therefore, the main objective of the current research is to fill this void and focus on the impact
of social media on fundraising. In particular, we will focus on the effect of Facebook likes on
charitable fundraising. We focus on Facebook likes as the primary social media vehicle, with
over 2.2 billion active users monthly (Zephoria 2018). Facebook likes can affect charitable
First, we build on the knowledge that consumer actions in the charity space inevitably
involve some social signalling (e.g., Jung et al., 2017). Specifically, aggregate Facebook likes
provide a signal or social proof about the quality, value, virtue, or credibility of the charity.
Establishing credibility is of great importance to charities, and emerging evidence suggests social
media can help provide that credibility (Curtis et al. 2010). A demonstration that Facebook likes
1
Saxton and Wang (2014) studied factors that influence donations to charity on Facebook causes. (Causes.com is a
social network environment created by Facebook where people can organize campaigns, support fundraisers, and
petitions.) They found that, in general, donations are small and are influenced by network effects (i.e., the number of
members on Facebook causes) and website specific variables, but not the charity’s financial capacity and efficiency
ratios. In addition, there is some evidence of how information sharing affects individuals’ contributions to
crowdfunding campaign (Tan, Lu, and Tan 2017).
3
can assist charities with establishing credibility is an important contribution to charity marketing.
Towards that end, in the first study, we isolate and study the effect of the signal of the total
guests towards environmental towel use resulted in subsequent substantial change in behavior.
Likewise, the act of clicking on the Facebook like may escalate one’s liking or commitment to
spend more than nonmembers (Lipsman et al. 2012). In Manchanda et al. (2015), customers were
invited to join a firm’s new online community. Comparing purchases before and after launch for
those who joined versus those who did not join revealed that approximately a fifth of post-launch
revenue was attributed to the act of joining. Although social media is particularly effective in
relationship escalation (Sosik and Bazarova, 2014), John et al. (2017) argue these proven effects
could be due to self-selection of those prone to commitment escalation being more likely to click
a Facebook like. In that study, controlling for self-selection, the authors found no commitment
escalation. This finding suggests that people who elect to like a cause might be predisposed to
acting in a manner favorable to that cause. Thus, distinguishing the impact of the Facebook like
from the self-selection requires careful analysis, which we conduct in the second study, where
we isolate the effect of self-selection from escalation of charitable intent due to clicking on the
Facebook like.
Third, Facebook likes may have an indirect social-contagion effect, as Facebook friends pass
on information, through their likes or messages. Clicking the Facebook like influences Facebook
information dissemination known as social contagion (Aral 2011; Aral and Walker, 2011;
We pursue these questions with auction experiments. Auction experiments (e.g., Bagchi and
Cheema, 2012; Schmidt et al. 2014) have been shown effective at assessing bidder sensitivities
to contextual and social stimuli. In Study 3 —an auction field study— we measure the indirect
effect as well as the direct effects and further disentangle the signaling effect from the escalation
effect when the two co-exist in the field. We distinguish the escalation effect of clicking a
Facebook like from the signaling effect of total likes, by accounting for the potential role of
Facebook impressions and bidding intensity in mediating the effect of Facebook likes on WTP.
We highlight several contributions. First, we show that the total number of Facebook likes
sends an informative signal to potential donors, resulting in (1) increased favorability towards the
promoted charity and (2) a higher willingness to contribute and pay to the charity. Second, we
test for and observe that clicking the Facebook like has a positive effect on one’s commitment to
the cause. This finding extends the work by John et al. (2017), who reported that clicking the
Facebook like does not increase brand favorability in a for-profit setting. Specifically, we show
that clicking the Facebook like results in a significantly higher willingness to contribute to
charity, and it marginally increases favorability. In testing for this effect, controlling for self-
selection is crucial, because as those who click on the Facebook like are generally more
favorable towards a cause. Third, we test for the direct and indirect effect of Facebook likes on
fundraising revenue and WTP in a field study. We find that the total number of likes has both a
direct signaling effect and an indirect effect through Facebook impression and bidding intensity
on WTP. In addition, we find a direct effect of clicking the Facebook like that is two to three
times greater than that for the total number of Facebook likes, depending on the category.
5
THE EXPERIMENTS
We conducted three studies with the simple objective of testing whether a social media
campaign on Facebook can improve fundraising outcomes. We used charity settings for all three
studies and focused on the impact of the Facebook likes. The first was an online study that
examined the impact of different Facebook like totals on consumers’ perceptions and intentions
to contribute to charity. This format allowed us to test for a potential signaling effect of the total
number of likes. The second study was an online study that examined the impact of clicking the
Facebook like on consumers’ intentions to contribute to the charity and their perceptions of the
charity. Finally, the third study was a field study that examined the influence of social media,
STUDY 1
Objectives and hypothesis of Study 1. The objective of this study was to test for a potential
signaling effect of the total number of Facebook likes on willingness to contribute. The number
of Facebook likes may provide a signal to potential buyers (Bikhchandani et al. 1998; Zhang and
Liu 2012), which could pertain to quality, value, or to social benefit (Mathwick et al. 2008;
These arguments suggest that the total number of Facebook likes sends a signal to visitors
about the quality or value of the charity, resulting in favorable perceptions towards the charity.
Our first hypothesis involves a direct relationship between social media activity and preference
for the entity being promoted. We operationalize preference towards the charity by measuring
respondents perceived favorability towards (1) the charity’s activities, (2) the charity’s efforts,
and (3) the charity’s organization, on a 7-point scale (1 = very unfavorable, 7 = very favorable).
H1: The total number of Facebook likes increases favorability towards the promoted charity.
We posit that this favorability is associated with intended contribution. We use four
measures to estimate respondents' willingness to contribute to charity: (1) intent to volunteer, (2)
intent to spread favorable WOM, (3) intent to pay attention to communication, and (4) intent to
donate (all measured based on a 7-point scale, where 1 = very unlikely, 7 = very likely).
Hypothesis 2 follows:
H2: The total number of Facebook likes increases willingness to contribute to charity.
Design of Study 1. To test the signaling effect of the total number of Facebook likes on
intention to donate (volunteer), we conducted an online experiment with a 2 x 2 design along the
dimensions {High Facebook like total, Low Facebook like total} x {High charity familiarity,
Low charity familiarity}. The number of Facebook likes was either 13 million or 13. The two
charities selected were St. Jude (high familiarity) and Rohinga (low familiarity), with the
expectation that a potential signaling effect of the total number of Facebook likes is stronger for
an unfamiliar charity. We recruited 400 participants from Amazon Mechanical Turk (MTurk)
and randomly assigned them to the four different conditions. Participants first answered a
number of demographic and charitable orientation questions. Next, we showed them a picture of
the front page from each charity, with the number of Facebook likes—either 13 million or 13—
clearly visible at the top of the page (see Figure 1). Participants then answered questions about
their familiarity with the charity, perceptions about the number of Facebook likes, and the
constructs measuring favorability and intent to contribute as stated earlier. Web Appendix A
The results of Study 1 are shown in Table 1. The top part of Table 1 provides the results for
7
the effect of the number of Facebook likes on respondents’ perceptions of the favorability of the
charity’s activities, efforts, and organization (H1). The t-tests show that perceived favorability of
Figure 1. Stimuli used for Study 1 --Different charities and number of Facebook Likes
------------------------------------------------------------------------------------------
8
activities, effort, and organization is significantly higher with a higher number of Facebook likes
than with a low number of likes. These results hold both for the familiar and the unfamiliar
charity. Although the favorability is higher for the more familiar charity (for both low and high
numbers of Facebook likes), the increase is somewhat higher for the unfamiliar charity.
The bottom part of Table 1 shows the results for intentions to volunteer, spread favorable
WOM, pay attention to communication, and donate to the charity (H2). The t-tests show that all
intentions are statistically significantly higher with a higher number of Facebook likes, in support
H1: Total number of Facebook likes increases favorability of the promoted entity
Low familiar charity – Rohinga
Perceived favor-
ability of charity's: Activities Efforts Organization
Low number of likes 4.98 (.15) 4.98 (.14) 4.84 (.14)
High number of likes 5.72 (.10) 5.69 (.10) 5.64 (.10)
T-test 4.17 4.11 4.66
p-value <0.001 <0.001 <0.001
High familiar charity -- St. Jude
Low number of likes 5.42 (.15) 5.46 (.16) 5.45 (.15)
High number of likes 5.86 (.11) 5.94 (.10) 6.01 (0.10)
T-test 2.33 2.61 3.11
p-value 0.021 0.009 0.002
H2: The total number of Facebook likes increases willingness to contribute.
Low familiar charity – Rohinga
Spread favorable Pay attention to Donate
Willingness to: Volunteer WOM communication
Low number of likes 3.56 (.18) 3.72 (.18) 3.86 (.19) 3.81 (.22)
High number of likes 4.22 (.20) 4.42 (.19) 4.78 (.19) 4.64 (.25)
T-test 2.47 2.68 3.42 2.48
p-value .015 .008 .001 .015
High familiar charity -- St. Jude
Low number of likes 3.95 (.20) 4.32 (0.21) 4.12 (.20) 4.42 (.24)
High number of likes 4.71 (.20) 5.15 (0.18) 4.93 (.19) 4.90 (.24)
T-test 2.69 3.01 2.96 1.41
p-value .008 .003 .004 0.162
9
of H2 for the unfamiliar charity. For the familiar charity, we find partial support for H2, because
all except for intentions to donate are significantly higher. Note that the intention to donate is
higher for the more familiar charity (with both low and high numbers of Facebook likes).
Thus, we find strong support for a signaling effect of the number of Facebook likes, and this
effect is stronger for the unfamiliar charity than for the familiar charity.
STUDY 2
Objectives and hypothesis of Study 2. The objective of Study 2 was to determine whether
clicking a Facebook like escalates a bidder’s commitment to a cause. We consider the possibility
that social media content may enhance one’s emotional connection to the item for sale, which
may lead to the act of clicking a Facebook like having a positive effect on one’s commitment to
the cause (Seidman 2013). The need to belong is a major motivator for Facebook use
(Baumeister and Leary 1995), as is self-presentation (Zhao, Grasmuck, and Martin 2008). Users
who wish to satisfy their need to belong tend to engage in strategic self-presentation and click
Facebook likes they deem as enhancing their image (Utz et al. 2012). Once an individual links
his self-image to an item, the value he or she places on this item may increase. In for-profit
situations, however, clicking a Facebook like does not cause a person to view a brand more
favorably (John et al. 2017). Specifically, John et al. (2017) asked participants recruited from
MTurk to click a Facebook like for a brand (three studies with three different brands: Coca-Cola,
Pepsi, and Burt’s Bees). They controlled incentives to click like (zero, 1 cent, 5 cents, or 50
cents2) as well as separating intent to click like from actually clicking like. The authors found
that clicking like for a brand on Facebook did not impact brand attitude or intent to purchase.
The major difference between the for-profit and cause-related setting we are investigating is that
2
We used the 50-cents incentives in our study, as in John et al. (2017)
10
in the current setting, consumers are more likely to have an emotional connection, and social
media is more beneficial when consumers have an emotional attachment (Bernritter et al. 2016).
community-related phenomenon. Manchanda et al. (2015) provide evidence for this community
effect. Specifically, they find a significant increase in customer expenditures after customers join
the firm’s community. They also discuss and largely rule out self-selection and other alternative
explanations.
Commitment escalation is not observed in all cases. For example, Kristofferson, White, and
Peloza (2014) studied token commitment in a cause-related setting. Although they did not study
Facebook likes explicitly, they showed that making a token commitment (publically wearing a
pin; signing a petition) to a charitable initiative did not increase people’s subsequent propensity
to provide meaningful support to that cause in terms of donation and volunteer labor.
Nonetheless, considerable evidence shows that involvement with charity has a positive impact on
The latter is also consistent with results by Popkowski Leszczyc and Rothkopf (2010), who
find that charitable bidders act as volunteer shills trying to drive-up prices in charity auctions.
Therefore, clicking the Facebook like is likely to escalate one’s commitment to the cause,
H3: Clicking the Facebook like increases favorability towards the promoted charity.
H4: Clicking the Facebook like increases willingness to contribute to the charity.
Design of Study 2. To examine the impact of clicking the Facebook like on a person’s
willingness to contribute to charity, controlling for self-selection is essential. That is, those who
3
Choi and Jing-Ann (2010) found that past volunteering significantly influences current volunteering, and Liu and
Aake (2008) reported that a higher donation follows higher volunteerism.
11
like the charity more are more likely to click like and are also more willing to contribute to the
charity. To do so, we used an approach similar to that used by John et al. (2017). In one of our
experimental conditions, we incentivized participants to click on the Facebook like, ensuring that
participants who did like the charity, as well as those who did not, clicked on the Facebook like,
clicked the like). This condition is compared to one in which participants did not leave a like.
We also included a condition that placed no restriction on clicking the Facebook like, to examine
the extent of self-selection—whether people who click are more likely to contribute.
We again recruited 190 participants from MTurk and randomly assigned them to three
conditions. The approach used was similar to that in John et al. (2017). Participants were asked
to visit a website for a charity and were told they needed to answer some questions.
The first condition placed no restrictions on clicking on the Facebook like (and participants
were told they were allowed to do so if they wanted to). In the second condition, participants
were asked to click on the Facebook like on the website, for which they received $0.50. Finally,
in the third (control) condition, they were asked to visit the webpage and not to touch anything
on the website. The random division resulted in allocations of 69, 62, and 59 to the three
conditions, respectively.4
Next, all participants answered several demographic questions and questions related to the
charity’s Facebook page, and completed the dependent variable constructs measuring
favorability and intent to contribute, identical to those in Study 1. Web Appendix B provides all
Results and discussion of Study 2. We included the first condition, without restrictions on
4
In condition 2, three of the 65 participants were deleted because they did not click the like button even though they
were asked to. In the third condition, 12 of the 71 subjects were excluded because they clicked on the like button
even though they were told not to.
12
clicking the Facebook like, to examine self-selection. We found evidence consistent with self-
selection (but not sufficient to show self-selection) as bidders who clicked like were more likely
to contribute. Out of the 69 participants in this condition, 19 did not click on the Facebook like
and 50 did. Those who did click on the Facebook like were more willing to volunteer (Mlike =
4.8 vs. MNoLike = 3.4, T= 3.23, p = .002), spread favorable WOM (Mlike = 5.4 vs. MNoLike = 3.6,
T= 4.12, p < .001), pay attention to communication (Mlike = 5.3 vs. MNoLike = 3.2, T=5.34 ,p <
.001), and donate (Mlike = 5.0 vs. MNoLike = 3.3, T= 2.05, p = .049).
These tests are all based on within-treatment differences and cannot be separated from
subjects self-selecting into clicking the Facebook like or not. Therefore, we next compare
conditions 2 (all participants clicked the Facebook like) and 3 (no participants clicked the
charity. The results of these comparisons and related tests for H3 and H4 are provided in Table
2. The results indicate that those who did clicked on the Facebook like had a higher willingness
to contribute, in support of H4. However, we only find marginal support for the assertion that
clicking the Facebook like results in higher favorability towards the charity. Hence, any
escalation of commitment obtained from clicking the Facebook like appears to be more intrinsic
Finally, comparing the willingness to contribute in condition 1 for those who self-selected to
click on the Facebook like with condition 2 (all clicked on the Facebook like), we observe an
insignificant difference (Mlike, condition 1 = 5.0 vs. Mlike, condition 2 = 4.6, T=1.15, p=0.254).
Studies 1 and 2 provide evidence of a signalling effect (the direct effect of the total number of
Facebook likes) and escalation commitment (a direct effect of clicking a Facebook like).
These results suggest that, at least in the short run, we can influence consumers’ or donors’
13
willingness to contribute. A different question is to what extent these results may have a longer-
term effect. We note that the signaling effect of the total number of Facebook likes also
increased participants’ favorability towards the charity, suggesting these effects may have a more
lasting impact. For clicking the Facebook like, the impact on favorability was only marginally
Overall, the results of the first two studies provide evidence of both a direct effect of the total
number of Facebook likes and of clicking the Facebook like. Next, we conducted a field study in
which we tested these hypotheses in a field setting, as well as testing the indirect social-
STUDY 3
Objectives and hypothesis of Study 3. The primary goal of Study 3 was to be able to increase
fundraising revenues through effective use of social media. That is, we are interested in whether
14
we can increase charity revenues by increasing total Facebook likes. We focused on both the
impact of the total number of Facebook likes and of clicking the Facebook like on bidders’ WTP
in auctions. The total number of Facebook likes may have both a direct signaling effect and an
indirect social-contagion effect on WTP, whereas we expected the effect of clicking the
Facebook like to have a direct effect on WTP by escalating one’s commitment to the cause.
Recall that we also examined the direct signaling effect in Study 1 and the impact of an
individual’s clicking the Facebook like in Study 2. In this study, we investigate both signalling
and commitment escalation and add the indirect social-contagion effect on WTP.
Conceptual framework. Our model explicitly captures three potential ways that Facebook
likes increase WTP. Specifically, we include terms for the following: the signalling effect as
captured by the direct effect of total Facebook likes, commitment escalation captured by the
direct effect of clicking the Facebook like, and the social contagion as captured by the mediating
role of Facebook impressions and bidder intensity in the relationship between Facebook likes
and WTP. Including all three terms allows us to determine whether the effect of the total number
We focus on two outcome measures: (1) WTP and (2) bidding intensity. We quantify the
impact of social media on each of these and examine a possible mediating role of impressions
and bidding intensity, defined as the number of bids divided by the number of bidders, in
affecting WTP. We illustrate these relationships and the resulting hypotheses in Figure 2.
Direct effect of total number of Facebook likes and clicking the Facebook like on WTP.
First, as in Study 1, we tested H2 to determine whether the total number of Facebook likes has a
positive effect on bidders’ WTP (parallel to willingness to donate in Study 1). We tested whether
Figure 2. A conceptual model of the impact of Facebook likes on bidders’ WTP in auctions.
Clicking
Facebook
a2* H5 (+) Likes H4 (+)
c' H2 (+)
b1
perception of the charity or, in the case of artwork, the expected future resale value. For art,
social media takes on an added dimension of providing a narrative on the artist herself. This
narrative has a direct impact on the value of the art, and when the narrative about the artist’s life
changes, so does the value of the paintings, sometimes drastically (e.g., Espinosa 2015). In the
present case, highlighting the artists’ plight could have added to the perceived value of the items.
That is, the direct information value from the number of Facebook likes—transmitted through
social media-- impacts the item’s quality or expected future resale value.
Next, as in Study 2, we tested H4—that clicking the Facebook like increases a bidder’s WTP
(parallel to willingness to donate in Study 2). As discussed above, we expected clicking the
Facebook like to increase a bidder’s emotional connection to the item for sale and thereby have a
Indirect social-contagion effect of total number of Facebook likes on WTP. Third, Facebook
likes tend to result in Facebook impressions, leading to information dissemination, online WOM,
16
and social contagion (Aral 2011; Aral and Walker 2011; Iyengar et al. 2011). We expected this
social-contagion effect to result in increased bidding activity and in turn higher WTP. We first
considered the impact on bidding intensity, followed by a test of the mediating role of Facebook
New Facebook likes show up as recommendations in the sidebar of some friends. The
Facebook news feed algorithm shows content to a fraction of friends (based on factors such as
friends, frequency of contact, common interests, etc.) who will see a particular post and may in
turn pass on this information to other friends. Higher bidding intensity is commonly observed
and is critical for revenue generation in charity auctions (Haruvy and Popkowski Leszczyc 2009;
Popkowski Leszczyc and Rothkopf 2010). The integration of social network activity into bidding
behavior has only recently received attention (Hinz and Spann 2008; Dass, Reddy and Iacobucci
2014). Social networks are in turn influenced by online WOM (Rui et al. 2013; Trusov et al.
2009). Facebook likes in particular result in more viral content and WOM as interested parties
interact (Berger and Schwartz 2011; Wojnicki and Godes 2008), which should attract more
bidders to the website. This effect will result in increased traffic to the auction website,
translating to additional bids and higher bidding intensity—a key mediator in our analysis—
resulting in increased WTP (Dholakia and Soltysinski 2001, Dholakia, Basuroy, and Soltysinski
2002). Our fifth hypothesis pertains to the impact of social media exposure on increased bidding
intensity, measured here as the number of bids divided by the number of bidders.
H5. The total number of Facebook likes increases bidding intensity in auctions.
potential mediators in the relationship between Facebook likes and WTP. The mediation tests are
central to our ability to draw conclusions. Depending on the outcome of the mediation tests, the
17
direct effect of social media activity on ending prices would imply social media activity by its
own virtue—perhaps the mere existence of a Facebook presence—enhances the expected value
of the item to the buyer, whereas a mediated effect suggests social media increases Facebook
bidding intensity affects auction outcomes, particularly value assessment (Ariely and Simonson
2003; Cheema, Chakravarti, and Sinha 2012; Dholakia and Soltysinski 2001; Dholakia et al.
2002). In our framework, we examine whether such bidding intensity mediates auction
antecedents. Although no extant research establishes such a mediation, some related evidence
Steinhart et al. (2013) found the degree of perceived risk of losing mediates the effects of
Facebook Impressions. We further consider the possibility that social media contributes to
ending prices by increasing impressions. Padilla-Meléndez and del Águila-Obra (2013) show
that museums can increase impressions with the use of social media, referred to as value
creation. The expectation is that the additional visibility (impressions) resulting from social
media will increase traffic to the auctions and therefore bidding intensity, which results in higher
WTP. This finding is consistent with findings by Bockstedt and Goh (2011), who analyzed a set
of eBay auction listings for identical products and found that enhancing visibility for auctions in
the auction listing affects auction outcomes (price premiums) through increased traffic.
Serial mediation. Next, we consider the process through which total Facebook likes
influence bidders’ WTP. Total Facebook likes result in viral activity (Facebook impressions) as
18
Facebook friends pass on information, through their likes or messages, which in turn influences
the bidding intensity, which results in increased WTP. We therefore test for serial mediation
whereby Facebook likes first influence Facebook impressions, which in turn influence bidding
intensity, which results in increased WTP. In the empirical analysis, we test for the possibility
that the results are only partially mediated and that both direct and indirect effects are present.
H6. The impact of Facebook likes on willingness to pay is serially mediated through
Mediating role of Facebook impressions and bidding intensity on the relationship between
clicking the Facebook like and WTP. Similar to the above discussion for total number of
Facebook likes, bidding intensity might mediate the impact of clicking the Facebook like on
WTP. Different from the total number of Facebook likes (which are expected to result in viral
activity), we do not expect clicking the Facebook like to have a significant impact on Facebook
impressions. However, because clicking the Facebook like increases the emotional connection to
the item for sale, bidders would likely increase the bidding intensity.
Design of Study 3. We harnessed the power of Facebook likes to raise funds for under-
privileged children in a major Canadian metropolitan area. Specifically, as part of the study,
and shared emotional tidbits about their life stories. Combined with additional donated products
and an outpouring of support from the community, we launched a Facebook campaign for
comparable groups of children, with particular aspects varied for experimental manipulation and
with Facebook likes varying across groups. Subsequently, we auctioned off the artwork (and
additional products), integrated with Facebook pages and Facebook likes. With this design, we
19
aimed to study the impact of social media on fundraising in charity auctions. Specifically, we
quantified the impact of Facebook likes and Facebook impressions5 on WTP, and we teased out
the impact of the total number of Facebook likes versus the effect of clicking the Facebook like.
As reported earlier, we found that aggregate Facebook likes provide a signal about the
quality or value of the charity (Study 1). By contrast, clicking the Facebook like escalates one’s
liking or commitment to the cause. We separated out the direct effects of signaling versus
escalation. We further looked at the indirect effect of Facebook likes and disentangled the social
value of Facebook likes due to increased perceived value of the charity (signal effect) versus
increased exposure (social contagion effect). We accomplished the latter by measuring Facebook
impression (due to Facebook likes) and accounted for the potential mediating role of Facebook
impressions and bidding intensity in mediating the effect of Facebook likes on WTP.
We collaborated with the Canadian branch of a major internationally known charity, which
focuses on children and youth from disadvantaged circumstances, with the aim of fostering
opportunities, confidence, and skills through community-based services. Their programs consist
In collaboration with this charity, we organized a series of art events with different groups of
children participating in off-school programs. Next, the artwork created by the different groups
of children were displayed on different Facebook pages (one for each different group of children)
especially created for this event, as well as additional items donated. Finally, we employed a
locally well-known website for charity auctions, and auctioned off the artwork (and additional
5
The term Facebook impressions captures the number of times a post is seen (on a user’s News Feed, ticker, or
directly on his or her Facebook page).
20
We carefully designed the study with different experimental cohorts and several control
cohorts. We organized the art events as part of the programs offered by clubs. Altogether, six
different clubs of the same charity were involved. The six clubs are located in distinct
neighborhoods within one major North American metropolitan area. A club consists of a
clubhouse and staff members that organize events for children to participate during school
holidays and after school. These programs are on a drop-by basis and are free of charge to all
children. Children pre-register for events during school vacations. The current event took place
The children’s legal guardians received an announcement for the art program with
information about the fundraiser several days before the event and provided consent forms for
the children to participate and for pictures to be used. The consent form included a short survey
asking the children about their favorite hobbies, their happiest memory, saddest story, their
favorite person, and their personal hero. We used this information later in the Facebook pages
We hired and trained a staff of art teachers and art students to run the art event. Each art
event lasted for approximately one and a half hours, at the start of which children received a
short lecture about different types of painters. Subsequently children divided into groups of
eight to ten children based on similar age groups (one or two groups per location). Twelve
different groups with a total of 107 children participated in the event. A few children did not
receive consent to participate and were provided with an alternative activity during the art event.
Each group of children created at least 10 artworks. Each artwork was the product of their group
effort rather than an individual’s effort and was promoted as such to have a broader appeal to
bidders. We randomly assigned the 12 different groups into two control groups and 10 treatment
21
groups. To measure the impact of social media, we created Facebook pages for the 10 treatment
Each Facebook page was designed such that the format and structure of each group’s page
were identical (the order and the type of posting on each page were identical, the artwork and
information was group specific). The cover or heading of each page consisted of a collage of the
Figure 3. Different Facebook page heading consisting of a collage of the group’s artwork.
Next, a Facebook post announced the event, indicating its purpose and the charity involved
and a link to the website where the artwork would be sold, including a video specifically created
for this event. An announcement followed, asking people to like the page. To generate
excitement and buzz, the announcement indicated 50 cents would be donated to the charity for
22
We next posted information about the specific group, including a group picture of the
children creating the art, followed by pictures of the teaching staff.7 Subsequent postings showed
each artwork, including a picture and a description of the artwork and the date the item would be
sold in an online auction (with a link to the website), as well as simultaneous auctions for
additional donated products for this group (see Figure 4). Each posting for artwork was followed
by quotes from the children (e.g., their happiest and saddest memories), obtained from the survey
on the consent form, and pictures of the children in action (see Figure 5).
We created Facebook pages to enhance the emotional connection with the event and to
encourage clicking the like button and to disseminate WOM about the fundraiser. The specific
data we obtained from social media include the number of Facebook likes and Facebook
6
Facebook likes, generated through a social media campaign, can be quite costly (Hoffman and Fodor 2010). In the
present settings, we acquired Facebook likes from the local community at a cost of 50 cents per like, following John
et al. (2017).
7
One of the Facebook pages did not have a picture. However, results were not significantly different, suggesting
results are not due to the pictures used.
23
information passed on through friends to others via a Facebook like. That is, Facebook
impressions occur when a person sees a post due to a like or a message by a friend.8
We created the Facebook pages several weeks before the start of the auctions to allow for the
social media effects, that is, for the number of Facebook likes to accumulate and WOM to
disseminate. During this time period, we contacted either by email or phone all legal guardians
of the children who participated, to inform them about the upcoming auctions and to encourage
them to spread the news, by clicking the Facebook like. We also distributed a flyer showing the
artwork and announcing the dates of the auctions at the different clubs of the charity.
The next step was to sell the artwork and additional items through a locally well-known
8
Facebook impressions fall into classifications of paid, organic, or viral. Paid impressions are the result of paid ads.
Organic impressions are direct exposures to our Facebook posts, whereas viral impressions are passed on by friends
to others via a Facebook like. Our focus is on the viral information that is passed on by friends.
24
website for charity auctions. We sent e-mails to all members of the auction website with
information about the upcoming auctions and details about the fundraiser, including links to the
Facebook pages. At the time of this study, the website had approximately 10,000 members. We
provided similar incentives to the members for clicking on the Facebook like (a 50-centsdonation
We conducted 480 auctions (120 auctions for artwork and 360 auctions for additional items)
during May and June 2016. The additional items consisted of multiple identical items and
included mostly gift certificates and some gift items. All items were randomly assigned and
counter-balanced across groups such that, for example, gift certificates of the same value were
sold for each group. Artworks were group specific but were randomly assigned to the different
time periods for the group. For each group, we sold 40 items (10 artworks and 30 other items).
We created 12 different seller accounts for each of the groups (10 experimental and two control
groups). Each seller had a special seller page with information about the charity and the specific
group and links to his or her Facebook page. For the control group, we only provided
information about the charity. In all instances, we clearly indicated all donations would go to the
local chapter of the charity (not to the individual groups). At any given time, we only ran
auctions for four different sellers simultaneously. At any given time, we ran 16 different
auctions, that is, four auctions for four different groups, consisting of one auction for an artwork
and three auctions for other items. Each auction lasted two days, all with starting and ending
times at 10:00 pm. When new auctions were listed, we also pre-announced the next set of 16
upcoming auctions (such that bidders could see the auctions and the starting dates but could not
The title of the auction clearly identified the specific group. In addition, each auction
25
provided information about the group, a picture of the group,9 a picture and description of the
artwork, and a link to the Facebook page that included a picture of the Facebook cover with the
number of Facebook likes the page had received (which was dynamically updated) and a button
9
Each artwork was a joint collaboration between group members—hence a group picture. For the cleanest
comparison between treatments, the pictures and descriptions of each group would have to be identical. They are as
similar as possible, but they are not identical. Differences between pictures might be an antecedent for the difference
between likes for different groups. The antecedents for likes are outside the scope of this paper. They could be part
historical accident, part idiosyncratic differences in families and social ties of the group members to bidders, or in
differences between the pictures. Although we cannot shed light on the cause of the differences in likes between
groups, other than historical accident, we can say they lead to differences in impressions and bids.
26
Experimental Manipulations
We have a panel of auctions for 12 groups (labeled groups 1-12) observed over 30 periods
(60 days). The main variable we manipulated in our experiment is the number of Facebook likes,
both between groups (within a period) and within groups (over time).
First, we varied the number of Facebook likes between groups through the presence (groups
1-10) or absence (groups 11 and 12) of a Facebook page and resulting Facebook likes. The two
control groups (groups 11 and 12) did not have a Facebook page, whereas the 10 treatment
Second, the initial starting points of Facebook likes for the 10 treatment groups varied by
design. We varied the initial number of Facebook likes by inducing a different number of likes
for different groups prior to the start of the experiment. We did so by sending out e-mails
Third, the actual number of Facebook likes increased naturally over time across the 10
treatment groups, as people left likes for different groups. Naturally, late auctions had more
Facebook likes than early auctions. Table 3 summarizes these three experimental manipulations
and shows the distribution across different groups. Following the actual distribution of initial
likes, groups are classified into groups with low and high numbers of initial Facebook likes. The
number of Facebook likes over the initial number, resulting in a classification of groups into
The experiment took place over a two-month period. As mentioned above, we always ran
auctions for four groups simultaneously. Auctions for any specific group had a duration of two
days, followed by four days of inactivity (during which we ran auctions for other groups) before
27
the next set of auctions for that group started. In total, we held 10 sets of auctions for each
group. The same four groups were always matched together (to keep the time between set of
auctions constant), but the order of the auctions varied (which influences the order in which
auctions were displayed on the website). Other donated products, consisting of gift certificates
and gift items, were counter-balanced across groups and time periods, such that the value of the
items was the same for each group and across time periods. For each group, we sold the
identical value in gift certificates and close to the identical value for the gift items. The value of
all items sold at any specific time period was also almost identical.
The information obtained from the auctions include the bidding histories (with number of
bids and bidders) and ending prices. We also obtained clickstream data showing the pages
28
visited by different visitors to the website. Critically, clickstream data provided us with
information on which bidders clicked the Facebook like (by clicking on the link in the auction).
We combine the auction data with the Facebook data in a panel data set. This data set
includes the timing and the amount of the bids, the total number of Facebook likes and Facebook
impressions at any point of time a bid was placed, as well as an indicator that the bidder clicked
The effect of self-selection in Study 3. As in Study 1, the issue of self-selection was a central
concern in Study 3, because bidders who liked the cause and were willing to pay a premium for
the cause were also more likely to click the Facebook like (Bapna and Umyarov 2015; John et al.
2017; Mochon et al. 2017). Although Study 1 demonstrated that escalation trumps self-selection,
distinguishing the effect of clicking the like button from a self-selection effect in Study 3 is
important. The most suitable approach to deal with self-selection is random assignment.
Although we randomly assigned groups and products over time, we were unable to randomize
propensity to bid in auctions for the current charity, and new member to control for self-
selection, that is, the likelihood that those bidders were more likely to leave a like and also bid
more. Charitable intent controls for the fact that more charitable bidders tend to be willing to pay
higher prices in charity auctions (Haruvy and Popkowski Leszczyc 2009; Popkowski Leszczyc
and Rothkopf 2010). Propenstity to bid in auctions for the current charity controls for the fact
that supporters of the charity involved were willing to pay higher prices. Finally, new member
controls for those who self-selected into the current event to bid in auctions for specific artwork.
Empirical Analyses
Summary Statistics. Before conducting analyses at the bid level, we want to determine the
29
impact of Facebook likes and impressions on the revenue to the charity, by focusing on auction
ending prices. We begin by examining the effect of the mere presence of a Facebook page on
ending price and bidding intensity, as a first test corresponding to H2 and H5 (we subsequently
The statistics and p-values for statistical significance are summarized in Table 4. Test
statistics are reported for three different product categories and show that ending prices were
significantly higher in auctions for groups with a Facebook page present for artwork and gift
items (consistent with H2) but not for gift certificates. Similar to H5, bidding intensity was
significantly higher in auctions for groups with a Facebook page present for artwork and
marginally significant for gift items but not for gift certificates.
Table 4. The impact of the Presence of a Facebook page on Ending prices and Bidding
Intensity in auctions.
H1: Ending Prices H3: Bidding Intensity
Product Facebook N Mean Price T-testa Mean Bid T-test
Category Page (Std. Error) Intensity
Yes 79 3.041 1.574
(100) (.360) p=0.002 (.125) p<0.001
Artwork
No 14 1.501 1.095
(20) (.347) (.074)
Yes 167 23.667 1.734
Gift (168) (1.526) p=0.468 (.061) p=0.400
Certificates No 31 23.854 1.702
(3.818) (.109)
Yes 140 5.489 1.780
(.415) p<0.001 (.081) p=0.091
Gift items
No 27 3.414 1.580
(.324) (.124)
a
p-values are based on one-sided t-test with unequal variances.
These hypotheses test for the positive effect of the presence of a Facebook page (rather than
30
the number of Facebook likes) on ending prices in auctions (H2) and on bidding intensity (H5).
The results provide strong initial support for H2 and H5 for artwork and gift items.
The finding that the presence of a Facebook page is not significant for gift certificates is not
surprising. Facebook likes are less informative due to the clear value of the certificates, and an
emotional connection is lacking because, unlike the artwork, the certificates are not associated
with the children (especially because the auction proceeds go to the charity rather than to the
specific groups).
Number of initial Facebook likes. We next examine the influence of the number of Facebook
likes (the second manipulation in Table 3). We manipulated the number of Facebook likes across
groups to induce variation in the number of initial likes. The treatment groups were divided into
five groups with a low number of initial Facebook likes and five groups with a high number of
initial likes (see Table 3). Table 5 shows significantly higher ending prices for auctions with a
high number of initial Facebook likes (as compared to a low number of initial likes) for artwork
and gift items (in support for H2). Whereas previous analyses compared the treatment versus the
control group, these analyses only compare the treatment groups, providing a (stronger) test of
Table 5. Average ending prices in auctions with a low vs. high number of initial
Facebook likesa.
No. of initial Artwork Gift Gift items
Facebook likes Certificates
Low $2.002 $23.668 $4.729
(0.274) (2.168) (0.304)
n=37 n=82 n=70
High $3.956 $23.769 $6.248
(0.601) (2.185) (0.764)
n=42 n=84 n=70
T-test p=0.002 p=0.487 p=0.034
a Only observations with positive likes are included (N=385); the control group is excluded
31
Influence of social media on ending price at the auction level. To test the signal effect of the
total number of Facebook likes (Total Likes) on ending prices, we start with simple regression
models that estimate the relationship between ending price and the number of Facebook likes, as
well as the potential mediators Facebook impressions and bidding intensity for the different
product categories used above. The dependent variable is the selling price in the auction, where
First, considering the results of the base model without potential mediators, we see that total
likes increased ending prices for the first two categories, consistent with the t-test reported earlier
in the paper. Furthermore, the results indicate both Facebook impressions and bidding intensity
influenced ending prices for the same categories. However, the significance of total likes
disappears when adding Facebook impressions and bidding intensity, whereas adding bidding
intensity does not affect the significance or magnitude of impressions or the other way around,
indicating either Facebook impressions or bidding intensity or both can serve as mediators for
Influence of social media on WTP. To test our hypotheses and to determine the impact of a
bidder clicking the Facebook like, we next focus on analyses at the bid level. The effect of
clicking the Facebook like is only observable in the bid-level analysis and is expected to be
significant because it enforces the bidders’ commitment to the charity. We are also concerned
with self-selection, which is primarily discernible through bid-level analysis. In addition, the
total number of Facebook likes is a signal10 that should have an impact on individual WTP as
10
By contrast, Facebook impressions are not a signal, but rather an indication of the amount of WOM, and thus
should have an impact on ending price but not necessarily on individual bids, or at least the effect will be weaker.
32
In this section, we report on bid-level analysis to estimate a formal model of the effect of
social media on WTP. The empirical analysis uses bid-history data and dynamically updates the
number of cumulative likes over time. We also integrate clickstream data, such that we can use
data on individual bidders clicking the Facebook like. As such, we distinguish between the effect
of a bidder clicking the Facebook like and the total number of displayed likes by others. Finally,
Self-selection. A potential concern with the current data is the issue of self-selection, because
bidders who like the cause and are willing to pay a premium for it are also more likely to leave a
like (Bapna and Umyarov 2015; John et al. 2017; Mochon et al. 2017). The most suitable
groups and products over time, we were unable to randomize bidders to different auctions
33
(treatment conditions). Instead, we include charitable intent, propensity to bid in auctions for the
current charity, and new member as control variables in our analysis. In particular, we include a
measure of charitable intent, because people who are more charitable tend to be willing to pay
higher prices in charity auctions (Haruvy and Popkowski Leszczyc 2009; Popkowski Leszczyc
and Rothkopf 2010). The measure of charitable intent is based on the proportion of bids in
charity auctions versus non-charity auctions (for the identical product) based on bidding histories
during the past 14 years on the website. Propensity to bid in auctions for the current charity
adjusts for the liking of the current charity. Propensity to bid is the proportion of bids on the
current charity compared to three other charities (based on a different study selling identical
items for four different charities). Finally, new member is a dummy variable that indicates
whether a bidder is a new member who just became a member for the current fundraising event,
We estimate a simple linear regression model with the bid amount as the dependent variable.
We include the self-selection variables Charitable Intent and New Member (propensity to bid in
the focal charity was dropped due to insignificance). The variables of interest include Total
Likes (the total number of likes at the time of the bid), Click Like (a dummy variable indicating
whether the bidder clicked on the Facebook like for the group of the auction they were bidding
in), and Facebook Impressions (the number of impressions at the time of the bid). The results
are displayed in Table 7. We also include Day, which is a measure of the number of days since
the start of the auction event. This time trend measures the impact on WTP, because bidders’
interest for auctioned items may vary over time. Finally, we include category-specific fixed
effects and interactions between the category-specific variables and the variables of interest.
Results show that, in support of H2, total likes have a major impact on WTP for paintings
34
and gift items (once we allow for bidding intensity in the model) but not for gift certificates with
clear monetary values and a lack of emotional connection (as opposed to artwork).
We also find a positive effect for clicking the Facebook like for paintings (in support of H4)
but not for gift certificates or gift items. The positive effect for paintings is expected because
social media enhances the emotional connection with the painting, which influences commitment
to the cause. This connection is likely to be even stronger for the paintings in the present study
Table 7. Results of Total number of Facebook Likes and Clicking the Facebook like on
WTP (N = 3400)
Base Model Base Model Base Model
Without potential Including Impression Including both
Mediators mediators
Effect Estimate P-value Estimate P-value Estimate P-value
Charitable Intent 0.152 0.205 0.186 0.117 0.018 0.869
New Member 0.143** 0.017 0.135** 0.022 0.151** 0.005
Category: Gift Items -0.320 0.156 -0.339 0.139 -0.416** 0.048
Category: Paintings -0.962* 0.060 -1.499** 0.003 -1.768** 0.000
Total Likes * Gift Items 0.004 0.268 0.008* 0.077 0.009** 0.028
Total Likes * Paintings 0.096** <.0001 0.153** <.001 0.168** <.001
Total Likes * Gift -0.002 0.633 0.002 0.676 -0.001 0.725
Certificates
FB impressions × Gift -0.0002* 0.076 -0.0004** 0.000
Items
FB impressions × -0.0028** <.001 -0.0037** <.001
Paintings
FB impressions × Gift -0.0002 0.105 -0.0002** 0.022
Certificates
Click Like* Gift Items -0.024 0.621 -0.022 0.643 -0.010 0.830
Click Like* Paintings 0.228** 0.016 0.202** 0.030 0.178** 0.038
Click Like* Gift -0.086** 0.015 -0.086** 0.014 0.016 0.613
Certificates
Day * Gift Items 0.000 0.927 -0.001 0.557 -0.002 0.301
Day * Paintings -0.005 0.194 -0.051** <.001 -0.068** <.001
Day * Gift Certificates -0.002 0.191 -0.003 0.041 -0.002 0.115
Bidding Intensity 0.347** <.001
We find that Facebook impressions have a negative effect on WTP for all three product
35
categories. At first glance, these results seem to be inconsistent with the results for the analysis
of ending prices reported in Table 6. However, although we expect a positive effect on ending
price, the effect on WTP need not be positive. Individual bidders are not necessarily going to be
number of Facebook likes, which is a signal—need not positively influence a bidder`s WTP.
Instead, the effect of Facebook impressions on final price more likely arises through bidding
intensity. That is, Facebook impressions disseminate information, which attracts more bidders
The bidding intensity, when it mediates a click on like directly without impressions-- appears
to have a negative effect on WTP, possibly because bidders are discouraged by the increased
competition and feel they have a lower chance of winning (consistent with Garcia and Tor 2009
and Häubl and Popkowski Leszczyc 2018). However, when impressions are added to the
mediation path along with bidding intensity, the combined path does positively mediate likes
towards higher WTP, consistent with the idea of impressions serving to increase information
Influence of total likes on bidding intensity. To test for the impact of social media exposure
on increased bidding intensity, we estimate a regression model with bidding intensity (the
number of bids divided by the number of bidders) as the dependent variable. We focus on the
effect of total likes on bidding intensity while including as control variables most of the same
variables included in the previous model. In support of H5, we observe a highly significant
effect of total likes on bidding intensity (β = 0.009, p=.002). In addition, as expected, charitable
intent is highly significant (β = 0.409, p<.001), indicating more charitable bidders bid more
aggressively; however, new members bid less aggressively (β = -0.210, p=.001). We also run a
36
similar model that includes Facebook impressions, which also significantly influenced bidder
intensity (β = 0.0004, p<.001). However, total likes become insignificant once we add Facebook
impressions to the model (β = 0.001, p=.703). These results show that social media exposure
through total likes and impressions both influence bidding intensity. However, the effect of total
Mediation Analysis
We next conduct mediation analysis to determine whether Facebook impressions and bidding
intensity mediate the direct effects of Facebook likes. This approach allows us to test whether
the direct effect (social proof or signal effect) of Facebook likes is still present after we control
To test H6—whether the impact of the total number of Facebook likes on WTP is serially
mediated through Facebook impressions and in turn by bidding intensity—we conduct serial
mediation analysis (Hayes 2013) using the bootstrap approach by Preacher and Hayes (2008).
More specifically, we test whether the positive effect of total likes on WTP goes through
Presenting the mediation analysis using the framework presented by Figure 2 is useful. The
indirect effect a1b1 denotes the impact of total likes on WTP as mediated by Facebook
impressions, the indirect effect a2b2 denotes the impact of total likes on WTP as mediated by
bidding intensity, and the indirect effect a1d21b2 denotes the impact of total likes on WTP as
We summarize the results of the mediation analyses in Table 8. We run different models
testing for single mediators and multiple (serial) mediators for both total likes and click like. The
results indicate a significant indirect effect of total likes on WTP through the two mediators,
37
Facebook impressions and bidding intensity, (a1 × d21 × b2 = 0.0003), and the 95% confidence
Hence, we find support for H6. After adding the two mediators, the effect of total likes for
paintings (Table 7) increases from 0.096 to 0.168. This finding suggests the impact of total likes
is partially mediated and that total likes has both a positive direct signaling effect on WTP, a
negative indirect effect of Facebook impressions through bidding intensity, and a positive
Table 8. Results of Mediation Analysis of effect of Total Likes and Own Likes on WTP
Looking at the mediation analysis for click like, we see that click like remains significant
after adding bidding intensity. This finding suggests click like also has both a direct and indirect
We separated out the signaling effect of the total number of Facebook likes and the
escalation effect due to a bidder clicking a Facebook like and find both have a direct positive
effect on WTP. This finding suggests the increase in WTP is due both to the signaling value of
Facebook likes—which positively influences the item’s quality or expected future resale value—
and to commitment escalation. We further identified the social contagion effect, by accounting
38
for the effect of Facebook impressions and bidding intensity in mediating the effect of Facebook
likes on WTP.
GENERAL DISCUSSION
Companies have long practiced social media marketing for the purpose of increasing
exposure and awareness (Mangold and Faulds 2009). Non-profit marketing soon followed suit.
From the early days of social media, charities and other non-profit organizations have asked
people to like their Facebook pages. When asked in interviews about the purpose of such social
media actions (Hou and Lampe 2015), managers of non-profit organizations cite public
media activism has resulted in some impressive social changes (Dewey 2014) but has been more
elusive in charity fundraising. In terms of cold hard cash, charities’ great expectations from
social media have resulted in somewhat of a letdown. This letdown resulted in the inevitable
backlash against social-media “slacktivism.” For example, UNICEF Sweden ran an ad campaign
in 2013 with the ad “Like us on Facebook and we will vaccinate zero children against polio. We
have nothing against likes but vaccine costs money” (Khazan 2013).
Our objective was first and foremost to improve our ability to raise funds for charity through
social media. We wanted to map the path from a Facebook like to money raised for charity.
Towards our goal, we first tried to isolate the value of a like in increasing charitable
intention. We isolated two effects—the signaling effect that comes from seeing a higher number
of likes (Study 1), and the commitment-escalation effect that arises from clicking on the
We next wanted to place these effects in the context of an actual fundraising effort so that we
could map social media influences to these two effects but the effects in turn could be mapped
into cash raised. For the specific fundraising environment, we selected an auction setting. Past
works (e.g., Haruvy and Popkowski Leszczyc 2009; Popkowski Leszczyc and Rothkopf 2010)
have demonstrated that auctions are very useful for raising funds for charity. This study was the
Overall, adding a Facebook feature in our auctions was successful. We found a positive
effect on ending prices with the introduction of Facebook to auctions. Average selling prices
were over 30% higher than for two control groups without Facebook pages (and likes).
We found that auction revenues were a function of bidder participation, bidding intensity,
and WTP—and we identify the impact of social media on each of these elements, with special
attention to the mediating role of Facebook impressions and bidding intensity. Interestingly, we
found a strong and positive direct effect of the total number of Facebook likes on WTP and
ending prices, as well as a positive indirect effect through Facebook impression and bidding
intensity, where the latter serves to mediate the effect of Facebook likes on WTP. This finding
suggests both increased exposure to the cause and an increased perceived value of the cause. In
addition, we found that clicking the Facebook like has a direct effect on WTP by escalating one’s
Lastly, we found the effect of social media was most beneficial for products with which
bidders had emotional attachments rather than universally, which is consistent with Bernritter et
al. (2016), who showed that consumer behavior with respect to Facebook likes for non-profit
brands is fundamentally different than for for-profit brands, because consumers on social media
Other results from the auction platform were more nuanced. Facebook impressions are
especially difficult to nail down. Most importantly, total likes and click like have both a direct
and indirect effect on WTP—total likes mediated through Facebook impression and bidding
Facebook impressions surprisingly had a negative effect on prices (both controlling for likes
and not controlling for them, and other modeling assumptions). They therefore do not directly
mediate total likes or click like. However, they clearly positively affect bidding intensity;
therefore, combined with bidding intensity, they mediate total likes in a positive way to affect
WTP. Click like, however, is mediated directly through bidding intensity. The effect of click
like is more than twice that of the total likes for paintings and more than three times bigger for
gift certificates. We can interpret the finding that click like is directly mediated by bidding
intensity to mean click like (after controlling for self-selection) led to increased commitment or
engagement, specifically commitment to visit the website and bid. Total likes provide a signal
that increase the ending prices for paintings and gift items but not for gift certificates, where the
value is established. However, the direct effect of click like remains significant, suggesting both
Managerial Implications
As our opening paragraph illustrated, if the goal of a social media campaign is fundraising,
charity managers should have a clear mapping from the efforts of the campaign to charitable
intent and, finally, to cash. A social media campaign without a clear mapping can easily backfire,
We started out with two direct effects that can convert Facebook likes to charitable intent.
The first was signaling and the second was commitment escalation. Signaling is most appropriate
41
when the quality of the charity, its efforts, or its benefits are not fully known to the prospective
donor. Then the number of likes can serve as a signal to reduce perceived risk or uncertainty.
Clearly, if this route is to be taken, the number of Facebook likes should be made prominent on
Our second investigated route was through commitment escalation. In recent years,
Facebook has attempted to facilitate donations directly through its websites, through a Donate
button, bypassing the like button to get directly to the action. Although this approach can be
charitable intent, and managers have expressed concern about this effort backfiring (McLaughlin
2017). Our findings are in line with the position that managers should maintain their own control
Facebook friends pass on information, through their likes or messages. Our results of the field
study show that leaving likes results in information dissemination and social contagion, which in
turn influences WTP. This finding indicates the importance to managers of using Facebook for
Having mapped efforts to charitable intent, the manager should next identify the exact
fundraising mechanism through which charitable intent is converted to cash. In this investigation,
we focused on charity auctions. Many other fundraising approaches exist, but auctions are a
In recent years, the charitable fundraising literature has found that an individual donor’s
WTP could depend on donations the charity is able to amass from others in a short period of
time. The literature has three sets of related social influences on fundraising: (1) social reference,
42
The first influence is a straightforward social reference. In that research stream, Shang and
Croson (2009), in a natural field experiment with phone banks that receive inbound calls from
public radio campaigns, explored the impact of upward social influence. They report that
contributions from recent donors had a big impact on donations. Croson and Shang (2008)
examined the impact of downward social information and found similar results.
The second stream pertains to challenge grants, whereby fundraisers use seed donations by
others to motivate future donors to reach a particular donation threshold. List and Lucking-
Reiley (2002) found that challenge gifts announced by the fundraiser influence charitable
contributions in university capital campaign. Landry et al.’s (2006) data from a door-to-door
fundraising drive suggested challenge gifts increased conditional contributions. Rondeau and
List (2008) used direct-mail solicitations to Sierra Club supports to explore the effect of
matching and seed money on donor behavior. They found that announcement of seed money
increased the participation rate of potential donors by 23% and total dollar contributions by 18%,
The third stream pertains to matching grants, where a donor, typically an organization,
Eckel and Grossman (2008) with Minnesota National Public Radio, found that matching grants
result in larger total donations to charities than comparable rebate subsidies. Thus, the
contributions by others, rather than the impact of a donation amount, is what influences giving.
Auctions are an interesting manifestation of this “other donor: effect, because of the known
effect of bidding intensity (also known as frenzy or competitive arousal). We showed that
bidding intensity mediates the Facebook-like effect, which is an important consideration for
43
managers. Auctions with a strong, established private value, in which bidding intensity is less
likely to play a role, are not likely to benefit to the same extent from a Facebook campaign.
Limitations
We conducted our study in a context of more non-durable items to be sold over a short
period of time. Although our campaign likely increased awareness for the organization for
which the funds were being raised, this increased awareness was not the focus of the
investigation (nor do we have measures for it). Rather, we were focused on the relatively short-
term impact on short-duration auctions for unique items. The short blitz of social media through
John et al. 2017; Mochon et al. 2017) but has not received as much attention as the better-known
Short-term social media is tricky, to put it mildly. It generates traffic, which is critical for
auctions, but it also shapes value expectations, in particular in a charitable setting. These two
effects have to be disentangled. Social media also results in multiple measurable intermediate
outcomes (i.e., prior to auction outcomes), namely, Facebook impressions and likes, and
Facebook likes in turn can be decomposed into own likes (an individual clicking the Facebook
like) and others’ likes (the total number of Facebook likes), with differing effects. This
likes are treated as exogenous variables, and one could argue that more charitable bidders might
leave more likes and also bid more. We controlled for this possibility first and foremost through
the randomized assignment of groups to treatments, and through pooling proceeds so that no one
group would benefit exclusively from the proceeds for that group (i.e., we reduced the incentives
44
Further, we note that bidders liking an auction because they like a specific artwork (e.g.,
because their child was in that group) does not explain why likes have a positive effect on ending
prices for both artwork and gift items but not for gift certificates.
Lastly, we looked at all past charitable choices of individuals in the auction platform prior to
joining this study and classify their charitable propensity in that manner. We find no evidence
that their charitable propensity affects their bids or likes in any way. As a result, although we do
not dismiss self-selection as an empirical concern, the overall evidence does not point to it being
Future Directions
Future research should look into additional ways to control for self-selection in field
experiments in general and in social media research in particular. Future research could delve
deeper into the types of products for which social media can influence consumer WTP. Clearly,
artwork cannot be the only category although it is deeply personal. The research on long-term
social media campaigns (e.g., Mangold and Faulds, 2009) has picked on high-involvement
References
Aral, Sinan (2011), "Commentary—identifying social influence: A comment on opinion
leadership and social contagion in new product diffusion," Marketing Science, 30 (2), 217-
223.
Aral, Sinan, and Dylan Walker (2011), "Creating social contagion through viral product design:
A randomized trial of peer influence in networks," Management science, 57 (9), 1623-1639.
Ariely, Dan and Itamar Simonson (2003), “Buying, bidding, playing, or competing? Value
assessment and decision dynamics in Internet auctions,” Journal of Consumer Psychology, 13
(2), 113–123.
Baca-Motes, Katie, Amber Brown, Ayelet Gneezy, Elizabeth A. Keenan, and Leif D. Nelson
(2013), “Commitment and Behavior Change: Evidence from the Field,” Journal of Consumer
Research, 39, 1070-1084.
45
Bagchi, Rajesh, and Amar Cheema (2012), “The effect of red background color on willingness-
to-pay: The moderating role of selling mechanism,” Journal of Consumer Research, 39(5),
947-960.
Bapna, Ravi, and Akhmed Umyarov (2015), "Do your online friends make you pay? A
Science, 61 (8), 1902-1920.
Baumeister, Roy F., and Mark R. Leary (1995), “The need to belong: Desire for interpersonal
attachments as a fundamental human motivation,” Psychological Bulletin, 117, 497–529
Berger, Jonah, and Eric M. Schwartz (2011), “What drives immediate and ongoing word of
mouth?,” Journal of Marketing Research, 48 (5), 869-880.
Bernritter, Stefan F., Peeter W.J. Verlegh, and Edith G. Smit (2016), “Why Nonprofits Are
Easier to Endorse on Social Media: The Roles of Warmth and Brand Symbolism,” Journal of
Interactive Marketing, 33, 27–42.
Bikhchandani, Sushil, David Hirshleifer, and Ivo Welch (1998), “Learning from the behavior of
others: Conformity, fads, and informational cascades,” Journal of economic perspectives 12
(3), 151-170.
Bockstedt, Jesse, and Huat Goh Kim (2011), “Seller strategies for differentiation in highly
competitive online auction markets,” Journal of Management Information Systems, 28 (3),
235-268.
Brandtzaeg, Petter Bae, and Ida Maria Haugstveit (2014), “Facebook likes: a study of liking
practices for humanitarian causes,” International Journal of Web Based Communities, 10 (3),
258-279.
Briones, Rowena, Beth Kuch, Brooke Fisher Liu, and Yan Jin (2011), “Keeping up with the
digital age: How the American Red Cross uses social media to build relationships,” Public
Relations Review, 37 (1), 37–43.
Campbell, David A., Kristina Lambright, and Christopher J. Wells (2014), “Looking for friends,
fans, and followers? Social media use in public and nonprofit human services,” Public
Administration Review, 74 (5), 655–663.
Cheema, Amar, Dipankar Chakravarti, and Atanu R. Sinha (2012), “Bidding behavior in
descending and ascending auctions,” Marketing Science, 31(5), 779-800.
Choi, Namkee G., and Rita Jing-Ann (2010), “Time and money volunteering among older adults:
the relationship between past and current volunteering and correlates of change and
stability,” Ageing and Society, 30(4), 559-581.
Croson, Rachel, and Jen Y. Shang (2008), “The impact of downward social information on
contribution decisions,” Experimental Economics, 11 (3), 221-233.
Curtis, Lindley, Carrie Edwards, Kristen L. Fraser, Sheryl Gudelsky, Jenny Holmquist, Kristin
Thornton, and Kaye D. Sweetser (2010), “Adoption of social media for public relations by
nonprofit organizations,” Public Relations Review, 36 (1), 90-92.
Dass, Mayukh, Srinivas K. Reddy, and Dawn Iacobucci (2014), “A Network Bidder Behavior
Model in Online Auctions: A Case of Fine Art Auctions,” Journal of Retailing, 90 (4), 445-462.
Dewey, Caitlin (2014), #Bringbackourgirls, #Kony2012, and the complete, divisive history of
‘hashtag activism,’ Washington Post, May 8, 2014, accessed online May 27, 2018.
https://www.washingtonpost.com/news/the-intersect/wp/2014/05/08/bringbackourgirls-kony2012-and-
the-complete-divisive-history-of-hashtag-activism/?noredirect=on&utm_term=.f77c4798e469
46
Dholakia, Utpal M., and Kerry Soltysinski (2001), “Coveted or overlooked? The psychology of
bidding for comparable listings in digital auctions,” Marketing Letters, 12 (3), 225-237.
Dholakia, Utpal M., Suman Basuroy, and Kerry Soltysinski (2002), “Auction or agent (or both)?
A study of moderators of the herding bias in digital auctions,” International Journal of
Research in Marketing, 19 (2), 115-130.
Eckel, Catherine C., and Philip J. Grossman (2008), “Subsidizing charitable contributions: a
natural field experiment comparing matching and rebate subsidies,” Experimental
Economics, 11 (3), 234-252.
Espinosa, Víctor M. (2015), Martín Ramírez: Framing His Life and Art. University of Texas
Press.
Garcia, Stephen M. and Avishalom Tor (2009), “The N-Effect: More Competitors, Less
Competition,” Psychological Science, 20 (7), 871-877.
Ganim Barnes, Nora (2014), “Picture This: Top Charities Master Visual and Social Media.”
Retrieved from the University of Massachusetts Dartmouth Center for Marketing Research website:
http://www.umassd.edu/cmr/socialmediare search/charitiessocialmedia/, Retrieved, May 14, 2018
Haruvy, Ernan E., and Peter T.L. Popkowski Leszczyc (2009), “Bidder Motives in Cause
Related Auctions,” International Journal of Research in Marketing, 26 (4) 324–331.
Häubl, Gerald and Peter T.L. Popkowski Leszczyc (2018), “Bidding Frenzy: Speed of
Competitor Reaction and Willingness to Pay in Auctions,” forthcoming in Journal of
Consumer Research.
Hayes, Andrew F. (2013), Introduction to Mediation, Moderation, and Conditional Process
Analysis: A Regression-Based Approach, Guilford Press.
Hinz, Oliver, and Martin Spann (2008), “The Impact of Information Diffusion on Bidding
Behavior in Secret Reserve Price Auctions,” Information Systems Research, 19 (3), 351–68.
Hoffman, Donna L., and Marek Fodor (2010), “Can you measure the ROI of your social media
marketing?” MIT Sloan Management Review, 52 (1), 41.
Hou, Youyang, and Cliff Lampe (2015), “Social media effectiveness for public engagement:
Example of small nonprofits,” In Proceedings of the 33rd Annual ACM Conference on
Human Factors in Computing Systems (pp. 3107-3116). ACM.
Iyengar, Raghuram, Christophe Van den Bulte, and Thomas W. Valente (2011), “Opinion
leadership and social contagion in new product diffusion,” Marketing Science 30 (2) 195–212.
John, Leslie K., Oliver Emrich, Sunil Gupta, and Michael Norton (2017), “Does “Liking” Lead
to Loving? The Impact of Joining a Brand’s Social Network on Marketing Outcomes,”
Journal of Marketing Research, 54 (1), 144-155.
Jung, Minah H., Leif D. Nelson, Uri Gneezy, and Ayelet Gneezy (2017), “Signaling Virtue:
Charitable Behavior Under Consumer Elective Pricing,” Marketing Science, 36(2), 187-194.
Khazan, Olga (2013), “UNICEF Tells Slacktivists: Give Money, Not Facebook Likes,” The
Atlantic, April 30, 2013. Accessed May 27, 2018.
https://www.theatlantic.com/international/archive/2013/04/unicef-tells-slacktivists-give-
money-not-facebook-likes/275429/
Kristofferson, Kirk, Katherine White, and John Peloza (2014), “The Nature of Slacktivism: How
the Social Observability of an Initial Act of Token Support Affects Subsequent Prosocial
Action,” Journal of Consumer Research, 40 (6), 1149–66.
47
Landry, Craig E, , Andreas Lange, John A. List, Michael K. Price, and Nicholas Rupp (2006),
“Toward an Understanding of the Economics of Charity: Evidence from a Field Experiment,”
Quarterly Journal of Economics, 121 (2), 747-782.
List, John A., and David Lucking-Reiley (2002), “The effects of seed money and refunds on
charitable giving: experimental evidence from a University capital campaign,” Journal of
Political Economy, 110, 215-233.
Lipsman, Andrew, Graham Mudd, Mike Rich, and Sean Bruich (2012), “The power of “Like”.
How Brands Reach (and Influence) Fans Through Social-Media Marketing,” Journal of
Advertising research 52 (1), 40-52.
Liu, Wendy and Jennifer Aaker (2008), “The Happiness of Giving: The Time-Ask Effect,”
Journal of Consumer Research, 35 (3), 543-557.
Madianou, Mirca (2013), “Humanitarian campaigns in social media: Network architectures and
polymedia events,” Journalism Studies, 14 (2), 249-266.
Manchanda, Puneet, Grant Packard, and Adithya Pattabhiramaiah (2015), "Social dollars: The
economic impact of customer participation in a firm-sponsored online customer community,”
Marketing Science, 34 (3), 367-387.
Mangold, W. Glynn, and David J. Faulds (2009), “Social media: The new hybrid element of the
promotion mix,” Business horizons, 52 (4), 357-365.
Mathwick, Charla, Caroline Wiertz, and Ko De Ruyter (2007), “Social capital production in a
virtual P3 community,” Journal of consumer research, 34 (6), 832-849.
McLaughlin, Milo (2017), “Facebook’s new fundraising button offers charities mixed blessings.”
Accessed May 27, 2018. https://cellosignal.com/insights/facebooks-new-fundraising-button-
mixed-blessing.
Mochon, Daniel, Karen Johnson, Janet Schwartz, and Dan Ariely (2017), “What are likes worth?
A Facebook page field experiment,” Journal of Marketing Research 54 (2), 306-317.
Novak, Melissa (2012), “Consumer Perception of the Efficacy of Social Media Branding by
Non-Profit and For-Profit Organizations,” Doctoral dissertation. Rowan University, U.S.
Padilla-Meléndez, Antonio, and Ana Rosa del Águila-Obra (2013), “Web and social media usage
by museums: Online value creation,” International Journal of Information Management, 33
(5), 892-898.
Popkowski Leszczyc, Peter T.L. and Michael H. Rothkopf (2010), “Charitable motives and
bidding in charity auctions,” Management Science, 56 (3), 399-413.
Preacher, Kristopher J. and Andrew F. Hayes (2008), “Asymptotic and Resampling Strategies for
Assessing and Comparing Indirect Effects in Multiple Mediator Models,” Behavior Research
Methods, 40 (3), 879-891.
Rondeau, Daniel, and John A. List (2008), “Matching and challenge gifts to charity: evidence
from laboratory and natural field experiments,” Experimental economics, 11(3), 253-267.
Rui, Huaxia, Yizao Liu, and Andrew Whinston (2013), “Whose and What Chatter Matters? The
Effect of Tweets on Movie Sales,” Decision Support Systems, 55 (4), 863–70.
Shang, Jen, and Rachel Croson (2009), “A field experiment in charitable contribution: The
impact of social information on the voluntary provision of public goods,” The Economic
Journal, 119 (540), 1422-1439.
48
Saxton, Gregory D., and Lili Wang (2014), “The social network effect: The determinants of
giving through social media,” Nonprofit and Voluntary Sector Quarterly, 43 (5), 850-868.
Schmidt, Klaus M., Martin Spann, and Robert Zeithammer (2014), “Pay what you want as a
marketing strategy in monopolistic and competitive markets,” Management Science, 61 (6),
1217-1236.
Seidman, Gwendolyn (2013), “Self-presentation and belonging on Facebook: How personality
influences social media use and motivations,” Personality and Individual Differences, 54 (3),
402-407.
Sosik, Victoria Schwanda, and Natalya N. Bazarova (2014), “Relational maintenance on social
network sites: How Facebook communication predicts relational escalation,” Computers in
Human Behavior, 35, 124-131.
Steinhart, Yael., Michael A. Kamins, David Mazursky, and Avraham Noy (2013), “Thinking or
feeling the risk in online auctions: the effects of priming auction outcomes and the dual
system on risk perception and amount bid,” Journal of Interactive Marketing, 27 (1), 47-61.
Toubia, Olivier , Andrew T. Stephen, (2013), “Intrinsic vs. Image-Related Utility in Social
Media: Why Do People Contribute Content to Twitter?,” Marketing Science, 32 (3), 368-392.
Trusov, Michael, Randolph E. Bucklin, and Koen Pauwels (2009), “Effects of word-of-mouth
versus traditional marketing: findings from an internet social networking site,” Journal of
marketing 73(5), 90-102.
Utz, Sonja., Martin Tanis, and Ivar Vermeulen (2012), “It is all about being popular: The effects
of need for popularity on social network site use,” Cyberpsychology, Behavior, and Social
Networking, 15, 37–42.
Valverde, Will, Cameron Lefevre, and Annaliese Hoehling (2013), eNonprofit Benchmarks
Study: An Analysis of Online Messaging, Fundraising, Advocacy, Social Media and Mobile
Metrics for Nonprofit Organizations. The Nonprofit Technology Network, New York.
Wang, Dian, Haipeng (Allan) Chen, Cexun (Jeff) Cai, and Marco Palma (2018), “Know Pain, No
Gain: How Loss-Framing of Incentives Affects Creativity,” working paper.
Waters, Richard D., Emily Burnett, Anna Lamm, and Jessica Lucas (2009), “Engaging
stakeholders through social networking: How nonprofit organizations are using Facebook,”
Public relations review, 35 (2), 102-106.
Wojnicki, Andrea, and David Godes (2008), “Word-of-mouth as self-enhancement,”
unpublished working paper.
Young, Jimmy A. (2017), “Facebook, Twitter, and blogs: The adoption and utilization of social
media in nonprofit human service organizations,” Human Service Organizations:
Management, Leadership & Governance, 41 (1), 44-57.
Zephoria (2018), “The Top 20 Valuable Facebook Statistics.” Accessed May 15, 2018.
https://zephoria.com/top-15-valuable-facebook-statistics.
Zhang, Juanjuan and Peng Liu (2012), “Rational Herding in Microloan Markets,” Management
Science, 58 (5), 892-912.
Zhao, Shanyang, Sherri Grasmuck, and Jason Martin (2008), “Identity construction on
Facebook: Digital empowerment in anchored relationships,” Computers in Human Behavior,
24, 1816–1836.