Professional Documents
Culture Documents
The Effects of Presence of Others on Online Charitable Donation Behavior
The Effects of Presence of Others on Online Charitable Donation Behavior
The Effects of Presence of Others on Online Charitable Donation Behavior
Thesis Title:
ABSTRACT
There are more and more non-profit organizations (NPOs) utilizing social
networking sites (SNSs) to attract public attention and raise funds. However, it is still an
unresolved problem and an underlying apprehension for NPOs whether the bystander
effect emerges when people perceive the presence of others on SNSs, which makes
people less likely to lend a helping hand. The present research proposed the following
hypothesis according to literature: The presence of others on SNSs will positively
(rather than negatively) affects helping intention and monetary donation through the
mediation effects of awareness of victims’ needs and perceived responsibility to help.
The results supported this inference. Besides, positive effects on the two mediators from
tie strength with message communicators and psychological closeness to victims are
also found. Psychological closeness further negatively moderates the main effect of
presence of others, which means that an effect of “offering fuel in snowy weather”
exists—the presence of others can make people feel higher level of need awareness and
perceived responsibility when the targets are psychologically distant than close. These
findings contribute a lot to not only theories but also practices.
1. INTRODUCTION
More and more NPOs have embraced the virtues of social networking sites (SNSs)
today. Through the well-designed interaction mechanism of social networking sits,
NPOs can now successfully spread out their philanthropic missions or events more
rapidly and more broadly. In fact, it is not until social networking sites sprang up in the
past twenty years that the information delivered by “group of others” has opportunities
to intervene the process of individuals’ decision making. To be more specific, these
third-party communicators, real-world friends and relatives or media channel itself in
1
the past research (e.g., Schervish & Havens, 1997; Bekkers, 2004), turn out to be
“group of others” who influence people’s behavior intention by liking, commenting, or
sharing messages on social networking sites, such as Facebook, Twitter, YouTube,
Google+, and Instagram.
Nevertheless, the results from empirical studies dealing with the influences of
real-world presence of others on helping behavior are mixed. While some
studies—majorly based on the well-established perspective of the bystander
effect—revealed that the presence of other people in a critical situation reduces the
likelihood that an individual will help (Darley & Latane, 1968), there were still other
current studies showed that sometimes the presence of others increases an individual’s
subsequent donations due to role modeling (e.g., Lincoln, 1977) or other social concerns
(e.g., Rutkowski, Gruder, & Romer, 1983; van Bommel, van Prooijen, Elffers, & Van
Lange, 2012). It is bewildering that which one is true, or closer to the truth, in the
circumstances of online helping. Whether the bystander effect appears on online social
networking sites is still an unsettled issue (Sproull, Conley, & Moon, 2005) and should
be a great concern to charities and other NPOs.
To bridge the research gap in understanding how the presence of others as a kind
of third-party communicator influences online helping behavior, this research tries to
investigate the effect of number of others on charitable helping intentions and monetary
donations on online social networking sites. Based on past research indicating that
people behaves differently in private self-focused situations and public social-focused
situations, the research put emphasis on how people perceived and behaved especially
in online social environments. In this research, it is hypothesized that the group sizes of
others’ presence (number of people) on social networking sites will positively affect
both intention (helping intention) and real donation behavior (monetary donation)
through the two possible psychological mechanisms (donors’ awareness of need and
perceived responsibility). Furthermore, the relationships between donors and recipients
(psychological closeness) or third parties (tie strength) may also play roles in the
process.
2
2. LITERATURE REVIEW
“Number of people” is the number of other people that the subject believes also
knowing the issue (Darley & Latane, 1968). On Facebook, number of people is usually
shown in the form of concrete number of likes, number of comments, and number of
shares. Considering that the meaning of number of likes and number of shares are more
and more divergent recently, the researcher chose number of shares on Facebook as the
major stimuli of the presence of others on social networking sites.
It is interesting that research has found that people behaves differently in private
self-focused situations and public social-focused situations (Fenigstein, Scheier, and
Buss, 1975). Fenigstein (1979) indicated that public (or social) self-consciousness make
people in position to infer how others perceive them and are highly sensitive to the
reactions of others in a social interaction situation. Froming and Carver (1981) also
showed that public self-consciousness would be positively correlated with compliance.
That is to say, attention to the private self may result in behavior that reflects personal
attitudes; attention to the public self may cause behavior to become more consistent
with societal expectations. When people’s public self-consciousness is triggered, they
tend to behave more prosocially.
2.1.3 The Effect of the Presence of others on Helping Behavior in Social Contexts:
Awareness of Need as a Mediator
Awareness of need, also known as “perceived need”, is the first prerequisite for any
kind of helping behavior. Awareness of need means an individual’s awareness of a
helping target’s need for support, including what kind of need it is, where the need is,
and who needs help (Bekkers & Wiepking, 2010). Past studies have shown that the
degree of need for help is positively related to the likelihood that help will be given
(Cheung & Chan, 2000). One study further examined the effects of awareness of need
on donations and revealed that not objective need but subjective perceptions of need are
crucial (Wagner & Wheeler, 1969).
There are some survey studies revealing that people adapt their giving to what
others in their environment are giving (Carman, 2003; Wu et al., 2004). From the
perspective of social learning, when seeing that others give to a charity, people can
consciously or subconsciously take this as a signal that the helping behavior is a right
thing to do. The behavior from role models may further catch people’s attentions and
then positively affect people’s awareness of victims’ need to be helped (Lincoln, 1977).
In fact, much research has supported a social learning explanation of different kinds of
prosocial behavior (for reviews of social learning theory, see Bandura, 1977; Rushton,
1980; Bandura, 1986).
To sum, based on social learning theory, the presence of others may generate
higher level of awareness of need to help others, which can indirectly and positively
affect an individual’s donation behavior. The researcher proposed the following
hypotheses.
2.1.4 The Effect of the Presence of others on Helping Behavior in Social Contexts:
Perceived Responsibility as a Mediator
Just as the discussion in previous section, attention to the public self will be easily
triggered due to the characteristic of social networking sites (Donath & Boyd, 2004). It
seemed that the second view of bystander effect was more appropriate for the helping
situations on social networking sites. Accordingly, the researcher developed the
following hypotheses.
“Friends” are one of the most important types of identities that composite the
presence of others on social networking sites. On Facebook, the “friend” relationships
can only be seen as a kind of contracts to get basic permission to see each other’s news
feeds. Because of the mixed characteristics of “friends”, it seems important to
understand the differences among different types of “friends”, especially in the context
of helping and giving. Thus, the present research tries to test whether these “friends”
have equal influences on an individual’s perception, decision making process, and
behavior.
5
Social tie is the fundamental element in social network. Tie strength is generally
considered as one of the best ways to classify the social relationships with these
collectively-called “friends” (Gilbert & Karahalios, 2009). Strong ties, usually
significant others like close friends or relatives, are people who are close and whose
social circles tightly overlap with an individual’s own; often, they are also the people
most like the individual (Levin & Cross, 2004; Gilbert & Karahalios, 2009). On the
contrary, loose acquaintances or friends of friends are known as weak ties.
Schervish and Havens (1997) found that people who are asked to give by a relative
or a friend donate a larger percentage of their income. Sokolowski (1996) analyzed the
data from a 1992 survey in the United States and also found that being asked by socially
significant others to volunteer for or contribute money to an NPO was a good predictor
of both the amount of time volunteered and the value of charitable donations. Indeed,
some people volunteer their time and contribute their money without being asked. But
for most givers, being asked is cited as a major reason for their charitable efforts
(Schervish & Havens, 1997).
There are two explanations: First, from an active view, information suggested by
donors’ relatives or friends is a symbol of endorsement (Bekkers & Wiepking, 2010;
Pan & Chiou, 2011). Therefore, donors are more likely to stand in the victims’ shoes
that are suggested by their relatives or friends, and higher level of awareness of victims’
needs can be perceived if the information is endorsed by their relatives or friends.
Second, from a passive view, social pressure about feeling the responsibility to take
actions tends to be stronger when a strong tie makes a request for a donation (Bekkers,
2004). Although donors sometimes deny the importance of social pressure (Polonsky,
Shelley, & Voola, 2002), survey studies have found that donations are rather strongly
related to social pressure (Smith & McSweeney, 2007). Based on the above inferences
about the effects of tie strength on helping behavior, the researcher proposed the
following hypotheses.
In line with the concept described above, the presence of others who are strong ties
in giving situations may lead to different outcomes from the presence of others who are
weak ties. Although large number of weak ties can at least make use of their greater
structural ability to provide novel insights, strong ties are more important sources of
useful knowledge and they can usually affect an individual’s perceptions and attitudes
(Levin & Cross, 2004), which may lead to different level of subjective awareness of
6
need to help others. From the perspective of social learning, it is also suggested that
individuals are more likely to choose significant others, who share higher
model-observer similarities, as their role models and these role models’ words and
deeds tend to be more influential than other weak ties (Bandura, 1986; Schunk &
Zimmerman, 1997). That is, an effect of “putting the icing on the cake” may exist: the
information endorsed by group of strong ties is possibly more vivid and trustworthy,
which make individuals feel higher level of awareness of victims’ needs, than endorsed
by group of weak ties.
H4a: (Perceived) Tie strength will positively moderate the relationship between
(perceived) number of people and awareness of need such that the positive relationship
is stronger in strong ties than in weak ties.
H4b: (Perceived) Tie strength will positively moderate the relationship between
(perceived) number of people and perceived responsibility such that the positive
relationship is stronger in strong ties than in weak ties.
Please note that the concepts of psychological closeness and tie strength used in
this research were different. Psychological closeness was used to describe the
relationships between donors and recipients, while tie strength was used to describe the
relationships between donors and third-party communicators, i.e., Facebook friends. To
distinguish the effects of these two constructs and to show a whole picture of different
7
relationships on social networking sites, the researcher also adds this variable of
interest.
People prefer their in-group to an out-group. Small and Simonsohn (2008) revealed
that the closer the personal relationship with a victim, the greater the degree of
sympathy and giving toward other victims suffering from the same misfortunes, which
is analogous to “intergroup contact theory”, a prominent theory in the domain of
prejudice and intergroup relations (see Pettigrew & Tropp, 2006, for a meta-analysis
review including 515 studies). When individuals view the recipients as in-group, they
are more willing to give a help.
8
level of empathy (Coke, Batson, & McDavis, 1978; Stocks, Lishner, & Decker, 2009),
so they tend to stand in the victims’ shoes and are aware of the needs more intensively.
Past research has also shown that psychological closeness increases moral
obligation (Van Boven, Kane, McGraw, & Dale, 2010; Williams et al., 2014). Levine
and Crowther (2008) had found that number of people encourages intervention when
bystanders and victim share social category membership. Rutkowski et al. (1983) also
suggested that closely-connected groups have greater social cohesion and as such
participants feel empowered by the presence of fellow bystanders and count themselves
able and responsible to intervene. That is, when people feel connected and similar with
the help recipients, the bystander effect will be reversed, so that helping is more likely
as group size increases. On the basis that the researcher has proposed that the presence
of others can positively affect both people’s awareness of need and perceived
responsibility in social contexts, this research further proposed the following
hypotheses.
3. METHODOLOGY
3.1 Design
Although this research mainly adopted an experimental design, the researcher also
measured participants’ perceptions about the stimulus just as a survey approach and
used them not only for manipulation checks but also for hypothesis testing. To be more
specific, the study adopted 2 (number of people: high vs. low) x 2 (tie strength: strong
vs. weak) x 2 (psychological closeness: close vs. distant) between subject design. Hence,
there would be 8 conditions in total conducted in laboratory. Participants evaluated their
perceptions about the situations, attitudes toward the people in need (awareness of need
and perceived responsibility) and willingness to help (helping intention and the amount
they would like to donate), which resembled usual surveys beside participants should
also interpret the manipulated and controlled stimulus.
9
3.2 Materials
3.2.1 Manipulation of Number of People
The researcher collected the data of 187 NPOs (49, 113, and 25 NPOs for each
category—“women and youngers”, “the disabled”, and “the elderly”, respectively) on
iGiving website according to the three types of charity recipients and manually searched
for the corresponding Facebook fan page for each of the NPOs. Of them, 167 NPOs (43,
102, and 22 NPOs for each category, respectively) had their own official Facebook fan
page. Then, the numbers of shares of the 50 newest posts (during August 2016 to
February 2017) from each NPO’s fan page were crawled by using a well-functioning R
package—Rfacebook. Of all sampled numbers of likes in each category, the 3th
percentile (1, 2, and 3 for each category, respectively) and the 97th percentile (406, 404,
and 112 for each category, respectively) were set for the low number and the high
number of post shares (representing the number of people involved in the post) in
present study. And to be authentic to the real, the numbers of likes were also visible.
They were calculated by the set numbers of post shares multiplied by rational ratios
according to the real ratios of the sampled mean number of likes over the sampled mean
number of shares.
According to the pilot study with 70 samples to test the manipulation of different
types of psychological closeness, spatial distances (country) could generate the most
significant differences (Mclose = 3.35 vs. Mdistant = 2.79, t = 2.045, p < 0.05) than
temporal distances (Mclose = 2.96 vs. Mdistant = 2.94, t = 0.065, p = 0.948) and social
distances (gender) (Mclose = 2.98 vs. Mdistant = 2.76, t = 0.795, p = 0.430). Therefore, the
researcher would adopt spatial distances (country) as the manipulation method in this
study.
With the Facebook algorithms, we have known that strong ties are usually shown
near the top of friend lists, while weak ties are generally shown near the bottom.
Therefore, the researcher would ask participants to log into their Facebook accounts and
fill in the top 5 (strong ties) or the last 5 (weak ties) names, which would later be
randomly shown on the posts of the fabricated charitable events.
As for dependent variables, both abstract helping intentions and concrete helping
behavior were considered so as not to fall into the common intention-behavior gap
(Sheeran, 2002). Participants’ helping intentions were simply measured with an item “I
would like to help this person” on a 7-point Likert scale. On the other hand, the helping
behavior was measured with the real monetary donation—participants were asked how
much money (up to the maximum of NT$1,000) they would like to donate to the certain
charitable event they had just read if they luckily got a windfall gain of NT$1,000. By
this way, participants could make real decisions to donate money to the charitable
events without unrelated concerns.
Please note that a control variable named “social desirability” was also included.
Social desirability is defined by the need of people to obtain approval by responding in
a culturally appropriate and acceptable manner (Crowne & Marlowe, 1960). Past
research has shown that the primary variables measured by self-report methods are
prone to be subject to the effect of social desirability, which may lead to problematic
results (Fisher & Katz, 2000). Hence, by adding this control variable, possible
confounding effects which may lead to inappropriate conclusions, such as the problem
of common method variance (CMV) and faking-good responses, could be statistically
eliminated. Five items selected from Social Desirability Scale-17 (SDS-17; Stöber, 2001)
based on the standard of highest factor loadings examined by past Taiwanese research
(Yang, 2015) were used to measure this construct by using 7-point Likert scales.
3.3 Participants
Participants were recruited from PTT Bulletin Board System (PTT BBS), which is
the largest bulletin board system in Taiwan and is widely used by college students and
non-students, and school-level student Groups on Facebook, some of these are only
open to students in our school while the others are only open to students in other 3
schools near our campus. At the end, 418 valid subjects were collected in total. Among
the participants, 39.2% (164/418) were male and 60.8% (254/418) were female; 93.1%
(389/418) were students and 6.9% (29/418) were non-student; 74.2% (310/418) were
between 20 to 29 years old, 22% (92/418) were 19 years old or below, and 3.8%
(16/418) were 30 years old or above.
3.4 Procedure
Each participant was then assigned a computer with an incognito window of
11
Google Chrome browser, showing the experimental website, and asked to complete the
tasks according to the instructions. First, participants needed to log into their Facebook
accounts and find the first or the last 5 people in their friend lists (according to the
assigned tie strength condition: strong tie or weak tie, respectively) and key in the
names on the experimental webpage. It is worth to mention that participants were
friendly informed that the study was operated in Chrome’s incognito window, so
unauthorized privacy data (especially Facebook’s id and password) would not be
recorded. Then, participants were presented a simulated screenshot of a Facebook post
including a name of their friend, the cause description, photos, as well as the number of
likes, just the same as a real post shown on Facebook timelines. After they finished their
reading, participants could click the button to the next webpage and started to report
their attitude and helping intention toward the post, the amount they were willing to
donate—if they won the NT$1,000 (about US$31) lucky draw prize afterwards—and
the assessments for the independent variables (manipulation check). The reading and
reporting process would repeat 3 times with three wholly different posts categorized as
different types of charitable causes. Finally, participants completed the demographic and
Internet usage pattern questionnaires.
Each participant spent about 11 minutes on average to complete the entire survey.
Participants who finished the experiment would receive NT$150 (about US$4.7) as an
incentive at the end of the study. After receiving the monetary reward, participants were
fully debriefed, thanked for participation, and dismissed.
4. RESULTS
12
close to the helping targets than participants assigned to distant-target conditions (Mclose
= 2.87 vs. Mdistant = 2.46, t = 5.74, p < 0.01). The above results indicated that
participants’ feelings about number of present others, tie strength with the message
delivers, and psychological closeness with the helping targets were all successfully
manipulated.
The Cronbach’s α values from the variables of the researcher’s interests fell
between 0.846 and 0.953, which were in the commonly acceptable range of reliability
(Nunnally, 1978). The Cronbach’s α value of control variable social desirability was
0.658, slightly lower than the general acceptance level of 0.7. However, it could still be
seen as a reliable measure because this result was concordant with the past studies,
which showed that the Cronbach’s α values of the SDS-17 were about 0.70 while the
test-retest reliability (4 weeks) was 0.82 (Tran, Stieger, & Voracek, 2012). Though the
scale lacked high internal consistency, Tran et al. (2012) argued that the test-retest
reliability might be more meaningful (vs. internal consistency) for indicating the
reliability of the SDS-17 because the scale developer intentionally sampled items with a
broad trait coverage. In all, with the reliability of measurement confirmed, the results’
reliability was satisfactory.
The correlations and other descriptive statistics were list in Table 1. To estimate the
validity of the constructs, the square root of average variance extracted (AVE) values
among the constructs measured with multi-items (i.e., tie strength, psychological
closeness, awareness of need, perceived responsibility, and social desirability as a
control variable) were also calculated and reported in Table 1. AVE measures the level
of variance captured by a construct versus the level due to measurement error, and all of
the AVEs from the variables of the researcher’s interests were larger than 0.5, which
indicated great convergent validities (Hair, Ringle, & Sarstedt, 2011). The square root of
AVE values were larger than the correlations with other constructs, so discriminant
validities of the constructs were also great (Hulland, 1999; Gefen & Straub, 2005). The
only one exception was again the social desirability scale. Although it had good
discriminant validity, its AVE was only 0.307 and less than the criterion of 0.5. However,
according to the suggestion from Hair, Black, Babin, Anderson, and Tatham (2006), an
13
AVE larger than 0.25 was actually acceptable. In all, the validities of measurement were
also confirmed to be satisfactory.
Mean S.D. 1 2 3 4 5 6 7 8
1 NP 3.79 1.70
2 TS 4.48 1.89 0.029 0.943
3 PC 2.65 1.26 0.198*** 0.056** 0.820
4 AN 5.11 1.07 0.229*** 0.100*** 0.288*** 0.780
5 PR 4.41 1.08 0.240*** 0.103*** 0.555*** 0.576*** 0.830
6 HI 4.72 1.23 0.204*** 0.131*** 0.470*** 0.587*** 0.735***
7 MD 450.6 322.2 0.084*** 0.112*** 0.248*** 0.269*** 0.421*** 0.345***
8 SD 4.26 0.85 0.062** -0.078*** 0.072** 0.039 0.167*** 0.138*** 0.177*** 0.554
Note: NP—Number of People; TS—Tie Strength; PC—Psychological Closeness; AN—Awareness of Need;
PR—Perceived Responsibility; HI—Helping Intention; MD—Monetary Donation; SD—Social Desirability (control
variable). The values on the diagonal and marked with underlined italics are the square root of AVE. * p < 0.1; ** p <
0.05; *** p < 0.01.
Table 1 Descriptive Statistics, Square Root AVEs and Correlations among Constructs
14
It was also noticed that the above results could only be found when participants’
self-report values of perceptions were used to analyze. If the manipulated dummy
variables were used (i.e., highNP = 1, lowNP = 0, strongTS = 1, weakTS = 0, closePC =
1, and distantPC = 0) and tested with ANOVA, few of the effects described above could
be verified despite that the manipulations had already been checked beforehand. Beside
the main effect of tie strength on awareness of need (SSTS = 10.43, F(1,51) = 5.46, p <
0.05), none of the other effects is significant when either awareness of need or perceived
responsibility served as the dependent variable.
To sum, the three-step regression analysis indicated that both awareness of need
and perceived responsibility mediated the effect of number of people on helping
intention. Perceived responsibility also mediated the effect of number of people on
monetary donation. However, it failed to prove that awareness of need mediated the
effect of number of people on monetary donation when controlling perceived
responsibility at the same time, perhaps for the strong correlation between these two
variables (r = 0.576).
15
Figure 2 Results from the Present Study
The conclusions from past research focusing on how presence of others affects
helping intentions or behavior were mixed. The researcher tried to summary from
literature that the differences between private personal situations and public social
situations could play an important role; furthermore, the researcher proposed that the
presence of others on social networking sites was more likely to affect people’s helping
intentions and behavior positively. Indeed, the results of the present study shows that the
presence of others increases people’s awareness of victims’ needs and perceived
responsibilities to help, and it also positively affects helping intentions and monetary
donations fully mediated by these two types of perceptions.
Besides number of the presence of other people, tie strength and psychological
closeness both have significantly positive influences on awareness of victims’ needs and
perceived responsibilities to help. These results replicate theory as well as past research
and support the notions that close friends and relatives can be important socializers of
prosocial behavior, and people do experience higher level of involvement for
psychologically close events than distant ones.
16
However, tie strength does not moderate the relationship between the presence of
others and awareness of need, nor does the relationship between the presence of others
and perceived responsibility; the moderation effects of psychological closeness are
contrary to the hypotheses and are negative. It seems that number of likes or shares and
tie strength may serve as two different sources of endorsement for charitable
information, but not for each other. In fact, from an individual’s aspect, it is not always
true that all of the presence of others shared the same tie strength. Perhaps this is the
reason why there is no synergistic effect from these two factors.
It is also noticed that not only the mediation but also the moderation effects hold
only when perceptions are used in hypothesis testing. If the dummy coding values of
manipulated groups are used, few of the above effects can be found despite the fact that
participants’ perceptions about the stimulus have been verified to be successfully
manipulated. These results supports the literature which has pointed out that perceptions
are not always the same as objective facts, and the former has much stronger relation
with people’s behavior intentions than the latter, particularly in helping situations
(Wagner & Wheeler, 1969).
The major limitations of the present research are that the composition of the
sample in the study was primary students, and participants reported their perceptions
only based on Facebook situations. Besides, there are still some related issues which
need to be further investigated. First, there are many other forms of presence of others
that can be seen on online communities. Can the effects of other forms of presence of
others reach the same conclusions as the present study? Second, besides the two
mediators discussed in this research, is there any other important variable
(psychological mechanism) also mediating the effect of stimulus of presence of others
17
on helping? Last, although the researcher has tried to summary from literature and
indicated that the differences between private (personal) situations and public (social)
situations could lead to either negative or positive effect from the presence of others, it
is still unclear due to lack of direct comparisons between them. There should be more
studies for further verifications.
All in all, this empirical research has provided a really good start for future
research, and it contributes profoundly to not only practices of NPOs but also some
recent developed concepts that have only been discussed theoretically. It is hoped that
there will be more studies further exploring related issues or extending the results to
other fields.
22