Look to Others Before You Leap: A Systematic Literature Review of Social Information Effects on Donation Amounts

Perceived Social Norms

Human decision-making is largely influenced by social norms as they provide cues about how to behave in a given situation (i.e., descriptive social norm), including charitable donations (Cialdini, Reno, & Kallgren, 1990). “Conforming,” a term often connected with social norms, refers to social comparison stimulated by a desire to blend in, resulting in behavior adjustment: it is a social phenomenon whereby people perceive other people’s behavior as cues for acceptable behavior and adjust accordingly (Meyer & Yang, 2016). Human behavior is strongly influenced by a desire to conform to social norms (Bernheim, 1994; Festinger, 1954).

Social norms have been related to donation behavior (Bekkers & Wiepking, 2011). By making social information available, researchers try to influence the decisions of potential donors by hinting at an answer to the fundamental question: what does a person like me do in this situation? Donors who desire to conform are likely to target the modal donation amount as a reference for the acceptable donation amount (Bernheim, 1994). Most of the studies that mention an explanation behind the social information effect suggest social norms are at work (see Figure 3).

Smith et al. (2015) analyzed online fundraising data (focused on giving by peers) and found a positive effect of previous donations: a £10 increase of past donations increases donation amounts by £2.50. The researchers explain their findings by stating that social information provides donors with “the appropriate amount to give.” However, the researchers also state that the observed large and small donations imply that donors are not simply following the group by matching the amount donated most often. We would expect that if social information functions as a social norm, individuals mirror the amount, because social norms increase the individual tendency to conform (Bernheim, 1994; Festinger, 1954). Van Teunenbroek & Bekker (2018) found a similar effect studying donors to crowdfunding projects in a large field experiment: donors increased their donation amounts in response to social information, but they did not mirror the suggestion amount of €82. The researchers argue that the suggestion amount was too atypical to mirror a suggestion amount, because €82 does not resemble a commonly chosen donation amount.

Sasaki (2019) used a similar design as Smith et al. (2015): an online fundraising page mainly supported by peers. The study found that the more the last five donations resembled each other, the more likely they were to influence a new donor. It is important to note that the studies of Smith et al. (2015) and Sasaki (2019) did not contain an explicit test to measure if social information functions as a social norm.

Croson et al. (2009) conducted a laboratory experiment with students. The participants read a scenario mentioning that they donated $25 to a radio campaign; during a conversation with a solicitor, they were told that another member donated $10 or $50. Next the participants indicated how much they thought the average radio listener would donate (i.e., hypothetical donation context). The researchers found a direct effect of social information on hypothetical charitable giving, which decreased—but did not disappear—after controlling for social norms as a mediating variable. While the strength of the relation between social information and hypothetical giving is reduced by including social norms as a mediator, the mediating variable only explains a part of the relation. Social norms may thus partially explain the relationship between social information and charitable giving.

Goeschl et al. (2018) conducted a modified dictator game to research causal mechanisms that link social information with donation behavior through influencing social norms. Contrary to the study of Croson et al. (2009), the study existed out of two interconnected experiments conducted among a non-student sample with actual donations to an existing public good. The researchers followed a similar research design as Croson et al. (2009), but this time including actual donation amounts instead of hypothetical giving. The researchers included three conditions, control, low (€1), and high (€7). Only the high condition resulted in higher donation amounts, but not more donors. In addition, the researchers asked the participants what they thought other donors would donate (that is the stated belief). The researchers found similar results: social information affects donation behavior through changing the perception of the social norm. The outcome shows that social information does not only increase the donation amount (in the case of a high suggestion amount) but also increases the belief about the donation amounts of other donors. However, the stated beliefs are not fully aligned with the suggestion amount through social information: in the case of the high and base condition, the stated beliefs are higher than the actual donation amount.

There is one paper (Bicchieri & Xiao, 2009) that specifies the social information effects even further by distinguishing between descriptive (i.e., what most people do) and normative (i.e., what should be done) norms. The social information described in the previous studies conveys a descriptive social norm, because it describes the most common behavior (on average donors donate X euros) instead of how individuals should behave (you should donate around the average of X euros). The researchers found that participants in the role of a dictator (the one in control of distributing the earnings) is more likely to offer an equal split, if they read that this is often done by others (i.e., a descriptive social norm), or that an equal split is most appropriate according to dictators of previous sessions (i.e., an injunctive social norm). Interestingly, social information in the form of a descriptive social norm trumps an injunctive social norm when both are presented at the same time. Social information as described in this article refers to descriptive norms: it provides individuals with information on typical behavior in a given situation (Cialdini et al., 1990), instead of an injunctive norm which refers to what is typically approved in society (i.e., what people ought to do). Social information can change the perceived descriptive social norm, which may then change the donation behavior. While such a mediation is suggested by researchers focusing on social information effects, the effect remains unclear as previous studies have not assessed it properly.

Awareness of Need

Numerous studies examining donation patterns have shown that people must be aware of there being a need for help before they feel motived to give (more) (Bekkers & Wiepking, 2011), also referred to as awareness of need. Several field studies demonstrated that the degree of need for help increases the likelihood of helping (Levitt & Kornhaber, 1977; Schwartz, 1974; Staub & Baer, 1974).

Awareness of need can be increased if beneficiaries communicate this need to potential donors (Bekkers & Wiepking, 2011). We would expect that the implementation of social information could increase awareness of need by communicating this need to donors: “If people are willing to donate large amounts of money, there must really be a need for help and I should donate a higher amount.” However, this mechanism has not received any attention from scholars examining effects of social information.

Expected Quality

It is a classic assumption in economics that people are rational and aim to increase their utility from social interaction. We expect that people prefer to donate to projects and practitioners of higher quality, as low-quality projects are less likely to provide utility to people. People find it hard to judge the quality of philanthropic projects and practitioners (Handy, 1995; Rose-Ackerman, 1980, 1981) possibly because there are so many charities, and proper quality signals are costly (Vesterlund, 2003). If other donors contribute, donors may perceive the donation as a signal that the charity is of a good quality, which enhances the perceived trustworthiness and efficacy of that charity (Bekkers & Wiepking, 2011). Also, social information could signal that a non-profit is associated with a positive organizational legitimacy. A quality signal is important for establishing enough trust to donate (Bekkers, 2003).

One study specifically suggested that social information works as a quality signal (Vesterlund, 2003). Vesterlund (2003) suggests that donors’ social information changes the way donors think about the value of the project. However, Vesterlund did not include an explicit test of expected quality. The explanation of Vesterlund builds on the assumption that individuals have imperfect information about the quality of the charity.

Smith et al. (2015) provided an indirect empirical test to determine if social information functions as a quality signal. The researchers used by-proxy reasoning to determine if social information influenced the perceived quality of a project. The researchers based their approach on the assumption that the content of social information would be more important for “unknown” charities that were characterized by being start-ups, small-sized, involving young people, and based on overseas (Heutel, 2014). These types of charities were more likely to require a quality signal, as charities about which donors know little are probably more in need of additional information (Heutel, 2014). Social information, therefore, should have a stronger effect (Smith et al., 2015). However, Smith et al. (2015) found that the effects of social information were stronger for larger and older charities and that there were no effects of geography.

The reasoning of Smith et al. (2015) was based on the untested assumption that small, young, and overseas charities are perceived as a “risky investment.” With recent scandals relating to some of the bigger and better-known charities, however, donors might be more skeptical about the quality of these bigger and well-known charities.

Van Teunenbroek & Bekkers (2018) found no evidence to support the idea that social information functions as a quality signal in their field experiment with donors to crowdfunding campaigns. They expected that social information would be most effective in the middle of the campaign, as donors in the middle were expected to base their donation decision on the perceived quality of the project. Contrary to their expectations social information was especially effective in the beginning and end of a campaign.

Social information could function as a quality signal, by providing donors with the idea that others think that this is a good project/non-profit. While findings by Smith et al. (2015) and Van Teunenbroek & Bekkers (2018) suggest social information does not work in this way, they did not explicitly measure perceived quality. Therefore, it is still uncertain if social information functions as a quality signal.

Individual Donation Impact

None of the reviewed studies mentioned that social information could relate to a lower donation impact. However, social information does not always increase donation behavior and has even been found to decrease donation behavior in some studies (e.g., Croson & Shang, 2008, 2013; Meyer & Yang, 2016). Economists describe that donors are influenced by the estimated impact of their donation (Duncan, 2004), such that potential donors think “If other people are already donating large amounts of money, my money is less needed to reach the donation goal, and I could donate a lower amount (or not at all) to save money without appearing less important.” The impact philanthropy model claims that donors are stimulated and enjoy personally increasing the output of a public good (Duncan, 2004), possibly to appear important. Our interpretation of our impact mechanism is that impact refers to a donor’s feeling of how much their donation contributed to successfully assembling enough money to produce the public good. As a result, the attractiveness of giving decreases with each donation as it decreases the donation impact (Duncan, 2004). Providing social information, therefore, could be harmful. The feeling of making a difference also matters in prosocial behavior (Aknin, Dunn, Whillans, Grant, & Norton, 2013; Bekkers & Wiepking, 2011).

While impact has not been examined directly, there is one study that showed that group size has a negative influence on the donation amount (Scharf & Smith, 2016). This means that if the donation amount of previous donors gives individuals the feeling that the project is supported by a large group, it could have a negative effect on the individual donation amount.

Who

When people receive information, it matters who provides it: they prefer to follow information from similar others (e.g., family and friends, that is, strong ties) or people with similar attributes (Festinger, 1954). If there is a shared sense of social identification (i.e., shared identity), this is particularly influential for someone’s own decision-making (Weber, Kopelman, & Messick, 2004).

Researchers identified several ways in which a shared identity can be manipulated to increase the effectiveness of social information on donation behavior. Hysenbelli et al. (2013) conducted a laboratory experiment with Italian students in a hypothetical context: students indicated how much they would donate after reading a scenario mentioning the donation amount of previous Italian (i.e., similar identity) or German donors. The researchers found that a high suggestion amount (€90) resulted in higher donation amounts, whereas a low suggestion (€10) had no effect. A suggestion made by a similar other increased the effectiveness of the high suggestion amount, but not of the low suggestion amount.

One study suggested that although donors who make multiple donations (“prior members”) are unaffected by social information, the effect can be re-established when the solicitor connects the term “prior members” to social information in their solicitation (Shang & Croson, 2009). We suggest that, by adding the word “prior,” the researchers have increased the focus on similarity in having made a prior donation and have thus increased the sense of shared identity between the source and the target of the social information.

Croson et al. (2010) found that if social information was connected to the same gender as that of the participant, the information had a stronger effect on donations than in the mismatched condition (e.g., showing a female the donation amount of a male). Unfortunately, the study did not use a randomized control design. Therefore, we cannot be sure whether the increasing participation rate of females solicited by a female was due to gender matching. It is also important to discuss the role of attractiveness and gender in relation to donation amounts, because the gender of the donor and solicitor is a partial determinant of the donation amount (Raihani & Smith, 2015). A study analyzing online donations showed that male donors donate more when solicited by attractive females. The study also shows that males do not conform to donation amounts made by previous male donors. It is important to note that in this study the researchers focused on relatively large donations (i.e., at least twice the average and more than £50).

What

Classic economic theories argue that people respond to relative prices at all times (Sargeant & Jay, 2004). This effect can also be found in philanthropy: when the costs of giving are lowered, donations increase (Eckel & Grossman, 2003; Eckel, Tech, & Grossman, 2004). If a suggested donation amount is too high, people may perceive the amount as unfair or excessive, which decreases charitable giving (Hysenbelli et al., 2013). A suggested donation amount that is perceived as too high is probably less effective in influencing giving.

Several studies support this line of reasoning: social information based on the 99th percentile is no longer effective in increasing the individual donation amount (Croson & Shang, 2013; Shang & Croson, 2006). While a high donation amount might not increase the individual donation amount, Smith et al. (2015) found that a large donation (at least 10 or more times the average donation) does increase the number of donors. Apart from using amounts that are too high, mentioning an amount that is too low can decrease the individual donation amount (Croson & Shang, 2008; Meyer & Yang, 2016). In a laboratory experiment with students, other studies found that suggesting a high donation amount (€90) is actually more effective than suggesting a low donation amount (€10) to increase hypothetical giving (Hysenbelli et al., 2013). Unfortunately, it is unclear why exactly donors do not respond to information about very high donation amounts. Also, it is unclear when exactly the information is perceived as too high.

The amount mentioned might influence the nature of the impact mediator, changing the effect from negative to positive. We would expect that a lower suggestion amount increases the perceived impact of a donation, as there is a higher percentage of the target amount left. This means that a donor can have a higher impact. Duncan (2004) states that donors enjoy personally increasing the output of the good. We expect that a low suggestion amount makes donors feel like their gift really would be important and could have a big impact.

Where

The philanthropic literature describes that the donation context influences donation amounts. For instance, individuals are more generous in a public setting where they are observed (Van Vugt & Hardy, 2010). Charitable givers are held in high regard, and refraining from giving could damage one’s reputation (Bekkers & Wiepking, 2011). If we apply this to the “perceived social norm” mediator, we would expect that a public donation context increases the effect as people will only benefit from behaving in line with social norms if their behavior is observable. While some people might refrain from giving after seeing social information because it decreases the impact of their donation, therefore, they might still donate if the context is public. Social information is expected to be less effective in a private context, as the behavior is less observable. An exception on this is online giving where a donor can choose to publish their donation amount.

Alpizar et al. (2008a, 2008b) collected field data at a national park and found that tourists who were informed about other people’s typical donation ahead of the solicitation donated 25% higher amounts, but the effect was not significantly different from an anonymous donation context. Unfortunately, social information and anonymity were not manipulated in separate conditions, and the influence of anonymity on the social information effect, therefore, remains unclear.

A second contextual factor that could increase the effectiveness of social information is an ambiguous context. Social psychologists perceive a situation as ambiguous if there is no obvious way to behave and objective standards are unavailable. From an economic perspective, a context is ambiguous if the realized event is unknown or unique (Izhakian & Benninga, 2011). In an ambiguous context, information is an important input for decision making (Frank, 2015).

If social information affects the perception of a social norm, it is likely to be more effective in an ambiguous context (Crutchfield, 1955; Festinger, 1954), as people are more likely to compare themselves with other people in ambiguous situations (Buunk & Mussweiler, 2001; J. R. Martin & Wheeler, 2002). In an ambiguous context, therefore, social information could function as a social guideline.

For example, new donors to a radio campaign were affected by social information, but renewing donors were not (Murphy et al., 2015; Shang & Croson, 2009). As renewing donors are familiar with the context, it is not an ambiguous context to them, and, as a result, social information has a limited or no influence on this situation (Bekkers, 2012). New donors, on the other hand, are unaware of such a reference amount, and so they are looking for a social signal about the “correct” donation amount.