Abstract
Charities often include low-value monetary (e.g., coins) and nonmonetary (e.g., greeting cards) pregiving incentives (PGIs) in their donation request letters. Yet little is known about how donors respond to this marketing strategy. In seven studies, including two large-scale field experiments, the authors demonstrate that the effectiveness of PGIs depends on the organization’s goals. People are more likely to open and read a letter containing a monetary PGI (vs. a nonmonetary PGI or no PGI). In addition, monetary PGIs increase response rates in donor acquisition campaigns. However, the return on investment for direct mail campaigns drops significantly when PGIs are included. Furthermore, average donations for appeals with a nonmonetary PGI or no PGI are similar, while those with a monetary PGI are actually lower than when a nonmonetary PGI or no PGI is included. This is because monetary PGIs increase exchange norms while decreasing communal norms. This effect remains significant when accounting for alternative explanations such as manipulative intent and the anchoring and adjustment heuristic.
Keywords
Not a week goes by that I don’t receive requests for monetary donations to one charitable organization or another.…However, I must ask these organizations if sending out hundreds of thousands of requests for donations with a nickel or dime attached inside an envelope full of address labels imprinted with my name and address is honestly worth it.
Most charitable organizations rely on individual donations to provide much-needed services, and over 90% of nonprofit organizations use direct mail as one of their primary fundraising methods (Non-Profit Research Collaborative 2016). However, most direct mail campaigns achieve modest success, with an average response rate of 1% to 3.7% (Data & Marketing Association 2015). To attract attention and encourage donations, charities often include pregiving incentives (PGIs), also known as unconditional gifts or front-end premiums, in their donation request letters (for examples, see Web Appendix A). A PGI is defined as the provision of a benefit or a favor before requesting compliance (Marwell and Schmitt 1967; Regan 1971).
In 2013–2015, PGIs such as coins and greeting cards were included in approximately 40% of the total nonprofit mail volume (Paradysz + PM Digital Research 2016). In fact, inclusion of monetary PGIs is so popular, the strategy has its own moniker: “the coin trick” (Jones 2010; Rieck 2010). Yet there are concerns over the effectiveness of this strategy (Taylor 2018). The goal of the present research is to address the question posed in the quote at the opening of this article—are PGIs in charitable campaigns worth it?
Several streams of research have examined the effect of incentives on prosocial behavior (for a review, see White, Habib, and Dahl [2020]), some yielding conflicting results. The literature on reciprocity shows that people feel a strong sense of obligation to repay benefits they have received, even if those benefits are unwanted (Cialdini and Goldstein 2004; Gouldner 1960). Accordingly, Falk (2007) found that including a small PGI, such as a postcard, in a charity appeal significantly increased the number of donations, though the average donation amount remained fairly constant. Motivation crowding theory, in contrast, proposes that external rewards diminish intrinsic motivation or sincerity to do good, which can result in reduced donations (Barasch, Berman, and Small 2016; Gneezy, Meier, and Rey-Biel 2011). For instance, Newman and Shen (2012) found that providing nonmonetary conditional/promised gifts (e.g., pens, reusable bags) decreased average donations because such external incentives “crowd out” altruistic motivations to donate.
One limitation of the aforementioned literature is the focus on just one aspect of donation behavior: intrinsic characteristics of the donor (i.e., sense of obligation, intrinsic motivation). Yet offering incentives can also affect the relationship between two entities. Relationship norms between consumers and organizations are important to understand as they play a critical role in consumer reactions, including attitudes and behaviors toward a brand’s marketing actions (Aggarwal 2004). To illustrate how incentives might influence a relationship, imagine a situation in which the person you have a burgeoning romantic relationship with gives you a $20 bill before asking you to attend a family function with them. We predict that the mere introduction of money is likely to lead you to question the nature of the relationship. But what if, rather than cash, your partner gives you a gift like a bottle of wine or heartfelt greeting card?
This scenario exemplifies a second limitation of prior research on incentives. It does not address how different types of incentives affect donations. The plethora of diverse gifts that charities send to potential donors can be classified into two groups: monetary and nonmonetary. In the current research, we draw on the interpersonal relationships literature to develop a framework for understanding consumer reactions to these two classes of PGIs. We propose that donors typically perceive charities as communal organizations and use communal norms when interacting with them. However, the inclusion of monetary PGIs diminishes communal norms while increasing exchange norms, resulting in lower average donations.
Our research makes a number of theoretical and practical contributions. Although the literature on brand relationships is substantial, relatively little of it has examined relationship norms and factors that influence their salience (for a review, see MacInnis and Folkes [2017]). Foundational research in marketing manipulated the salience of communal and exchange norms using hypothetical scenarios grounded in personal relationships (Aggarwal 2004; Aggarwal and Law 2005). In contrast, we show that organizations can unwittingly influence the salience of these norms simply by including gifts or incentives in their donation solicitations.
The current research also extends recent work on how superficial elements of a donation appeal can influence donation behavior, even when the content is held constant (Townsend 2017). We show that PGIs are tangible cues that influence people’s communal and exchange perceptions of the charitable organization, leading to sizable differences in donation behavior. In doing so, we answer calls for investigation of PGIs in the donation context, with an emphasis on “the influence of the type of gift” (Alpizar, Carlsson, and Johansson-Stenman 2008, p. 1059).
Our work has managerial implications for the hundreds of nonprofit organizations that include PGIs in their direct mail campaigns. As stated previously, including PGIs such as coins and greeting cards in direct mail campaigns is a common strategy for nonprofits (Paradysz + PM Digital Research 2016). We use different methods, samples, and incentives to examine the effect of PGIs on multiple outcomes of interest to charitable organizations, including increasing awareness, procuring future donors, influencing consumer perceptions, and fundraising. Results indicate that PGIs have different effects on different outcomes, and that the best strategy depends on what the charity wants to achieve. For example, monetary incentives may not be effective at increasing the average donation amount but may help charities gain visibility and awareness by increasing the opening and response rate among people who have not donated before. Thus, the results of this research allow nonprofits to make more informed cost–benefit analyses in deciding whether strategies such as “the coin trick” are worth it.
Theoretical Development and Hypotheses
Relationship Norms in the Marketplace
Extant research has identified two primary relationship norms that guide how people give and receive benefits: communal norms and exchange norms (e.g., Clark and Mills 1993). Communal norms dictate that individuals attend to other’s needs and demonstrate concern for one another. People following communal norms are motivated to care for others and are willing to incur costs to do so, regardless of whether they will receive anything in return. Most family relationships, romantic relationships and friendships are governed by communal norms (Clark and Mills 1993). In comparison, exchange norms are those in which benefits are given with the expectation of receiving comparable benefits in return. Those who follow exchange norms are less likely to help others without a benefit or reward. A prototypical relationship example following exchange norms is that of business partners.
Relationship norms influence interactions not only with individuals but also with marketplace entities. Aggarwal (2004) showed that people who followed communal norms evaluated a brand more positively when the brand offered a noncomparable (vs. comparable) reward in return for help, while those who followed exchange norms evaluated the brand more positively when given a comparable (vs. noncomparable) reward. Wan, Hui, and Wyer (2011) found that when communality was low, people expected businesses to provide services for their payments and viewed providers negatively during service failures. In contrast, when communality was high and self-obligation was highlighted, people behaved in a more caring and understanding manner, and thus reacted less negatively to service failures.
Beyond the situations manipulated in laboratory experiments, few marketplace interactions have been found to operate according to communal norms. This may be because previous research has largely focused on for-profit companies, which are generally perceived as providing benefits with the expectation of receiving comparable benefits in return. We propose that, unlike for-profit companies, nonprofits are perceived as high communality organizations that care for others’ welfare and give benefits without expecting anything in return. In support of this, Aaker, Vohs, and Mogilner (2010) found that nonprofits such as charities (vs. for-profit companies) are seen as warm entities that care about the welfare of others. McGraw, Schwartz, and Tetlock (2012) identified additional organizations that consumers expect to have communal obligations (i.e., religious and pharmaceutical), and showed that people respond to the marketing efforts of these organizations differently than those of typical businesses.
To further lay the foundation that nonprofits are seen as communally oriented, we conducted a pilot test on perceptions of a wide range of both for-profits and nonprofits (for details, see Web Appendix B). In addition, we examined the potential role of organizational familiarity, which has been shown to affect donors’ inferences about the charity (Smith and Schwarz 2012). Results from the pilot study revealed that communal perceptions are higher (Mcharities = 5.54, Mbusiness = 3.68; F(1, 199) = 95.97, p < .001), and exchange perceptions are lower (Mcharities = 2.37, Mbusiness = 4.54; F(1, 199) = 99.37, p < .001), for charities than for businesses, regardless of how familiar they are. We propose that this high degree of communality is a foundation for donors’ willingness to give benefits (e.g., money) generously.
The Role of Incentive Type on Charity Perception and Donation Behavior
We predict that relationship norms are influenced by the type of PGI (monetary vs. nonmonetary) commonly included with charity letters. Money is more likely to evoke marketplace norms and lead people to behave in a quid pro quo manner (i.e., give comparable benefits for any benefits received) compared with nonmonetary goods (Heyman and Ariely 2004). Activating the concept of money leads people to infer that they are in a businesslike or exchange relationship and behave as though they are interacting with a business party (e.g., Kouchaki et al. 2013). Including a monetary PGI along with a donation appeal thus has the potential to increase exchange norms. At the same time, a monetary PGI should decrease communal norms. Reminders of money lead people to be more self-centered and eschew strong relational ties, such that they prefer not to rely on or be relied on by others (Vows, Mead, and Goode 2008). They may therefore be less inclined to think of relationships in communal terms after receiving a monetary PGI. We hypothesize that the increase in exchange and decrease in communal norms due to the presence of the monetary PGI will jointly lead to lower average donations.
Nonmonetary PGIs are preferred over monetary ones in communal relationships, as they do not evoke marketplace norms in a payment context (Heyman and Ariely 2004) and are consistent with communal norms. A nonmonetary gift may thus have the potential to increase communality. However, unlike exchange norms, communal norms do not dictate the prompt repayment of benefits given or received (Clark and Mills 1993). Indeed, previous research shows that communal participants evaluate a brand more positively if the request for help from the brand is delayed compared with when it is made immediately (Aggarwal 2004). Giving a nonmonetary PGI and then asking for help straightaway is more aligned with exchange norms and could offset any communal increase from receiving a small gift, resulting in no overall change in communality. In addition, communal norms dictate noncontingent, need-based giving (Clark and Mills 1993). Thus, whether and how much people donate should not depend on the presence or the value of nonmonetary PGIs. Taken together, this suggests that a nonmonetary PGI is unlikely to increase net communality (i.e., communal norms relative to exchange norms) or average donations in a direct mail campaign. It is relevant to note that our conceptualization suggests that a monetary PGI will lead to lower net communality but not necessarily flip the overall relationship from communal to exchange. Previous research typically manipulates communal versus exchange relationships, while less attention has been paid to how people react to communality variations on a continuum. Yet a review of the literature reveals that even manipulations of relationship norms, which clearly delineate their differences, do not always produce effects at the extreme ends of the continuum. For instance, Wan, Hui, and Wyer’s (2011; Experiment 2) manipulation check showed that both the communal and exchange scenarios elicited communality scores above the midpoint of the seven-point scale (4.83 vs. 4.08, respectively), with a significant difference between the means. Despite the fact that both scores were on the communal end of the scale, their relative difference produced significant hypothesized effects in consumer behavior. Conceptually, this is explained by research showing that communal relationships vary along a continuum, with one’s willingness to devote resources toward promoting the other’s welfare increasing with higher levels of communality (Clark and Aragón 2013).
Importantly, charities have campaign objectives other than maximizing individual contributions. Pregiving incentives may be useful for goals that are lower on the consumer response hierarchy such as increasing public awareness and enlarging the donor list (Lavidge and Steiner 1961). Research has shown that monetary incentives are more effective in eliciting survey responses than nonmonetary ones or no incentives, and monetary incentives are particularly useful for people who have no interests in the survey topic (Singer and Ye 2013). Moreover, lay beliefs and laws guiding people on how to treat money may lead them to be reluctant to throw away cash (e.g., Title 18, Chapter 17, of the U.S. Code prohibits the debasement of coins). Thus, monetary PGIs could be effective in promoting initial engagement with a piece of mail, especially among individuals who are unfamiliar with the charity. In addition, opening rate is correlated with response rate, or the likelihood to act on the mailing (Feld et al. 2013). Charities can add these respondents to their mailing list for future campaigns. Thus, although our conceptualization predicts lower average donations with monetary PGIs (vs. nonmonetary and no PGIs), they may be effective for increasing opening and response rates, especially among people who are unfamiliar with the charity and thus need an incentive to open the envelope. Formally, we hypothesize:
Overview of Studies
We test our hypotheses in seven studies. Study 1 examines the effect of PGIs on the opening rate of donation letters. Study 2 is a field study in which we partner with a charity on a 9,000 household donor acquisition campaign to examine the effect of PGIs on response rate, average donation amount, and return on investment (ROI). Study 3 examines the effect of PGIs on average donations using real incentives and contributions. Study 4 examines the mediating role of relationship norms on average donations while testing the alternative explanations of anchoring, manipulative intent, and charity inefficiency. Both the incentives and donations in Study 4 are hypothetical. Study 5 examines the impact of PGIs of varying value, and Study 6 examines the effect of phrasing monetary PGIs in different ways on average donations. These studies use hypothetical incentives but ask participants to donate from their potential bonus payment to increase realism while curtailing study costs. Finally, Study 7 is a field experiment that examines the anchoring effect on response rate, average donations and ROI for a year-end campaign for recurring donors.
Study 1
Charities often have the goal to increase awareness and achieve high visibility through their donation campaigns (Smith and Schwarz 2012). Encouraging potential donors to open the charity letter may be especially critical to the success of donor acquisition campaigns, in which appeals are sent to people who are unfamiliar with the charity or have never donated before. In this study, we explore the crucial question of how enclosures of PGIs affect how these individuals handle the letter. We predict that people are reluctant to throw away a piece of mail with money in it, which results in them opening and reading the letter (H1). We sample a diverse population by recruiting staff and faculty at a major U.S. university, who are likely to have more disposable income than convenience samples such as college students and online survey takers.
Participants and Procedure
Two hundred forty university staff and faculty members from a large Midwestern university (187 female; Mage = 46.02 years; Mincome = $75,000 to $99,000) participated in this study in exchange for either a $5 prepaid Visa card or a university branded gift of a similar value (e.g., hat, wine glasses). The study was conducted in six buildings on campus (School of Business, School of Education, University Health Center, University Administration, University Endowment, and Payroll and Human Resources). Participants in each building were approached individually for oral consent to participate in the study and signed up for a time slot of their choice. Participants were taken one at a time to the experiment room. A research assistant who was unaware of the study hypotheses administered the study and recorded the data. The study consisted of a mail sorting task along with a short survey of demographic information.
Mail sorting task
The study utilized a between-subjects design. Participants were randomly assigned to one of three PGI conditions: a charity letter with a clear window presenting a monetary PGI (quarter), a nonmonetary PGI (a greeting card), or no PGI. All participants were presented with six pieces of sealed mail in the same order, including the focal charity letter from a fictitious charity called “Help Fight Cancer Society” (the same charity used in one of the supplemental studies in the Web Appendix). The other five pieces of mail were the same across all three PGI conditions. There was one solicited letter (dental bill), and four unsolicited letters (SmartShopper ad, credit card offer, car insurance ad, and retirement insurance ad). We included both solicited and unsolicited letters to compare the treatment of the charity letter with mail of varying importance. The cover design and content for each letter were created based on actual letters.
Four boxes were placed on a table with the following labels: unopen throw away, open without reading, open and read, and unopen but keep. Participants were asked to imagine that they just got home from getting their mail and were sorting them. They were asked to sort the mail as they normally would and place each letter into the box corresponding to their decision. An “other” option was provided in case they intended to handle the letter differently. After the mail sorting task, participants answered a few demographic questions, chose either a Visa card or a university gift as their payment, and were thanked and dismissed.
Results and Discussion
We compare the opening rate of the three PGI conditions next and report the comparison of each appeal against the solicited and unsolicited letters in Web Appendix C.
Unopen throw away
Consistent with H1, the number of people in the monetary PGI condition who chose to throw away the charity letter without opening it (16.46%, N = 13) was significantly lower than those in the nonmonetary PGI (45%, N = 36, χ2(1) = 15.19, p < .001) and no-PGI (49.38%, N = 40, χ2(1) = 19.57, p < .001) conditions. No difference was found between the monetary and nonmonetary PGI conditions (χ2(1) = .31, p = .58).
Open without reading
We found no difference across the three PGI conditions (38.46%, Nmonetary = 20; 32.69%, Nnonmonetary = 17, 28.85%, Ncontrol = 15; all ps > .29).
Open and read
The number of people who chose to open and read the charity letter was significantly higher in the monetary PGI (48.10%, N = 38) condition than that in the nonmonetary (26.25%, N = 21, χ2(1) = 8.13, p = .004) and no-PGI (17.28%, N = 14, χ2(1) = 17.31, p < .001) conditions. No difference was found between the nonmonetary and no-PGI conditions (χ2(1) = 1.90, p = .17).
Unopen but keep
We found no difference across the three PGI conditions (30.77%, Nmonetary = 8; 23.08%, Nnonmonetary = 6; 46.15%, Ncontrol = 12; all ps > .14).
Discussion
This study examined the effect of PGIs on opening rate in a donor acquisition campaign. Confirming H1, fewer participants chose to throw away the charity letter without opening it in the monetary PGI condition than the nonmonetary and no-PGI conditions. Moreover, more people chose to open and read the letter in the monetary PGI condition than in the nonmonetary and no-PGI conditions. Enclosing a nonmonetary PGI was no more effective than not enclosing any PGI in leading people open and/or read the charity letter. Having assessed the impact of PGIs on opening rate, we turn our attention to other important campaign outcomes, including the response rate and average donation amount and ROI.
Study 2
A primary objective of donor acquisition campaigns is to raise awareness for the organization and enlarge the donor pool, which can lead to more profitable campaigns in the future (Smart Annual Giving 2013). Study 1 suggests that a monetary PGI promotes this goal by persuading recipients to open and read the letter. In Study 2, we partner with a local mental health and suicide prevention nonprofit to examine more downstream consequences.
As discussed in the “Conceptual Development” section, we predict that a monetary PGI (vs. nonmonetary and no PGI) will lead to lower average donations. We also examine how different types of PGIs affect the response rate, or number of replies to the mailing. Previous research has shown that monetary payments lead people to behave in a quid pro quo manner (Heyman and Ariely 2004). Thus, after receiving a coin in the mail, people may respond by sending the money back to the organization. For low-value monetary PGIs, this would lead to lower average donations, but a higher response rate.
Because charitable organizations often lose money in donor acquisition campaigns, an important goal is to minimize losses (Smart Annual Giving 2013). Thus, we measure the ROI for each type of PGI. We predict that enclosing PGIs, both monetary and nonmonetary, will lead to a lower ROI than not enclosing any PGI—although the response rate may be somewhat higher for monetary PGI appeals, average donations will be lower, and the total amount raised is unlikely to offset the high cost of sending PGIs.
Given the extremely low response rate for typical direct mail campaigns (.65%; Smart Annual Giving 2013), it is difficult to draw conclusions from nonresponses about whether PGI had an impact on donation behavior. Nonresponses could be due to indifference toward the PGI, or because people never received or saw the letter. Thus, to calculate average donation amount, we analyzed data only from respondents who made a donation. This is consistent with prior research on donation appeals with low response rates (Fajardo, Townsend, and Bolander 2018). However, in estimating net loss per mailing, we included all those who had been solicited (N = 9,000), because ROI addresses whether the benefits of enclosing PGIs justify the total cost.
Materials
PGIs
An examination of over 100 charity letters revealed that common monetary PGIs range from a few cents to $2.50, and common nonmonetary PGIs include address labels, greeting cards, and note pads. In consultation with the charity, we decided to use a quarter ($.25) for the monetary PGI and a greeting card for the nonmonetary PGI. To control for the potential confounds of perceived overhead costs of the charity (Gneezy, Keenan, and Gneezy 2014) and subjective incentive value (Raghubir 2006), we conducted a pretest to ensure that the perceived cost (to the charity) and value (to the recipient) are similar for the monetary and nonmonetary PGIs we chose (both ps > .15; see Web Appendix D).
Mailing list
A city-wide mailing list of 11,000 people—from a total population of about 91,000—was acquired by the charity. The mailing list contained people’s names, addresses, zip codes, and phone numbers. Nine thousand people were randomly selected from the mailing list and assigned to one of the three PGI conditions: monetary ($.25), nonmonetary (a greeting card), or no PGI. The charity ensured that none of them had donated to the charity before.
Business reply return envelopes
We included a standard business reply return envelope with each letter. Participants also had the option of using their own stamps to help the charity reduce costs. A code was stamped on each return envelope to keep track of PGI condition.
Donation request letters, incentives, and return card
Three versions of the donation request letter were created to correspond to the monetary, nonmonetary, and no-PGI conditions, along with a perforated return card containing donation instructions at the bottom of the letter (see Web Appendix E, Study 2). For both the monetary and nonmonetary conditions, an additional sentence at the top front page of the donation request letter stated, “Please accept the attached quarter [greeting card] as our gift to you.” A quarter was glued to the top front page in the monetary PGI condition, and a blank greeting card with accompanying envelope was placed within the folded donation request letter in the nonmonetary PGI condition. Three distinct online donation links were created to correspond to each incentive condition and placed clearly at the top of the perforated return card to give donors the option of donating online.
Dependent Measures
Response rate
Response rate was defined as the number of participants who made a donation in each condition.
Average donation
The average donation in each condition was computed by dividing the total amount of money donated by the number of donors who contributed.
ROI
The costs associated with incentives, postage, and printing were similar to or lower than the costs of comparable campaigns the charity has run in the past. We calculated the ROI for each condition by taking into consideration total donations relative to associated costs [ROI = donations received − variable costs (e.g., material cost, incentive cost [if any], business return envelope cost [for cost details, see Web Appendix F]). We then divided this number by the number of participants in each condition.
Results and Discussion
Campaign overview
The response rate seven weeks after mailing (December 18, 2015, to February 5, 2016) was .56% (N = 50). The charity received a total of $1,559.50: $654.50 from the monetary PGI condition (range: $.25 to $200), $335 from the nonmonetary PGI condition (range: $10 to $50), and $570 from the no-PGI condition (range: $10 to $100), with an average donation amount of $31.19. The response rate and average donation are comparable with the national average for donor acquisition campaigns, at .65%, and $15–$45, respectively (Smart Annual Giving 2013).
Response rate
The response rate from the monetary PGI (54%, N = 27) was significantly higher than the nonmonetary PGI (22%, N = 11, χ2(1) = 6.78, p = .009) and no PGI (24%, N = 12, χ2(1) = 5.81, p = .016). No difference was found between the nonmonetary PGI and no-PGI conditions (χ2(1) = .04, p = .83).
Of note, over half the responses (N = 15) in the monetary PGI condition were returns of the quarter. Such a response is predicted by the relationship norms framework, as lower communal norms and higher exchange norms lead to a one-to-one exchange mindset and less focus on others’ needs. Charities recognize that initial contributions tend to be small, and they hope to cultivate first-time donors to give more over time; thus, the contact information for everyone who “acts upon the mailing” (MobileCause 2019), including people mailing back the quarter, may be added to the donor pool for future campaigns. For these reasons, we include donors who mailed back the quarter in the calculation of overall response rate and average donations. Nevertheless, some charities may view these responses as fundamentally different from standard donations and be interested in the results of the campaign when they are removed. We briefly discuss the donation results excluding those responses here and provide more details in Web Appendix G.
Average donation
One person in the monetary PGI condition donated $200, which far surpassed the average (z > 3) and is considered an outlier (MacKinnon, Lockwood, and Williams 2004). For completeness, we report the results both excluding and including the outlier. A Shapiro–Wilk normality test showed that the donation amount was not normally distributed (p < .001), so we log-transformed the data. Excluding the outlier, results from a one-way analysis of variance (ANOVA) revealed that PGI type had a significant impact on donation amount (in log-transformed values: Mmonetary = .30, SD = 1.09; Mnonmonetary = 1.42, SD = .26; Mcontrol = 1.56, SD = .35; F(2, 46) = 12.50, p < .001; in dollars: Mmonetary = $17.48, SD = $25.88; Mnonmonetary = $30.45, SD = $16.35; Mcontrol = $47.50, SD = $34.54; see Web Appendix H). Consistent with H2, the monetary PGI led to significantly lower donations than the nonmonetary PGI (p = .001) and no PGI (p < .001). No difference was found between the nonmonetary and no-PGI conditions (p = .70). Similar results were found when the outlier was included (in log-transformed values: Mmonetary = .38, SD = 1.14; Mnonmonetary = 1.42, SD = .26; Mcontrol = 1.56, SD = .35; monetary PGI vs. nonmonetary PGI: p = .002, monetary vs. no PGI: p < .001, nonmonetary PGI vs. no PGI: p = .70; F(2, 47) = 10.20, p < .001; in dollars: Mmonetary = $24.24, SD = $43.34; Mnonmonetary = $30.45, SD = $16.35; Mcontrol = $47.50, SD = $34.54). Notably, this effect is erased when the 15 donors who merely returned the quarter are included in the analysis (all ps > .18; for all comparisons, see Web Appendix G).
ROI
The ROI for the entire campaign was −$3,766.48, with −$1,562.32 in the monetary PGI condition, −$1,477.46 in the nonmonetary PGI condition, and −$726.79 in the no-PGI condition. A Shapiro–Wilk normality test indicated that ROI was not normally distributed (p < .001), so we log-transformed the data. Due to the negative values of the data, we added a constant number a (a = .741) so the new ROI data became Y + a, where min (Y + a) = .001 (Box and Cox 1964). We found that PGI had a significant impact on ROI (in log-transformed values: Mmonetary = −2.97, SD = .34; Mnonmonetary = −.84, SD = .14; Mcontrol = −.50, SD = .13; F(2, 8,997) = 107,748.64, p < .001,
Discussion
The results of this field experiment provide initial support for our hypothesis that a monetary PGI (vs. a nonmonetary PGI or no PGI) will decrease average donations, while a nonmonetary PGI (vs. no PGI) will have no effect. The monetary PGI also elicited a higher response rate, driven by people who returned the $.25. Finally, enclosures of both monetary and nonmonetary PGIs led to a significantly worse ROI than no PGI. A caveat to these results is that no average donation amount difference is observed when people who merely returned the quarter are excluded in the analysis.
In Studies 3–6, we further examine the effect of PGIs on donation behavior by testing our predictions in a more controlled setting. An important methodological departure from the field experiment is that all participants are asked to read the charity appeal in these subsequent studies. People do not have the option of ignoring the letter, so a major predictor of response rate (whether one opens or reads the appeal) is controlled for. The design of the lab experiments does not lend itself to accurate calculations of ROI, so we return to ROI in Study 7, in which we conduct a large-scale field experiment for a year-end campaign.
Study 3
In Study 3, we again partnered with a local mental health clinic to provide student participants with a physical charity letter to open and read. To measure actual behavior, participants were allowed to donate money from their study payment. We utilized monetary and nonmonetary PGIs of a different value to increase the generalizability of the findings.
Participants and Procedure
One hundred thirty-two students (71 women; Mage = 20.77 years) from a large public Midwestern university participated in this study in exchange for $5. Participants were randomly assigned to one of three conditions: a charity appeal with a monetary PGI ($.50), nonmonetary PGI (a higher-quality greeting card), or no PGI. We conducted a pretest to ensure there was no difference in either the perceived cost or subjective benefit of the two PGIs (both ps > .43) (for details, see Web Appendix D). All instructions and responses were on paper. Participants received five dollars as well as course credit for their participation and were instructed to read the charity letter (see Web Appendix E) and respond to the questions in the survey booklet. Responses from two participants were unusable—one participant refused the payment before the study started and another participant did not complete the survey—leaving a final sample of 130.
Dependent Measures
Response rate
Response rate was defined as the number of participants who made a donation in each condition.
Average donation
Participants were asked how much of the money they received from their participant payment and charity appeal, if applicable, they would like to donate to the charity. The average donation in each condition was computed by dividing the total amount of money donated by the number of participants in each condition.
Average donation percentage
Because the total amount of money available for donation differed across conditions—participants in the monetary PGI condition had $5.50 to donate (their $5.00 study payment, plus the $.50 PGI), whereas those in the nonmonetary and control conditions only had $5.00—we calculated donation percentage (donation amount divided by the total amount of money available for donation) in addition to the average donation amount.
Results and Discussion
Response rate
No response rate difference was found between any of the three conditions (Nmonetary = 30, Nnonmonetary = 29, Ncontrol = 32; all ps > .64).
Average donation
Results from a one-way ANOVA showed a marginal effect of PGI on average donations (Mmonetary = $1.58, SD = $2.07; Mnonmonetary = $2.37, SD = $2.14; Mcontrol = $2.50, SD = $2.14; F(2, 127) = 2.39, p = .096,
Average donation percentage
Results from a one-way ANOVA revealed a main effect of PGI type on donation percentage (Mmonetary = 28.75%, SD = 37.58%; Mnonmonetary = 47.44%, SD = 42.77%; Mcontrol = 50%, SD = 42.81%; F(2, 127) = 3.44, p = .035,
Discussion
Results from Study 3 provide further evidence that enclosing PGIs does not result in higher donations. Consistent with Study 2, nonmonetary PGIs did not increase donations relative to no PGI, and enclosing monetary PGIs resulted in lower average donations than enclosing nonmonetary PGIs and no PGI. However, the latter effect is significantly only for donation percentage and merely marginal for donation amount, which may limit the inferences for this study. In addition, we did not find an effect of PGI on response rate, perhaps because everyone was forced to read the letter. We discuss this possibility subsequently.
Studies 4–6 examine the effect of PGIs on donation behavior as well as the underlying mechanism of relationship norms by utilizing both mediation and moderation techniques, while ruling out alternative explanations. Unlike other the studies in this article, Studies 4–6 use hypothetical scenarios whereby participants imagine receiving a PGI. A substantial amount of research, especially in economics, has reported substantively little difference between hypothetical and real scenarios (Dai, Galeotti, and Villeval 2018; Falk and Heckman 2009). For example, Ben-Ner, Kramer, and Levy (2008) conducted two experiments, one in which participants were given actual money to spend, and another in which they were asked to imagine they were given money. They found that the “the average subject behaves essentially the same” in these two conditions (p. 1783). Such findings suggest it is valuable to take a multimethod approach to decision-making tasks, particularly when limited resources would otherwise constrain the number of studies that can be run and hypotheses tested.
Study 4
In Study 4, we examine the mediating role of relationship norms on donation behavior. As the pilot test demonstrates, charities are inherently perceived as communal organizations. On the one hand, we propose that monetary PGIs diminish communal norms and increase exchange norms, resulting in lower average donations. On the other hand, nonmonetary PGIs are expected to produce no net change in communality. We also test several alternative explanations in this study. First, previous research suggests that inferences of manipulative intent decrease message persuasiveness (Campbell 1995). It is possible that people perceive greater manipulative intent when charities include monetary (vs. nonmonetary) PGIs, which leads them to be less supportive of the charity. Second, PGI type may influence perceived charity efficiency, which has been shown to affect donation decisions (Winterich and Barone 2011). Specifically, people may believe that charities that send money with their donation appeals use their resources less efficiently than those that do not. Finally, it is possible that manipulative intent or charity efficiency influence perceived communality in serial, which results in lower donations.
Participants and Procedure
One hundred fifty-three students (96 women; Mage = 20 years) from a major Midwestern university participated in the study in exchange for course credit. Participants were randomly assigned to read a charity letter (Web Appendix E) enclosing a monetary PGI ($.25), a nonmonetary PGI (a low-value greeting card), or no PGI. After reading the charity letter, participants responded to the relationship norm items, followed by the donation request, and then the measures of manipulative intent and charity inefficiency.
Dependent Measures
Response rate was calculated the same way as in Study 3.
Average donation
Participants indicated their donations by dragging a slider anchored at $0 and $50, with a write-in option labeled “other” for anyone willing to donate more than $50.
Relationship norms
Though some research suggests that communal and exchange norms are orthogonal (Johnson and Grimm 2010), others conceptualize them as opposite ends of the same scale (Aggarwal 2004; Clark and Aragón 2013). In our research, communal and exchange norms are always highly negatively correlated (rs < −.46). We thus use Aggarwal’s (2004) net communality for our mediation analyses and report results with each scale separately in Web Appendix I for completeness.
We measured communal norms (e.g., “This organization is concerned about other people’s welfare”; α = .79; for all items, see Web Appendix J) and exchange norms (e.g., “Whenever this organization gives or offers something, it expects something in return”; α = .88) using four items each. We reverse-coded the exchange norms and combined them with communal norms (Pearson’s r = −.56) to form the net communality score (Cronbach’s alpha = .87). The higher the net communality score, the higher the communal relative to exchange norms. Manipulative intent (α = .87) and charity inefficiency (α = .75) (Web Appendix J) were measured on a seven-point Likert scale (1 = “completely disagree,” and 7 = “completely agree”).
Results and Discussion
Prior to running the analyses, we eliminated participants who failed the attention check (N = 24; Web Appendix D), leaving a final sample of 129 participants (84 women; Mage = 20.32 years).
Response rate
Chi-squares tests showed that the number of participants who indicated a nonzero donation did not differ between any of the three conditions (Nmonetary = 36, Nnonmonetary = 41, Ncontrol = 42; all ps > .43).
Average donation
Results from a one-way ANOVA revealed that PGI type had a marginal effect on average donations (Mmonetary = $11.83, SD = $10.89; Mnonmonetary = $17.93, SD = $17.41; Mcontrol = $19.82, SD = $18.24; F(2, 126) = 2.84, p = .062,
Relationship norms
We found that PGI type had a significant impact on the net communality score (Mmonetary = 4.63, Mnonmonetary = 5.12, Mcontrol = 5.44; F(2, 126) = 5.56, p = .005,
Manipulative intent
We found that PGI type had a marginal effect on manipulative intent (Mmonetary = 3.46, Mnonmonetary = 2.98, Mcontrol = 2.88; F(2, 126) = 2.70, p = .071,
Charity inefficiency
Results revealed a significant effect of PGI on perceptions of charity inefficiency (Mmonetary = 4.39, Mnonmonetary = 3.45, Mcontrol = 3.55; F(2, 126) = 7.98, p = .001,
Mediation Analysis
Relationship norms
We conducted mediation models with 10,000 bootstrap samples (Hayes 2017; PROCESS v3.1. Model 4). Because there were three different incentive conditions (monetary, nonmonetary, and no PGI), we dummy-coded the three groups by creating two new dummy variables, MNM (monetary PGI vs. nonmonetary PGI) and MC (monetary PGI vs. no PGI). Both variables used the monetary PGI condition as the reference group (Hayes 2017). Consistent with H4, results showed that the indirect effect of net communality was significant for both comparisons (monetary vs. no PGI: B = 4.54, 95% confidence interval [CI] = [1.56, 8.47]; monetary vs. nonmonetary: B = 2.75, 95% CI = [.22, 5.89]). We also performed single/dual mediation analyses for communal norms and exchange norms (see Web Appendix I).
Alternative mediators
Perceived manipulative intent alone did not mediate the effect of PGI type on donation (for details, see Web Appendix I). In the dual mediation model including net communality and manipulative intent, only the indirect effect of net communality was significant (monetary vs. no PGI: B = 3.41, 95% CI = [.85, 6.99]; monetary vs. nonmonetary: B = 2.07, 95% CI = [.09, 4.65]). Perceived charity inefficiency did produce a significant indirect effect on its own (monetary vs. no PGI: B = 3.41, 95% CI = [1.10, 6.49]; monetary vs. nonmonetary: B = 3.82, 95% CI = [1.45, 6.80]). However, the indirect effect of charity inefficiency became nonsignificant once net communality was included in the model, whereas net communality remained a significant mediator (monetary vs. no PGI: B = 3.75, 95% CI = [1.16, 7.66]; monetary vs. nonmonetary: B = 2.27, 95% CI = [.16, 5.25]). We also conducted multiple mediation and serial mediation analyses for PGI type → charity inefficiency/manipulative intent → net communality score → donation (see Web Appendix I).
Discussion
Study 4 examined potential underlying mechanisms for the effect of PGIs on donation behavior. Our results suggest that a monetary PGI, compared with a nonmonetary or no PGI, leads people to perceive the charity as lower on net communality, resulting in lower average donations. However, that the average donation amount difference is marginal between the monetary and nonmonetary PGI conditions, so H2 is only partially supported. We also examined manipulative intent and charity inefficiency as alternative explanations. Net communality remained a significant mediator after controlling for these constructs, while manipulative intent and charity inefficiency did not, suggesting that the simultaneous change in communal norms and exchange norms predicts donation behavior above and beyond the effects of manipulative intent and charity inefficiency. Further analyses on the relationship between these constructs reveal evidence of serial mediation, such that perceived higher charity inefficiency leads to lowered communality, resulting in lower donations.
Results from Studies 2–4 have provided converging evidence that enclosing low-value monetary PGIs in charity appeals decreases communal norms and increases exchange norms, leading to lower average donations compared with nonmonetary PGIs or no PGI. In Study 5, we provide another test of our theory by varying the value of the PGI.
Study 5
According to theory on relationship norms, when exchange norms are dominant, the degree of benefits or favors returned by the recipients is contingent on the level of benefits received, leading people to behave in a quid pro quo manner (Clark and Mills 1993). When communal norms are dominant, donations are need based rather than incentive based, so incentives should not play as pivotal a role in donation decisions (Miller et al. 2014). Because monetary PGIs evoke more exchange norms and less communal norms, they should lead people to make donations based more on how much they received than the perceived needs of the charity (Miller et al. 2014). In other words, high- (vs. low-) value monetary PGIs should result in higher donations. However, the value of a nonmonetary gift should have little impact on donations because, as explained previously, no change in net communality is expected. This is true regardless of the value of the gift. Moreover, considering that the amount of money to be reciprocated increases while need-based donations remain stable, it is possible that donations elicited by monetary PGIs will be greater than those elicited by nonmonetary PGIs at higher incentive values. Formally, we propose the following hypothesis:
In testing this hypothesis, we also address an alternative explanation for the effect—that reminders of money lead people to be less prosocial (e.g., Vows, Mead, and Goode 2006). Consistent with this literature, receiving monetary PGIs should lead people to be less charitable and thus donate less. However, this theory does not make different predictions for high- versus low-value monetary PGIs. If the effect of PGI type on donation behavior is solely driven by the psychological consequences of money rather than the simultaneous movement of communal and exchange norms, we should expect a high-value monetary PGI to lead to lower donations than a comparable value nonmonetary PGI. The relationship norms framework, in contrast, would predict the opposite finding.
Participants and Procedure
One-hundred thirty-nine participants (73 women; Mage = 37.07 years) were recruited from Amazon Mechanical Turk (MTurk). The study used a 2 (incentive type: monetary vs. nonmonetary) × 2 (PGI value: low [$.25] vs. high [$2.50]) between-subjects design. We selected the high-value incentive after an examination of over 100 charity appeals found a maximum monetary PGI amount of $2.50. Participants in the low-value monetary condition imagined receiving a quarter ($.25) while those in the high-value monetary condition imagined receiving two $1 coins and two quarters (Web Appendix E). Participants in the low-value nonmonetary condition imagined receiving one greeting card and those in the high-value nonmonetary condition imagined receiving eight greeting cards. A pretest showed that the two low-value monetary and nonmonetary PGIs and the two high-value monetary and nonmonetary PGIs did not differ on subjective value or perceived cost to the charity (see Web Appendix D).
In the main study, participants read a charity letter from a fictitious food pantry (for stimuli, see Web Appendix E). After reading the charity appeal, participants were told that they would automatically be entered into a lottery for $10 as an extra show of appreciation for participating in the study. Then they were asked how much they would be willing to donate from their winnings by dragging a slider anchored at $0 and $10.
Response rates and average donations were calculated the same way as previous studies.
Results and Discussion
Response rate
No interaction was found between PGI type and value on response rate (p = .78). No difference was found between low versus high-value PGI conditions or within high- and low-value PGI conditions (low value: Nmonetary = 32, Nnonmonetary = 31; high value: Nmonetary = 30, Nnonmonetary = 31; all ps > .20).
Average donation
Consistent with H5, results from a two-way ANOVA revealed a significant interaction between PGI type and PGI value (F(1, 135) = 10.30, p = .002,
Note that average donation for the high-value nonmonetary PGI appeal was directionally lower than the low-value nonmonetary PGI appeal (p = .14). One possible explanation is that, unlike money, greeting cards have diminishing marginal utility, leading to scope insensitivity (e.g., Kogut, Slovic, and Västfjäll 2015). If this is the case, a set of disparate incentives should lead to greater perceived gift value and may increase donations. We tested this hypothesis in a follow-up study with three conditions (N = 273): one greeting card, eight greeting cards, or eight different gifts of similar cost including a pen, a note pad, a binder clip, a card, and so on (see Web Appendix E, Study 5). Results revealed no donation difference among the three conditions (Mone card = $4.52, Meight cards = $3.95, Meight gifts = $4.69; all ps > . 14). Thus, the inability of high-value nonmonetary incentives to increase donations is less likely to be due to scope insensitivity than the failure to increase communal (and decrease exchange) norms.
Comparing between PGI types, $.25 led to significantly lower donations than one greeting card (p = .026). The effect is reversed for higher-value monetary and nonmonetary PGIs, such that $2.50 elicited significantly more donations than eight greeting cards (p = .024), although we suspect that this effect can be either attenuated or reversed depending on the value of the incentives. In practice, enclosing a high-value monetary PGI may not be desirable or feasible for charities, considering the high cost as well as the low average return.
Study 5 helps establish relationship norms as a primary contributor to donation decisions. In the next study, we consider a potential low-cost way to improve average donations. Charitable organizations are highly strategic about the choice of wording in their appeals. So far, we have used the common phrasing of the PGI as a gift (e.g., Smile Train uses the statement “We have enclosed a world map as our free gift to you.”) Yet charities may be interested in the comparative effectiveness of framing the gift in a different way. For example, Obis (an international eye care charity) encloses a nickel in its charity appeal and states, “This nickel can help restore a child’s vision.” Food for the Poor says in its charity appeals, “Please return these coins along with your gift.” A reasonable question is whether such statements mitigate or enhance the changes in exchange and communal norms from monetary PGIs. In the next study, we explore the effect of framing on relationship norms, response rate, average donations, and the alternative explanations of charity inefficiency and manipulative intent. Doing so also allows us to test the generalizability and external validity of the focal effect.
Study 6
Study 6 examines the effect of framing the monetary gift in various ways. We approached this study in an exploratory manner with the goal of testing two competing hypotheses. It is possible that phrasing a monetary PGI in a more communal way will prevent exchange norms from increasing and communal norms from decreasing. For instance, the phrasing “this nickel can help restore a child’s vision” may lead people to interpret the enclosure of the monetary PGI as a demonstration to illustrate that even a little money can help the cause, which is compatible with the communal nature of the charity. However, this is an empirical question, as previous research suggests the effect of money on marketplace norms is strong and consistent (Heyman and Ariely 2004; Kouchaki et al. 2013; Vows, Mead, and Goode 2008). Phrasing the gift in a communal way may do little to change donor behavior if a monetary mindset has already been activated. Indeed, McGraw, Schwartz, and Tetlock (2012) demonstrated that it is possible to improve consumer sentiment by reframing commercial marketing strategies as communal, unless consumers were already in a persuasion frame of mind. Thus, we suspect that, regardless of the phrasing for the monetary PGI, its mere inclusion will consistently lead to lower donations compared with a nonmonetary PGI and no PGI.
Participants and Procedure
Five hundred seven MTurk workers (282 women; Mage = 37.11 years) were randomly assigned to one of five conditions in which they saw a charity appeal from the fictitious food charity in Study 5. Three conditions manipulated phrasing of the monetary PGI: (1) “This $.25 can help provide a meal” (monetary-help), (2) “Please return this $.25 along with your donation” (monetary-return), and (3) “This $.25 is a gift to you” (monetary-gift). We chose these three monetary PGI phrasings on the basis of those used in existing appeals. The other two conditions were the same charity appeals enclosing either a low-value nonmonetary PGI (“This greeting card is a gift to you”) or no PGI. All aspects of the charity appeals, besides the PGI, were identical across conditions (Web Appendix E). As in Study 5, participants were told they would be automatically entered into a lottery for $10 and asked how much they would be willing to donate if they won ($0 to $10). Afterward, they answered the same measures from Study 3 in the following order: perceived communality (communal [α = .85] and exchange [α = .83] norms were negatively correlated; Pearson’s r = −.46), manipulative intent (α = .86), and charity inefficiency (α = .80). Note that rather than measuring net communality before the dependent variable (as in Study 4), we measured it after the donation dependent variable to rule out potential order effects.
Response rates and average donations were calculated the same way as previous studies.
Results and Discussion
Response rate
A chi-square test showed an unexpected higher response rate for the nonmonetary PGI condition (Nnonmonetary = 99) than for all three monetary phrasing conditions (Nmonetary help = 87, Nmonetary return = 86, Nmonetary gift = 90; all ps < .04). The response rate for the no-PGI condition (Ncontrol = 93) was marginally higher than that of the monetary-return condition (p = .052). No other response rate difference was found (all ps > .17).
Average donation
An ANOVA revealed a significant main effect of PGI type on average donations (Mmonetary help = $3.86, SD = $3.04; Mmonetary return = $4.04, SD = $3.37; Mmonetary gift = $4.28, SD = $3.03; Mnonmonetary = $5.22, SD = $3.39; Mcontrol = $5.13, SD = $3.53; F(4, 502) = 3.72, p = .005,
Relationship norms
An ANOVA showed that PGI type had a significant impact on the net communality score (Mmonetary help = 5.07, SD = 1.20; Mmonetary return = 4.82, SD = 1.29; Mmonetary gift = 4.71, SD = 1.15; Mnonmonetary = 5.20, SD = 1.16; Mcontrol = 5.29, SD = 1.11; F(4, 502) = 4.45, p = .002,
Manipulative intent
The effect of PGI type on manipulative intent was significant (Mmonetary help = 2.83, Mmonetary return = 3.75, Mmonetary gift = 3.37, Mnonmonetary = 2.66, Mcontrol = 2.46; F(4, 502) = 12.71, p < .001,
Charity inefficiency
We found that PGI type had a significant impact on perceived charity inefficiency (Mmonetary help = 4.03, Mmonetary return = 4.44, Mmonetary gift = 4.23, Mnonmonetary = 3.64, Mcontrol = 3.50; F(4, 502) = 6.59, p < .001,
Mediation Analysis
Relationship norms
We conducted mediation analyses using the same mediation procedure in Study 4. The indirect effect of net communality was significant collapsing across the three monetary phrasing conditions: (monetary vs. no PGI: B = .34, 95% CI = [.13, .59]; monetary vs. nonmonetary PGI: B = .27, 95% CI = [.05, .52]; (for mediation analyses for each monetary phrasing, see Web Appendix I).
Alternative mediators
Combining the three monetary phrasings, both manipulative intent and charity inefficiency mediated the effect of PGI type on donations (see Web Appendix I). However, their indirect effects were insignificant when net communality was included in the model, while net communality remained a significant mediator (net communality and manipulative intent: monetary vs. no PGI: B = .28, 95% CI = [.09, .52]; monetary vs. nonmonetary PGI: B = .22, 95% CI = [.04, .45]; net communality and charity inefficiency: monetary vs. no PGI: B = .29, 95% CI = [.10, .52]; monetary vs. nonmonetary: B = .23, 95% CI = [.04, .46]). For serial mediation results, see Web Appendix I.
In summary, this study demonstrates that describing monetary PGIs in various ways does not make a difference in terms of donations. We did find an unexpected effect in the monetary-help condition, in that it leads people to perceive the charity as more communal than those in the monetary-gift condition. A follow-up study (N = 166 MTurkers) revealed no differences among the two phrasings on perceived impact of their donations, perceived efficacy, perceived self-concept (moral/ethical), positive/negative affect, and empathy (all ps > .18). We suspect that the phrasing in the monetary-help condition maintains the perceived communality of the charity by focusing on the victims and emphasizing the compassionate nature of the organization. However, the negative effect of charity inefficiency produced by the coin was strong enough to overcome this perception and lower overall donations compared with the nonmonetary PGI and no-PGI conditions, as demonstrated by the serial mediation results (PGI type → charity inefficiency → net communality → donations; see Web Appendix I). These findings suggest that enclosing monetary PGIs leads to lower average donations, regardless of how monetary PGIs are phrased.
Study 7
In Study 7, we again partnered with the mental health clinic in Study 2 to launch a large-scale direct mail campaign. Unlike Study 2, this study targets existing donors in a year-end campaign. The goals of this study are threefold. First, we examine whether the PGI effects generalize to a warm mailing list (i.e., people who already have a relationship with the charity). We expect that such a sample will result in a higher overall response rate than Study 2 because people are more likely to open mail from a known sender. However, effects on relationship norms, and thus average donations, should be the same. Second, we test another alternative explanation for the effect of PGIs on donations—the anchoring and adjustment heuristic (Tversky and Kahneman 1974). This hypothesis suggests that people who receive a monetary PGI use the coin(s) as a reference point or anchor and thus donate less than when a low-value numerical anchor is not present. In this study, we include the same anchor in all the charity appeals. If the results are due to anchoring rather than relationship norms, this should erase the effect of PGI type on donations. Third, we include a different nonmonetary PGI to make sure the effect is not an artifact of the gift we chose. Although greeting cards are an externally valid PGI currently used by charities, they are also thin and made out of paper, making them vulnerable to being overlooked or accidentally discarded. Furthermore, a greeting card may be seen as a communal gift. Therefore, in this study, we use a magnet as the nonmonetary PGI, which is more of a neutral incentive and less likely to go unnoticed in an envelope. A pretest confirmed that the magnet and coin did not elicit differences in perceived costs or benefits (Web Appendix D).
Materials
Mailing list
We randomly selected 2,643 donors from the charity’s existing donor list and assigned them to one of the three PGI conditions: monetary ($.25), nonmonetary (a magnet), or no PGI.
Business reply return envelopes
Consistent with Study 2, a standard business reply return envelope was enclosed within each letter. We printed a unique code on each return envelope to differentiate the three PGI conditions.
Window envelopes, donation request letters, and incentives
We held the content of all three versions of the letter constant. To examine the anchoring effect, we included a sentence at the top left corner above each person’s name: “Even $.25 can help prevent suicide” (see Web Appendix E, Study 7). This sentence replaced the one used in the previous experiments to introduce the PGI (e.g., “Please accept the enclosed $.25 [magnet] as a gift to you”). To ensure that participants saw the incentives, both the quarter and the magnet were displayed through the clear windows of the envelopes. As in Study 2, the charity created three distinct online donation links for each incentive condition so that donors could choose to give online.
Dependent Measures
Response rate
Response rate was defined as the number of participants who made a donation in each condition.
Average donation
The average donation in each condition was computed by dividing the total amount of money donated by the number of donors who contributed.
ROI
As in Study 2, we calculated the ROI for each condition by considering total donations in relation to associated costs, divided by the number of recipients (for cost details, see Web Appendix F).
Results and Discussion
Campaign overview
The response rate a little over eight weeks after mailing (November 4, 2019, through December 31, 2019) was 2.50% (N = 66). We received a total of $7,541.25 ($1,756.25 from the monetary PGI condition (range: $.25 to $500), $2,415 from the nonmonetary PGI condition (range: $10 to $550), and $3,370 from the no-PGI condition (range: $20 to $500), with an average donation amount of $114.26.
Response rate
We found no difference in response rate among the three conditions (33.33%, Nmonetary = 22; 37.88%, Nnonmonetary = 25; 28.79%, Ncontrol = 19; all ps > .35). Two people in the monetary PGI condition merely returned the quarter. We report the results excluding these two participants in Web Appendix G.
Average donation
A Shapiro–Wilk normality test revealed that the donation data were not normally distributed (p < .001), so we log-transformed the data. A one-way ANOVA showed a significant effect of PGI type on donation amount (in log-transformed value: Mmonetary = 1.49, SD = .80; Mnonmonetary = 1.78, SD = .42; Mcontrol = 2.03, SD = .48; F(2, 63) = 4.34, p = .017,
ROI
The amount raised by the entire campaign was $4,597.04, with $663.31 in the monetary PGI condition, $1,349.66 in the nonmonetary PGI condition, and $2,584.07 in the no-PGI condition. A Shapiro–Wilk normality test showed that the ROI data were not normally distributed (p < .001), so we log-transformed the data. As in Study 2, we added a constant number a (a = 1.241) to offset the negative values. Results showed that PGI had a significant impact on ROI (in log-transformed values: Mmonetary = −2.89, SD = .71; Mnonmonetary = −1.42, SD = .55; Mcontrol = −.40, SD = .37; F(2, 2640) = 4,376.73, p < .001,
Discussion
The results of this study suggest that the effect of PGI type on donations is unlikely to be driven by the anchoring and adjustment heuristic. The effect held when an explicit anchor was present in the appeal. Consistent with the results in previous studies, enclosing a monetary PGI led to lower average donations, and enclosing a nonmonetary PGI was no more effective than not including a PGI. In addition, as in Study 2, including either a monetary or nonmonetary PGI led to a lower ROI than not including a PGI. Unlike Study 2, however, the response rate did not differ by PGI condition. We postulate that this is because recipients in this study were already familiar with the charity, having donated to them before. We discuss this possibility in the general discussion. A weakness of this study is that we did not include a manipulation check of the anchoring manipulation due to the difficulty of surveying donors in the field. However, an additional test of the anchoring effect using a different nonmonetary PGI also failed to find support for this alternative (Web Appendix L).
General Discussion
Charities utilize different marketing strategies in hopes of fundraising and enlarging their donor base (Newman and Shen 2012). Enclosing low-value PGIs is one strategy that has been extensively practiced by charities (Paradysz + PM Digital Research 2016). Through a combination of field and lab studies, the present research examines the effectiveness of utilizing this strategy on opening rate, response rate, donation behavior, and relationship norms.
In seven studies, we examine how and why donors respond to different types of PGIs. Study 1 examines the effect of PGI type on opening rate and shows that enclosing a monetary PGI (vs. nonmonetary or no PGI) leads to significantly more people opening and reading the charity letter. Study 2, a donor acquisition campaign, finds that (1) a monetary PGI leads to a higher response rate than both a nonmonetary PGI and no PGI, (2) a monetary PGI leads to lower average donations compared with a comparable value nonmonetary PGI and no PGI, and (3) both monetary and nonmonetary PGIs lead charities to suffer higher net financial loss than no PGI in donor acquisition efforts. Results from Study 3 provide further evidence that enclosing a monetary PGI leads people to donate less money than a nonmonetary PGI and no PGI, and enclosing a nonmonetary PGI is no more effective than when no PGI is included. Results from Study 4 show that reduced donations in the monetary incentive condition are due to decreased net communality levels. Study 4 also tests manipulative intent and charity inefficiency as alternative explanations and demonstrate a significant effect of net communality over and above their influence. Consistent with the relationship norms framework, Study 5 shows that the effect of PGI type is moderated by incentive value. A high-value monetary PGI leads to more donations than a low-value monetary PGI, whereas there is no difference for high- versus low-value nonmonetary PGIs. Study 6 shows that phrasing a monetary PGI in various ways consistently leads to lower donations than a nonmonetary PGI or no PGI. Finally, Study 7 shows that a donation campaign for recurring donors utilizing a different nonmonetary PGI supports the conclusions that a monetary PGI decreases average donations compared with a nonmonetary PGI or no PGI. Study 7 also tested and ruled out the anchoring and adjustment heuristic by including an explicit anchor in all the charity appeals.
Theoretical Implications
The current research contributes to the literature on relationship norms in two ways. First, research on relationship norms has focused on for-profit entities (Aggarwal 2004; Wan, Hui, and Wyer 2011). In contrast, the current research examines how these norms affect consumer perception and behavior in the context of nonprofit organizations. For-profits and nonprofits are governed by different relationship norms, such that for-profit companies operate on exchange norms, whereas the default relationship norms for nonprofit companies are communal. This fundamental difference affects how people respond to these organizations at baseline, and indicates that using incentives as a marketing strategy has disparate effects on consumer behavior for these two categories of organizations. Second, instead of manipulating the salience of the communal and exchange norms through hypothetical scenarios, as is common in previous research, we examine how strategies that nonprofits currently use and have full control over organically influence the salience of these norms. Our results suggest that PGIs affect communal and exchange norms simultaneously and that these norms have a dual impact on donations.
We add to recent work on how people make inferences about an organization based on superficial elements in direct mail campaigns (Townsend 2017). We show that PGIs serve as cues about the organization, influencing communal and exchange norms. Lower net communality from monetary PGIs leads people to behave in a quid pro quo manner and be less likely to donate based on perceived need. This is true even though the net communality score remained significantly above the midpoint of the scale (Study 4: Mmonetary = 4.63, p < .001; Study 6: Mmonetary = 4.86, p < .001). Finally, we contribute to the PGI literature (e.g., Marwell and Schmitt 1967) by moving beyond general conclusions about all incentives to a nuanced account of how and why different types of PGIs affect donation behaviors.
Managerial Implications
This research addresses the question of whether including PGIs is a worthwhile strategy for charities. The ultimate answer is that it depends on the goal of the donation campaign. Next, we describe how PGIs affect various campaign related outcomes.
Opening rate
Charities often need to find ways to enter potential donors’ consideration set (Smith and Schwarz 2012). If the goal of the campaign is to raise awareness and to help the charity gain exposure, enclosing a monetary PGI appears to be an effective strategy. Our research suggests that enclosing a monetary PGI persuades recipients not only to open the letter, but to read it. Notably, the charity we used to test this hypothesis was fictional and thus unfamiliar to all participants. It is reasonable to assume that the decision to open a letter will be multiply determined when donors are already aware of the charity. A PGI (or its absence) may not be as powerful of a contributing factor for existing donors. However, given that there are over a million charities in the United States (Internal Revenue Service 2019), many of which are unknown to a significant number of Americans, this effect is an important consideration for many.
Response rate
Enclosing a monetary PGI may help organizations achieve the goal of enlarging the donor pool for future campaigns. Results from the donor acquisition campaign (Study 2) showed that a monetary PGI led to a significantly higher response rate than both a nonmonetary PGI and no PGI, whereas there were no differences between the nonmonetary and no-PGI conditions. However, 15 of 27 people in the monetary PGI condition returned/donated the enclosed $.25, and charities need to decide whether to consider them potential future donors.
The results for response rate were not consistent across all studies. We found an unexpectedly high response rate for the nonmonetary PGI condition in Study 6 and no response rate difference in any other study, including the annual campaign (Study 7). These results should be interpreted in light of the results in Study 1, in which we found that the opening rate was higher for letters containing a monetary PGI for an unfamiliar charity. This could have led to the higher response rate in the first field study, which utilized a cold mailing list. Donors in Study 7 were already familiar with the charity, having donated to them before, so opening rate for this segment should be higher overall, and less likely to be dependent on the presence of a gift. The same line of reasoning applies to the lab experiments. All participants were forced to open and read the letter, negating any effect of opening rate on donation likelihood. In other words, opening the letter is a necessary condition for responding to it. However, as discussed previously, the number of people who open the letter is partially determined by whether the donor is familiar with the organization. Thus, monetary PGIs may be especially beneficial for charities that are lesser known or just starting to build their donor list.
Average donation
If the primary goal of the campaign is to maximize the contribution of each donor, results consistently show that inclusion of monetary PGIs is a bad idea, regardless of how the incentive is couched. Nonmonetary PGIs perform no better than no incentives. In fact, including them, even when we increased their value—from 4 to 8 to 12 cards (M1 card = $5.73, M4 cards = $4.84, M8 cards = $4.84, M12 cards = $4.73; all ps > .20) or offered a greater variety of gifts (Study 5) led to the same average donations as when no PGI was included. Note that both Studies 3 and 4 (student lab studies) yielded marginally significant donation differences between the monetary and nonmonetary PGI conditions, which could be because participants were not donating their own money. The effects are more pronounced in other studies.
Total donations
Charities may have the goal of raising the most money possible. Across all seven studies, the no-PGI appeals resulted in the most money raised ($5,454.88), followed by the nonmonetary PGI ($4,178.57), and the monetary PGI ($3,392.67). The effectiveness of PGIs on total donations may depend on who the recipients are. In Study 2, the most donations came from the monetary PGI appeal, whereas in Study 7, the no-PGI appeal raised the most money. This suggests a monetary PGI is more effective for nondonors (perhaps due to a higher opening rate), while no PGI is needed to maximize overall donations from existing donors.
ROI
If the goal of the campaign is to minimize losses or yield a higher ROI, it would be more effective to not include a PGI. Both monetary and nonmonetary PGIs resulted in a significantly lower ROI than no PGI in both field experiments. Specifically, in the donor acquisition campaign, where ROIs for all conditions were negative, enclosing a monetary PGI led to an additional $.27 net loss per mailing, which is more than twice the net loss than when no PGI was included. Enclosing a nonmonetary PGI resulted in an additional $.24 net loss per mailing, which is also close to twice the net loss compared with a mailing with no PGI. Extrapolated to the 9,000 individuals solicited in this campaign, inclusion of a monetary PGI (vs. no PGI) resulted in an additional loss of $810 and adding a nonmonetary PGI resulted in an additional loss of $720. We found similar results in the donation campaign for recurring donors. Enclosing a monetary PGI resulted in a $2.19 lower ROI per person compared with no PGI ($1,929.39 in total), while a nonmonetary PGI (vs. no PGI) resulted in a $1.41 lower ROI per person ($1,242.21 in total).
Relationship norms
Results suggest that enclosing a monetary PGI leads people to perceive the charity as less communal and more exchange oriented, which directly harms donations. Enclosing a nonmonetary PGI does not affect net communality levels. We suggest that this is because an immediate ask for help accompanying the gift offsets any increase in communality from the gift itself. Indeed, results from an unreported MTurk study (N = 152) showed that net communality was significantly lower when participants were asked for a donation immediately after receiving a PGI than when they were given a gift without a donation request (Mimmediate donation = 4.17, Mgift only = 4.83; p < .001). Further analyses showed that participants in the immediate donation condition scored significantly lower on communal norms and significantly higher on exchange norms compared with those in the gift-only condition (both ps < .01). This finding suggests that the norm of reciprocity may be less applicable when the default relationship norm is communal. Thus, if charities want to enhance communal norms, we suggest sending a gift with a delayed request for help.
Limitations and Future Research
One limitation of this research is the small sample sizes in some of the studies. This is partly due to the nature of direct mail campaigns (e.g., a low response rate is common in similar campaigns). However, the inclusion of multiple studies in the paper and the web appendix showing consistent effects, some with relatively large sample sizes, should help alleviate this concern.
Another limitation of this research is that it does not explore the long-term consequences of PGIs. For example, charities may enjoy additional benefits from nonmonetary gifts if people find items with the names and logos of the organization practical and use them regularly. Cognitive dissonance theory (Festinger 1957) predicts that, over time, this action will strengthen recipients’ feelings of connection to the charity. In addition, appraisals of their own behavior (“If I am associating myself with this charity, I must think it is a worthwhile cause”) may lead them to make donations in the future. Public display of the charity name and logo could also serve as advertising for people who were previously unaware of the organization. A longitudinal study could be used to test these hypotheses.
Future research might also find it fruitful to explore the effect of PGIs on other outcomes. In Web Appendix M, we examine the effect of PGIs on donations of time. Unlike money, PGI type had no impact on time donations. Research insights from a 2014 survey may shed light on why this is. Among donors, “financial support usually comes first” and “few volunteer for sectors that they are not also supporting financially” (Fidelity Charitable 2014). In addition, time is personally more meaningful than money, and motivations and beliefs can lead to very different effects on time versus money donations (Kulow and Kramer 2016). Thus, people who are willing to donate money may not be willing to donate time. We hope this research will spur further investigation on PGIs and prosocial behavior of all kinds.
Supplemental Material
Supplemental Material, UPDATED_Web_Appendix_JM.18.0163 - Coins Are Cold and Cards Are Caring: The Effect of Pregiving Incentives on Charity Perceptions, Relationship Norms, and Donation Behavior
Supplemental Material, UPDATED_Web_Appendix_JM.18.0163 for Coins Are Cold and Cards Are Caring: The Effect of Pregiving Incentives on Charity Perceptions, Relationship Norms, and Donation Behavior by Bingqing (Miranda) Yin, Yexin Jessica Li and Surendra Singh in Journal of Marketing
Footnotes
Acknowledgments
The authors would like to thank the JM review team for their invaluable input, and Prakash Shenoy and Min Zhao for their helpful comments.
Associate Editor
Vanitha Swaminathan
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the University of Kansas General Research Fund allocation # 2301009 granted to the first and the second author, and the DSRF Dissertation Fund granted to the first author.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
