How the compatibility between negative mood and single (vs multiple) representative victim(s) increases charitable giving

The devastating Turkey-Syria earthquake claimed over 50,000 lives, and Morocco’s floods displaced thousands in 2023. Disasters like these highlight the crucial role of charities in mobilizing aid for victims. To maximize support, charities must find compelling ways to inspire generosity and action from individuals. One notable finding from past research on charitable giving is that people tend to be more generous toward a single, identified victim than toward an unidentified victim or multiple unidentified victims (Cryder et al., 2013; Erlandsson et al., 2015; Jenni and Loewenstein, 1997; Kogut and Ritov, 2005a; Lee and Feeley, 2016, 2018; Moche and Västfjäll, 2021; Small, 2015; Small and Loewenstein, 2003; Small et al., 2007). For example, when a charity appeal reveals a victim’s name, age and/or photograph, people are more likely to donate their money to help that one specific victim than when a victim or victims are not identified (Kogut and Ritov, 2005a). Past research shows that this identifiable victim effect is useful only for a single victim, and not for multiple victims (Kogut and Ritov, 2005a). At the same time, raising donations for a single victim (vs multiple victims), without any identifying information, does not increase donations (e.g. see Kogut and Ritov, 2005a; Kogut and Ritov, 2007, and Small et al., 2007). In other words, identifiability and singularity of a victim are both necessary to effectively solicit donations.

While focusing on a single identified victim can be highly effective for driving donations, it has practical limitations. When a victim is identified, donors are typically asked to contribute specifically to help that individual, and the intended beneficiary is clearly defined in most cases. In contexts such as natural disasters or widespread poverty, it can be costly or even infeasible to identify each affected person. Moreover, it is often unnecessary to identify every individual beneficiary.

A charity often needs to address broader needs, such as improving infrastructure or providing education. For this reason, it must develop persuasive appeals that work even without featuring a specific identified victim. In such cases, it often uses the image of a representative victim, someone who is one of the many potential beneficiaries impacted by the cause. The message does not claim that the person pictured is the direct recipient of the donor’s support, but instead signals that this person represents those who will benefit from the charity’s work. Under these circumstances, identifying the individual is not necessary, as donors understand their contributions will support a wider group of affected people. Accordingly, the present research addresses this gap by examining the impact of featuring a representative victim while soliciting donations for a general funding pool.

A representative victim can be depicted as either a single individual or a group of people who symbolize those affected by the cause. Specifically, a charity appeal might depict children suffering from malnutrition using either an image of an individual child or a group of children. When an image depicting a single victim is presented without indicating that the donation will go directly to that individual, it functions as a symbolic representation of a broader group of victims. Sometimes, the organization discloses the identity of the person in the picture and details their circumstances to draw attention to that individual and their community’s plight. At other times, it includes only an image of a person, whom viewers often assume to be potential beneficiaries of the charity. Whether or not a charity appeal shares detailed information about the victim, like their name and age, its main purpose is to encourage support for the wider community or group that the victim represents, rather than focusing solely on the individual shown. Thus, the purpose of our research is to identify a method to enhance the effectiveness of charity appeals in encouraging support for a group of victims, beyond any specific individual depicted, by leveraging the well-established identified single victim effect.

There is an important distinction between the willingness to help the individual depicted in an image and the willingness to assist the larger group that individual represents. One reason a single, identified victim is often more effective in driving donations is the reference group effect. According to Jenni and Loewenstein (1997), people are more likely to give when their contribution benefits a larger proportion of those in need. For example, helping 1 out of 10 feels more impactful than helping 1 out of 100. Supporting a single identified victim (1 out of 1) seems to fulfill the entire need, reinforcing the identified victim effect through the reference group effect. However, when the featured victim merely symbolizes a broader community, donations extend beyond the individual, diminishing the sense of fully meeting the need and reducing the likelihood of the reference group effect. As a result, presenting a single representative victim may yield less impact than highlighting an individually identified victim, although direct comparisons are challenging due to differences in the intended beneficiaries: either a specific individual or a broader group represented by the victim. Given this potential absence of a reference group effect, it is not anticipated that featuring a single representative victim will be notably more effective than presenting multiple representative victims in terms of increasing donations.

In the present study, we propose that the effectiveness of a single representative victim, as opposed to multiple representative victims, is moderated by the donor’s mood. Specifically, negative moods such as sadness are known to increase sympathy and promote the deep, systematic processing of emotional stimuli (Bodenhausen et al., 1994; Schwarz, 1990; Small and Verrochi, 2009). When individuals feel sad, they are more likely to engage with emotionally resonant content and focus on the suffering of others. A single representative victim offers a clear, emotionally focused target for sympathy, allowing donors to form a personal connection and experience a sense of emotional resonance. In contrast, multiple representative victims may diffuse attention and emotional engagement, leading to compassion fatigue or psychic numbing, a phenomenon where people feel less emotional impact and are less motivated to help as the number of victims increases (Small et al., 2007). Consequently, a single representative victim is more compatible with a negative mood because it aligns with that state’s cognitive tendencies: focused attention and heightened sympathy. We therefore expect that appeals pairing a single representative victim with a negative mood will be processed more fluently, thereby increasing sympathy and donation behavior more effectively than appeals featuring multiple representative victims. We will discuss the theoretical framework in greater detail in the literature review.

Unlike the single identified victim effect, using a single representative victim may offer greater strategic flexibility for charitable organizations. Most notably, detailed identification of the featured individual may not be required, as donations are directed toward the broader community or the general cause the charity supports rather than a specific person. This approach facilitates a more equitable allocation of resources, helping organizations avoid “winner-take-all” scenarios where a particularly compelling victim receives disproportionately high support at the expense of others with equal need. Beyond these practical advantages, this research advances the literature on charitable giving, specifically the stream of research focused on optimizing support for large groups and organizations. While past research comparing single and multiple victims has largely focused on the role of identifiability (i.e. comparison between identified vs unidentified single victim or identified vs unidentified multiple victims) (e.g. Erlandsson et al., 2015; Kogut and Ritov, 2005a; Lee and Feeley, 2016; Sabato and Kogut, 2021; Small and Loewenstein, 2003) (see Table 1 for a summary of past studies), we demonstrate the distinct utility of the representative victim. We show that this format can more effectively motivate giving when paired with a donor’s mood state that aligns with the appeal’s structure.

In the following sections, we review the prior research that informs our predictions. We first examine the differential responses to individual versus group victims, followed by an exploration of the relationship between mood, cognitive engagement and sympathy. Drawing on these theoretical foundations, we hypothesize that a single representative victim is more effective than multiple representative victims when donors are in a negative mood. We then present five empirical studies and one meta-analysis that provide support for this prediction.

Research consistently shows that individuals react more emotionally and generously toward identified victims than toward unidentified ones (e.g. Kogut and Ritov, 2005a; Small and Loewenstein, 2003). This disparity is often attributed to the distinct information processing approaches used for individual versus group cases (Hamilton and Sherman, 1996; Sherman et al., 1999). Specifically, Kogut and Ritov (2005a) argue that the identified victim effect is driven by the victim being a single person. Because an individual is perceived as a more coherent and stable entity than a group, people tend to engage in more extensive, “on-line” processing to form a consistent impression (Hamilton and Sherman, 1996). Consequently, victim information is processed more deeply for a single individual, leading to higher donation compared to multiple victims.

Beyond these cognitive processing differences, past research highlights a difference in emotional reactions to individuals versus groups. Cameron and Payne (2011) explain the decrease in group-targeted empathy through the “collapse of compassion.” They suggest that when faced with multiple victims, donors engage in motivated emotion regulation, proactively suppressing their emotional engagement to avoid the anticipated financial and emotional distress of helping a large group. These findings have significant implications for the representative victim effect. While the singularity of a single representative victim may initially evoke engagement, its effectiveness may diminish if donors perceive the individual as a surrogate for a larger group, potentially triggering the same regulatory mechanisms that suppress compassion. Therefore, the singularity of a single representative victim might not be sufficient, on its own, to significantly outperform appeals featuring multiple representative victims.

For this reason, the reference group effect is unlikely to manifest in this context. The single identified victim effect is partially driven by the perception that a contribution has a greater impact when it helps a larger proportion of the group in need (e.g. 1 out of 1) (Cryder et al., 2013). However, in a representative victim appeal, donors understand that their contribution is intended for the broader community rather than the featured individual alone, which could diminish the reference group effect. In fact, Erlandsson et al. (2015) showed that perceived impact did not contribute to the identifiable victim effect in helping situations, particularly when donations were directed to a group represented by the identified victim. To address the limited perceived impact, we suggest that an additional condition may be necessary to strengthen donor engagement. Specifically, we suggest that negative moods may play a critical role in fostering the emotional connection and motivation necessary to drive charitable behavior. The following section explores how negative moods moderate the effectiveness of the single representative victim appeal.

Prior research indicates that negative moods significantly influence donors’ sensitivity to suffering and their propensity for prosocial action. For instance, Cao et al. (2017) found that individuals experiencing sadness perceive others’ pain as more distressing, leading to heightened emotional reactivity and sympathy. Similarly, Xiao et al. (2021) found that negative moods are associated with a greater propensity to invest time and resources in assisting others when compared with a neutral mood. In charitable giving, negative moods may decrease emotional regulation but increase attention to victim information, leading to greater sympathy and more helping behavior. This observation is consistent with emotional contagion theory, which posits that donors may mirror the emotions presented in charity appeals, particularly when sad facial expressions are depicted, resulting in increased donor sadness and sympathy (Small and Verrochi, 2009).

Negative moods also influence cognitive processing in systematic ways. According to affect-as-information theory, negative affect signals that a situation requires careful attention, prompting bottom-up, detail-oriented processing and deeper integration of incoming information (Schwarz and Clore, 1983). This approach is considered more effective when charity appeals are processed fluently,—that is, when information is presented clearly and coherently, enabling donors to focus directly on emotional content and thereby potentially increasing emotional resonance and sympathy (Bae, 2019; Dickert and Slovic, 2009). Appeals featuring a single representative victim are generally perceived as more coherent and less distracting than those featuring groups, which may result in higher fluency. Cognitive load theory also suggests that donors experience increased mental effort when processing information about multiple individuals (Paas and van Merriënboer, 2020). Presenting a single, representative victim, thus, makes charity appeals easier to comprehend and draws more attention to the individual. This increased attention is associated with greater support for victims (Bebko et al., 2014). Therefore, featuring one individual in a charity appeal can make it easier for people to engage with the message, which may lead to increased donations.

Furthermore, people are generally more likely to regulate their emotions when confronted with multiple victims. This early emotional regulation can act as a barrier, making it harder to initiate engagement even in negative moods. In contrast, a single representative victim lowers that initial hurdle. The singularity of the figure makes it easier to emotionally connect, and negative moods further facilitate this engagement. We expect that individuals in a negative mood are more likely to engage with the content of a charity appeal when a single representative victim is featured rather than multiple representative victims. This emotional engagement can lead people to process the information more fluently, deepening their sympathetic response and allowing them to extend those feelings to the broader group the victim represents. We, thus, expect this process effectively to offset the absence of the reference group effect, reducing the psychological resistance often associated with donating to a group and strengthening the persuasive impact of the single representative victim.

In contrast, when individuals are in neutral or positive moods, emotion regulation mechanisms may suppress sympathetic responses toward both single and multiple representative victims, because they understand that the donation benefits a group rather than a uniquely identified individual. Compared to negative moods, these moods are less effective in fostering emotional engagement. Consequently, individuals may find it more difficult to overcome emotional resistance toward helping a group represented by the victim. Therefore, we expect that featuring a single representative victim (vs multiple representative victims) will be more effective in increasing donation behavior primarily when individuals are in negative moods, whereas this advantage will be attenuated under neutral or positive moods.

Therefore, we formally hypothesize:

Because both negative moods and the use of a single representative victim serve as drivers of emotional engagement and extended information processing, we propose that these two factors are highly compatible. As mentioned earlier, this compatibility enhances processing fluency for the information presented in charity appeals such as the victim’s specific needs and the urgency of the situation. We further expect that this enhanced fluency will lead to increased sympathy for the victims. When individuals process information fluently and experience cognitive ease, they become more receptive to emotional resonance. This allows them to focus directly on the emotional content of the appeal (Schwarz, 2004), which is expected to increase sympathetic responses.

On the other hand, when an appeal is processed with disfluency, it can function as a cognitive “alarm” that prompts scrutiny and caution (Alter et al., 2007). This disfluency can activate persuasion knowledge, leading donors to interpret the sadness depicted in the appeal as a manipulative tactic. Such skepticism dampens sympathy and ultimately reduces charitable intentions (Choi and Park, 2021). Therefore, fluently processed appeals are more likely to foster sympathy by enabling a direct, undistracted emotional response, whereas disfluency creates cognitive barriers to sympathetic engagement.

There is also a well-established relationship between sympathy and donation behavior (e.g. Choi and Park, 2021; Erlandsson et al., 2015; Small and Verrochi, 2009; Small et al., 2007). Research consistently shows that sympathy plays a critical role in motivating giving, as these feelings heighten a donor’s emotional investment and willingness to assist. For instance, Small et al. (2007) found that donors are more likely to give when they feel a strong personal connection or sympathy toward the individuals represented in an appeal. Similarly, Dickert et al. (2011) showed that heightened sympathetic feelings drive an increased likelihood of giving. Accordingly, we propose the following mediation hypothesis:

We tested the proposed relationship between donors’ mood and the type of representative victim across five empirical studies and one meta-analysis. In the majority of these experiments, specifically Studies 1, 2, 4 and 5, we induced the donor’s mood independently from the charity appeal, referred to as incidental mood, to avoid potential confounding effects that might arise from manipulating both mood and victim type within a single appeal. In contrast, Study 3 elicited the donor’s mood directly through the charity message to enhance the practical utility of our findings for charitable organizations. The results of these studies provide robust support for our hypotheses. Study 1 reveals that participants experiencing incidental negative moods donate significantly more after viewing an appeal featuring a single representative victim than one featuring multiple victims; however, this difference disappears among participants in positive moods, supporting H1. Study 2 identifies and demonstrates the underlying psychological mechanism of processing fluency and sympathy driving these effects as proposed inH2. Study 3 replicates these findings while inducing the donor’s emotional state within the charity message itself. In Studies 4 and 5, we used refined methods for manipulating representative victim types to further minimize confounding effects. Consistent with our previous studies, we provided limited identification, such as a photo, in Study 4, while Study 5 included more specific details such as the names and ages of the victims. The difference appeared only in negative moods, not positive moods, regardless of whether representative victims were fully identified. Finally, Study 6 presents a meta-analysis of the combined data from Studies 4 and 5, confirming the overall robustness of the observed effects across different levels of victim identification.

In Study 1, we examine the relationship between a donor’s mood type and representative victim type on donation behavior to test H1. We expect that when participants experience negative moods, a charity appeal displaying a single representative victim will be more effective in increasing the donation amount compared to an appeal displaying multiple representative victims. However, when participants experience positive moods, we expect this difference in donation amount will not be observed between the two representative victim types.

Initially, 238 participants were recruited via CloudResearch (Litman et al., 2017) to participate in a preliminary marketing study about purchase intentions for Broadway shows, specifically Aladdin, which was not related to the main study. We then asked if they would like to participate in an additional study for a bonus payment of $0.50 (Choi et al., 2016; Sharma and Morwitz, 2016). One hundred and ninety-four participants agreed to participate and were randomly assigned to one of four conditions in a 2 (mood: negative, positive) × 2 (representative victim type: single, multiple) between-subjects design. Incidental mood was induced through a recall task where participants in the positive [negative] mood condition recalled and vividly described the happiest [unhappiest] day of their life (Labroo and Patrick, 2009; Schwarz and Clore, 1983). We instructed them to relive these feelings so a stranger reading the description could feel the same way. Afterwards, we conducted a mood manipulation check using a single item: “How does recalling your story make you feel?” (1 = very unhappy to 7 = very happy; Labroo and Patrick, 2009; Schwarz and Clore, 1983).

Next, all participants viewed a charity appeal ^[1] from World Vision. Participants in the single representative victim condition saw an appeal displaying a single child, while those in the multiple representative victim condition saw an appeal displaying multiple children. The text message used in the two ads were identical, except for adopting either singular or plural forms when addressing the victim(s). Participants were then asked if they were interested in donating some of the bonus payment ($0.50) to World Vision. They could donate any amount between 0 and 50 cents. Each participant’s pledged amount was donated to World Vision, and participants received the remaining amount as their bonus payment. After specifying the donation amount, participants responded to a few questions that measured charity familiarity ^[2] (α = 0.96) and charity credibility on a nine-point scale. Because past studies show that donors’ familiarity with a charity (e.g. Choi et al., 2020) and perceived credibility of a charity (e.g. Bekkers, 2003) affect donors’ donation decisions, we included these measures as covariates in our analyses.

Note that to ensure that the two charity appeals differed only in the victim type and not in the amount of positive or negative mood they elicited, we conducted a pretest on a different group of 79 participants from the same population (M_age = 35.3 years, 38 males and 41 females). As intended, the two victim-type appeals did not differ in the extent to which they evoked happiness [1 = not at all, 9 = very much; M_single = 2.64, M_group = 2.95, F (1, 77) < 1] or sadness [M_single = 6.33, M_group = 5.90, F (1, 77) < 1].

Incidental Mood Manipulation Check. As expected, participants who recalled their happiest moment felt happier than those who recalled their unhappiest moment [6.13 vs 1.75, F (1, 192) = 548.36, p  < 0.001].

Donated Amount. Four participants, outside of 2.5 standard deviations from the mean, were identified as outliers (Puzakova and Aggarwal, 2018; Meyvis and Van Osselaer, 2018) and removed. We used 190 qualified participants for our analyses (M_age = 35.9 years, 76 males and 114 females). We conducted a two-way ANCOVA with incidental mood and representative victim type as predictors and charity familiarity and credibility as covariates. We found a significant and positive impact of charity credibility on the amount donated [F (1, 184) = 11.52, p  = 0.001, η_p² = 0.059] ^[3]. There was no significant main effect found for the representative victim type [F (1, 184) = 2.87, p  = 0.092, η_p² = 0.015]. However, the interaction effect of victim type and mood on donation amount was found [F (1, 184) = 4.47, p  = 0.036, η_p² = 0.024]. In line with H1, planned contrasts showed that, under the negative mood condition, the charity appeal featuring a single representative victim resulted in a higher donation amount than the appeal showing multiple victims [M_single = 14.29¢ (SE = 2.24¢) vs M_multiple = 5.63¢ (SE = 2.43¢), F (1, 184) = 6.81, p  = 0.01, η_p² = 0.036, CI = 2.2112 ∼ 15.207] ^[4], but this difference was not observed in the positive mood condition [F (1, 184) < 1] (Figure 1).

Study 2 tests the mechanism underlying the effect (H2). Specifically, we examine whether the compatibility between negative moods and a single representative victim enhances donors’ fluency in processing the charitable message, and as a result, donors feel more sympathetic toward victims and become more willing to donate. This study also aims to rule out the possibility that self-efficacy accounts for our observed effect. Past research shows that positive moods could increase self-efficacy compared to negative moods (Kavanagh and Bower, 1985; Thelwell et al., 2007) though the relationship between moods and self-efficacy was not consistent (Cunningham, 1988; Kavanagh and Hausfeld, 1986). According to this finding, it may be possible that negative moods (vs positive moods) could reduce participants’ perceived efficacy, which may in turn demotivate them from giving when they view multiple victims (vs a single victim) in need, even if the single victim represents a group of victims. Hence, we measure perceived efficacy in the study to test this alternative account. Finally, this study includes a neutral mood condition to examine whether the increased compatibility with a single representative victim is unique under a negative mood state.

Three hundred and eight participants were recruited from CloudResearch (M_age = 37.9, 128 males and 180 females) and were randomly assigned to one of the six conditions in a 3 (mood: negative, positive, neutral) × 2 (representative victim type: single, multiple) between-subjects design.

The study consisted of two ostensibly unrelated studies. In the first study, participants watched one of the three short films used in past research that induced the intended mood types: Friends (positive), The Champ (negative) and a documentary film about Van Gogh (neutral) (Kandrack and Lundberg, 2013; Lenton et al., 2013; Rottenberg et al., 2002). A pretest conducted on a different sample of 175 participants from the same population (M_age = 36.4 years, 117 males and 58 females) confirmed that the incidental mood manipulation was successful (i.e. How do you feel when you watch the movie? 1 = very unhappy, 7 = very happy). Specifically, participants felt least happy when they watched the Champ, whereas they felt most happy when they watched Friends (M_champ = 4.02 vs M_Gogh = 5.92, p  < 0.001; M_champ = 4.02 vs M_Friends = 6.81, p  < 0.001; M_Gogh = 5.92 vs M_Friends = 6.81, p  = 0.02). After watching their assigned movie, participants responded to several filler questions related to the movie (e.g. What is the genre of the movie?). We included these filler questions to ensure that the video watching was perceived as a separate, unrelated study to the next study on charitable giving.

After completing the filler questions, all participants were introduced to another study on charity appeals. All participants viewed an appeal from World Help that solicited donations for children living in poverty. Participants in the single representative victim condition saw an appeal displaying a picture of a single victim, whereas those in the multiple representative victim condition saw another appeal displaying multiple victims. Apart from including a photo of the victim(s), the appeals did not provide any information that could identify the individual(s), such as their name or age. The text message used in the two appeals were identical, except for adopting either singular or plural forms when addressing the victim(s). Another pretest was conducted on a separate sample of 109 participants from the same population (48 males and 61 females, M_age = 36.9 years) to ensure that the two appeals did not differ in the level of mood they induced. As intended, the two appeals did not induce different levels of sadness (M_single = 5.45, M_multiple = 5.11, F (1, 107) < 1) or happiness [M_single = 2.55, M_multiple = 2.48, F (1, 107) < 1].

After viewing their assigned charity appeal, participants reported their donation intention to the charity (α = 0.93) ^[5]. Participants then responded to three items that measured processing fluency on nine-point scales (α = 0.93; Biswas and Pechmann, 2012; Lee and Feeley, 2016). They also reported their feelings of sympathy (α = 0.97; Choi and Park, 2021; Choi et al., 2016). Next, perceived efficacy was measured [α = 0.71, modified Ellen et al. (1991) to fit the donation context]. Finally, participants responded to the same questions that measured charity familiarity (α = 0.93) and charity credibility, both of which were included as covariates in our analyses.

Donation Intention. We conducted a two-way ANCOVA with incidental mood [negative mood (coded 0) vs neutral (coded 1) vs positive mood (coded 2)] and representative victim type [single (coded 1) vs multiple (coded 0)] as predictors and charity familiarity and charity credibility as covariates. We found a significant and positive impact of charity credibility on participants’ intention to donate [F (1, 300) = 109.17, p  < 0.001, η_p² = 0.267]. On the other hand, charity familiarity did not affect donation intention [F (1, 300) = 1.55, p  > 0.21]. While the main effect of incidental mood was not observed [F (1, 300) < 1], a significant main effect of representative victim type was found [M_single = 5.27, M_group = 4.74, F (1, 300) = 5.00, p  = 0.026, η_p² = 0.016].

The omnibus 2 × 3 interaction effect was not significant [F (1, 300) = 2.04, p  > 0.13]. Importantly, however, the aforementioned main effect was qualified by an interaction that tested whether the moderating effect of representative victim type was unique in the negative mood condition, compared to the other mood conditions. Specifically, we tested an interaction that compared the effect of representative victim types in the negative mood condition, to its effect in the other two mood conditions combined, and found a two-way interaction [F (1, 302) = 4.06, p  = 0.045, η_p² = 0.013]. This interaction indicates that the impact of representative victim type on participants’ donation intention depended on whether participants were feeling negative mood or not (i.e. neutral or positive). The partial interaction contrast that directly compared the incidental negative mood condition with only the incidental positive mood condition was not significant [F (1, 204) = 2.83, p  = 0.094, η_p² = 0.014], and another partial interaction contrast that directly compared the incidental negative mood condition with only the neutral mood condition was not significant [F (1, 192) = 2.69, p  = 0.102]. However, as expected, planned contrast revealed that when participants were in negative moods, they were more likely to donate after viewing a single representative victim than multiple representative victims [M_single = 5.61 (SE = 0.31), M_multiple = 4.40 (SE = 0.28), F (1, 300) = 8.47, p  = 0.004, η_p² = 0.027, CI = 0.392 ∼ 2.029]. In contrast, this difference was not observed when participants were under neutral or positive moods [Fs (1, 300) < 1] (Figure 2).

Processing Fluency. Another ANCOVA with the same predictors and covariates revealed significant positive effects of both charity credibility [F (1, 300) = 5.83, p  = 0.016, η_p² = 0.019] and charity familiarity [F (1, 300) = 20.84, p  < 0.001, η_p² = 0.065] on participants’ processing fluency. No significant main effect for mood [F (1, 300) < 1] or representative victim type was found [M_single = 8.20, M_multiple = 7.91, F (1, 300) = 3.57, p  = 0.06, η_p² = 0.012].

The omnibus 3 × 2 interaction effect was not significant [F (1, 300) = 1.40, p  > 0.24]. More important, a 2 × 2 (negative mood vs combined neutral/positive mood) interaction that directly tested whether the difference in the representative victim type was unique in a negative mood state, compared to the other mood states, was not statistically significant though the direction was consistent with our expectations [F (1, 302) = 2.82, p  = 0.094, η_p² = 0.009]. The partial interaction contrast that directly compared the incidental negative mood condition with only the incidental positive mood condition was not statistically significant [F (1, 204) = 2.89, p  = 0.091, η_p² = 0.014], and another partial interaction contrast that directly compared the incidental negative mood condition with only the neutral mood condition was not significant [F (1, 192) = 1.66, p  = 0.200]. Following the recommendation of Rosenthal and Rosnow (2008), we performed planned contrasts, as specific directional hypotheses regarding the treatment effects were developed prior to data collection. Planned contrasts revealed that viewing a single representative victim (vs multiple representative victims) in a charity appeal enhanced participants’ processing fluency significantly only when they were in negative moods [M_single = 8.38 (SE = 0.20), M_group = 7.73 (0.18), F(1, 300) = 5.95, p  = 0.015, η_p² = 0.02, CI = 0.126 ∼ 1.18], but not when they were feeling neutral or positive moods [Fs (1, 300) < 1] (Figure 3).

Sympathy. The same ANCOVA revealed a significant and positive impact of charity credibility on the amount of sympathy participants felt after viewing the appeal [F (1, 300) = 63.22, p  < 0.001, η_p² = 0.174]. However, charity familiarity did not affect sympathy [F (1, 300) = 1.71, p  > 0.19]. A significant main effect of representative victim type was observed [M_single = 7.27, M_group = 6.46, F (1, 300) = 14.77, p  < 0.001, η_p² = 0.047].

This main effect was qualified by a significant omnibus 3 × 2 interaction between mood type and the representative victim type [F (1, 300) = 3.37, p  = 0.036, η_p² = 0.022]. The 2 (incidental mood: negative mood vs other mood conditions combined) × 2 (representative victim type) interaction was also significant [F (1, 302) = 5.71, p  = 0.018, η_p² = 0.019]. Note that the partial interaction contrast that directly compared the incidental negative mood condition with only the incidental positive mood condition was significant [F (1, 204) = 6.85, p  = 0.010, η_p² = 0.032], but another partial interaction contrast that directly compared the incidental negative mood condition with only the neutral mood condition was not significant [F (1, 192) = 2.27, p  = 0.134]. Planned contrasts revealed that when participants were in negative moods, they felt significantly more sympathy after seeing the appeal featuring a single representative victim than after seeing the appeal featuring multiple representative victims [M_single = 7.80, M_multiple = 6.29, F (1, 300) = 17.05, p  < 0.001, η_p² = 0.054, CI = 0.791 ∼ 2.232]. However, when participants were feeling neutral [M_single = 7.29, M_multiple = 6.59, F (1, 300) = 3.63, p  = 0.058, η_p² = 0.012] or positive [F (1, 300) < 1], the difference was not observed (Figure 4).

Moderated Serial Mediation. We conducted a moderated serial mediation analysis to examine whether processing fluency and sympathy sequentially mediated the effect of representative victim type on charitable giving when participants were in a negative mood but not when they were in a neutral or a positive mood. Ten thousand bootstrap samples were generated on the entire data, including representative victim type as the predictor (X: 1 = single, 0 = multiple), donation intentions as the outcome (Y), processing fluency as the first mediator (M₁), sympathy as the second mediator (M₂) and incidental mood as the moderator (W: 0 = negative, 1 = neutral and positive combined) (Blanchard et al., 2016; Hayes, 2015, the macro PROCESS Model 6). The results supported the proposed multichain mechanism (H2). Specifically, when participants were in negative moods, presenting a single representative victim (vs multiple victims) increased processing fluency (b = 0.6531, p  = 0.015). Higher processing fluency then increased sympathy (b = 0.1888, p  = 0.013), which subsequently increased donation intention (b = 0.5075, p  < 0.001). The 95% bootstrap confidence interval for the indirect effect through both mediators did not include zero (B = 0.0626, CI = 0.0035 ∼ 0.1662), supporting the proposed mechanism involving processing fluency and sympathy. In contrast, when participants were in neutral or positive moods, no indirect effects were found (see Figure 5 and Table 2).

Alternative explanation based on perceived efficacy. The same ANCOVA revealed a positive impact of charity credibility on perceived efficacy [F (1, 300) = 58.82, p  < 0.001, η_p² = 0.164]. However, neither the interaction nor the main effects of the predictors were significant (p’s > 0.49), indicating that self-efficacy is not the mechanism that drives our effects. More specifically, mood did not affect self-efficacy [M_negative = 5.62, M_neutral = 5.40, M_positive = 5.65, F(2, 300) < 1].

To enhance practical applicability, Study 3 replicates these effects by inducing mood directly through emotional appeals in charity appeals. We anticipate that negative appeals featuring a single representative victim will be more effective than those featuring multiple victims, whereas no such difference is expected for positive appeals. Following Study 1, we used a bonus payment to assess donation behavior. This study was pre-registered on AsPredicted (#123700; Link to As PredictedLink to the cited article) ^[6].

Procedure. We used a procedure similar to Study 1 to assess donation behavior. A total of 513 participants were recruited for an unrelated survey in which we asked questions about occupations. For instance, participants listed jobs considered “low status” that could potentially earn either a small or large income. Once they completed the survey, we invited them to join our main study. Of these, 377 agreed to participate; however, four did not finish the study, resulting in 373 participants (178 men, 195 women, Mage = 40.45 years). These individuals were randomly assigned to a 2 (emotion appeal: negative, positive) × 2 (representative victim type: single, multiple) between-subjects design. Negative appeals featured distressed victims and messages about the adverse consequences of poverty, while positive appeals featured smiling victims and messages about successful outcomes. No specific identification information was provided. After viewing the ad, participants were told they received a 50-cent bonus and were asked how much they would donate. Charity familiarity (α = 0.93) and credibility were measured as covariates.

Donation Amount. A two-way ANCOVA revealed a significant interaction between emotion appeal and representative victim type on donation amount [F (1, 367) = 7.64, p  = 0.006, η² = 0.02]. No main effects were found [F (1, 367) < 1.44]. Both familiarity [F (1, 367) = 8.61, p  = 0.004, η² = 0.023] and credibility [F (1, 367) = 60.13, p  < 0.001, η² = 0.14] of the charity significantly affected the donation amount. In line with H1, planned contrasts showed that for negative appeals, the single representative victim was more effective [M_single = 17.38¢ (SE = 1.54¢) vs M_multiple = 11.80¢ (SE = 1.52¢), F (1, 367) = 6.65, p  = 0.01, η² = 0.018, CI = 1.326 ∼ 9.837]. For positive appeals, no significant difference was observed [F (1, 367) = 1.78] (Figure 6). To address the limitation of using different victim faces across conditions, subsequent studies use matched stimuli.

Study 4 strengthens methodological rigor by following Kogut and Ritov (2007), ensuring the victim in the single representative victim appeal is also included in the multiple representative victim appeal. This study was pre-registered. (AsPredicted #205153; Link to As PredictedLink to a PDF of the cited article).

Procedure. A total of 498 participants were recruited from CloudResearch. However, one participant did not answer the donation amount question, resulting in a final sample of 497 participants (238 males, 258 females and 1 who preferred not to answer; M_age = 41.0 years). Participants were randomly assigned to one of four conditions in a 2 (incidental mood: negative, positive) × 2 (representative victim type: single, multiple) between-subjects design. To induce incidental mood, each participant watched the same films used in Study 2. Those in the positive mood condition viewed a short film from Friends, while those in the negative mood condition watched an equally-timed film from The Champ. After viewing their assigned film, participants answered a question related to the film. For the negative mood group, the question was “Aside from ‘Dad,’ what nickname does the son use for his father?” For the positive mood group, the question asked why Ross did not talk much on his date. These questions served two purposes: first, to foster the belief that the study was unrelated to the main study about donation behavior; second, as potential attention checks. The pre-registration noted that we might explore whether removing participants who failed these questions would affect the results.

The representative victim type was manipulated using a charity appeal featuring either a single child or multiple children. For the multiple-victim condition, a group portrait of six children was shown, while the single-victim condition displayed six separate pictures of the same children, each presented an equal number of times using segments from the group portrait.

After seeing the appeal, participants were told they would be automatically entered for a chance to win an additional $10 as a reward for participation. They then indicated how much of this $10 they would be willing to donate to the featured charity if selected. Following this, participants completed the same familiarity (α = 0.94) and charity credibility measures as in previous studies.

A two-way ANCOVA, controlling for charity familiarity and credibility, showed a directional interaction between incidental mood and victim type on donation amount, which was not statistically significant [F (1, 491) = 3.81, p  = 0.052, η2 = 0.008]. No main effects were observed [Fs (1, 491) < 0.247]. Both familiarity [F (1, 491) = 9.25, p  = 0. 002, η² = 0.018] and credibility [F (1, 491) = 41.63, p  < 0.001, η² = 0.078] of the charity had a significant impact on donation amounts. In line with recommendations for theory-driven a priori tests (Rosenthal and Rosnow, 2008), planned contrasts were examined even though the interaction was not significant. Consistent with our hypothesis (H1), planned contrast analysis indicated that under the negative mood condition, the appeal featuring a single representative victim elicited higher donations than the one with multiple representative victims [M_single = $5.25 (SE = 0.31) vs M_multiple = $4.50 (SE = 0.31), F (1, 491) = 2.87, p  = 0.091, η² = 0.008, CI = −0.120 ∼ 1.612]. Although this difference did not reach statistical significance (p  = 0.091), the direction of effect aligned with our hypothesis and replicated patterns observed in our previous studies. On the other hand, in the positive mood condition, no significant difference was found [F (1, 491) = 1.11].

In a subsequent analysis, we examined whether the effect was enhanced by excluding participants who did not respond appropriately to the film-related questions. For the positive mood condition, 10 individuals who failed to indicate that Ross spoke little during the date because of overly bright white teeth were excluded. In the negative mood condition, we removed 16 participants whose answers were clearly incorrect (e.g. papa) or who said they did not know ^[7]. No main effect was found [Fs (1, 465) < 0.419], but the interaction effect was significant [F (1, 465) = 4.93, p  = 0.027, η² = 0.01]. Consistent with our predictions, in the negative mood condition, the appeal featuring a single representative victim led to higher donations than the appeal featuring multiple representative victims [M_single = $5.36 (SE = 0.32) vs M_multiple = $4.46 (SE = 0.33), F(1, 465) = 3.87, p  = 0.050, η² = 0.008, CI = 0.001 ∼ 1.801]. However, there was no difference in donation amount in the positive mood condition [F (1, 465) = 1.31]. Both familiarity [F (1, 465) = 7.68, p  = 0.006, η² = 0.016] and credibility [F (1, 465) = 40.18, p  < 0.001, η² = 0.080] significantly increased donation amount (see Figure 7).

The results provide additional support for the hypothesis. By using incidental mood induction and carefully matched stimuli, along with a pre-registered design, this study offers a more rigorous test of the effect.

In our previous four studies, we offered minimal identification details for the victims, displaying only representative victims’ image. We posited that a single representative victim may not yield substantially greater effectiveness compared to multiple representative victims, even when their identification information is disclosed, as the reference group effect might not contribute to the overall impact. In Study 5, we investigated whether our findings persist when more detailed identification information about the representative victims is provided.

We recruited 479 participants (242 males, 229 females, 8 unspecified; M_age = 39.4 years) from CloudResearch and randomly assigned them to one of four conditions in a 2 (incidental mood: negative, positive) × 2 (victim type: single, multiple) between-subjects design. In Study 5, using the same images and procedure as Study 4, we added individual victim identification (age, name) and increased the bonus payment from $10 to $20 for donations. Other than these changes, all procedures matched those in Study 4.

The two-way ANCOVA did not yield a statistically significant interaction term [F (1, 473) = 1.18, p  = 0.278]. However, given our a priori hypothesis that a single (vs multiple) representative victim would lead to higher donations specifically in the negative mood, but not in the positive mood condition, we conducted a planned contrast test. As expected, adverts featuring a single representative victim led to significantly higher donations than those with multiple victims in the negative mood condition [M_single = $9.24 (SE = 0.57) vs M_multiple = $6.92, (SE = 0.57), F(1, 473) = 8.167, p  = 0.004, η² = 0.017, CI = 0.724 ∼ 3.91], but not in the positive mood condition [F (1, 473) = 1.818], supporting H1. A main effect showed that the appeal featuring a single representative victim (vs multiple representative victims) resulted in higher donations [F (1, 473) = 8.89, p  = 0.003, η² = 0.018]. Familiarity with the charity did not significantly affect donations [F (1, 473) = 2.90, p  = 0.089], while credibility did [F (1, 473) = 55.27, p  < 0.001, η² = 0.11]. There was no main effect of mood [F (1, 473) = 0.56].

We tested whether excluding participants who gave incorrect answers about the films impacted results, using the same criteria as in Study 4. After removing 5 from the positive mood group and 9 from the negative mood group, there was no significant change. The interaction effect [F (1, 459) = 0.95] and mood main effect [F (1, 459) = 0.60] were not significant. The representative victim type had a significant main effect [F (1, 459) = 9.68, p  = 0.002, η² = 0.021]; specifically, in the negative mood condition, donations were higher for the appeal featuring a single victim (vs multiple victims) [M_single = $9.30 (SE = 0.59) vs M_multiple = $6.94 (SE = 0.58), F(1, 459) = 8.18, p  = 0.004, η² = 0.018, CI = 0.740 ∼ 3.992]. No difference was found in the positive mood condition [F (1, 459) = 2.33]. Familiarity did not affect donation amount [F (1, 459) = 2.21] but credibility increased donation [F (1, 459) = 53.46, p  < 0.001, η² = 0.10] (see Figure 8).

To address the limitations of individual study power and provide a more definitive test of our hypothesis, we conducted an internal meta-analysis using an integrative data analysis (IDA) approach (Curran and Hussong, 2009) by combining data from Study 4 and Study 5. In the meta-analysis, we included all participants regardless of whether they answered the film-related questions correctly or incorrectly. Because the dependent variables in Study 4 and Study 5 used different scales (out of $10 vs $20), we standardized the donation amounts by converting them to Z-scores within each study. We then pooled the data and conducted a global ANCOVA including “Study” as a fixed factor.

To test our hypotheses across the combined sample (n = 976), we first conducted an integrative ANCOVA. The three-way interaction between incidental mood, victim type, and study type (Study 4 and Study 5) was not significant [F (1, 966) = 0.344], indicating that the observed patterns were consistent across both studies. Consequently, we combined data across studies and found a significant two-way interaction between incidental mood and victim type on standardized donation amounts, [F (1, 970) = 4.76, p  = 0.029, η² = 0.005]. Planned contrasts revealed that in the negative mood condition, a single representative victim elicited significantly higher donations [M_single = 0.158 (SE = 0.061) vs M_multiple = −0.118 (SE = 0.061), F (1, 970) = 10.31, p  = 0.001, η² = 0.011, CI = 0.107 ∼ 0.444], while no difference emerged in the positive mood condition [F (1, 970) = 0.031]. A main effect of victim type was found [F (1, 970) = 5.89, p  = 0.015, η² = 0.006]. However, mood main effect was not found [F (1, 970) = 0.415]. Both familiarity [F (1, 970) = 11.50, p  < 0.001, η² = 0.012] and credibility [F (1, 970) = 96.18, p  < 0.001, η² = 0.09] increased donation amount (see Figure 9). By increasing statistical power and demonstrating consistent patterns across studies, results from this meta-analysis provided further evidence that the persuasive advantage of featuring a single representative victim over multiple representative victims emerges specifically under negative mood conditions, but not under positive mood conditions.

The current research provides a comprehensive examination of how a donor’s mood interacts with the number of representative victims to influence charitable giving. Across six studies, including an internal meta-analysis, we consistently found that the effectiveness of a single representative victim is contingent upon the donor experiencing negative moods. This effect persists consistently, irrespective of changes in the identification details of representative victims. While the interaction between mood and representative victim was not always statistically significant, planned contrasts dependably confirmed that soliciting donations when donors are in negative moods is more effective when referencing a single representative victim compared to multiple representative victims. This effect appears to follow a serial mediation process, such that greater compatibility between negative mood and a single representative victim is associated with enhanced processing fluency, which is in turn linked to higher sympathy and donation behavior.

Specifically, Study 1 provides the empirical test of the primary hypothesis by demonstrating that, under negative moods, charity appeals featuring a single representative victim elicit higher actual donation amounts than appeals featuring multiple representative victims. By using real monetary consequences rather than hypothetical intentions, this study establishes the behavioral relevance of the proposed effect. Building on Study 1, Study 2 examines why the effect occurs by testing the proposed serial mechanism involving processing fluency and sympathy. This study also introduces a neutral mood condition, allowing us to clarify that the effect is specific to negative moods rather than a general advantage of single representative victims. In addition, Study 2 includes a measure of perceived efficacy to address a plausible alternative explanation, providing further support for the proposed process account.

Although Studies 1 and 2 rely on incidental mood inductions, Study 3 replicates the effect using mood elicited directly by the charity appeal. This design increases the practical relevance of the findings by demonstrating that the effect holds in realistic contexts, where donors’ emotional states are shaped by the appeal. Study 4 increases methodological rigor by ensuring that the victim depicted in the single-victim condition is also included in the multiple-victim condition. This design follows prior practices and eliminates the possibility that differences in victim identity, rather than victim numerosity, drive the observed effects. The study thus provides a more conservative test of the hypothesis while maintaining the same underlying logic as earlier studies. Study 5 examines whether the focal effect persists when representative victims are accompanied by identifying information (e.g. name and age). The results show that even when identification information is present, a single representative victim is more effective than multiple representative victims only under negative moods, reinforcing the claim that mood–message compatibility, rather than identifiability alone, drives the effect. To address concerns about statistical power and variability across individual experiments, Study 6 conducts an internal meta-analysis combining data from Studies 4 and 5. This analysis confirms the robustness and consistency of the effect across studies with stringent controls and different donation scales, providing a more definitive test of the theoretical predictions.

This research contributes to the literature on charitable giving in several important ways. First, by introducing a novel approach to soliciting donations on behalf of groups of victims, our findings enrich existing research on strategies for increasing charitable giving to multiple beneficiaries or organizations (e.g. Ein-Gar and Levontin, 2013; Kogut and Ritov, 2005b; Kogut and Ritov, 2011; Sharma and Morwitz, 2016; Small et al., 2007; Smith et al., 2013).

Second, although examining the reference group effect was not our central objective, our findings suggest that this mechanism plays a meaningful role in shaping donation behavior. We do not expect the reference group effect to be fully activated when a representative victim stands in for many unidentified victims. In Study 5, we replicated prior findings on the single identified victim effect by incorporating identifying information, such as age, name and photographs, into the appeal, while directing donations to a supporting charity rather than to the identified victim. The results showed that featuring a single representative victim was more effective than presenting multiple victims when participants were in a negative mood, but not when they were in a positive mood. This suggests that negative moods may compensate for the diminished reference group effect in representative-victim appeals.

Third, the current research enriches our understanding of how different victim types (identified vs unidentified) relate to the processing modes (Chaiken, 1980; Petty and Cacioppo, 1986). Prior research suggests that the concrete details provided by an identified victim can increase affective involvement and encourage more systematic or central processing (Sherman et al., 1999). Building on this work, we suggest that negative moods may make donors more receptive to charity appeals featuring a single representative victim. Because negative moods have been associated with more systematic, detail-oriented processing (Schwarz, 1990; Gasper and Clore, 2002), such appeals may be perceived as more coherent or easier to process. This increased processing fluency may, in turn, be associated with greater sympathy and higher levels of charitable giving.

Finally, this research contributes to the literature on emotional appeals in charity advertising (Choi and Park, 2021; Septianto and Tjiptono, 2019; Small and Verrochi, 2009) by identifying a critical boundary condition: the number of victims presented. While negative emotions can encourage helping behavior through both altruistic and egoistic motives (Batson et al., 1981; Small and Verrochi, 2009), we demonstrate that presenting multiple victims may become emotionally overwhelming or cognitively taxing, thereby reducing the effectiveness of such appeals.

From a practical perspective, our findings offer valuable guidance for charities and nonprofit organizations seeking to raise funds for causes involving multiple victims. Specifically, the results suggest that highlighting a single representative victim, rather than presenting multiple victims simultaneously, can be a more effective strategy for encouraging donations, particularly when donors are likely to be in a negative mood.

Moreover, charities should be cautious when using emotionally intense imagery involving multiple victims. Although such approaches may be intended to evoke sympathy, they can inadvertently overwhelm potential donors or increase cognitive burden, thereby diminishing donation intentions. Instead, focusing on a single representative victim may improve processing fluency and emotional engagement, leading to greater sympathy and higher levels of giving. Taken together, these insights underscore the importance of carefully tailoring victim presentation strategies to donors’ emotional states and cognitive processing tendencies. By doing so, charities can design more effective appeals.

We note that our findings partially overlap with those of Sabato and Kogut (2021), particularly in showing no difference in donation amounts between a single identified victim and unidentified children under positive mood conditions. However, several key distinctions warrant emphasis. First, Sabato and Kogut (2021) compared a single identified victim against unidentified multiple victims without controlling for identifiability and singularity, making it unclear whether the observed effects stemmed from singularity, victim identification or both. This design could also send mixed signals about donation allocation: participants in the identified single-victim condition might assume their contribution directly benefited that individual, whereas those in the unidentified multiple-victim condition might view it as a general donation to the charity. In contrast, our study controlled for victim identification and varied only the number of representative victims (single vs multiple). Moreover, by explicitly soliciting donations to charity, we minimized the likelihood that participants would believe their funds were earmarked for a specific victim. Most importantly, unlike Sabato and Kogut (2021), our research does not aim to test the identifiable victim effect; instead, we focus on developing effective charity appeals for unidentified multiple victims by leveraging a compelling representative victim. Second, our studies diverge in theoretical rationale. Sabato and Kogut (2021) invoked mood maintenance theory to explain why individuals in positive moods are less inclined to donate to a single identified victim. By comparison, we examined why charity appeals featuring a single representative victim prove more effective than those with multiple representatives, but only under negative mood conditions. While Sabato and Kogut (2021) observed donation differences in neutral moods, we did not, a discrepancy we attribute to the methodological differences outlined above. Despite these overlaps in effects, we believe our findings offer meaningful practical contributions by providing actionable insights into enhancing the effectiveness of charity appeals for groups of victims.

Another limitation of the current research concerns the evidence for the proposed underlying mechanism. Although the findings are consistent with a process account involving processing fluency and sympathy, the present studies do not comprehensively test or rule out all possible alternative mechanisms that might also contribute to the observed effects. As with much prior work in this domain, fully disentangling multiple, potentially overlapping psychological processes remains challenging. Accordingly, our results should not be interpreted as providing a definitive account of the mechanism. Rather, they suggest that the proposed process offers a plausible and partial explanation of how negative mood and victim singularity jointly shape donation behavior, while leaving open the possibility that other processes may also play a role.

Future research could explore whether mood serves as a boundary condition for the entitativity effect, which has been shown to increase donations to multiple victims (Smith et al., 2013). Specifically, Smith et al. (2013) demonstrated that enhancing the perceived entitativity (i.e. group coherence) of multiple victims can motivate donations to them at levels comparable to those for a single victim. However, this effect was tested only with identified human victims, leaving it unclear whether it extends to unidentified human victims. Although the effect has been observed for unidentified animal victims, prior studies indicate that animal and human victims are perceived differently (Ein-Gar and Levontin, 2013; Fujita et al., 2008), raising the possibility that entitativity alone may not suffice to drive donations when human victims remain unidentified. This concern is further supported by Kogut and Ritov (2007), who found that victim singularity by itself does not motivate giving, implying that perceived coherence in an entitative group of human victims may also fall short. Therefore, future studies could examine whether negative moods amplify the entitativity effect, potentially due to the compatibility between negative moods and perceptions of group coherence.

In addition, future research might investigate how negative mood moderates the adverse impact of victim identifiability on prosocial behavior. Kogut (2011) identified scenarios where identification reduces prosocial actions, particularly when the victim is perceived as responsible for their own plight. It would be worth investigating whether negative moods exacerbate this negative effect by prompting deeper engagement with the victim’s information, thereby intensifying adverse feelings and leading to reduced helping behavior.

In closing, we highlight a potential ethical concern with using a single victim’s image to represent a larger group of victims. Such representations may lead donors to mistakenly assume that their contribution is intended solely for the depicted individual, especially if accompanied by identifying details like a name or age. To mitigate this risk, we recommend that charitable organizations clarify donation allocation, for example, by explicitly stating that funds will support specific causes or projects, or by prompting donors to confirm their contribution to the organization or cause after viewing the advertisement, as implemented in our studies. Furthermore, using animations or drawings of a single victim to symbolize the broader group, rather than real photographs, could help prevent misconceptions about the donation’s target.

The supplementary material for this article can be found online.