How can I help you? Multiple resource availability promotes generosity with low-value (but not high-value) resources

Abstract

People commonly possess multiple, differentially-valued resources they can use to benefit those in need: contributing money, volunteering time, donating unwanted possessions, posting on social media to raise awareness, and more. But the majority of experimental work on generosity and helping behavior has studied giving when only a single valuable resource is available to give. This project considers: when people have multiple, differentially-valued resources to use to benefit a dependent other, which resources will they use to help, and how much? Results from an experiment show, first, that generosity is higher with lower-value resources. More importantly, when multiple, differentially-valued resources are available to use to benefit others, helping is higher than when a single resource is available, all else being equal. This is because when multiple resources are available, people are especially willing to give away their relatively lower-value resources. Put differently, when people can benefit others using multiple resources, they are more likely to consider how they should help, rather than whether they should.

Keywords

Exchange experiments generosity prosocial behavior

Potential benefactors are often able to help others in multiple different ways. For instance, people may behave generously by giving money or volunteering their time (Borgonovi, 2008; Brooks, 2005; Duncan, 1999; Lee et al., 1999; Lengfeld and Ordemann, 2016; Manatschal and Freitag, 2014; Schervish and Havens, 1997; Wiepking and Maas, 2009; Wilson and Musick, 1997; Zak et al., 2007). In lieu of money or time, they could opt to contribute unwanted items around the home, such as gently used clothing or extra food from the pantry (Companion, 2010; Laitala, 2014; Molnar et al., 2001; Park et al., 2017; Rezaei Arangdad et al., 2019), to a family or charity in need. Or they might choose to engage in still different (and lower-cost) behaviors like “sending thoughts and prayers,” posting on social media, wearing a ribbon, or signing an online petition to bring awareness to the needs of an individual, cause, or charitable organization (Kristofferson et al., 2013; Lane and Dal Cin, 2018; Moore, 2008; Thunström, 2020; Wallace et al., 2017).

These examples illustrate that people typically can and do make decisions about which, if any, of the multiple resources they have available—money, time, possessions, energy, and more—to use to benefit others (Glanville et al., 2016; Schaefer, 2009; Smith and Davidson, 2014). Some of those resources are more valuable, to both benefactor and beneficiary (e.g. money), compared to others (e.g. ribbon wearing—a form of prosocial engagement that critics have referred to as “slacktivism” for satisfying people’s desire to engage in prosocial behaviors but providing significantly less help (Hill and Thompson-Hayes, 2017; Kristofferson et al., 2013)).

But with a few important exceptions described below, whether and how generosity patterns are shaped by the availability of multiple resources that can be used to help has received less attention in the literature. This project explores the question: when people have multiple, differentially-valued resources to use to benefit a dependent other, which of these resources will they use to help, and how much? When differentially-valued resources can be deployed to help others, the result may not only be different helping rates across the different resource types. Multiple resource availability may also alter overall helping patterns compared to situations where only one resource type is available to give. Thus knowing more about the relationship between multiple resource availability and generous behavior is critical in understanding the generous acts we commonly observe in daily life.

First consider how helping patterns might be altered across multiple resources that can be used to help, but that differ in their values. Economic models have long predicted that people are more willing to sacrifice their own resources to benefit others the lower is the cost of doing so (Fehr and Fischbacher, 2003; Oliver, 1984; Rabin, 1993). And both self-report measures and behavioral experiments lend support to these arguments: perceived cost is an important factor in people’s decision-making about helping (Ferguson et al., 2019; Stewart-Williams, 2007); generally, helping decreases the higher is the cost (Bode et al., 2015). A straightforward extension of this argument is that when multiple resources may be deployed to benefit a dependent third party—as when a charity is willing to accept monetary donations, volunteer work, gently used clothing, and/or a social media post to help raise awareness—people will be more likely to help using those resources that are less costly to them compared to those that are more costly, all else being equal.

One relevant stream of literature is work on the effects of stake size on giving in the Dictator Game, which asks whether the size of an endowment affects people’s decisions about distributing it between oneself and a dependent other. Some of this prior research has found that people behave similarly regardless of endowment size, sharing equivalent proportions from resource pools of $5 versus $10 (Forsythe et al., 1994), $10 versus $100 (Carpenter et al., 2005), and other variations (Diekmann, 2004; List and Cherry, 2008). But others have demonstrated that higher stakes are linked with relatively lower levels of generosity (Novakova and Flegr, 2013; Raihani et al., 2013). And meta-analyses reveal that, overall, the value of the resource pool is associated with a reduction in the proportion of the resource given (Engel, 2011; Larney et al., 2019). These findings are suggestive that generosity is affected by the value of the resource to be distributed.

Another informative line of work is that on the “denomination effect,” which demonstrates that decisions about apportioning even identical amounts of a resource are altered by how the value of the resource is represented. For example, the likelihood of spending money is lower—and the preference for receiving money is higher—when the same amount of money is represented as one $20 bill than when it is represented as 20 $1 bills (Raghubir and Srivastava, 2009). This finding is aligned with other work showing that mental representations of money, and spending behaviors, are altered by the way value is presented—for example, as cash compared to credit cards or gift certificates (Raghubir and Srivastava, 2008).

Consider, though, that this prior research either varied the amount of a single resource that was available to give (e.g. $5 vs $10), or examined the same amount of a resource, but varied how it was presented (e.g. 20 $1 bills or 1 $20 bill). But as noted above, people not only experience differences in the amount or denomination of a single resource they may use to benefit others. They also are commonly able to use one or more of the multiple resources they have available to benefit a third party (Glanville et al., 2016; Schaefer, 2009; Smith and Davidson, 2014)—as when they may, to revisit the examples described above, help by making a financial contribution, donating unwanted possessions, and/or changing their social media profile picture to raise awareness for a cause. Thus the work described here seeks to answer a related, but distinct question: how helping patterns are altered in the presence of multiple resources that differ in their value, and can be used to help. As discussed in more detail in the next section, the experiment conducted to address this question held constant the size and denomination of an endowment (it always contained 10 tokens; benefactors could give any whole token amount, from 0 to 10), but manipulated the presence of differentially-valued tokens in the endowment (e.g. high-value tokens only, low-value tokens only, or a mix of both) and examined the proportion of the resource that was given.

One particularly relevant study on the distribution of differentially-valued resources is Blake and Rand’s (2010) finding that children donate a higher proportion of an endowment consisting of their least favorite sticker type (i.e. a lower-value resource) than an endowment consisting of their favorite sticker type (i.e. a higher-value resource) to another child, suggesting that prosocial behavior is indeed moderated by the value of the different resources available to give. Yet because this work was designed to answer different questions, it does not address the second question considered here: how giving when multiple, differentially-valued resources are available differs from giving patterns when only one resource type is available (in the Blake and Rand context, for instance, an endowment consisting of both sticker types, vs an endowment of only one’s favorite stickers or only one’s least favorite stickers), all else being equal.

The current project assessed (1) whether generosity patterns are altered when the resource that can be used to benefit others is of relatively higher or lower value. Still further, it explored (2) whether and how generosity toward a dependent third party differs when multiple, differentially-valued resources are available to give compared to when only a single resource is available to give. Both questions were addressed in an experiment conducted with 200 Amazon Mechanical Turk users located in the United States (Buhrmester et al., 2016; Burleigh et al., 2018; Kennedy et al., 2018; Shank, 2016; Weinberg et al., 2014). The procedures are described in the next section.

Methods

The study procedures were reviewed and approved by the Institutional Review Board at the author’s university prior to beginning the research. There was no deception in the study, which was conducted via Amazon’s Mechanical Turk (Buhrmester et al., 2016; Shank, 2016; Weinberg et al., 2014). In order to proceed past an initial invitation screen, participants were required to be 18 years of age, located in the US, and, to confirm the US location requirement, not using a virtual private server (VPS), virtual private network (VPN), or proxy to mask their location (Burleigh et al., 2018; Kennedy et al., 2018). Data was collected from 200 participants who met these qualifications. They completed the experiment in exchange for a flat payment of $2 plus the chance to receive a bonus based on their and others’ decisions in the task described below.

After answering several demographic questions, the main task was introduced as follows. Participants were told that they would be making decisions about how much of a 10-token endowment to transfer to another participant in each of several “rounds.” These tokens were valuable: they would be translated into money at the end of the study, and earnings from one randomly selected round of the study would make up their bonus payment.

At the start of a given round the participant would receive the 10-token endowment and could give anywhere from 0 to 10 of those tokens to an other. Tokens given to others would be doubled, though tokens kept for oneself were not. In this way, it was costly to help others, but when help was provided the benefit received was greater than the cost, as is common when people provide benefits to others. So that the study did not require deception, participants were also told that they would be in the receiver role with a different other participant, that is, receiving tokens from different others, rather than sending them. The instructions made clear that participants would never be paired with the same other more than once. Further, no one would find out how many tokens they had received until the study was completed. As a result, the study consisted of a series of anonymous, one-shot decisions to provide benefits to another, and posed a tension between individual and collective interests: if all study participants gave away all of their tokens, all would double their earnings in a given round; at the same time, any given person would benefit from keeping their own tokens and receiving benefits from others.

Prior to the start of the first round, participants received one additional piece of information. Specifically, they were informed ahead of time that while they would always be making decisions about the distribution of a 10-token endowment between themselves and an other, there were several different types of tokens in the study. The tokens differed in how much money they were worth, that is, how valuable they were. And the token type that they would have in their 10-token endowment would change across the rounds of the study. Red tokens, they were told, were the most valuable resource in the study (thus, they are referred to here as “high-value” resources), at 10 cents per token. Orange tokens (“moderate-value” resources) were 7.5 cents per token. Blue tokens were the least valuable resource at 5 cents per token (“low-value” resources). Several examples and quiz questions about the instructions followed, to ensure participants’ understanding of the task and the different types of tokens. Incorrect answers were followed by a detailed description of the correct answer.

Participants then went on to make decisions to transfer any whole number amount of the 10-token endowment in several rounds. Each round was one decision condition: they completed a decision with 10 high-value tokens, a decision with 10 low-value tokens, and a decision with a 10-token endowment consisting of 5 high-value and 5 low-value tokens. The latter decision is referred to here as the multiple resources decision, compared to the two former decisions, which contain a single resource, of high or low value, only.

Note that, while each endowment consisted of the same amount of tokens (10), the entire value of the endowments in the three decisions, and thus participants’ wealth in the three decisions, necessarily differed. The 10-token endowment of only high-value tokens was worth double the 10-token endowment of only low-value tokens (10 tokens × 10 cents = $1.00 total endowment in the single resource, high-value decision; 10 tokens × 5 cents = $0.50 total endowment in the single resource, low-value decision), while the endowment with half high-value and half low-value resources fell in between (5 tokens × 10 cents + 5 tokens × 5 cents = $0.75 total endowment in the multiple resources decision). To compare giving patterns when multiple resources versus a single resource were available to give while holding constant the total value of the endowment (i.e. initial wealth), the third single resource decision consisted of giving 10 “moderate-value” (“orange” in the study instructions) tokens. These tokens were valued directly in between high-value and low-value tokens (at 7.5 cents); thus the value of the entire endowment, and participants’ initial wealth, for this single resource decision was the same as the decision with multiple (half high-value and half low-value) resources ($0.75 cents total in both decisions). This decision condition allowed for a test of whether generosity was altered in the presence of multiple resources versus a single resource, holding constant the total value of what was available to give. This is particularly important given that, with the within-subjects nature of the design, participants likely experienced different subjective senses of wealth across the decisions—an issue I consider further in the discussion section. Table 1 gives an overview of the study conditions.

Table 1.

Overview of study conditions.

Decision condition	Endowment received (tokens)	Each token is worth (cents)	Total value of the endowment (dollars)
Single resource, high-value	10 red	10	1.00
Single resource, moderate-value	10 orange	7.5	0.75
Single resource, low-value	10 blue	5	0.50
Multiple resources	10–5 red and 5 blue	10 (red) and 5 (blue)	0.75

Thus the main study manipulation was the presence of single or multiple resources (token types) when participants made giving decisions, as well as the value of the resource. The manipulation was conducted within-subjects; all participants made decisions in each condition (each round was one condition) in random order. During the single resource decisions, participants decided how many of 10 same-value resources to give. They made one decision each with high-value (moderate-value, low-value) resources, where they decided how many of 10 red (orange, blue) tokens to give to the dependent other. In the multiple resource condition, they decided how many of 10 tokens to distribute to a dependent other, but this time, 5 of the tokens were high-value and 5 were low-value. For this decision, participants indicated how many red tokens and how many blue tokens they would send in separate questions on the same screen. Whether the red token decision or blue token decision was listed first on this screen was counterbalanced.

Once the entire study was completed, participants were randomly paired and their bonuses were calculated based on actual decisions in a randomly selected round. The entire procedure was advertised as taking no more than 15 minutes (actual average time of completion was 11.78 minutes). The average payment including the bonus was $2.75; the result was an average pay rate of approximately $14/hour. A codebook and dataset for replicating the analyses are available at the Open Science Framework: https://osf.io/py9dv/?view_only=23bef857c53544eab07773fc8cb969ca.

Results

Analytic strategy

Analyses relied on multilevel linear models given that the data structure was nested: participants completed multiple decisions. These models account for the dependencies that typically occur in nested data (Snijders and Bosker, 2011). To ensure that the fixed effect estimates were robust to violations of assumptions of multilevel linear models, bootstrapped confidence intervals are provided around all fixed effect estimates presented in the tables (percentile method, N = 2000 samples).

The outcome variable of interest in the models described in this section was the proportion of tokens given out of the 10-token endowment. I used proportion of tokens given as the outcome variable for several reasons. First, the number of tokens in the endowment (10) was the one factor that was held constant across all conditions (see Table 1) and thus allows for the most straightforward comparison across conditions. Second, using the monetary value of what was given as the outcome variable, rather than the proportion of the available resource that was given, produces misleading conclusions. Consider that giving one high-value token would yield double the value to the beneficiary than would giving one low-value token; in this sense it could appear to be a more generous behavior. But at the same time, giving one high-value (vs one low-value) token also entails the benefactor keeping double the value for the self; that is, from this perspective the same behavior is more selfish. Third, I follow past work in using proportion given as the outcome variable. This past work has manipulated resource pool size (as described above, e.g. $5 vs $10, Forsythe et al., 1994; $10 vs $100, Carpenter et al., 2005; and other variations, Diekmann, 2004; List and Cherry, 2008; Novakova and Flegr, 2013; Raihani et al., 2013) and compared giving patterns across the very different endowments by examining the proportion of what was available to give that was actually given (here, how many tokens were given, out of 10).

Single-resource giving

First consider giving patterns in the single-resource decisions. As described above, participants were told before making any decisions that there were several types of resources (tokens) in the study, which differed in their value. As a result, when making decisions about sharing the 10-token endowment consisting of a given token type, they were aware of how the value of the single resource they had compared to the value of the other resources in the study. Results demonstrate that these differences in resource type mattered when people were making giving decisions—they gave a significantly higher proportion of the 10-token endowment when it contained low-value tokens, compared to endowments consisting of moderate-value (B = 0.05, p = 0.001) or high-value (B = 0.08, p < 0.001) tokens. The difference between the latter two decisions was not statistically significant, though it followed the general trend: high-value resource giving was somewhat lower than moderate-value resource giving (B = −0.02, p = 0.13). Figure 1 displays these patterns (as well as giving in the multiple resource condition, which is discussed in the next section). See also Table 2, Model 1.

Figure 1.

Mean proportion of the 10-token endowment given, by resource type. The three single resource decisions are on the left (solid color bars) and the multiple resource decision is on the right (patterned bar). Error bars denote the standard error of the mean. ^a,bDifferent subscripts indicate the means significantly differ, p < 0.050; shared subscripts indicate the means do not significantly differ, p > 0.100. N = 200 participants (800 decisions).

Table 2.

Models predicting proportion of the 10-token endowment given, by resource type.

	Model 1: single resource decisions only		Model 2: all decisions, overall		Model 3: all decisions, by resource type in the multiple resources decision
	B (SE)	Bootstrapped 95%CI	B (SE)	Bootstrapped 95%CI	B (SE)	Bootstrapped 95%CI
Single resource, high-value decision^a	−0.08 (0.02)***	−0.11, −0.05	−0.07 (0.02)***	−0.10, −0.03	−0.14 (0.02)***	−0.18, −0.10
Single resource, moderate-value decision	−0.05 (0.02)**	−0.09, −0.02	−0.04 (0.02)*	−0.07, −0.01	−0.11 (0.02)***	−0.15, −0.07
Single resource, low-value decision			0.01 (0.02)	−0.02, 0.04	−0.06 (0.02)**	−0.10, −0.02
Multiple resources decision—high value portion					−0.14 (0.02)***	−0.18, −0.11
Intercept	0.46 (0.03)***	0.40, 0.51	0.44 (0.03)***	0.39, 0.50	0.51 (0.03)***	0.46, 0.57
Variance components—variance (standard deviation)
Level 1	0.03 (0.16)		0.03 (0.17)		0.04 (0.19)
Level 2	0.14 (0.37)		0.14 (0.37)		0.14 (0.37)

Note: Multilevel linear models (maximum likelihood estimation) with random intercepts for participants. N = 600 decisions nested in 200 participants in Model 1, N = 800 decisions nested in 200 participants in Model 2, N = 1000 decisions nested in 200 participants in Model 3. Supplementary models controlling for the order in which the decisions were made did not affect results.

Single resource, low-value decision is the reference category in Model 1 (the multiple resources decision was not included in this model); Multiple resources decision is the reference category in Model 2. Multiple resources decision, low value portion is the reference category in Model 3.

***

p < .001. **p < 0.010. *p < 0.050.

These findings suggest that people tended to be more selfish with endowments made up of high-value resources and more generous with endowments of low-value resources, holding constant the ability to distribute those resources (0–10 tokens were available to be distributed in all conditions). Follow-up analyses considered two additional issues. First, some past work suggests there are gender differences in sensitivity to the costs of prosocial behavior (Cox and Deck, 2006)—finding, for instance, that women are more generous when it is expensive while men are more generous when it is cheap (Andreoni and Vesterlund, 2001). It is then at least possible that the relationship between resource type and giving may have been moderated by gender. But gender did not interact with resource type to predict giving in follow-up models, nor did gender predict giving overall in this or the other models discussed below.

Second, alternative models examined not the proportion of the 10-token endowment given, as in the results above, but the likelihood of giving anything to the other, versus giving nothing. After all, one alternative explanation for the above results is that people may have been strategically sending different proportions of tokens to the other in order to send approximately identical dollar values to the other. Given that the tokens differed in their value, sending one high-value token was the dollar equivalent of sending two low-value tokens (though the former is more proportionally “selfish” in that only 1/10th, rather than 2/10ths, of the total endowment is given).

These alternative analyses on giving anything versus giving nothing were in line with those on the proportion of the endowment shared. People were significantly more likely to keep all of the high-value tokens (45% kept all of the high-value resource), compared to their likelihood of keeping all of the low- (36%) or moderate-value (39.5%) resource (generalized linear mixed models, B = 2.78, p < 0.001 and B = 1.63, p = 0.008 respectively). Patterns in the latter two conditions also trended in the expected direction: people were more likely to keep all of their moderate-value resources than they were to keep all of their low-value resources (B = 1.15, p = 0.052). That is, not only did people share fewer of their high-value resources than their low-value resources, holding constant the number of resources (10 tokens) available to give; they were also increasingly less willing to benefit a dependent other at all as the value of the resource increased.

Multiple-resource giving

Results thus far examined giving behavior when a single (relatively more or less valuable) resource was available to give. But as described above, in daily life people typically possess multiple resources they can use to benefit others, and those seeking help can often accept it in more than one form. The multiple resources decision gave participants a 10-token endowment consisting of half high-value and half low-value tokens. How did giving differ when multiple resources were available to give compared to when only a single resource was available?

Initial analyses considered the proportion of the 10-token endowment given (regardless of the specific token type—high-value or low-value—given in the multiple resources condition, discussed next). People were significantly more generous with their 10-token endowment when it consisted of multiple resources, compared to when only high-value resources were available to give and, more importantly (given that wealth was equivalent in the multiple resources decision and the single resource, moderate-value decision), when only moderate-value resources were available to give (B = 0.07, p < 0.001 and B = 0.04, p = 0.01, respectively; see also Figure 1 and Table 2, Model 2). The difference in resource-sharing when multiple resources were available to give and when only low-value resources were available was not significant (p = 0.45). Again, these patterns were also reflected in ancillary analyses on the likelihood of giving nothing (vs anything) to the other: significantly fewer people gave nothing to the other when multiple resource types were available to give (34.5%) compared to when only high-value (45%, p < 0.001) or moderate-value (39.5%, p < 0.01) resources were available.

That generosity with the multiple resource endowment was higher than generosity with the single resource, moderate-value endowment (and likewise, that giving anything at all was more likely in the former decision, compared to the latter) is particularly important. This is because the total dollar value of these endowments, and thus participants’ wealth in these two decision rounds, was identical: giving (keeping) the entire 10-token endowment in these two decisions entailed giving away (keeping) the exact same dollar amount (75 cents in total; see Table 1). Why was giving enhanced when multiple, differentially-valued resources were available to give? The answer appears in analyses considering the type of resource people chose to give when multiple resources were available. After all, there were more options than simply giving all, or keeping all, of the endowment: a decision-maker could also give a portion of the multiple resource endowment and, in doing so, could distinguish between two types of resources, one that was relatively higher-value and one that was relatively lower-value, when deciding how much to give.

First consider giving patterns with the high-value portion of the multiple resource endowment. People were significantly less generous with the high-value portion of the resources compared to the low-value portion (B = −0.14, p < 0.001, see also Table 2, Model 3). The proportion of high-value resources given was similar regardless of whether low-value resources were also available to give or not (B = −0.01, p = 0.78). And high-value giving in the multiple resources decision also did not differ from moderate-value resource giving (B = −0.03, p = 0.12). Enhanced giving in the multiple resources decision, then, was not driven by higher giving levels of the high-value portion of the endowment.

Instead, high overall giving patterns in the multiple resource decision were driven by high levels of generosity with the low-value resources: as already noted, people were more generous, within the multiple resources endowment, with the low-value portion compared to the high-value portion. They were also more generous with the low-value portion of the multiple resource endowment compared to the single resource endowments of both high-value resources (B = 0.14, p < 0.001) and moderate-value resources (B = 0.11, p < 0.001). Intriguingly, people were even more generous with their low-value resources when multiple resources were available to give (i.e. when high-value resources were also present), compared to when only low-value resources were available to give (i.e. when high-value resources were not present, B = 0.06, p < 0.01, see also Table 2, Model 3).

To probe further the relationship between multiple resource availability, the value of the resource, and giving, follow-up models displayed in Table 3 compare the proportion of low- versus high-value resources given when both resource types versus a single resource type were available to give. They show that while giving was higher, on the whole, with low-value resources (compared to high-value resources, see the main effect of low-value resources in Model 1 and Model 2, ps < 0.001), the difference was particularly pronounced when multiple resources were available to give (the interaction effect in Model 2, p < 0.01). That is, people were especially generous with their low-value resources in the presence of a high-value resource. Figure 2 shows the interaction effect graphically.

Table 3.

Models predicting proportion of resources given, by multiple resource availability and resource type.

	Model 1		Model 2
	B (SE)	Bootstrapped 95%CI	B (SE)	Bootstrapped 95%CI
Giving decision involved multiple resources (MR)^a	0.03 (0.01)⁺	−0.03, 0.06	−0.01 (0.02)	−0.05, 0.03
Giving decision was with low-value resources (LV)^b	0.11 (0.01)***	0.08, 0.14	0.08 (0.02)***	0.04, 0.12
MR × LV			0.06 (0.03)*	0.01, 0.12
Intercept	0.36 (0.03)***	0.31, 0.42	0.38 (0.03)***	0.32, 0.44
Variance components—variance (standard deviation)
Level 1	0.04 (0.21)		0.04 (0.21)
Level 2	0.14 (0.37)		0.14 (0.37)

Giving decision involved a single resource is the reference category.

Giving decision was with high-value resources is the reference category. (The moderate-value, single resource decision was omitted from these models.) Multilevel linear models (maximum likelihood estimation) with random intercepts for participants. N = 800 decisions nested in 200 participants. Supplementary models controlling for the order in which the decisions were made did not affect results.

***

p < 0.001. *p < 0.050. ⁺p < 0.100.

Figure 2.

The availability of multiple resources to give promotes low-value resource giving. Estimated proportion of resources given, as predicted from Model 2 in Table 2, in the single resource and multiple resource decisions, by resource type (low vs high value). Error bars denote the standard errors.

Again, ancillary analyses on the probability of giving nothing to the other (vs giving anything) supported the results on proportion of the resource shared. People were significantly less likely to give none of their low-value resources (generalized linear mixed model with random intercepts for participants, B = −1.15, p < 0.01, compared to their moderate- or high-value resources), and this was particularly the case in the multiple resource decision (multiple resource availability × low-value resource decision interaction, B = −1.56, p < 0.01). Thus people were especially less likely to be maximally selfish with the low-value portion of the endowment when multiple resources were available to give.

Critically, these results also show that though multiple resource availability promoted giving of low-value resources, there was not an equivalent decrease in high-value resource giving. After all, one possibility for the increased giving of low-value resources when multiple resources were available was that people wanted to keep high-value resources for themselves, and used the low-value resources as a “substitute” in order to engage in less costly acts of generosity. But this was not observed in the pattern of results here: the difference in the proportion of high-value resources shared in the single versus multiple resource conditions was not significant (B = −0.01, p = 0.79, Table 3, Model 2, see also Figure 2). Why multiple resource availability promoted low-value resource giving and did not result in an equivalent reduction of high-value resource giving is an intriguing area for future exploration. More generally, that the availability of multiple, differentially-valued resources to help a dependent other prompts generosity with low-value (but not high-value) resources has important implications for how individuals, groups, or charitable organizations in need might request help from others to boost giving. These issues are considered in the next section.

Discussion

People often possess multiple, differentially-valued resources they can use to benefit others in need. Perhaps as a result, those in need of help—from individuals seeking aid during a time of crisis to larger-scale charitable organizations aiming to cure a disease—commonly request or accept assistance via multiple modes. Someone wanting to help a sick neighbor could organize a “meal train” on her behalf, contribute to a crowdfund for her medical costs, and/or drive her to a doctor appointment. A community member might spend his Saturday volunteering at a local homeless shelter and/or post a flyer about the shelter’s call for volunteers on his neighborhood bulletin board. The Ice Bucket Challenge for ALS research that went viral on social media in the mid-2010s provides another real-life instructive example: participants in the Challenge posted videos of themselves dumping buckets of icy water on their heads and challenging a friend to either do the same, donate money, or both with the goal of bringing awareness to ALS and fundraising for ALS research (Sohn, 2017; Van der Linden, 2017). In this way, participants could help by raising awareness or money, two different resource types.

This project sought to understand what resources people use to benefit others when they have different-valued resources at their disposal, and how much they give. Generosity was higher—both in terms of proportion of the endowment shared and the likelihood of sharing anything at all—with relatively low-value resources, while higher-value resources prompted more selfish behavior. Still further, when a mix of resource types (some low-value and some high-value) were available, people were more generous than when a single resource was available, even when holding constant the total value of the endowment. Subsequent results demonstrated that this was because multiple resource availability allowed people to distinguish between resource types to share, and they were particularly willing to share their relatively lower-value (but not their higher-value) resources. The Ice Bucket Challenge example remains instructive: the viral challenge successfully raised significant funding for ALS research, though only about 20% of those who made Ice Bucket Challenge social media videos actually donated to the cause (Moore, 2014; Sohn, 2017). Given the availability of multiple resources that could be used to help—donating money or raising awareness on one’s social media—many more people opted to post a video (a lower-value method of helping) than to donate money (a higher-value method).

Importantly, while multiple resource availability promoted giving via generosity with low-value resources, it also did not significantly suppress high-value giving below giving patterns when only high-value resources were available. That one (less valuable) form of helping did not “crowd out” generosity with the other (high-value) resource is important. After all, the results from this study may at first glance call into question whether those in critical need of a relatively higher-value resource in real life (e.g. blood donation during a shortage caused by a global pandemic) should even offer relatively lower-value methods of helping (e.g. posting a downloadable image, urging others to give blood, to use as one’s social media profile photo): why offer multiple ways of helping if doing so will primarily promote helping via a resource the recipient needs less? But instead the results suggest that offering this lower-value method of helping will not simultaneously reduce high-value giving, and will actually yield greater benefits on the whole. That low-value and high-value helping did not crowd each other out in this study is in line with other past work, which has found that formal volunteering versus informal helping (Burr et al., 2005; Lee and Brudney, 2012; Plagnol and Huppert, 2010; Wilson and Musick, 1997), volunteering versus donating (Bryant et al., 2003; Freeman, 1997; Hartmann and Werding, 2012; Schervish and Havens, 1997), and “sending thoughts and prayers” versus donations to natural disaster victims (Thunström, 2020) are not substitutes for each other but are instead positively related.

Understanding the factors driving generous and prosocial behavior is a critical task for the social sciences (Simpson and Willer, 2015). The vast majority of past work has examined helping behavior with a single valuable resource; what happens in the presence of multiple resources that can be used to help has remained underexplored in the literature. Importantly, this study demonstrated differential giving patterns with low-value resources, especially when multiple resource types were available to give. But the potential mechanisms behind this finding, as well as critical moderators, remain future avenues for work in this area.

One important potential mediating factor is personal wealth. In the experiment described here, participants were aware at the beginning that there were multiple, differently-valued resources in the study. Thus when faced with an endowment containing low-value tokens (as they were in the single resource, low-value decision or the multiple resources decision), they were fully aware that they were relatively less wealthy than in the decision where only high-value tokens were present. The experience of having more or less wealth would have fluctuated across the decision-conditions and thus final generous decision-making may have been impacted by initial personal wealth in the round, rather than resource type alone. Indeed, past work suggests that lower socioeconomic status is associated with more prosocial behavior, compared to those who are more well-off (Piff et al., 2010, 2016). Note, however, that the results from this experiment also demonstrated that giving was higher when multiple resources were available compared to when a single, moderately-valued resource was available; in these two conditions total personal wealth was held constant. Nevertheless, how the experience of personal wealth impacts generosity when benefactors possess multiple, differentially-valued resources remains an important area for future work.

Next, note that in the study described here, giving behavior was completely anonymous. But many of the real-world helping scenarios described above contain actions that are publicly visible—to the recipient, to third parties, or both. Onymity (Wang et al., 2017) and the availability of reputational rewards (Wu et al., 2016) are well-known factors in promoting other-regarding behavior. Introducing identifiability, or the prospect for receiving reputational benefits, in a study like the one described here may not only cause increased giving levels overall—it may also promote giving especially of one’s higher value resources. These questions could be addressed in an extension of the study described here.

Additionally, in this project, resource value was equivalent to the giver and the receiver: high value resources were the same (high) value, and low value resources were the same (low) value. But in the real world, resource value is commonly not this symmetrical. Future work could test whether people are especially generous with their low-value resources when those low-value resources are higher in value to the person in need, or whether the value of the resource from the perspective of the giver is more important than the value to the receiver in determining what benefits people are willing to give to others. Independently manipulating the value of the resource to both givers and recipients could yield alternative patterns that would better show when and why multiple resource availability is particularly likely to foster the distribution of beneficial resources to others.

In closing, this work contains some practical implications for charitable organizations or individuals in need. It may seem prudent to request only high-value resources when they are needed. But the results show that people are much more willing to part with their lower-value resources, especially when they have high-value resources at their disposal as well. The presence of relatively lower-value resources prompts higher levels of giving, including increasing the portion people who are willing to give anything at all. Still further, offering these low-cost methods of helping does not simultaneously reduce higher-value giving, yielding greater average benefits than when only one resource is requested. Would the Ice Bucket Challenge have received as many benefits if only monetary donations had been requested, rather than monetary donations, a social media video, or both? This work says no. Offering people more low-cost methods of helping alongside those higher-cost methods appears to allow people to consider how they should help, rather than whether they should, and helping is promoted. After all, something is better than nothing, and from the eyes of the beneficiary in need, every little bit of endowment-sharing helps.

Footnotes

Acknowledgements

I thank Bhavika Garg, Anna Greenleaf, Aspen Martin, Prayo Smitasin, and Angela Wu for valuable research assistance.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The author acknowledges financial support from The Department of Sociology at Duke University.

ORCID iD

Ashley Harrell

References

Andreoni

Vesterlund

(2001) Which is the fair sex? Gender differences in altruism. The Quarterly Journal of Economics 116(1): 293–312.

Blake

Rand

(2010) Currency value moderates equity preference among young children. Evolution and Human Behavior 31(3): 210–218.

Bode

Miller

O’Gorman

, et al. (2015) Increased costs reduce reciprocal helping behaviour of humans in a virtual evacuation experiment. Scientific Reports 5: 15896.

Borgonovi

(2008) Divided we stand, united we fall: Religious pluralism, giving, and volunteering. American Sociological Review 73(1): 105–128.

Brooks

(2005) Gifts of Time and Money: The Role of Charity in America’s Communities. Lanham, MD: Rowman & Littlefield.

Bryant

Jeon-Slaughter

Kang

, et al. (2003) Participation in philanthropic activities: Donating money and time. Journal of Consumer Policy 26(1): 43–73.

Buhrmester

Kwang

Gosling

(2016) Amazon’s mechanical Turk: A new source of inexpensive, yet high-quality data? Perspectives on Psychological Science 6(1): 3–5.

Burleigh

Kennedy

Clifford

(2018) How to screen out VPS and international respondents using Qualtrics: A protocol. Working paper. Available at: https://ssrn.com/abstract=3265459 (accessed 10 February 2021).

Burr

Choi

Mutchler

, et al. (2005) Caregiving and volunteering: Are private and public helping behaviors linked? The Journals of Gerontology Series B: Psychological Sciences and Social Sciences 60(5): S247–S256.

10.

Carpenter

Verhoogen

Burks

(2005) The effect of stakes in distribution experiments. Economics Letters 86(3): 393–398.

11.

Companion

(2010) Constriction in the variety of urban food pantry donations by private individuals. Journal of Urban Affairs 32(5): 633–646.

12.

Cox

Deck

(2006) When are women more generous than men? Economic Inquiry 44(4): 587–598.

13.

Diekmann

(2004) The power of reciprocity: Fairness, reciprocity, and stakes in variants of the dictator game. Journal of Conflict Resolution 48(4): 487–505.

14.

Duncan

(1999) Modeling charitable contributions of time and money. Journal of Public Economics 72(2): 213–242.

15.

Engel

(2011) Dictator games: A meta study. Experimental Economics 14(4): 583–610.

16.

Fehr

Fischbacher

(2003) The nature of human altruism. Nature 425(6960): 785.

17.

Ferguson

Zhao

O’Carroll

, et al. (2019) Costless and costly prosociality: Correspondence among personality traits, economic preferences, and real-world prosociality. Social Psychological and Personality Science 10(4): 461–471.

18.

Forsythe

Horowitz

Savin

, et al. (1994) Fairness in simple bargaining experiments. Games and Economic Behavior 6(3): 347–369.

19.

Freeman

(1997) Working for nothing: The supply of volunteer labor. Journal of Labor Economics 15(1, Part 2): S140–S166.

20.

Glanville

Paxton

Wang

(2016) Social capital and generosity: A multilevel analysis. Nonprofit and Voluntary Sector Quarterly 45(3): 526–547.

21.

Hartmann

Werding

(2012) Donating time or money: Are they substitutes or complements? CESifo Working Paper No. 3835, Available at: https://ssrn.com/abstract=2084100 (accessed 10 February 2021).

22.

Hill

Thompson-Hayes

(2017) From Awareness to Commitment in Public Health Campaigns: The Awareness Myth. Lanham, MD: Rowman & Littlefield.

23.

Kennedy

Clifford

Burleigh

, et al. (2018) The shape of and solutions to the MTurk quality crisis. Political Science Research and Methods 8(4): 614–629.

24.

Kristofferson

White

Peloza

(2013) The nature of slacktivism: How the social observability of an initial act of token support affects subsequent prosocial action. Journal of Consumer Research 40(6): 1149–1166.

25.

Laitala

(2014) Consumers’ clothing disposal behaviour–a synthesis of research results. International Journal of Consumer Studies 38(5): 444–457.

26.

Lane

Dal Cin

(2018) Sharing beyond Slacktivism: The effect of socially observable prosocial media sharing on subsequent offline helping behavior. Information, Communication & Society 21(11): 1523–1540.

27.

Larney

Rotella

Barclay

(2019) Stake size effects in ultimatum game and dictator game offers: A meta-analysis. Organizational Behavior and Human Decision Processes 151: 61–72.

28.

Lee

Piliavin

Call

(1999) Giving time, money, and blood: Similarities and differences. Social Psychology Quarterly 62(3): 276–290.

29.

Lee

Brudney

(2012) Participation in formal and informal volunteering: Implications for volunteer recruitment. Nonprofit Management and Leadership 23(2): 159–180.

30.

Lengfeld

Ordemann

(2016) The long shadow of occupation: Volunteering in retirement. Rationality and Society 28(1): 3–23.

31.

List

Cherry

(2008) Examining the role of fairness in high stakes allocation decisions. Journal of Economic Behavior & Organization 65(1): 1–8.

32.

Manatschal

Freitag

(2014) Reciprocity and volunteering. Rationality and Society 26(2): 208-235.

33.

Molnar

Duffy

Claxton

, et al. (2001) Private food assistance in a small metropolitan area: Urban resources and rural needs. Journal of Sociology and Social Welfare 28(3): 187.

34.

Moore

(2014) The ice bucket challenge data debrief: Who dumped, who donated, and was it all worth it? RJMetrics TheDataPoint. Available at: https://blog.rjmetrics.com/2014/09/03/the-ice-bucket-challenge-data-debrief-who-dumped-who-donated-and-was-it-all-worth-it/ (accessed 10 February 2021).

35.

Moore

(2008) Ribbon Culture: Charity, Compassion and Public Awareness. London: Palgrave Macmillan.

36.

Novakova

Flegr

(2013) How much is our fairness worth? The effect of raising stakes on offers by proposers and minimum acceptable offers in dictator and ultimatum games. PLoS One 8(4): e60966.

37.

Oliver

(1984) “If you don’t do it, nobody else will”: Active and token contributors to local collective action. American Sociological Review 49(5): 601–610.

38.

Park

Cho

Johnson

, et al. (2017) Use of behavioral reasoning theory to examine the role of social responsibility in attitudes toward apparel donation. International Journal of Consumer Studies 41(3): 333–339.

39.

Piff

Kraus

Côté

, et al. (2010) Having less, giving more: The influence of social class on prosocial behavior. Journal of Personality and Social Psychology 99(5): 771–784.

40.

Piff

Stancato

Horberg

(2016) Wealth and wrongdoing: Social class differences in ethical reasoning and behavior. In: van Prooijen

van Lange

PAM

(eds) Cheating, Corruption, and Concealment: The Roots of Dishonesty. Cambridge: Cambridge University Press, pp.185–207.

41.

Plagnol

Huppert

(2010) Happy to help? Exploring the factors associated with variations in rates of volunteering across Europe. Social Indicators Research 97(2): 157–176.

42.

Rabin

(1993) Incorporating fairness into game theory and economics. The American Economic Review 83(5): 1281–1302.

43.

Raghubir

Srivastava

(2008) Monopoly money: The effect of payment coupling and form on spending behavior. Journal of Experimental Psychology: Applied 14(3): 213–225.

44.

Raghubir

Srivastava

(2009) The denomination effect. Journal of Consumer Research 36(4): 701–713.

45.

Raihani

Mace

Lamba

(2013) The effect of $1, $5 and $10 stakes in an online dictator game. PLoS One 8(8): e73131.

46.

Rezaei Arangdad

Thoney-Barletta

Joines

, et al. (2019) Influence of demographics and motivational factors on US consumer clothing and shoes disposal behavior. Research Journal of Textile and Apparel 23(3): 170–188.

47.

Schaefer

(2009) Resource variation and the development of cohesion in exchange networks. American Sociological Review 74(4): 551–572.

48.

Schervish

Havens

(1997) Social participation and charitable giving: A multivariate analysis. Voluntas: International Journal of Voluntary and Nonprofit Organizations 8(3): 235–260.

49.

Shank

(2016) Using crowdsourcing websites for sociological research: The case of Amazon Mechanical Turk. The American Sociologist 47(1): 47–55.

50.

Simpson

Willer

(2015) Beyond altruism: Sociological foundations of cooperation and prosocial behavior. Annual Review of Sociology 41: 43–63.

51.

Smith

Davidson

(2014) The Paradox of Generosity: Giving We Receive, Grasping We Lose. New York, NY: Oxford University Press.

52.

Snijders

Bosker

(2011) Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling. London: Sage Publications.

53.

Sohn

(2017) The ice bucket challenge delivers. Nature 550(7676): S113–S114.

54.

Stewart-Williams

(2007) Altruism among kin vs. nonkin: Effects of cost of help and reciprocal exchange. Evolution and Human Behavior 28(3): 193–198.

55.

Thunström

(2020) Thoughts and prayers–Do they crowd out charity donations? Journal of Risk and Uncertainty 60:1–28.

56.

Van der Linden

(2017) The nature of viral altruism and how to make it stick. Nature Human Behaviour 1(0041): 1–3.

57.

Wallace

Buil

De Chernatony

(2017) When does “liking” a charity lead to donation behaviour? Exploring conspicuous donation behaviour on social media platforms. European Journal of Marketing 51(11/12): 2002–2029.

58.

Wang

Jusup

Wang

R-W

, et al. (2017) Onymity promotes cooperation in social dilemma experiments. Science Advances 3(3): e1601444.

59.

Weinberg

Freese

McElhattan

(2014) Comparing data characteristics and results of an online factorial survey between a population-based and a crowdsource-recruited sample. Sociological Science 1: 292–310.

60.

Wiepking

Maas

(2009) Resources that make you generous: Effects of social and human resources on charitable giving. Social Forces 87(4): 1973–1995.

61.

Wilson

Musick

(1997) Who cares? Toward an integrated theory of volunteer work. American Sociological Review 62(5): 694–713.

62.

Balliet

Van Lange

(2016) Reputation, gossip, and human cooperation. Social and Personality Psychology Compass 10(6): 350–364.

63.

Zak

Stanton

Ahmadi

(2007) Oxytocin increases generosity in humans. PLoS One 2(11): e1128.