Introduction

The methods used in the typical social psychology study have substantially changed in recent years. Studies published in top journals in the field are more frequently conducted online, are shorter in duration, and rely more heavily on self-report measures as compared to studies published in these same journals 15 years ago [1, 2]. This shift towards online, low-cost, high-volume methods is believed to have resulted from a combination of factors including calls for greater statistical power and replicability in the field [2], incentives for publication and competition over limited resources in academia [1, 3], and a prioritization of internal over external validity in decisions about what sorts of research to publish [4]. The trend toward online studies in particular has likely been accelerated by the heightened need to conduct research remotely brought on by the COVID-19 pandemic. Meanwhile, more effortful field research involving the study of cognition and behavior in naturalistic settings [e.g., 5] has become increasingly scarce in psychology, creating concerns about whether results obtained in the lab or online would generalize to real-world settings [3, 4, 6–12]; in other words, whether these findings are ecologically valid in the sense that this term has recently been used in experimental psychology [9, 13–16]. These competing concerns and practical constraints create a clear need for realistic behavioral measures and manipulations that would facilitate higher levels of ecological validity without compromising ease of implementation or internal validity.

Perhaps nowhere has this need been felt more acutely than in the field of moral psychology, which has traditionally relied heavily on vignettes, questionnaires, games, and thought experiments such as the trolley problem as opposed to the observation of moral behavior itself [e.g., 17–26]. These approaches are often highly appropriate, such as when researchers are primarily interested in unobservable cognitive phenomena (e.g., judgments of right and wrong) or when ethical concerns preclude the naturalistic study of the behavior of interest (e.g., decisions in life-or-death moral dilemmas) [22, 27]. Yet a growing number of studies make clear that procedures relying solely on hypothetical situations and self-reports about behavior often fail to predict moral behavior in more realistic settings. Clear gaps between intentions and behavior have been shown in domains including responses to racism and sexual harassment [28, 29], the desire for retributive punishment [30], willingness to forego personal gain to prevent harm to others [31], interventions in moral violations such as the theft of personal property [32], judgments in sacrificial harm dilemmas like the trolley problem [33], compliance with unethical requests made by authority figures [34], and of particular relevance to the current research, monetary donations [35]. Thus, what participants say with regard to morality is sometimes a poor guide to what they actually do, suggesting that moral psychologists in particular should be attentive to the extent to which the experimental procedures they use are representative of the naturalistic settings in which the behaviors of interest unfold [36, 37].

One of the most heavily studied forms of moral behavior in psychology and behavioral economics is the allocation of resources in the form of charitable donations, no doubt due to the ubiquity and importance of this particular form of prosociality. Indeed, American citizens alone currently donate over 400 billion dollars to charity each year [38]. Researchers have operationalized charitable donations by a variety of means, including the use of preexisting data sets on real-world giving [e.g., 39, 40], reports about past donation behavior [41], self-reported intentions or attitudes related to donating [42, 43], economic games [44], and actual donations made in laboratory or field studies [45, 46]. Decisions about which of these measures to use often involve tradeoffs between various methodological concerns. For example, measuring hypothetical donations in a lab setting allows for greater ease of use and internal validity but may be lacking in the real stakes and rich contextual details present in real-world donation decisions [47]. On the other hand, exploring real-world donations might mean relying on preexisting data sets that preclude causal inference or field studies where internal validity and ease of data collection are more variable.

Researchers in psychology and economics have been aware of these issues for some time [4, 6, 48, 49] and the “gold standard” approach for addressing them has been the field experiment, in which an experimental manipulation is embedded directly into a real-world charitable appeal, reconciling concerns between internal and external validity [e.g., 50–52]. Such studies remain relatively rare in social psychology, however, where self-report and hypothetical measures are still the norm [6, 7, 53]. The availability of a wider variety of ecologically valid measures of charitable giving–particularly those bridging methods psychologists are comfortable and familiar with at this point (e.g., online survey studies) [2] with those typically employed in field experiments–may thus be of value to psychology, a field in which concerns about the generalizability of findings to real-world settings is widespread [3, 4, 6, 7, 12].

To this end, the current research introduces a method which reflects the expanding application of technology to helping others in need: the use of online platforms originally designed for the creation of gift registries to allow participants in the lab or online to donate needed items to real-world charitable organizations, which we created and refer to here as a relief registry. Whereas charitable donations are highly common [38], charitable giving can often be inefficient and wasteful due to factors such as a lack of information about the available donation options and their relative efficacy [54, 55], and the personal preferences and social biases that influence donation decisions [56, 57]. As a result, well-intentioned donors may nonetheless end up donating items that are not actually needed [58, 59], as has been the case in the aftermath of recent disasters including Hurricane Sandy [60, 61] and the 2004 Indian Ocean tsunami [62]. Relief registries resolve these issues by connecting charitable organizations directly with donors, providing an easy-to-use mechanism by which high-priority donation items can be shipped directly to the places and people in the greatest need, and in the current context, allowing psychological research to have immediate, real-world impact. Indeed, similar relief registries to those used in the current research were successfully implemented in the aftermath of Hurricane Sandy [60–62].

These relief registries possess a number of desirable properties: They can be easily implemented in the laboratory or online, including alongside experimental manipulations and sampling procedures allowing for high levels of internal validity, while retaining the essential features of realistic donations in terms of overt characteristics (as charitable giving increasingly occurs online) [63] and the realistic psychological stakes involved, as participants are asked to give up real resources with immediate, tangible prosocial consequences. In this way, we see relief registries as overlapping with and complementary to standard field experimental approaches to charitable giving while modifying and building off of these approaches in several ways which may make them more useful or accessible within certain research contexts. First, relief registries were created to be relatively easy to incorporate into the sorts of studies which have become increasingly common in psychology (i.e., survey studies conducted online [2]) and thus, may be more readily accessible for many researchers as compared to “true” field experiments (i.e., experimental manipulations carried out in naturalistic settings) which often require a greater amount of effort, planning, and coordination to successfully carry out [3]. Thus, relief registries offer a possible alternative for researchers wishing to investigate charitable giving online rather than in person and in this way, complement similar research in experimental economics which has investigated naturalistic giving in online samples (e.g., [64]). The fact that relief registries operate via the exchange of needed goods rather than monetary donations also sets them apart from much preexisting research [e.g., 46, 65, 66], broadening the sorts of experimental designs in which they can be used (e.g., the two original experiments described below, which rely on the donation of specific items rather than money). Further, the clear link between participants’ actions (i.e., the selection of donation items in the relief registry) and prosocial outcomes (i.e., these same items being sent directly to charities) may yield greater psychological benefits for participants as compared to studies relying on monetary donations, which involve further intermediary decisions about how to spend said money occurring outside of participants’ awareness. In this way, the current research complements a growing literature within experimental economics focused on how framing donation requests in terms of tangible goods vs. money impacts total donation amounts, individuals’ propensity to give, and their motivations when giving (e.g., [64, 67, 68]). For these reasons, we see relief registries as a useful addition to the methodological toolbox available to those wishing to heighten the ecological validity of their research while acknowledging that standard field experimental approaches may still be preferable in many cases.

To demonstrate the use of relief registries in action, we coordinated efforts with non-profit organizations doing a variety of charitable work in the Greater Boston area, asking these organizations to provide a list of objects and items that they needed to best serve their community and help people in need. We then uploaded these objects onto an adapted wedding registry site (myregistry.com), used images of these objects as stimuli and provided subjects the opportunity to donate these items using the registry. Donated items could then be directly shipped to the places and people that need them the most. We conducted two preliminary experiments demonstrating the use of relief registries to test the external validity of a recent finding from our labs that people are more likely to help others in need after they have imagined themselves doing so [20, 69] and that the vividness and detail of imagined scenes partially mediates this relationship [70]. Whereas prior research testing these hypotheses has relied primarily on a measure of prosocial intention (i.e., participants’ self-reported willingness to help a target individual in need) [20, 71] or donations to hypothetical individuals [70], we tested whether imagination would exert similar effects on people’s actual donation behaviors when using relief registries. Specifically, we tested whether people would be more likely to donate items they had imagined themselves using to help others as compared to conditions controlling for exposure to donation items and descriptions of others in need (Study 1), and whether manipulating the strength of the visuospatial context (via the familiarity of the location where imagined events took place) causally affected the strength of this relationship (Study 2).

Study 1

Study 2

General discussion

The current research introduces relief registries–online gift registries adapted in collaboration with charitable organizations–as a measure of prosocial behavior for use in psychological and behavioral economic research. Relief registries build upon prior work exploring real-world donation behavior using correlational [75] and field experimental methods [50–52], offering researchers an easy to use method for studying charitable donations in a way that is easily adaptable to a variety of research designs and strives to balance concerns about internal validity, ecological validity, and the ability to recruit large samples of participants in the lab or online.

To demonstrate this method, we explored whether a prior finding from our labs–that imagination can be leveraged as a tool to foster intentions to help others [20, 69–72]–would generalize to prosocial behavior in the form of actual donations of items participants had imagined using to help others. In other words, relief registries allowed us to directly probe whether the intention-behavior gap found in some moral psychological research [28–35] would reemerge in the context of a specific phenomenon from our own research. Consistent with previous work on imagination, we found effects of episodic simulation on both self-reported prosocial intentions as well as donation behaviors during the relief registry task; people who had imagined themselves using specific items to help targets in need expressed greater willingness to help those targets but were also more likely to actually give up their own money to donate those items in a subsequent task. Whereas willingness to help did not significantly predict donations on a trial-level basis, this may be due in part to the mismatch between the referent of the hypothetical and behavioral measures (i.e., the fact that the hypothetical measure referred to one’s willingness to help someone in an imagined scenario rather than willingness to donate items). Whereas we would once again remind readers not to overinterpret these effects due to the small sample sizes used here, we reinforce that these findings add to mounting evidence for a relationship between memory and imagination and prosocial intentions, emotions (e.g., empathy), and behavior [20, 69–72, 76–79].

Whether tradeoffs between internal and external validity are an inherent feature of research design in the social sciences has been the subject of some debate [80–83]. Although we by no means intend to resolve this dispute, relief registries allow for the direct observation of prosocial behavior in a setting which involves real stakes [6, 7] and has real-world impact all without sacrificing the level of internal validity required for causal inference. In this way, we see relief registries as complementary to related field experimental approaches to charitable giving from psychology and behavioral economics [e.g., 50–52] while being potentially more easily accessible and useful for different types of experimental designs than these existing approaches. For example, relief registries may be preferable over standard field experiments measuring monetary donations [e.g., 46, 65, 66] when researchers wish to allow participants to freely choose and donate items according to their own preferences (e.g., [84]) and may also work better than existing alternatives for certain types of research designs–for example, the studies reported above, which required participants to imagine themselves using specific items to help others before deciding whether they would like to donate these items. Relief registries are thus a useful addition to researchers’ methodological toolbox, particularly in light of recent debates about the extent to which self-reported attitudes and intentions predict actual behaviors [3, 4, 6, 7].

We acknowledge that relief registries still possess some of the limitations present in other lab-based work (e.g., possible demand characteristics or the fact that participants only just received the money they are donating, potentially leading to less careful decisions) [85] and this is one of the reasons we would stress that relief registries are intended to supplement rather than replace complementary approaches such as field experiments. Specific elements of relief registries could also be easily adapted to suit different researchers’ needs and address potential shortcomings in the way we utilized this method here. For example, we matched all donation items at the low price of $0.50 to avoid floor effects in giving and ensure that participants’ choices would not be influenced by differences in item costs; however, researchers may wish to use prices which better reflect the actual costs of donation items in order to heighten the mundane realism of this procedure and avoid the potential confound of participants’ choices being influenced by their perceptions of the value of the items (e.g., selecting more expensive items because it would benefit the charities more while costing the participant the same amount of money). At the same time, the bonus compensation provided to participants for use in the relief registry task itself ($20 and $10 in Studies 1 and 2, respectively) may be unrealistically high for researchers working with limited budgets, particularly those wishing to implement relief registries in larger samples of participants. However, prior research is optimistic with regard to the comparability of findings when using larger versus smaller-stakes donations [86, 87], suggesting that the validity of the relief registry approach would be maintained even when using smaller compensation amounts. Ultimately, though we have demonstrated the use of relief registries in just one format and one context here, we view the potential applications of this procedure as quite broad and encourage future researchers to explore ways to amend, adapt, and improve upon it in their own research.

The methods and results from our two experiments contribute to an important ongoing conversation about ecological validity in the fields of psychology and behavioral economics [3, 6–8, 11, 37], and moral psychology more specifically [9, 12, 31, 88]. As the field of psychology began to initially take shape in the early 20^th century, it was not uncommon for researchers to build new apparatuses for studying sensation, perception, and memory. Over a century later, it might behoove moral psychologists to return to our tinkering experimental roots and develop new tools for measuring and manipulating behavior, and in turn expand the ecological scope and ultimate impact of our lab-based work. Whereas relief registries are just one example of this process, we would argue that similar logic and strategies to those utilized here could be applied to the creation of other manipulations and measures in moral psychology and related fields. The theoretical rubber of moral psychology is beginning to hit the road of everyday life, and it seems to us an exciting frontier bridging traditional, in-lab manipulations with impact in the real-world lies ahead.

Supporting information