ASKING FOR HELP OR SUPPORT?
SEMANTIC EFFECTS OF WORDING IN PROMOTIONAL CHARITY MESSAGES ON PEOPLE’S INTENTION TO DONATE
BY
NICOLE VERGIN
Completion Date:
June 15, 2020
ASKING FOR HELP OR SUPPORT?
SEMANTIC EFFECTS OF WORDING IN PROMOTIONAL CHARITY MESSAGES ON PEOPLE’S INTENTION TO DONATE
BY
NICOLE VERGIN
Faculty of Economics and Business Department of Marketing
Master Thesis - Marketing Management
Completion Date:
June 15, 2020
Am Sandfeld 20 41564 Kaarst, Germany
+491711583487 n.vergin@student.rug.nl Student Number: S3854388 1st Supervisor: Dr. Marijke Leliveld
2nd Supervisor: Dr. Judith de Groot University of Groningen
ABSTRACT
Semantic Effects of Wording in Promotional Charity Messages on People’s Intention to Donate
In advertising appeals, charities often use the words “help” and “support” interchangeably to encourage donations. This experimental study (N = 453) investigates how these two words can be effectively used in promotional messages to improve people’s attitude towards the charity and increase the donation intention. The findings indicate that “help” compared to
“support” creates a significantly more positive attitude and a higher donation intention.
However, this is only the case when the campaign refers to a local cause, meaning the proximity between the donor and the recipient is high. This effect is mediated by the higher perceived efficacy (and not by the perceived effort) associated with “help” as opposed to
“support”. In contrast, no significant differences in efficacy, attitude towards the charity and donation intention occur between “help” and “support” when the cause is distant (i.e., low proximity), or when no information regarding proximity is provided. Besides efficacy, the exploratory analysis reveals a second potential important mediator between the type of wording and the attitude/donation intention for high and low proximity levels: the perceived importance of the cause. To conclude, the findings have important implications for managers of charities to promote individual donations and improve the wellbeing of society.
Keywords: semantics, linguistics, wording, help, support, charitable giving, philanthropy, non-profit organizations, charities, donations
Research Theme: Philanthropy and charitable giving Seminar Supervisor: Dr. Marijke Leliveld
PREFACE
As the final deliverable of my master’s degree in Marketing Management at the University of Groningen, I have completed my master thesis in the field of charitable giving that I present to you on the following pages. On the one hand, acting in a prosocial manner has always been very important to me. Nowadays, climate change and other natural catastrophes have detrimental consequences for millions of people. On the other hand, I show a strong desire and passion for the creative and visual side of marketing. For these reasons, I decided to combine both interests by discovering how marketing messages could best be designed to increase donations and correspondingly improve the welfare of society. While observing many different donation campaigns, I noticed that charities regularly use the words “help”
and “support” interchangeably. However, the role of wording has not received much attention in this field from a research side, yet. Thus, my curiosity was triggered, and I was interested in finding out if such a single small word could make a difference concerning people’s donation behavior.
During the entire journey, several people strongly supported and accompanied me. First, I would like to express my gratitude to my supervisor Dr. Marijke Leliveld. She always had an open ear for my questions and provided detailed feedback, which enabled me to extend my research skills and to develop myself continuously. Further, I would like to thank my second supervisor Dr. Judith de Groot for taking the time to read my report and for delivering constructive feedback to gain a broader perspective. Generally, the University of Groningen taught me that not only the economic outlook is relevant when implementing marketing concepts and theories, but what counts is the power it has on influencing people to do better. I am passionate about convincing others to help create a better environment and world that we live in. As one would say: we all have the opportunity to make an impact.
TABLE OF CONTENTS
ABSTRACT ... 3
SEMANTIC EFFECTS OF WORDING IN PROMOTIONAL CHARITY MESSAGES ON PEOPLE’S INTENTION TO DONATE ... 3
PREFACE ... 4
TABLEOFCONTENTS ... 5
INTRODUCTION... 6
THEORETICALBACKGROUND ... 9
WHY WORDING MATTERS IN CHARITABLE GIVING ... 9
POTENTIAL UNDERLYING DIFFERENCE BETWEEN “HELP” AND “SUPPORT” ... 10
THE MEDIATING ROLE OF DONATION EFFICACY ON CHARITABLE GIVING ... 12
THE MEDIATING ROLE OF EFFORT ON CHARITABLE GIVING ... 13
THE MODERATING ROLE OF DONATION PROXIMITY ON DONATION EFFICACY & EFFORT... 15
CONCEPTUAL MODEL ... 17
METHODOLOGY ... 18
PARTICIPANTS AND DESIGN ... 18
STUDY PROCEDURE ... 19
MANIPULATION AND MEASUREMENT OF VARIABLES ... 19
RESULTS ... 23
MANIPULATION CHECK ... 23
MAIN ANOVA ANALYSES ... 23
MODERATED MEDIATION ANALYSES ... 27
EXPLORATORY ANALYSES ... 31
DISCUSSION ... 35
THEORETICAL IMPLICATIONS ... 35
LIMITATIONS AND FUTURE RESEARCH ... 38
MANAGERIAL IMPLICATIONS ... 41
FINAL CONCLUSION ... 41
REFERENCES ... 42
APPENDICES ... 51
APPENDIX 1 - PRETEST ... 51
APPENDIX 2 - MANIPULATION OF CHARITY MESSAGES ... 55
APPENDIX 3 - CRONBACH’S ALPHA FOR ALL VARIABLES ... 56
APPENDIX 4 - SURVEY DESIGN ... 57
APPENDIX 5 - RESULTS FOR SECOND SCALE OF INTENTION ... 62
INTRODUCTION
Poverty, hunger, diseases, war, and natural disasters - These are crucial challenges that the world faces every day, causing death and suffering among the population. Charities are very important parties, attending to these serious developments with an attempt to create a difference in people’s life (Leliveld & Risselada, 2017). Their social, economic, and political responsibility is key to benefiting the society as a whole, now and in the future (Flack, 2007;
Ware, 1989).
To prevent death and suffering by providing value to societies, charities strongly depend on people’s willingness to donate to a supported cause (Leliveld & Risselada, 2017). In 2018, Americans donated a total of $427.71 billion, which is equivalent to 2.1% of the gross domestic product (GDP) (Giving USA, 2019). Although this number is relatively large, charities have to deal with highly competitive environments, whereas charitable fundraising has become much harder (Das, Kerkhof & Kuiper, 2008; Sudhir, Roy & Cherian, 2016). As such, this underlines the importance for charities to rely on persuasive advertising messages that encourage individuals to help or support others (Das et al., 2008; Goering, Connor, Nagelhout & Steinberg, 2011; Sudhir et al., 2016).
An important question is thus how promotional communication drives people’s intention to give to charities. So far, research has developed a variety of factors in advertising messages that have an impact on donation behavior. To those belong the number of victims shown (Kogut & Ritov, 2005), the proximity of the donor to the victim (donation proximity) (Bar- Tal, 1976), message framing (e.g., Chang & Lee, 2010; Das et al., 2008; Levin & Gaeth, 1988), price effects (Karlan & List, 2007), or social pressure (DellaVigna, List &
Malmendier, 2012; Shang & Croson, 2006) for example. Furthermore, the effect of deadlines (Damgaard & Gravert, 2017) and donation descriptions (Das, Guha, Biswas & Krishnan, 2016) is also of relevance. However, there is no research on whether using a request for
“help” or a request for “support” works best. This is surprising as charity appeals frequently use both words. For example, the two specific charity campaigns of UNICEF either show
“help” or “support” (see Figure 1). These two words are often presented interchangeably in messages, but do they really provoke the same effect? Or could this little difference in wording between “help” and “support” change people’s willingness to donate and ultimately increase the funds of charities to do good?
FIGURE 1
UNICEF’s Charity Campaigns “Help” & “Support”
We know from linguistics that words can matter, depending on how we interpret them and how they affect emotions (Lindquist, Gendron, Satpute, Barrett, Lewis & Haviland-Jones, 2016; Lindquist, MacCormack & Shablack, 2015; Lupyan & Ward, 2013). On the one hand, the mere presence of a single word influences the degree to which we consciously detect a visual stimulus (Lupyan & Ward, 2013). On the other hand, even neutral words are often linked to different emotional words, evoking specific feelings in people’s minds (Barrett &
Lindquist, 2008). Thus, a single word can impact either people’s perceptions or feelings and their behavior.
While considerable importance is attributed to word choice in other domains (e.g., Lindquist et al., 2016), hardly any research has dealt with the influence of wording in charities’
advertisements on people’s intention to donate. Some researchers have outlined the importance of specific words like “small” (Cialdini, 2007), “caring”, or “helping” (Croson &
Shang, 2013) to influence people to give. However, no knowledge exists yet regarding the relative impact of very similar words and how or why one might work better than the other.
Thus, further research is needed to understand whether the insights on words from different domains can also be applied to the donation setting. This might have important implications for charities’ design of persuasion strategies.
Help
Support
Hence, the present study takes on the pioneering role in examining how charities can improve people’s attitude towards them and simultaneously increase their donation intention through the use of wording (“help” or “support”) in a promotional message. It also aims at touching upon how and why (i.e., the underlying processes) and in which context (i.e., the contextual circumstances) one word has a more significant effect than the other.
THEORETICAL BACKGROUND
Why Wording Matters in Charitable Giving
Charitable giving is a form of prosocial behavior that is defined as a wide range of activities accomplished by an individual whose focus lies on benefiting the society itself and less on oneself (Eisenberg, Losoya & Guthrie, 1997; Penner, Dovidio, Piliavin & Schroeder, 2005).
In order to analyze the act of donating, people’s attitude towards the charity and their donation intention serve as main behavioral variables.
Effective commercial messages are used by charities to positively influence an individual’s willingness to donate (e.g., Das et al., 2008; Goering et al., 2011; Sudhir et al., 2016). A communication element that receives considerable attention in donation literature is the framing of the message that impacts a specific behavior (e.g., Chang & Lee, 2010; Das et al., 2008; Grau & Folse, 2007). Framing refers to the type of words, images, and phrases that are used to portray a message to someone (Druckman, 2001). For example, people’s donation likelihood changes with different frames (Smith, 1987). More specifically, people are more likely to donate when being asked about help for “poor people” than for “people on welfare”
(Smith, 1987). While framing appeals to significant changes in the structure and content of words as well as phrases, wording often only requires a relatively minor modification like the replacement of one single term with another (Farrow, Grolleau & Mzoughi, 2018). Despite the vast amount of literature on framing, less is known about the effect of single words like
“help” or “support” within the context of charitable giving.
In order to theorize about the influence the two words might have, we turn to the literature on semantics (Cann, Kempson & Gregoromichelaki, 2009). Semantics relates to the sense and significance of language, including words, sentences, or other terms (Cann et al., 2009).
Research from different fields emphasizes the impact of semantics on people’s expectations and emotions (Barret & Lindquist, 2008; Lupyan & Ward, 2013). For example, reading a single word like “chair” can affect the degree to which we visually recognize and detect a chair in our environment and how we expect the world around us (Lupyan & Ward, 2013).
On the other hand, being faced with a neutral word like “mother” might create an instant feeling of love towards the mother as the word itself is associated with “love” (Barrett &
Lindquist, 2008). Expectations and emotions regarding behavior in turn strongly shape the behavior itself (Mayo, 2004; Schwarz, 2011). Thus, single words have the unique power in a campaign to generate diverse perceptions and emotions, affecting judgment and behavior.
However, can single words still make a difference even when they are relatively similar, like
“help” and “support”? Previous research underlines that very comparable words can indeed also lead to different perceptions (e.g., Bavelas & Coates, 2001). For example, violently describing a sexual offense as “assault” or “penetration” compared to “intercourse”, which is considered to be a more sexual, erotic term, affects how we perceive the malignity of a crime (Bavelas & Coates, 2001). The findings reveal that “sexualized descriptions minimize the inherent violence of sexual assaults and hide the survivors’ experience” (Bavelas & Coates, 2001, p. 29). Supporting the idea of different word associations, people also interpret related probability words, referring to change (e.g., “likely”, “rarely”, “possible”) differently (Willems, Albers & Smeets, 2020). Hence, we already learn from other contexts that similar words can be understood in various ways.
More related to consumer research, the study on restaurant inspections stresses that similar words like “critical” or “priority” are perceived differently by consumers in terms of the seriousness regarding the violations of rules by a restaurant (Kim, Ma & Almanza, 2017).
Here, the word “critical” provokes a higher seriousness than “priority”. Similarly, we know that the usage of the word “compensation” compared to “payment” can significantly increase respondents’ belief in a suggested environmental policy (Clot, Grolleau & Méral, 2017). The same can be applied to labeling an additional income as “bonus” rather than “rebate”, which leads to a higher perceived gain, motivating people to spend more of their income and save less (Epley, Mak & Idson, 2006). Even in the donation context itself, and as such most relevant for the current research question, previous studies point out the positive consequences of using the word “reimbursement” instead of “payment” to explain the perceived costs related to organ donation on donation behavior (Bowles, 2008). All this strongly suggests that the words “help” and “support”, despite being similar, elicit different perceptions or feelings that can change the donation behavior.
Potential Underlying Difference Between “Help” and “Support”
In order to better understand the difference between the two words, we first take a look at the dictionary. Helping someone is defined as “making it possible or easier for someone to do something by doing part of the work yourself” (Cambridge Dictionary, n.d.). In contrast, support refers to “giving encouragement to someone or something because you want him or her or it to succeed” (Cambridge Dictionary, n.d.). The comparison already implies that there is a major difference between the two words.
When an individual helps someone, he or she is highly involved and carries out an activity for the person that requires help. However, support is strongly related to assistance. Although one person accompanies the other one in achieving a task, the person still completes the activity himself or herself in the end. Thus, the difference in taking over a rather active (“help”) as opposed to a passive role (“support”) could lead to a different degree of an individual’s interference with a specific task. In other words, helping someone could be perceived as a more energetic, more vivid act than providing support. This potential difference might provoke two varying perceptions regarding efficacy and effort. Both efficacy and effort are seen as essential aspects in donating decisions (e.g., Ahn & Lee, 2019; Carroll
& Kachersky, 2019; Sharma & Morwitz, 2016; Shaw, Batson & Todd, 1994). We will discuss these two constructs in more detail below.
Donation Efficacy Associated With “Help” as Opposed to “Support”
Efficacy or, more specifically, response efficacy is defined as “the degree to which people believe their actions can make an impact” (Sharma & Morwitz, 2016, p. 46). In the domain of donations, response efficacy is labeled as perceived donation efficacy, describing a donor’s belief that his contribution has an impact on solving the supported cause (Carroll &
Kachersky, 2019). Therefore, we refer to perceived donation efficacy in this study.
Based on this definition, one could argue that if people link “help” to a more vivid and active role than “support”, they might have the feeling that they can create a bigger change through their behavior, leading to higher perceived efficacy. Indeed, past research shows that using a single target (compared to multiple) in message frames increases the perceived efficacy among individuals because a single target appears to be more vivid (Sharma & Morwitz, 2016). Hence, we assume that there is a connection between “help” and donation efficacy.
Since this is all very speculative, we conducted a small pilot test to get a better understanding of the potential differences. Here, we asked people to indicate to what extent they associate
“help” and “support” with (amongst others) the level of activeness and efficacy. The results of this study can be found in Appendix 1. The findings illustrate that people significantly relate “help” towards a more active level than “support”. Besides, the pilot test shows that
“help” invokes a higher efficacy in people’s mind compared to “support”. Hence, we expect for our main study that people exposed to the word “help” (compared to “support”) in the charity message perceive it as more active, ultimately associating it with a higher donation efficacy.
H1: The word “help" results in a higher perceived donation efficacy than the word
“support”.
Effort Associated With “Help” as Opposed to “Support”
Participation effort relates to “the degree of effort and resource input required from consumers to participate in companies’ CSR activities” (Ahn & Lee, 2019, p. 3). Following the previous concept, we define effort in a charity campaign as the degree of energy and input of resources required from donors to support the mentioned cause.
Referring to this definition, one could claim that if people expect to play an active role in a task, provoked by “help”, they might simultaneously feel that they need to put more energy into it. Correspondingly, this higher perceived investment of energy might be linked to a higher participation effort. The same pilot test, as referred to above, underlines that “help” is indeed more substantially linked to effort than “support”. Hence, we assume that using the word “help” will generally invoke a higher perceived effort in following a charitable cause, shown in the charity ad than the word “support”.
H2: The word “help" results in a higher perceived effort than the word “support”.
To conclude, “help” as opposed to “support” can encourage different associations in people’s minds, either an increased donation efficacy or an increased effort. This is important because past research stresses how both aspects are relevant in a donor’s decision to give to charities.
Below, we will provide a detailed overview of what we know about the underlying process of charitable choices in terms of efficacy and effort.
The Mediating Role of Donation Efficacy on Charitable Giving
Perceived donation efficacy seems to serve as an important link between the type of wording and people’s attitude towards the charity or their intention to donate. The vast amount of research supports this, underlining the essential role of perceived efficacy in people’s donation decisions (e.g., Bendapudi, Singh & Bendapudi, 1996; Cryder, Loewenstein &
Scheines, 2013; Fiske, Rosenblum & Travis, 2009; Sharma & Morwitz, 2016). For example, people with a low (compared to a high) willingness to give are more worried that their donation would not affect the strengthened cause (Diamond & Kashyap, 1997). Likewise, providing tangible information to increase the perceived impact of a donation positively influences generosity (Cryder et al., 2013; Sharma & Morwitz, 2016). Vice versa, a perceived
lacking donation efficacy decreases the chances of giving (Radley & Kennedy, 1992). This underlines people’s strong wish “to personally make a difference” while donating (Duncan, 2004, p. 2159). Hence, as we hypothesized that “help” vs. “support” affect perceived efficacy, we also expect that the effect of wording in a promotional charity message on the attitude and donation intention is mediated by perceived donation efficacy.
H3: “Help” (compared to “support”) leads to a more positive attitude towards the charity and a higher intention to donate.
H4: This effect is mediated by perceived donation efficacy.
The Mediating Role of Effort on Charitable Giving
In addition to donation efficacy, many studies emphasize the importance of the perceived effort in influencing donation behavior (e.g., Ahn & Lee, 2019; Howie, Yang, Vitell, Bush &
Vorhies, 2018). Thus, the effort could be an essential variable, connecting the type of wording and people’s attitude towards the charity or their intention to donate. Previous studies, however, reveal that the perceived effort can result in both positive and negative effects on donation behavior (Habel, Schons, Alavi & Wieseke, 2016; Howie et al., 2018).
Positive Perception of Effort
Positively speaking, participation effort can be related to the satisfying feeling and emotion that is derived from helping others according to the theory of warm glow (Ahn & Lee, 2019;
Andreoni, 1989; Habel et al., 2016; Howie et al., 2018). Thus, engaging people more intensively in the donation process could increase their personal benefits, achieving a more favorable attitude towards the charity and a higher participation intention.
In empirical terms, research from other fields shows that final products are perceived more favorably if the consumer has been actively involved in the production process (Franke, Keinz & Steger, 2009). Accordingly, consumers who feel highly engaged in the participation of a loyalty program portray a higher attractiveness towards the loyalty program (Kivetz &
Simonson, 2002). In the donation setting, consumers who associate the level of effort to act in a prosocial manner with warm glow or personal benefits, experience a higher willingness to pay and a higher customer loyalty (Habel et al., 2016) or a higher participation likelihood in the campaign (Howie et al., 2018). Thus, a higher perceived effort required by an
individual in a charity campaign might be positively associated with higher engagement, provoking a more positive attitude and an increased donation intention.
H5a: “Help” (compared to “support”) leads to a more positive attitude towards the charity and a higher intention to donate.
H6a: This effect is mediated by perceived effort.
Negative Perception of Effort
In contrast, high participation effort can also be perceived as personal costs (Howie et al., 2018). The simultaneous occurrence of personal expenses, together with the goal of benefiting the society, creates a dissonance in consumers’ minds (Chatzidakis, Kastanakis &
Stathopoulou, 2016; Festinger, 1962). Consumers are motivated to solve this dissonance by devaluing the importance of the activity to reduce their perceived moral responsibility (Schwartz & Howard, 1980). In the donation context, this means that people would start devaluing the importance of the cause so that they can justify that the higher perceived effort is not worth the supported cause (Howie et al., 2018). Thus, the higher perceived effort would negatively impact an individual’s attitude and donation intention.
Empirically, several authors stress the negative effect of perceived effort on prosocial behavior (Piliavin, Piliavin & Rodin, 1975; Tyler, Orwin & Schurer, 1982; Wagner &
Wheeler, 1969). For instance, the likelihood of saving energy among individuals decreases when the effort to do so is high (Tyler et al., 1982). Similarly, individuals are less likely to support a stranger when helping him would require a considerable amount of time (Shaw et al., 1994). This idea of high required efforts, negatively impacting the participation intention, is especially true for individuals that consider participation efforts as costs (Ahn & Lee, 2019). Consequently, when the effort is perceived as high, the attitude and donation outcomes might be less favorable. Hence, if the previously mentioned hypothesis does not prove true, there could also only be a negative perception.
H5b: “Help” (compared to “support”) leads to a more negative attitude towards the charity and a lower intention to donate.
H6b: This effect is mediated by perceived effort.
To summarize, based on existing research, an associated high effort can either produce positive or negative outcomes, depending on an individual’s perception.
The Moderating Role of Donation Proximity on Donation Efficacy & Effort
Both efficacy and effort are known to be affected by the common contextual factor, proximity (Balcetis, Cole & Bisi, 2015; Latané & Bourgeois, 2001). Donation proximity explains how close donors consider themselves to be towards the donation activity or the recipient (Varadarajan & Menon, 1988). Previous research differentiates between local, regional, and national donations to characterize the specific level of donation proximity (Varadarajan & Menon, 1988). Due to the link between donation proximity and the two underlying processes, we argue that donation proximity might serve as a highly relevant contextual cue, impacting the degree to which associations of efficacy and effort are created in consumers’ minds in response to the type of wording used. Consequently, on top of the already established hypotheses that are based on the assumption that people are not informed about donation proximity, we add two other conditions (low or high donation proximity) to further understand the word effect.
Donation Proximity & Donation Efficacy
To start, we dig deeper into the relationship between donation proximity and efficacy. Both signaling theory (Spence, 1974) and dynamic social impact theory (Latané & Bourgeois, 2001) suggest that the closer the donor feels towards the person that receives the donation, the higher the perceived efficacy and the higher the chances for engaging in prosocial behavior. Starting with signaling theory, it deals with the processing of cues in an exchange of information between a signaler and receiver to better evaluate the other party (Spence, 1974). By specifying that the donation is made to a recipient close to the donor’s environment, the message becomes more feasible. This creates the impression that a donor’s contribution will really make a difference, strongly influencing the perceived donation efficacy (Spence, 1974). Following the signaling theory, a similar idea is proposed by the social impact theory, indicating that consumers categorize others into social spaces. Those that are close towards the own social area are perceived to be directly impacted, increasing the perceived efficacy (Latané & Bourgeois, 2001). Taking into account both theories, we predict that donations with higher proximity achieve a higher feeling of perceived donation efficacy in people’s minds.
Due to the relationship between donation proximity and efficacy, we forecast that the effect of the type of wording on efficacy is strongly dependent on the context of donation proximity. More precisely, high donation proximity is expected to already elicit a high donation efficacy. Hence, adding “help” compared to “support” would not lead to a considerable change in efficacy. Therefore, we expect that the effect of words on efficacy becomes stronger when the donation recipient is either in a distant compared to a close location or when no information is provided about the location.
H7: The word effect on perceived efficacy is strongest in the low donation proximity/no information condition and weakest in the high donation proximity context.
Donation Proximity & Effort
Proximity also proves to impact the perceived effort (Balcetis et al., 2015). More specifically, the effort is considered to be an important underlying phenomenon of proximity in the food environment (Hunter, Hollands, Couturier & Marteau, 2018; Maas, de Ridder, de Vet & de Wit, 2012). For example, obtaining a snack that is 70 cm compared to 20 cm away is evaluated as a more effortful activity in people’s minds (Hunter et al., 2018; Maas et al., 2012). However, the effect of effort also occurs with the perceived distance by an individual regardless of how far the object actually is (Balcetis et al., 2015). Participants who judge the cone from where they currently stand as closer, perceive the task as less effortful (Balcetis et al., 2015). This shows that a distance perception itself is already sufficient to shape specific effort expectations independent of the actual distance. Similarly, if a donation is made to recipients within a donor’s environment, the recipients might be perceived as physically closer. Consequently, we predict that charity ads, focusing on higher donation proximity, correspond to a lower perceived effort.
Due to the vital role of donation proximity on effort, we propose that the effect of the type of wording on effort is strongly reliant on the context of donation proximity. More precisely, low donation proximity already provokes a high perceived effort. Hence, adding a specific word would not lead to significant differences in perceived effort. In contrast, high donation proximity is expected to evoke a relatively low effort. Thus, the term “help”, related to effort, could increase donors’ perceptions regarding the required effort. For these reasons, the impact of words on perceived effort is predicted to be stronger either for a close compared to a distant donation recipient or when no information about donation proximity is provided.
H8: The word effect on perceived effort is strongest in the high donation proximity/no information condition and weakest in the low donation proximity context.
Conceptual Model
Drawn from the hypotheses in the theoretical background, the conceptual model looks as follows (see Figure 2). While the type of wording acts as an independent variable with the two levels “help” and “support”, the dependent variables comprise the attitude towards the charity and the donation intention. The effect of wording on the attitude towards the charity and the donation intention is mediated by donation efficacy or effort. Furthermore, a relevant moderator, the donation proximity, is added with the three levels being defined as no information, high donation proximity (local), and low donation proximity (international).
FIGURE 2
Conceptual Model of Attitude Towards the Charity & Donation Intention
METHODOLOGY
Participants and Design
Since the present study aimed at inferring a causal relationship between the type of wording and the donation behavior, we conducted an online experiment as a suitable method (Aronson, Wilson & Brewer, 1998). We carried out the experiment with the help of a structured online questionnaire. A questionnaire is relatively easy to manage and reduces the inconsistency of the results through fixed-response questions (Malhotra, 2009). Due to the anonymity provided by the questionnaire, honesty among participants could be increased, leading to more reliable answers (Malhotra, 2009). The experimental design included a 2 (wording: “help” vs. “support”) x 3 (donation proximity: no information, high vs. low) between-subject design. Thus, six different experimental groups were established (see Table 1).
TABLE 1
Six Experimental Groups
In total, 468 US citizens were recruited on MTurk, a large online panel service. Out of the 468 people, 15 failed the attention check (for details about this check, see procedure below), resulting in 453 valid cases. The sample consisted of 226 (49.9%) men and 227 (50.1%) women with a mean age of 41 years (M = 40.92, SD = 12.10). Generally, the likelihood of donating to charities among the sample was relatively neutral (M = 4.50 (1 = extremely unlikely; 7 = extremely likely), SD = 1.88). On average, people donated 11 times (N = 449, M = 10.54, SD = 1.36) to charities in the last year (excluding 4 outliers). All participants were randomly assigned to one of the six experimental groups to ensure that only the experimental condition and no other variables differed between the participant groups.
Study Procedure
Members of the online panel were invited to take part in a study for a charity organization while disguising the specific research purpose. Those that agreed-upon were shown one of the six charity advertisements. To avoid biases that might result from the perceived importance of one particular charity cause, a relatively neutral and broad goal was chosen for this ad, meaning “providing aid for children in need”. Since a single victim in ads already creates a strong efficacy, which might influence the results, multiple victims were used (Sharma & Morwitz, 2016). Simultaneously, the charity logo of the Child Poverty Action Group, registered only in England and Wales, was selected to ensure that US participants lack familiarity with the charity. This prevented respondents to already have associations regarding the charity’s operating environment before testing in mind, from which they might unconsciously infer the level of donation proximity, biasing the results.
After participants were given some time to read the advertisement, they were asked to fill out the questionnaire in the next section. The survey contained all measures regarding the mediators, dependent variables, and the manipulation check (see Appendix 4). At the end of the questionnaire, demographic data concerning age, gender, and general donation behavior was collected. Furthermore, an attention check was added. Finally, participants were thanked for participation.
Manipulation and Measurement of Variables
Manipulation: Type of Wording & Donation Proximity
To manipulate the type of wording, participants were either confronted with the word “help”
or “support” in the charity ads. Moreover, the degree of donation proximity was varied by creating a local (high donation proximity) compared to an international environment (low donation proximity) to achieve the most substantial gap and corresponding effect. While the local environment was characterized by the term “community”, the international setting focused on the “world”. Also, the no information condition did not contain any information regarding the donation proximity and served as a control condition. As an example, the headlines comprised “help children in need from the community” or “support children in need” (see Figure 3).
FIGURE 3
Two Examples of Manipulated Advertisements
Help – High Proximity Support - No Proximity Information
Generally, the messages only varied in terms of wording and donation proximity and were designed in an attempt to resemble actual charity campaigns. All other factors like content, color, design, and the length of words were held constant to cancel out any extraneous effects. The six different messages can be found in Appendix 2.
Measurement: Efficacy
Donation efficacy was measured using four items, based on existing research (Carroll &
Kachersky, 2019; Sharma & Morwitz, 2016; White, MacDonnell & Dahl, 2011). As an example, respondents were asked the following: “If you would donate to the charity in the ad, to what extent do you think your donation will help (support) children in need (from the community/around the world)?” (seven-point Likert scale: 1 = absolutely not; 7 = absolutely). Since the factor analysis resulted in one factor (λ = 3.07, total explained variance
= 76.68%), a mean score of all four questions was computed (α = .89). Additional information regarding the reliability test for all variables can be found in Appendix 3.
Measurement: Effort
Three questions were used to assess the perceived effort required by individuals, adapted from previous research (Ahn & Lee, 2019; Maas et al., 2012; Xia, Kukar-Kinney & Monroe, 2010). For instance, participants had to indicate, to what extent they think based on the charity ad that it will be effortful to help (support) children in need (from the community/around the world) in the ad (seven-point Likert scale: 1 = absolutely not; 7 =
absolutely). Responses towards the three questions were combined to an average score (α = .91) after one main factor could be revealed (λ = 2.54, total explained variance = 84.59%).
Measurement: Cause Importance
How much importance respondents assigned to the supported cause, was measured using five semantic differential seven-point scales similar to prior work (Maheswaran & Meyers-Levy, 1990). In more detail, they had to indicate whether the cause in the campaign is unimportant/important to them, means nothing to them/means a lot to them, is personally irrelevant/relevant to them, doesn’t/does matter a great deal to them, and is of no/great concern to them. Higher average scores, summarizing all five items into one suitable factor (λ
= 4.45, total explained variance = 88.94%), provoked higher perceived cause importance (α = .97).
Measurement: Activeness
In order to assess the level of activeness, participants were, for instance, asked to what extent they agree that helping (supporting) children in need (from the community/around the world) in the charity ad would require an active role from their side (seven-point Likert scale: 1 = very strongly disagree; 7 = very strongly agree) (Saris & Gallhofer, 2014). In total, two items, in which one was reversely coded, were represented by one factor (λ = 1.65, total explained variance = 82.38%), leading to a combined Cronbach’s alpha score of .79.
Measurement: Attitude Towards the Charity
The attitude towards the charity organization was evaluated based on six semantic differential seven-point scales (Smith & McSweeney, 2007). Respondents referred to the following statement: “I perceive Child Poverty Action Group as unpleasant-pleasant, useless-useful, unsatisfying-satisfying, unfavorable-favorable, inconsiderate-considerate, and pointless- worthwhile”. After one factor could be identified for all six items (λ = 4.93, total explained variance = 82.10%), higher average scores suggested a more positive attitude towards the charity organization (α = .96).
Measurement: Intention to Donate
Participants had to indicate their intention to donate through two items, adopted from earlier studies (Grau & Folse, 2007; Sharma & Morwitz, 2016; Sheeran, Trafimow & Armitage, 2003). Their level of willingness to donate to the charity in the ad was, for example,
investigated on a seven-point Likert scale from extremely unlikely (1) to extremely likely (7).
Together with the other item, it summed up to one factor (λ = 1.88, total explained variance = 93.80%, α = .93). In order to get a broader picture of the donation intention, participants additionally had to state how much they would be willing to give to the charity out of five dollars that they could spend (second scale of the intention to donate).
Manipulation Check: Type of Wording and Donation Proximity
To ensure that participants carefully read the stimuli, they had to share whether they were asked for “help” or “support” in the charity ad at the end of the survey. Second, they had to indicate whether any information was given regarding the location of the donation. More specifically, they had to reveal where the donation would go to, either to children from the community or to children around the world. They could also choose that no information concerning the location was provided.
1Removing participants who failed the manipulation check from the data analysis, yielded approximately similar results
RESULTS
Manipulation Check
The manipulation check of the type of verb and the level of proximity was successful. First, a chi-square test indicated that 76.3% of respondents in the “help” condition answered the manipulation question regarding the type of verb correctly. Similarly, 76% provided the right response in the “support” condition. This confirmed that the majority of people correctly indicated in which condition they were, X2 (1, N = 453) = 123.98, p < .001. Second, 78.4% of participants in the high proximity, 82.8% in the low proximity, and 77.9% in the no information condition gave the correct answer concerning the manipulation question related to the level of proximity. This underlined that most people properly remembered the condition to which they were assigned according to the chi-square test, X2 (4, N = 453) = 446.66, p < .001. All participants, even those who failed the manipulation check, were included in the data analysis to keep the corresponding experimental power sufficiently high1.
Main ANOVA Analyses
In order to gain a general overview of the data, all variables were first analyzed by a 2 (wording: “help” vs. “support”) x 3 (donation proximity: no information, high vs. low) ANOVA. A summary of the cell means and standard deviations for each variable is provided in Table 2. The detailed analyses per variable are described in the following.
TABLE 2
Cell Means (Standard Deviations) for all Variables
Efficacy
Contrary to hypothesis H1, the results of the ANOVA yielded that the type of verb had no significant main effect on perceived efficacy, F(1, 447) = 0.17, p = .68. The same non- significance was true for the level of proximity, F(2, 447) = 1.25, p = .29. However, there was a marginally significant interaction effect between the type of verb and the level of proximity as expected, F(2, 447) = 2.97, p = .05, η2 = .01.
In order to test the interaction effect in more detail, simple contrast analyses were conducted.
These analyses illustrated that the proximity effect was only significant in the “help”
condition, F(2, 447) = 3.90, p < .05, η2 = .02. Participants in the high donation proximity (M
= 5.16, SD = 1.10) perceived the efficacy as significantly higher than participants in the low proximity condition (M = 4.60, SD = 1.30) but not significantly different from participants in the no information condition (M = 4.87, SD = 1.33). The proximity effect was not significant within the “support” condition, F(2, 447) = 0.33, p = .72. Furthermore, we predicted in hypothesis H7 that the word effect on perceived efficacy was strongest in the low donation proximity/no information condition and weakest in the high donation proximity context. As opposed to our prediction, the word effect was significant within the high proximity condition, F(1, 447) = 4.41, p < .05, η2 = .01, but neither in the low proximity condition, F(1, 447) = 1.67, p = .20, nor in the no information condition, F(1, 447) = 0.01, p = .91.
Respondents in the high proximity condition associated the word “help” (M = 5.16, SD = 1.10) with a higher efficacy than “support” (M = 4.73, SD = 1.23) (see Figure 4 for an overview of cell means).
FIGURE 4
Means for Efficacy (1 = absolutely not; 7 = absolutely)
2Normality assumption was not fully met but square root transformation yielded similar results whereas the original variable was used for better interpretation purposes
Effort
Partly finding evidence for hypothesis H2, the type of verb only had a marginally significant influence on perceived effort, F(1, 447) = 3.73, p = .05, η2 = .01. Participants associated the word “help” (M = 3.79, SD = 1.63) with a higher effort than the word “support” (M = 3.49, SD = 1.69). In contrast to the type of verb, no significant main effect was found for the level of proximity, F(2, 447) = 1.37, p = .26. Contrary to hypothesis H8, the interaction effect between the type of verb and the level of proximity on effort was not significant, F(2, 447) = 0.47, p = .62.
Attitude Towards the Charity
Both main effects of the type of verb, F(1, 447) = 0.00, p = .99, and the level of proximity, F(2, 447) = 0.17, p = .85, on attitude2 were not significant, rejecting part of hypothesis H3 and H5ab. Nevertheless, the analysis detected a marginally significant interaction effect between the two variables, F(2, 447) = 3.13, p = .05, η2 = .01. This showed that the effect of the type of wording on the attitude was dependent on proximity as moderator.
Specific contrast analyses revealed that the differences regarding proximity were mainly found within the “help” condition. Still, these differences were not sufficiently large enough to find a significant proximity effect, F(2, 447) = 2.39, p = .09. The proximity effect was also not significant within the “support” condition, F(2, 447) = 0.92, p = .40. Looking at the interaction from the other perspective, the word effect was only marginally significant in the low proximity, F(1, 447) = 3.44, p = .06, η2 = .01, but not significant in the high proximity, F(1, 447) = 2.77, p = .10, and not in the no information condition, F(1, 447) = 0.04, p = .84.
Respondents who saw the word “support” (M = 5.79, SD = 1.22) instead of “help” (M = 5.44, SD = 1.31) in the low proximity condition had a more positive attitude. These results suggested that “help” did not always create a better attitude than “support”, but it was dependent on the proximity condition (see Figure 5 for an overview of cell means).
2Normality assumption was not fully met but square root transformation yielded similar results whereas the original variable was used for better interpretation purposes
FIGURE 5
Means for Attitude Towards the Charity (1 = very negative; 7 = very positive)
Intention to Donate
The ANOVA analysis showed that the type of verb, F(1, 447) = 1.94, p = .16, and the level of proximity, F(2, 447) = 0.31, p = .73, had no significant main effect on the intention to donate2, rejecting the other half of hypothesis H3 and H5ab. However, there was a significant interaction effect, F(2, 447) = 4.40, p < .05, η2 = .02. This explained that the effect of the type of wording on the intention to donate was dependent on proximity as moderator.
Simple contrast analyses did not reveal any significant results for the proximity effect in the
”help” condition, F(2, 447) = 1.99, p = .14, and only marginally significant results in the
“support” condition, F(2, 447) = 2.76, p = .06, η2 = .01. Respondents in the “support”
condition were more willing to donate when they were part of the low donation proximity group (M = 3.99, SD = 1.73) compared to the high donation proximity condition (M = 3.30, SD = 1.94) but not towards the no information condition (M = 3.80, SD = 1.80).
Comparatively, the word effect was only significant within the high proximity condition, F(1, 447) = 8.51, p < .01, η2 = .02, but not significant in the low proximity condition, F(1, 447) = 1.61, p = .21, and not in the no information condition, F(1, 447) = 0.55, p = .46. Here, we identified a significant difference in the willingness to donate between “help” (M = 4.19, SD
= 1.88) and “support” (M = 3.30, SD = 1.94) for the high proximity condition (see Figure 6 for an overview of cell means).
FIGURE 6
Means for Intention to Donate (1 = extremely unlikely; 7 = extremely likely)
As an additional check, the second scale of intention, where participants had to indicate how much money they would be willing to donate out of 5 dollars, confirmed approximately similar results (see Appendix 5).
To conclude, we found general differences between “help” and support”, interacting with the different levels of proximity for perceived efficacy, but not for perceived effort. The differences were mainly visible in the high proximity condition and less in the low proximity or no information condition. In a similar vein, the attitude towards the charity and the donation intention also differed among the wording groups in the high proximity condition.
Still, these differences were only significant for donation intention. However, we found identical patterns between efficacy and the two dependent variables whereas moderated mediation analyses provided further insights.
Moderated Mediation Analyses
Using two PROCESS analyses (model 7), we tested whether perceived efficacy and effort mediated the effect of the type of verb (“help” = 1, “support” = 2) on the attitude towards the charity organization, and the intention to donate. Simultaneously, we investigated whether the level of proximity (high = 1, low = 2) moderated the impact of the type of verb on the two mediator variables. Since previous 2 x 3 ANOVA analyses demonstrated no significant differences between “help” and support” in the no information about proximity condition for each variable, we only focused on high versus low proximity in this analysis (N = 299). The
findings for the attitude towards the charity and the intention to donate can be found in Figures 7 and 8.
FIGURE 7
Moderated Mediation Analysis on Attitude Towards the Charity (Mediators: Efficacy & Effort)
FIGURE 8
Moderated Mediation Analysis on Intention to Donate (Mediators: Efficacy & Effort)
Efficacy
As shown in Figures 7 and 8, the results revealed that the type of verb significantly affected the perceived donation efficacy, β = -1.11, CI = [-1.99; -0.23]. Hence, hypothesis H1 was supported. The same was true for the level of proximity, β = -1.25, CI = [-2.13; -0.37]. The significant interaction effect of the type of verb and the level of proximity on perceived efficacy, β = 0.69, CI = [0.13; 1.24], confirmed that the level of proximity indeed moderated the relationship between the type of verb and efficacy. We expected the word effect on perceived efficacy to be the strongest in the low donation proximity/no information condition and weakest in the high donation proximity context in hypothesis H7. However, in line with the results from the ANOVA, the conditional effects of the type of wording at the values of proximity indicated the opposite and only showed significant results for the high proximity, β
= -0.43, CI = [-0.82; -0.03], but not for the low proximity condition, CI = [-0.13; 0.65]. Thus, we rejected the hypothesis H7.
Furthermore, the main effect of efficacy on the attitude towards the charity, β = 0.64, CI = [0.57; 0.72], and on the intention to donate, β = 0.92, CI = [0.78; 1.06], was positive and significant. However, there was no significant direct effect of the type of verb on the attitude towards the charity, CI = [-0.14; 0.25], and the intention to donate, CI = [-0.50; 0.18]. Even though no direct effect was found, the index of moderated mediation indicated that the level of proximity significantly moderated the indirect effect of the type of verb on the attitude towards the charity, CI = [0.07; 0.82], and the intention to donate, CI = [0.11; 1.18]. Hence, one could assume full moderated mediation, underlining that hypothesis H3 and H4 only proved true under the condition of moderation through proximity. Nevertheless, the indirect effect of the type of word on the attitude through efficacy was only significant for the high proximity, β = -0.27, CI = [-0.52; -0.02], but not for the low proximity condition, CI = [-0.09;
0.44]. Likewise, the intention to donate was significantly impacted by the type of word in the high proximity condition, β = -0.39, CI = [-0.76; -0.05], but not in the low proximity condition, CI = [-0.14; 0.63]. In general, similar results were found for the second scale of intention (see Appendix 5).
Effort
Contradicting hypothesis H2, the level of effort perceived by an individual, was not significantly influenced by the type of verb, CI = [-1.88; 0.56], and the level of proximity, CI
= [-1.57; 0.86]. Further, we found no significant interaction effect, CI = [-0.49; 1.05], similar
to the ANOVA analysis. Moreover, the main effect of effort on the attitude towards the charity, CI = [-0.11; 0.00], and on the intention to donate, CI = [-0.01; 0.19], was not significant. Similarly, the index of moderated mediation revealed non-significant results for both, the attitude towards the charity, CI = [-0.07; 0.03], and the intention to donate, CI = [- 0.05; 0.14]. For these reasons, we rejected hypothesis H5ab and H6ab as well as H8 regarding moderated mediation. Again, we found consistent results for the second scale of intention (see Appendix 5).
In conclusion, the findings provided sufficient empirical evidence that the level of proximity moderated the wording effect on efficacy, which in turn shaped the attitude and the intention to donate through mediation. However, this was not true for the perceived effort. Hence, differences between “help” and support” occurred for the level of efficacy but not for effort.
2Normality assumption was not fully met but square root transformation yielded similar results whereas the original variable was used for better interpretation purposes
Exploratory Analyses
As the theoretical framework suggested, the perceived importance of the cause and the level of activeness could be two additional variables that might be influenced by the word and proximity effects. Thus, we also performed further exploratory 2 x 3 ANOVA analyses on these variables to gain a broader perspective. The cell means and standard deviations for both variables are included in Table 2 (p. 23).
Cause Importance
Exploratory analyses yielded a significant interaction effect, F(2, 447) = 7.23, p = .001, η2 = .03., but no significant main effects of the type of verb, F(1, 447) = 0.07, p = .80, and the level of proximity, F(2, 447) = 0.38, p = .68, on the perceived importance of the cause2.
Based on simple contrast analyses, the proximity effect was significant in the “help”
condition, F(2, 447) = 4.71, p < .01, η2 = .02, and marginally significant in the “support”
condition, F(2, 447) = 2.87, p = .06, η2 = .01. For “help”, participants in the high donation proximity condition (M = 5.84, SD = 1.24) perceived the cause as significantly more important than participants in the low proximity condition (M = 5.14, SD = 1.49) but only marginally significantly more important than participants in the no information condition (M
= 5.40, SD = 1.49). In contrast, for “support”, participants in the high donation proximity condition (M = 5.19, SD = 1.36) perceived the cause as less important than participants in the low proximity condition (M = 5.73, SD = 1.36) but not than participants in the no information condition (M = 5.35, SD = 1.45).
Alternatively, there was a significant word effect within the high, F(1, 447) = 7.79, p < .01, η2 = .02, and low proximity condition, F(1, 447) = 6.68, p = .01, η2 = .02. People within the high proximity condition assigned a higher value to the cause when they saw the word “help”
(M = 5.84, SD = 1.24) compared to “support” (M = 5.19, SD = 1.36). The opposite was true for those in the low proximity condition, where “support” (M = 5.73, SD = 1.36) was associated with higher cause importance than “help” (M = 5.14, SD = 1.49). We found no significant word effect in the no information condition, F(1, 447) = 0.04, p = .83 (see Figure 9 for an overview of cell means). Thus, combining the word “help” with high donation proximity and “support” with low donation proximity led to a higher perceived importance of the cause.
FIGURE 9
Means for Cause Importance (1 = extremely unimportant; 7 = extremely important)
Significant differences occurred in the importance of the cause, and these results strongly resembled those for the attitude and the donation intention. For these reasons, we additionally tested for a possible moderated mediation effect of cause importance between the type of wording (“help” = 1, “support” = 2) as well as the level of proximity (high = 1, low = 2) and attitude/donation intention.
The two PROCESS analyses (model 7) for the attitude and donation intention (see Figures 10 and 11) showed that the type of verb, β = -1.88, CI = [-2.87; -0.89] significantly affected the perceived cause importance. The same was true for the level of proximity, β = -1.93, CI = [- 2.92; -0.95]. The significant interaction effect of the type of verb and the level of proximity on perceived cause importance, β = 1.23, CI = [0.61; 1.86], illustrated that the level of proximity indeed moderated the relationship between the type of verb and cause importance.
The conditional effects of the type of wording on cause importance at the values of proximity indicated significant results for the high proximity, β = -0.64, CI = [-1.09; -0.20], and for the low proximity condition, β = 0.59, CI = [0.15; 1.03].
FIGURE 10
Moderated Mediation Analysis on Attitude Towards the Charity (Mediator: Cause Importance)
Moreover, the direct effect of cause importance on the attitude towards the charity, β = 0.50, CI = [0.42; 0.57], and on the intention to donate, β = 0.76, CI = [0.63; 0.88], was positive and significant. However, there was no significant direct effect of the type of verb on the attitude towards the charity, CI = [-0.18; 0.24], and the intention to donate, CI = [-0.59; 0.12]. Even though no direct effect existed, the index of moderated mediation indicated that the level of proximity significantly moderated the indirect effect of the type of verb on the attitude towards the charity, CI = [0.29; 0.98], and the intention to donate, CI = [0.47; 1.44]. Hence, one could assume full moderated mediation. More specifically, the indirect effect of the type of word on attitude was significant for the high proximity, β = -0.32, CI = [-0.56; -0.11], and for the low proximity condition, β = 0.29, CI = [0.07; 0.54]. Likewise, the intention to donate was significantly impacted by the type of word in the high proximity condition, β = -0.49, CI
= [-0.82; -0.18], and in the low proximity condition, β = 0.45, CI = [0.12; 0.81]. In general, similar results were found for the second scale of intention (see Appendix 5). Hence, an additional mediator was detected, explaining the underlying process of “help” versus
“support” on the attitude towards the charity and donation intention.
FIGURE 11
Moderated Mediation Analysis on Intention to Donate (Mediator: Cause Importance)
Activeness
Results of the ANOVA showed that the level of activeness did not significantly differ between “help” and “support”, F(1, 447) = 0.03, p = .85. The same non-significance could be applied to the level of proximity, F(2, 447) = 0.84, p = .43. Moreover, the effect of the interaction between the type of verb and the level of proximity was not significant, F(2, 447)
= 0.59, p = .55. Hence, the difference in the perceptions of the word “help” versus “support”
did not derive from the contrast of taking over an active or passive role.
DISCUSSION
Theoretical Implications
The current study provided insights into how charities can effectively make use of specific words in their marketing campaigns to encourage people to donate. Existing research within the donation setting has mostly dealt with the influence of additional words on people’s donation intention so far (Cialdini, 2007; Croson & Shang, 2013; Langer, Chanowitz &
Blank, 1985). However, this paper focused on replacing very similar single words like “help”
and “support”. It was the first one, providing evidence that even this replacement shaped the response in the donation context. Using “help” compared to “support” provoked a more positive attitude and donation intention but only for advertisements that have dealt with high donation proximity. While the level of efficacy mediated this effect, the effort did not.
Further, we revealed the importance of the cause as a new, additional mediator beyond our conceptual model through exploratory analysis. We will go into more detail regarding all mediators below.
The Role of Efficacy
As expected, people associated the word “help” with higher efficacy than “support” (H1).
This result resembled the findings from the pretest. Supporting our predictions (H3, H4), these different efficacy perceptions mediated the effect of the type of wording on the attitude towards the charity and the donation intention, but the mediation effect was dependent on the proximity condition. Even though we found no direct word effect on the attitude or donation behavior, the significant role of efficacy on the donation response was confirmed in line with previous research (e.g., Bendapudi et al., 1996; Cryder et al., 2013; Fiske et al., 2009; Sharma
& Morwitz, 2016), emphasizing the validity of our measurement.
Further, we detected a significant relationship between proximity and efficacy that was approved by earlier studies (Latané & Bourgeois, 2001; Spence, 1974). Nevertheless, opposite to our expectations (H7), respondents only showed a significantly higher efficacy, more positive attitude, and donation response in the high proximity condition when the word
“help” was used compared to “support”. Instead, no significant differences existed for the low proximity and no information condition.
First, this contrary finding of the interaction effect could have derived from the fact that respondents might have gained the impression that they are not able to “help” children from around the world as they are too far away. Examination through further studies could provide insights on whether this combination would indeed provoke a perceived lacking ability, leading to a reduced feeling of efficacy, and a correspondingly less positive attitude and donation intention. Simultaneously, it would be interesting to investigate if such a combination could even create feelings of skepticism or distrust as participants might perceive this tactic as manipulative intent if it seems to be unrealistic. Other research on charity donations, for instance, has underlined that a sad facial expression in a campaign evoked stronger emotions of manipulative intent, lowering its effectiveness (Kang, Leliveld
& Ferraro, 2018).
Second, another reason for the interaction effect, opposite to what we expected, is described as follows. Existing studies have emphasized the essential role of local donations in creating favorable donation outcomes (e.g., Bar-Tal, 1976; Grau & Folse, 2007). Thus, when the donation proximity is high, an already positive image is created in donors’ minds. Since both
“help” and “support” appear to be realistic as the recipient is very close, “help” might have contributed to achieving even higher proximity through the stronger personal bond between the donor and the recipient. This would explain the increased efficacy, more positive attitude, and donation intention compared to “support” in the high proximity condition. Future research should test for this reasoning.
The Role of Effort
In contrast to our expectations (H2), “help” did not provoke a significantly higher perceived effort in participants’ minds than “support”. Note that in the pilot test we did find a difference, but in that study, respondents were confronted with both words, which enabled a comparison. Since participants only saw one word (and not the other) in the main study to best adapt to a real-life scenario, this might explain the non-significant results for the effort.
Likewise, the perceived effort also did not mediate the effect of the type of wording, interacting with proximity on attitude and donation intention, neither in a positive, nor a negative way, contradicting H5ab and H6ab. Even though existing papers drew attention to the importance of perceived effort regarding donation outcomes, their findings were highly polarized between positive and negative effects of higher levels of effort (Ahn & Lee, 2019;
