Introduction

Volunteerism provides a crucial component of civil society. For example, participation in voluntary work is widespread in the United States. In 2005, about 65.4 million American citizens were engaged as volunteers. The median amount of time spent for voluntary work per year was 50 h (Corporation for National and Community Service 2006). Furthermore, US volunteers contributed about 75 billion dollars to gross national product (Mowen and Sujan 2005). In the United Kingdom, around 22 million people participate in voluntary projects in their leisure time. In doing so, individual commitment amounts to 90 million hours of voluntary work per week (Volunteering England 2007). In 2004, approximately 36% of the German population older than 14 years volunteered (Gensicke et al. 2006).Footnote 1 Volunteerism takes place in a wide range of domains, ranging from voluntary service in social and cultural facilities or locals sports clubs to honorary appointments in ecology groups or human rights organizations. Even though there are arguments against volunteering (e.g., crowding-out of regular jobs),Footnote 2 one should not underrate voluntarism’s relevance for the sustainability of civil society. Volunteering provides solutions for social problems which cannot be solved by the market or state regulations. Because globalization has generated the need for the rebuilding of industrialized countries’ welfare systems, voluntarism will presumably become even more important in the near future.

This study is an attempt to develop managerial recommendations for the design of voluntary organizations’ communications policies. After presenting the theoretical foundations for the study, the results of two experimental studies will be given. Within the scope of the first experiment, constitutive criteria of persuasive communication (message framing and self-efficacy), which are also relevant in, for example, health promotion contexts (Chang 2007), will be considered. The second experiment is based on the so-called “arousal: cost-reward” model as well as on findings on persuasive communication. The “arousal: cost-reward” model originates from the analysis of prosocial behavior (e.g., Dovidio et al. 1991). Furthermore, gender-specific main and interaction effects are tested within the context of both experiments. In the concluding section, managerial recommendations for voluntary organizations as well as possibilities for future research are discussed.

This paper contributes to scientific progress with respect to the following points: To the best of the author’s knowledge this paper is the first to validate the effect of the “self-efficacy” by “message framing” interaction on behavioral intentions within the context of prosocial behavior, the effect of message framing on the willingness to volunteer, and the moderating effect of advertisement-induced arousal on the causal relationship between the costs of volunteering and the willingness to volunteer. This paper is one of a small number of studies that explicitly focus on the development of managerial recommendations for the design of voluntary organizations’ communication policy. Furthermore, this paper provides deeper insights into gender-specific effects on willingness to volunteer.

Theoretical Background

Voluntarism

According to Mowen and Sujan (2005), volunteering can be regarded as a kind of charitable activity which bears resemblance to consumer behavior. This is because taking over or carrying out voluntary work goes along with deciding how much of one’s leisure time one is willing to spend on volunteering in lieu of other (leisure) activities. Furthermore, volunteering can be subsumed under the superordinate concept of altruistic or philanthropic behavior. Due to the fact that voluntarism is closely connected with striving for social goals, voluntary work may be also considered a form of collective behavior. Within this context, individual free-riding can result in a collapse of social cooperation (Kollock 1998).

Voluntarism has been examined from an array of different angles (Dutta-Bergman 2004) including the economic, psychological (Unger 1991), the sociological (McPherson and Rotolo 1996), and organizational behavioral (Schaubroeck and Ganster 1991). Like this paper, a significant part of earlier multidisciplinary studies were based on behavioral science. For example, Wyman and Samu (2002) compared volunteering with the concept of symbolic consumption, with an alternative to other time-consuming activities (e.g., hobbies), and with donating behavior. Thus, Wyman and Samu (2002) differentiated between three important conceptual approaches to the analysis of volunteering behavior. Whilst some of the corresponding studies deal with the importance of demographics (e.g., Wilson 2000), other studies focus on the analysis of motivations for volunteering (e.g., Clary et al. 1998; Rehberg 2005; Yeung 2004). According to Marta et al. (2006), one has to distinguish between egoistic and altruistic motivations for volunteering. Within this context it has to be mentioned that volunteering encompasses a productive element. Thus, voluntary work should not be understood as merely a consumptive leisure activity (Erlinghagen 2003). Because of this and in addition to altruistic motives, citizens may be motivated to participate in voluntary work due to the fact that they (co-)produce services that otherwise would have had to be bought on the market.Footnote 3 For example, parents may volunteer in their children’s kindergarten to ensure an expansion of opening hours from the wee hours of the morning to late evening. Besides the consideration of volunteers’ motivations, behavioral approaches to volunteering take into account the impact of personality traits (Mowen and Sujan 2005), free-riding incentives (Unger 1991), and the impact of social pressure groups (Fisher and Ackerman 1998).

This paper addresses the question of how people can be attracted to voluntary work by means of voluntary organizations’ communications policies and a review of the scientific literature addressing the topic follows. As Fisher and Ackerman (1998) stated, papers focusing on voluntary work from a marketing-centric perspective are rare. One of the few papers which did so and one which validated the hypothesis that functional motivations influence willingness to volunteer is that of Clary et al. (1994). Clary et al. (1994) concluded that voluntary organizations should use advertising messages which are congruent with recipients’ functional motivation to volunteer. Fisher and Ackerman (1998) showed that the communication of (disadvantaged or fringe) groups’ specific needs as well as an emphasis on the fact that participating in voluntary work will be socially respected has a positive impact on people’s willingness to volunteer. In an empirical study, Clary et al. (1994) verified the relevance of a wide range of individual motivations to volunteer among retirees. Hence, different segments of retirees have to be approached with varying (communication) strategies.

Self-Efficacy

According to Wood and Bandura (1989), self-efficacy represents the self-assessment of one’s skills in performing specific tasks. One can also think of self-efficacy as the confidence in one’s own skills. The self-assessment of one’s own cognitive, social, linguistic, and technical capacity, in particular, may influence an individual’s self-efficacy perception. Likewise, and according to Bandura (1977), the observation of how other people cope with specific tasks or the influence of social pressure groups may impact perceived self-efficacy. Taken altogether, self-efficacy is a result of complex information processing (Gist and Mitchell 1992). Moreover, the related cognitive process is influenced by, for example, personality traits or situative factors.

Even though self-efficacy is mostly considered a moderating variable, one can assume a direct causal relationship between self-efficacy and intention to perform specific activities (Bandura 1977). Individuals with a higher perceived self-efficacy are more likely willing to carry out challenging tasks. A number of social scientific papers partially support this assumption. For example, Troutman et al. (2000) show—in a human resource management context—that the named hypothesis holds true for men, but not for women. In contrast, the findings of Troutman et al. (2000), Rodgers and Gauvin (1998) and Yzer et al. (1998), for example, indicate a positive effect of men’s and women’s perceived efficacy on the intention to perform specific behaviors or actual behaviors, respectively. Eden and Kinnar (1991) refer to this phenomenon as the Galatea effect.Footnote 4 Hypothesis H1 thus reads as follows:

H1: A higher (lower) perceived self-efficacy results in a higher (lower) willing ness to volunteer.

Message Framing

Everyday life is made up of a multitude of decision-making situations (e.g., consumption decisions). According to normative decision theory decision-making situations encompass the existing decision options, the possible consequences of choosing one of those options, as well as the consequences’ conditional probability of occurrence. The work of Tversky and Kahneman (1981) indicates that individuals do not always behave in accordance to the predictions of normative decision theory. Instead of behaving according to normative decision theory’s prediction, individuals are often geared to mental representations of decision-making problems. Within this context, one alludes to decision-making frames, which have a significant impact on decision making. The point of origin of Tversky and Kahnemann’s idea is a situation in which an individual is confronted with two decision options. One decision option carries risk; the other does not. The majority of people choose the riskless decision option. In contrast to this, individuals choose the risky decision option if decision problems are presented in a gain frame.

The concept of message framing, which is relevant within the context of this study, is closely related to the phenomenon of decision framing. It is assumed that differently framed persuasive communication have varying effects on individual behavior. Meyers-Levy and Maheswaran (1990) distinguish between several types of message framing. For example, message framing encloses the accentuation of positive versus negative characteristics of products, items, or subjects, verbal versus numeral descriptions, and depiction of negative versus positive consequences of specific behaviors. The last-mentioned type of message framing is of relevance for this study.

Empirical studies on the effects of message framing have been conducted regarding different domains of individual behavior. For example, Grewal et al. (1994) considered framed messages within the context of promoting the purchase of consumer durables. Buda (2003) incorporated aspects of message framing with regard to human resource management problems. Furthermore, many of the existing studies on the impact of message framing have focused on health-related behavior (e.g., Jones et al. 2003; O’Connor et al. 2005). Tversky and Kahneman (1981) came to the conclusion that gain frames (loss frames) promote riskless (risky) decision behavior. Correspondingly, and in a health-related context, Meyerowitz and Chaiken (1987) showed that an illustration of the positive consequences of a specific behavior (gain frame) had a stronger impact on actual behavior than loss frames. In contrast to these findings, Ganzach and Karsahi (1995), for example, demonstrated first that the loss frames’ effect on intentions or actual behavior may be more pronounced. Second, they showed that message framing may also have no behavioral impact under certain circumstances. At first glance, empirical findings on the behavioral effects of framed messages appear disparate. But these divergent findings are conditional by virtue of diverse factors which have an impact on the strength and direction of message framing effects. According to Block and Keller (1995), loss frames have a stronger (less strong) impact on behavioral intention when individuals (do not) process information with cognitive elaboration. Since voluntary work does not involve distinct monetary, financial, or social risks, it is assumed that people do not process voluntary organization’s advertising messages with pronounced cognitive elaboration. Thus, hypothesis H2 is:

H2: Gain frames have a stronger impact on the willingness to volunteer than loss frames.

Besides these two main effects, an interaction effect of message framing and perceived self-efficacy is assumed. There are only a small number of empirical studies that deal with the combined effect of message framing and perceived self-efficacy. For example, Williams et al. (2001) show that the effect of message framing on the intention to self-perform breast cancer checkups is mediated by perceived self-efficacy. Williams et al. (2001) argued that perceived self-efficacy is elevated by loss frames. Meyerowitz and Chaiken (1987) verified this effect empirically. In opposition to Meyerowitz and Chaiken (1987), Banks et al. (1995) could not demonstrate the described effect of message framing on self-efficacy perceptions.

Block and Keller (1995) provided an argument which can be used to justify the consecutively presumed interaction effect theoretically. It should be noted that Block and Keller (1995) did not take perceived self-efficacy into account but considered the general efficacy of health-related behavior (e.g., immunization against tick-borne diseases) instead. It is assumed that negatively framed information has a stronger behavioral impact when communication messages are processed with stronger cognitive elaboration. Furthermore, low perceived efficacy of a specific behavior leads to a low perception of the probability of success of the corresponding behavior and subsequently to a stronger cognitive elaboration of information processing. Thus, it is assumed that negatively framed messages have a stronger behavioral impact than positively framed messages when perceived self-efficacy is low. Since this is contrary to hypothesis H2’s prediction, one can expect mitigation or over-compensation, respectively, of the effect of gain frames on the willingness to volunteer when self-efficacy perception is low. In the case of strong self-efficacy perceptions the predictions of hypothesis H2 should persist thoroughly. Thus, hypothesis H3 is:

H3: The positive effect of loss frames on the willingness to volunteer is (is not) dampened by low (high) self-efficacy perceptions.

“Arousal: Cost-Reward” Model

Voluntary work constitutes a form of prosocial behavior. Prosocial behavior can be explained by the so-called “arousal: cost-reward” model (e.g., Dovidio et al. 1991). According to this approach, each potential helper evaluates the situational severity, upon meeting the person needing assistance. The potential helper is activated to a greater or lesser degree in correspondence to the assessed severity of need. Potential helpers trade off costs and rewards of helping, based on this activation. The higher the net rewards of helping, the higher is the individual’s willingness to help. Additionally, it is assumed that a stronger emotional activation enhances the willingness to help. This conforms to the empathy–altruism hypothesis which presumes that individual willingness to help is elevated by an emotional activation caused by empathy. This is because it is assumed that the willingness to help is more pronounced when individuals are able to put themselves in someone else’s position (Barnett et al. 1981). Several scholars presume or demonstrate that ad-induced emotional arousal positively impacts advertising’s persuasive power (e.g., Stout and Leckenby 1988). Even though empirical findings regarding the effects of ad-induced arousal are to some extent disparate (Brown et al. 1998), hypothesis H4 reads as follows:

H4: The stronger ad-induced arousal is the higher is the willingness to volunteer.

According to Bolnick (1975), the willingness to volunteer is dependent not only on the impact of social pressure groups and individual rewards of conducting voluntary work. Rather the costs of volunteering also influence the decision to take on voluntary work. Callow (2004) enumerates several categories of costs, which can be traded off against the rewards of volunteering. It can be assumed that lost leisure time is one of the major expenses of doing voluntary work. Of course, the opportunity costs of time vary among different persons. Thus, 1 h spent volunteering can be linked to strongly differing opportunity costs, when different people or categories of people (e.g., students enrolled at a university versus employed person) are considered. Nevertheless and according to the “arousal: cost-reward” model, hypothesis H5 is:

H5: Higher costs of volunteering lead to a lower willingness to volunteer.

If one complies with hypothesis H5 uncritically, one would recommend voluntary organizations to promote voluntary work by communicating minor expenditure of time. However this would be disadvantageous, because volunteers would come to recognize that voluntary organizations understate the costs of volunteering, at the very latest when they begin to do their volunteer work. This would damage voluntary organizations’ credibility and volunteers’ loyalty. The “arousal: cost-reward” model provides a loophole in this regard as it is predicted that the negative effects of the costs of volunteering on the willingness to volunteer can be dampened by (ad-)induced arousal (Klein et al. 2004). Therefore, hypothesis H6 is:

H6: Ad-induced arousal dampens the negative effects of the costs of volunteering on the willingness to volunteer.

Gender-Specific Effects

According to Wilson (2000), women are on average more willing to volunteer.Footnote 5 Corresponding to Brunel and Nelson (2003) there is a received opinion—probably outmoded—that women tend to be more altruistic than men, because of a natural disposition to being more caring and affectionate. Contrary to this, it is assumed that men are more competitive and as a result less affectionate. It has to be kept in mind that social influences may alter these gender-specific dispositions. Within an advertising context Brunel and Nelson (2000) showed that women reacted more positively to advertising which promoted charitable actions. This effect is explained by the notion that women adhere to a more moral weltanschauung. Therefore, hypothesis H7 states:

H7: Women have a more pronounced willingness to volunteer than men.

In addition to the above-mentioned direct effect of gender, gender-specific interaction effects are considered in the subsequent experiments. One must be aware that there is no comprehensive theoretical foundation for these interaction effects. Consequently, the interaction effects’ examination is of an exploratory nature. But one can presume that the effect of the costs of volunteering, for example, is less strong for women than it is for men since women probably have a more pronounced moral weltanschauung.

Figure 1 presents a figurative summary of the hypotheses derived above. It becomes clear that the subsequent experiments primarily focus on altruistic motives at the societal (macro) level and do not take self-centered motives on the individual (micro) level into account.

Fig. 1
figure 1

Summary of proposed research hypotheses Note: within the context of gender-specific effects on the willingness to volunteer merely different levels of willingness to volunteer are expected

Full size image

Research Design and Results

This paper’s concrete research object is the design of a print advertisement for promoting voluntary reading and writing coaching for children. The experiment took place in November 2006 at an University in Southwestern Germany. Coaching was picked because it can be assumed that providing voluntary private lessons is equally appealing to women and men. Other activities may be preferred by one or the other gender. For example, home care activities (athletics training activities) may be more appealing to women (men). Altogether two experimental studies based on two different student samples were conducted. A print advertisement promoting volunteering was presented to each of the studies’ participants. The corresponding print ads were manipulated experimentally. After seeing the ads each test person indicated her/his willingness to volunteer. Thereafter, questions regarding the success of the experimental manipulations were considered.

Experiment 1

Research Design

This study had a 2 (“Self-efficacy”) * 2 (“Message framing”) factorial design. Subjects were students of economics enrolled in a German university. Overall, 140 students were randomly assigned to experimental conditions. Due to incomplete responses the number of useable questionnaires was ultimately reduced to 131 (Appendix B). Respondents’ average age was 21.5 years (minimum age: 18 years; maximum age: 37 years). Furthermore 48.5% of the test subjects were women. No significant differences with regard to the distribution of gender (χ²(3, n = 131) = 3.58, p > .10) and age (F(3, n = 128) = .47, p > .10) between the four experimental groups were detected.

Independent Variables

Message framing was manipulated by presenting the positive (negative) consequences of (not) participating in voluntary coaching (Loss frame: “Without your help, a child won’t learn to read!” versus “With your help, a child will learn to read!”). A depiction of laughing or crying children, respectively, supported message framing visually. Self-efficacy was manipulated by modifying the print ads’ headline (Low self-efficacy: “Without your help a child won’t learn to read!” or “With your help a child will learn to read!”; High self-efficacy: “Without your help five children won’t learn to read!” or “With your help five children will learn to read!”).

Dependent Variable

Willingness to volunteer was considered as dependent variable. Three 5-point rating scale items were used to measure willingness to volunteer (Appendix A). Because Cronbach’s alpha amounts to .87, internal reliability can be assumed. It has to be kept in mind that the willingness to perform a specific behavior, which test persons indicate during surveys, does not correlate perfectly with subsequent behavior.Footnote 6 Thus, the current study tells more about the determinants of people expressing that they would like to (start to) volunteering than about the determinants of actual volunteering behavior.

Manipulation Check

Subjects appeared to perceive the message framing and self-efficacy manipulations as intended. The self-efficacy perceptions of low self-efficacy subjects were significantly lower than those of the high efficacy ones (F(1, n = 131) = 4.06, p < .05).Footnote 7 Subjects in the (loss) gain frame condition perceived that the message stressed (did not stress) the pros of volunteering rather than the cons of not volunteering to a greater extent (F(1, n = 131) = 9.82, p < .01).Footnote 8

Effects on Willingness to Volunteer

As a first result, significant main effects of message framing (F(1, n = 131) = 2.39, p > .10) as well as of perceived self-efficacy (F(1, n = 131) = .34, p > .10) were not detected. Therefore hypotheses H1 and H2 have to be disapproved. Contrary to this, the interaction of message framing and perceived self-efficacy was found to be significant (F(1, n = 131) = 7.63, p < .05). Figure 2 depicts this two-way (self-efficacy * massage framing) interaction’s disordinal nature. Contrary to hypothesis H3, low (high) perceived self-efficacy has a stronger (less strong) effect on willingness to volunteer within the scope of gain frames (loss frames). On the one hand, therefore, hypothesis H3 cannot be confirmed when high self-efficacy perception persist. On the other hand, the predictions of hypothesis H2 hold true under the condition of low self-efficacy.

Fig. 2
figure 2

“Self-efficacy” by “Message framing” interaction on Willingness to volunteer (estimated marginal means)

Full size image

Moreover, a significant main effect of the respondents’ gender was detected in a 2 (“Gender”) * 2 (“Perceived self-efficacy”) factorial design (F(1, n = 131) = 10.28, p < .01). Because the willingness to volunteer was higher for women than it was for men, hypothesis H7 can be confirmed. Again the main effect of self efficacy was not significant (F(1, n = 131) = .92, p > .10). In contrast, the interaction between respondents’ gender and perceived self-efficacy was significant (F(1, n = 131) = 4.76, p < .05).Footnote 9 The detected interaction effect of gender and self-efficacy can be classified as ordinal (see Fig. 3). Furthermore, the mean difference in willingness to volunteer between women with low and those with high self-efficacy perceptions was not significant (F(1, n = 64) = .90, p > .10). Manipulation of self-efficacy had an impact only on men’s willingness to volunteer.

Fig. 3
figure 3

“Self-efficacy” by “Gender” interaction on Willingness to volunteer (estimated marginal means) Note: there is no significant mean difference between women in the high and low self-efficacy condition

Full size image

Discussion

The surprising study results can probably be explained by the particularity of voluntary work. In contrast to diverse health-related behaviors, for example, volunteers struggle to attain collective goals (in this case, reduction of the rate of illiteracy among children). Thus, individual behavior may be based on a different decision calculus. First, it can be presumed that people with high self-efficacy perceptions rely on other members of society more heavily. This idea originated in work conducted by Bandura. According to Bandura (1977), an individual’s self-efficacy perceptions are influenced by the evaluation of other people’s efficacy in performing specific behaviors. Looked at from the other side, one can presume that individuals with high self-efficacy perceptions may think that other people also exhibit high efficacy in volunteering, for example. Second, voluntary work constitutes a public good. A crucial characteristic of public goods is that they are public—that is, they are available to everyone including those who did not contribute to their production. Thus, individuals are confronted with an incentive to take a free ride within the scope of voluntary work (Diekmann 1985; Unger 1991).

With reference to helping behavior, Darley and Latané (1968) illustrate a so-called phenomenon of diffusion of responsibility. Diffusion of responsibility comes about when people think that not only they but also others can help persons in need. One can also pose the argument that people always try to shift responsibility onto someone else when they think that that person has the same or even better helping skills. Therefore, free-riding incentives should be more pronounced when individual self-efficacy perceptions are high, because high self-efficacy is assigned to other persons. If the idea that loss frames result in deeper cognitive information processing is maintained, the detected interaction effect of message framing and self-efficacy can be explained. Whereas the incentives to free-ride impact greatly upon the willingness to volunteer under the condition of gain frames, people with high self-efficacy perception are more aware of their responsibility under the condition of loss frames. Thus, free-riding incentives or the diffusion of responsibility should be less distinct when loss frames are communicated.

Besides these empirical findings it has to be highlighted that the effect of message framing on willingness to volunteer conforms to the predictions of hypothesis H2 under the condition of low self-efficacy. First, this can be explained by the fact that incentives to free-ride should be less distinct when individuals hold low self-efficacy perceptions. Second, and contrary to hypothesis H3, it appears that low self-efficacy does not lead to deeper processing of advertising stimuli. Hence, the findings of Block and Keller (1995) regarding the interaction effect of general efficacy and message framing do not apply to this paper.

Interestingly, the manipulation of self-efficacy perceptions has an impact only on men’s willingness to volunteer. One factor which might explain this finding is that women show a fairly high tendency to volunteer as result of a supposedly more distinct moral weltanschauung. It seems that this pronounced inclination to voluntarism cannot be elevated further by the manipulation of self-efficacy perceptions.Footnote 10 Another possible explanation might be that women process advertising stimuli with higher cognitive elaboration (Meyers-Levy and Maheswaran 1991). In contrast, men often use heuristics to process communication messages (Darley and Smith 1995). Therefore, the result that self-efficacy perceptions primarily influence men’s willingness to volunteer can be explained by men’s superficial and uncritical processing of advertising stimuli.

Experiment 2

Research Design

The second experiment had the same research object: voluntary coaching of children in reading and writing skills. In order to validate the deduced research hypotheses, a 2 (“Ad-induced arousal”) * 2 (“Costs of volunteering”) factorial design was used. Another sample of 140 test persons was interviewed. The experiment took place in December 2006 at an University in Southwestern Germany. One hundred and thirty-one useable questionnaires were collected (Appendix C). Respondents’ mean age was 21.5 (minimum age: 18 years; maximum: 30 years). Fifty percent of the interviewees were women. As in the first experiment, there were no significant differences regarding the distribution of gender (χ²(3, n = 128) = .26, p > .10) and age (F(3, n = 129) = .39, p > .10) among the four experimental groups.

Independent Variables

Ad-induced arousal was manipulated by either depicting crying children together with an emotionalizing slogan (“Illiteracy equals poverty!”) or depicting a stylized reading person together with a neutral slogan (“Better reading, better writing.”). Costs of volunteering were manipulated by communicating different amounts of time that people would have to spend for voluntary work (1 h per week versus 3 h per week). Within this context it has to be made clear that individual opportunity costs of spending 1 h of free time doing voluntary work may vary between different social groups or even between members of the same group. However, confounding effects of varying opportunity costs are not of pronounced relevance here. First, by using a between-subjects experimental design, which generally assures internal validity, this problem was dampened. Randomization of subjects should have led to an equal distribution of test persons with different opportunity costs across each of the four experimental groups. Second, because all test persons are students enrolled in university, one can presume that the opportunity costs would not vary strongly.

Dependent Variable

Intention to volunteer was again chosen as a dependent variable. A three-item measure that was based on a 5-point rating scale was utilized (Appendix A). The value of Cronbach’s alpha, which amounts to .91, indicates high internal reliability.

Manipulation Check

The manipulation check regarding ad-induced arousal was based on a two-item measure (“The presented print advertising ...” (a) “... is not emotional!” = 1 to “... is emotional!” = 5 and (b) “... does not trigger compassion!” = 1 to “... triggers compassion!” = 5; Pearson’s correlation coefficient = .76, p < .01). Ad-induced arousal’s manipulation was successful (F(1, n = 125) = 58.02, p < .01). The same holds true for the manipulation of perceived costs of volunteering (F(1, n = 125) = 4.90, p < .05). A single-item approach was used to capture perceived costs of volunteering (“To spend one (three) hours of my free time volunteering ...”: “... is not laborious for me!” = 1 to “... is laborious for me!” = 5).

Effects on Willingness to Volunteer

The main effect of ad-induced arousal was significant F(1, n = 131) = 8.44, p < .01). Contrary to this, the main effect of the perceived costs of volunteering had no significant impact on the willingness to volunteer (F(1, n = 131) = .71, p > .10). In addition to this, a significant interaction effect of ad-induced arousal and perceived costs of volunteering was found (F(1, n = 131) = 2.76, p < .10). Figure 4 illustrates this ordinal interaction effect. Furthermore, no significant mean difference regarding the willingness to volunteer between the experimental groups “High ad-induced arousal”/”Low costs of volunteering” and “High ad-induced arousal”/”High costs of volunteering” was detected. This means that the perceived costs of volunteering impact solely on willingness to volunteer when ad-induced arousal is low. Hence, hypotheses H4 and H6 can be confirmed.

Fig. 4
figure 4

“Costs of volunteering” by “Ad-induced arousal” interaction on Willingness to volunteer (Estimated marginal means) Note: there is no significant mean difference between test persons in the “High arousal/High costs”-condition and “High arousal/Low costs”-condition

Full size image

Within the context of an additional two-factorial ANOVA, a significant main effect of gender (F(1, n = 131) = 21.22, p < .01) as well as of ad-induced arousal (F(1, n = 131) = 9.82, p < .01) was detected.Footnote 11 Hypothesis H7 was confirmed again. Over and above this, a significant interaction of gender and ad-induced arousal was found F(1, n = 131) = 3.01, p < .10). Figure 5 depicts the corresponding ordinal interaction effect. One can easily see that men have a lower willingness to volunteer. Furthermore, and based on supplementary one-factorial ANOVAs, it turned out that ad-induced arousal impacts men’s willingness to volunteer (F(1, n = 63) = 9.47, p < .01), on the one hand. On the other hand, women’s willingness is not affected by ad-induced arousal (F(1, n = 65) = 1.12, p > .10).

Fig. 5
figure 5

“Ad-induced arousal” by “Gender” interaction on Willingness to volunteer (estimated marginal means)

Full size image

Discussion

Contrary to the first experiment, the second experiment’s results were somewhat less surprising. With regard to the effects of ad-induced arousal (perceived costs of volunteering) research hypotheses derived from the “arousal: cost-reward” model were (partially) supported. While ad-induced arousal strengthened willingness to volunteer, costs of volunteering negatively impacted test persons’ inclination to take on voluntary work in the case of low ad-induced arousal exclusively. The non-significance of the costs of volunteering as a main effect might have been caused by testing the research hypotheses on the basis of a student sample. As mentioned above, and in comparison to employed persons for example, a student enrolled in university apparently has more leisure time at her/his disposal. Therefore one can expect the effect of the costs of volunteering to be more pronounced when a representative sample is considered.

Regarding gender-specific effects, it can be stated that ad-induced arousal impacts men’s willingness to volunteer to a greater extent than it does women’s. Similar to the first experiment, a ceiling effect caused by a distinct moral weltanschauung that got in the way of a stronger effect of ad-induced arousal may be presumed.

Summary of Results and Study Limitations

First, one interesting result of the study was the converse of message framing’s effect on the willingness to volunteer in the case of different levels of perceived self-efficacy. In the case of low (high) self-efficacy, gain frames strengthened (dampened) individuals’ inclination to take over voluntary work. It can be presumed that this converse was caused by reinforcement (attenuation) of incentives to free-ride. The second empirical finding was that willingness to volunteer was affected by the costs of volunteering only among male test persons not confronted with arousing print advertisement. Thus, when only men were considered, ad-induced arousal dampened the impact of the costs of volunteering on the willingness to volunteer. Another interesting gender-specific result and the study’s third finding was that the manipulation of perceived self-efficacy affected only men’s willingness to volunteer. Fourth, women showed a higher inclination towards voluntary work. At the same time, however, the impact of the considered print ads’ manipulations was less distinct for women than it was for men. Figure 6 summarizes the results of both experiments.

Fig. 6
figure 6

Summary of study results

Full size image

There were some factors which might curtail the studies’ explanatory power. First, the study was not based on a representative sample; a student sample was used. However, this procedural method is widely accepted in experimental research. Furthermore and according to Gensicke et al. (2006), the rate of volunteerism among German students is not distinguishable from the rate among the total German population. Besides this, there is no apparent reason to presume that students’ decision calculi differ distinctly from that of non-students’. Because of their constrained time budget and the therefore high opportunity costs of time it can be assumed that the manipulation of the costs of volunteering will have a more pronounced effect on employed persons’ willingness to volunteer than on non-employed persons. External validity of results might be limited in this regard.

Second, the study focused on voluntary coaching of children in writing and reading skills. Gender-specific effects might vary over different domains of voluntary work. For example, the inclination to volunteer as a soccer coach or at auxiliary fire brigade might be stronger among men than among women. Thus, gender-specific effects may vary between different kinds of voluntary work.

Third, the study was based on a manipulation of print advertisements. One cannot be sure to what extent the empirical findings apply to other types of advertising (e.g. radio ads, television ads, or public relation measures). On the one hand, an application of visually supported manipulations of ad-induced arousal is not possible within the scope of radio ads. The medium can spawn emotional arousal only through verbal elements, music, and sound. On the other hand, visually supported manipulations of ad-induced arousal might be more powerful in television commercials. However, there are no obvious limitations to the utilization of the other factors manipulated (message framing, perceived self-efficacy, and costs of volunteering) within the context of voluntary organizations’ communications policies. Hence and regarding these aspects, the empirical findings of this study can probably be applied to all types of advertising.

Managerial Recommendations, Future Research, and Outlook

Several managerial recommendations for voluntary organizations can be developed from this paper’s empirical findings. It must be reiterated that this study primarily focused on the determinants of the willingness to volunteer. Hence the determinants of actual volunteering behavior are considered only implicitly, if at all.

The first recommendation is that self-efficacy perceptions should not be strengthened when gain frames are used to promote volunteerism. An elevation of perceived self-efficacy would probably lead to pronounced incentives to take a free ride. Second, if voluntary organizations strive to motivate potential volunteers with the use of loss frames, self-efficacy perceptions should be enhanced concurrently. Apparently, people with higher confidence in their own skills seem to accept more responsibility when the negative consequences of not volunteering are highlighted. Third, people with special skills should be motivated by communicating loss frames and accentuating high individual self-efficacy.Footnote 12 Another relevant aspect of this study’s findings relates to people who try to improve their own competencies during times of unemployment by taking part in voluntary projects. It can be assumed that individual self-confidence and subsequent self-efficacy perceptions are lower when unemployed than when employed. The fourth recommendation, then, is that gain frames should be used if voluntary organizations want to appeal to the unemployed.

When voluntary organizations are thinking about gender-specific aspects of advertisement design, it should be kept in mind that the study found ad-induced arousal as well as the manipulation of self-efficacy perceptions had an impact primarily on men’s willingness to volunteer. Given that finding, the fifth recommendation is that these communication design options should be considered when male volunteers are being sought (e.g., placement of print ads in magazines with high proportion of male readers). Ad-induced arousal seems to dampen the perceived costs of volunteering in this context. This aspect might be even more important for male employees who have constrained time budgets. Because women cannot—to the same extent as men—be motivated to volunteer by the means of persuasive advertisement, the sixth recommendation is that the advertising budget should not be spent on persuasive print or television advertisements targeted to women. It might be more advantageous to motivate women by means of informative communication strategies. Following this idea and within the scope of television advertisement, it might be beneficial to place two separate ads during an advertising break. For example, an informative advertisement for women could be placed at the beginning of a commercial break with an additional emotion-arousing ad broadcasted at the end.

There remain several aspects in this field well worth researching. First, a generalization of this study’s empirical findings could be brought forward by conducting replica studies which consider different forms of voluntary work as well as different types of advertising. Second, the research framework of this paper could be applied to other forms of prosocial behavior, such as the promotion of donation behavior or ecology-minded behavior. Third and in the context of the “arousal: cost-reward” model, the proposed framework could be enhanced by considering volunteers’ motivations for volunteering. An interesting research project would be to examine whether ad-induced arousal strengthens the effect of egoistic and altruistic motivations on the willingness to volunteer to different extent. The “arousal: cost-reward” model would predict that the effect of individual motivations would be enhanced by ad-induced arousal. Voluntary organization could take advantage of this moderating effect of ad-induced arousal within the context of their communication strategies. Fourth, it would be interesting to analyze how different types of advertisement could be designed against the background of this study’s gender-specific results; that is, how informative and emotionally arousing communication elements, for example, can be combined optimally. Fifth, another important area for future research would be a more in-depth analysis of those cognitive processes which were not considered explicitly in this study. The converse of the effect of message framing, especially, which is probably due to a variation of free-riding incentives, should be taken into account.

Philanthropic behavior like volunteerism is a fundamental component of civil society. Volunteering strengthens social coherence and sustainability. This holds especially true for Western European nations that are currently undergoing a process of radical social change which has triggered the need to reconstruct social welfare systems. Because of these social and economic circumstances, it is crucial for voluntary organizations to know how to motivate potential volunteers. For that purpose an empirical study was conducted based on the “arousal: cost-reward” model of prosocial behavior and findings on persuasive advertisement. In this connection, this study’s findings contribute to a more in-depth understanding of individual motivation to volunteer as well as to the question of how voluntary organizations’ communication policy is designed optimally.