Slacktivism is a portmanteau of “slacker” and “activism.” This term is used to describe “lazy activism,” or low-cost and low-risk support for a cause, which has almost zero impact on the cause itself (Morozov, 2009). Slacktivism can be expressed through social media (e.g., liking a post), although in this paper we also use the term slacktivism for real-life display of support, such as wearing a badge. Kristofferson et al. (2014) define slacktivism as token support that is marginally helpful toward for a cause, in contrast with meaningful support (e.g., tangible contributions such as giving money or time). One of the core implications of slacktivism, also known as the slacktivism effect, is that such low-cost display of support for a cause does not lead people to actively become involved in it and can even reduce subsequent commitment to acting for the same cause. The advertisement from the 2013 UNICEF campaign “Likes don’t save lives” shows that charities are aware of this problem: “Like us on Facebook, and we will vaccinate zero children against polio. We have nothing against likes, but vaccine costs money. Please buy polio vaccine at unicef.se. It will only cost you 4€, but will save the lives of 12 children.”

The slacktivism effect has been explained by a unique desire to fulfill a need in self-esteem (Lee and Hsieh, 2013; Kristofferson et al., 2014). By engaging in a low-cost action, especially in front of others, individuals show endorsement for a good and charitable cause. This places them in a favorable light, which can in itself boost their self-worth. In turn, since self-enhancement needs have been adequately satisfied, efforts for subsequent, more costly behavior appear useless to them. Studies on symbolic self-completion have shown that participants prevented from completing a task describing them positively tried harder to give a good impression to others (Wicklund and Gollwitzer, 1981; Gollwitzer et al., 1982). Thus, participants who are not given the opportunity to give a good impression to others can be predicted to try harder subsequently, compared to participants who had the opportunity to display that they are “a good person” by having performed a low-cost act of support.

Despite the popularity of the slacktivism phenomenon narrative (Morozov, 2009; Robertson, 2014), there is little empirical evidence of actual decrease in behavioral engagement after an initial low-cost act, as highlighted by Jones (2015). Strong evidence for the slacktivism effect would be the observation that participants who had publicly displayed their support for a cause engaged in less subsequent behavioral engagement than participants in a control condition of no initial act of support. Among the few empirical studies examining this question, Kristofferson et al. (2014) have interpreted the difference between private and public conditions as due to a slacktivism effect, but they found no significant difference between the initial act of support condition and the control condition (no initial act of support) in two of their studies (Study 1 and 3). One of the rare research studies showing reduction in activism after an initial act of support can be found in Schumann and Klein (2015). Compared to participants in the control condition (exposed to a website and then re-directed to the post-questionnaire), participants in the experimental condition (asked to post a comment on the website) were less willing to join an offline action.

Interestingly, a number of experiments studying slacktivism showed evidence of an increase in subsequent behavioral engagement after an act of support, especially when the subsequent action is related to the first cause (Lee and Hsieh, 2013; Kwak et al., 2018; Lane and Dal Cin, 2018). In fact, the prediction that an initial act of support should reduce the likelihood of a subsequent act is at odds with predictions raised by the commitment theory, which we will delve into in the next section.

According to the commitment theory (Kiesler, 1971; Joule and Beauvois, 1989), the fact that someone freely performs an initial action, in our case wearing a ribbon in support of blood donations, should increase the probability that he or she engages in a second related behavior such as volunteering for an event promoting blood donation. Studies provide support for the commitment-theory effect in various domains (Dillard et al., 1984), including blood donation (Hayes et al., 1984; Fonte et al., 2017), although this has not been consistently replicated in all studies [e.g., Gamian-Wilk and Dolinski, 2019; see Foss and Dempsey (1979) for an example in the field of blood donation].

Close to the subject at hand in this paper is the foot-in-the-door effect (Freedman and Fraser, 1966). This phenomenon, predicted by the commitment theory, occurs when making someone accept to engage in an initial low-cost behavior (called initial act of support) increases the chances of acceptance of a second, more costly behavior (Joule et al., 2007). The circumstances in which the behavior is engaged in are decisive for the individual to feel committed, notably that the person feels they are free to accept or refuse, and has no external justification for engaging in the behavior other than goodwill (Joule et al., 2007), costs of requests (Reingen and Kernan, 1977), the pro-social character of the request (Dillard and Hale, 1992) and delay between the two requests (Chartrand et al., 1999). In contrast to the slacktivism effect, commitment theory and the foot-in-the-door effect predict that a low-cost supportive action (a like on a post) should increase the likelihood that the person will adopt other behaviors to support the same cause. According to Chartrand et al. (1999), “the effect appears more often than would be expected by chance. Hence, the foot-in-the-door phenomenon first demonstrated by Freedman and Fraser (1966) is real. Unfortunately, this is only part of the story. […] The literature is filled with failures to replicate and occasional reversals of the effect” (p. 212). The same observation was made by Burger (1999), who noted that a large number of studies either failed to show increase, or found a decrease in compliance after a foot-in-the-door request compared to a control condition.

Research describes two antagonist effects, namely an increase or a decrease in behavioral engagement following an initial support action. These opposite predictions, opposing commitment theory to the slacktivism effect, have already been highlighted in previous studies (Lee and Hsieh, 2013; Kristofferson et al., 2014; Schumann and Klein, 2015; Lane and Dal Cin, 2018). One possibility for these predictions of opposite effects is that both slacktivism and commitment effects appear concomitantly. This can lead to neutralizing the impact of the low-cost action. But in certain circumstances it is possible that one of the two processes becomes more powerful than the other and causes an observable effect (positive vs. negative) of the low-cost action on subsequent behavior. In this paper, we propose to test the idea that when people are motivated to truly care for the cause, the commitment effect should occur more than the slacktivism effect, while, on the other hand, when they accept to carry out the initial act with the motivation of giving a good impression, then the slacktivism effect (i.e., a decrease in subsequent behavioral engagement) would arise more than commitment. We refer to impression-management motives in the latter case, and prosocial motivation for the former.

Impression-management (IM) motives, as discussed by Leary and Kowalski (1990), correspond to the desire to control one’s personal image and to show oneself to other people in a desirable way. This motivation has been mostly studied in the context of organizational citizenship behavior such as helping co-workers (Bolino, 1999). Interestingly, Hui et al. (2000) showed that employees who considered that organizational citizenship behavior would help them reach a goal (a promotion) made more such behavioral engagement, but also cut back on them after the goal was attained. Thus, a seemingly helping behavior can be enacted with a motivation other than helping.

The other side of the coin is prosocial motivation (PSM), which signifies that the reason behind engagement in a behavior is to benefit someone (Grant and Berg, 2010). The closeness between PSMs and altruism, as well as other concepts, have been discussed elsewhere (Schott et al., 2019). Given the extended debate on those terms (Cialdini et al., 1997; Batson, 2017), including in the field of blood donation (Evans and Ferguson, 2014), we take here the position of sticking to the term “prosocial.” According to literature in organizational psychology, PSM is opposed to IM as an explanation underlying the engagement in a given behavior, in one case to “do good” and in the other case to “look good” (Grant and Mayer, 2009).

Two papers have examined the role played by IM motives on the slacktivism effect. First, Lane and Dal Cin (2018) measured a score of IM before and after having asked participants to share a video promoting a social cause on Facebook, in two experimental conditions (i.e., public sharing vs. anonymous sharing). Contrary to their hypothesis based on slacktivism, they found that publicly sharing a video on social media increased willingness to volunteer. They also found that IM was not a mediator of the effect of publicly vs. anonymously sharing the video on willingness to support the cause. Second, Kristofferson et al. (2014) argued that a private condition (i.e., participants receiving a pin in a small envelope) should activate consistency motivation, while a public condition (i.e., participants being requested to wear a pin conspicuously on their coat or shirt) should activate IM motivation. The idea is that participants who would have already displayed their support for the cause in public would no longer need to engage in subsequent behavior (i.e., donate money). This reasoning is not consistent with commitment theory, which predicts that if an initial act is made in public, people are more committed to act in line with it and to engage further in subsequent related behavior. Thus, the public vs. private experimental manipulation does not make it possible to determine which of the competing predictions is accurate, i.e., whether there is increase or decrease in subsequent behavioral engagement after an initial act of support. Moreover, the Kristofferson et al. (2014) study did not test the effect of PSM, which we address in the present research.

Our research aims to explain why an initial low-cost, low-impact act of support for a cause leads to either more or less behavioral engagement. Our hypothesis is that the type of motivation by which the person engages in an initial act of support acts as a moderator. We predict that, if an individual is motivated to help a cause for prosocial reasons when agreeing to wear a ribbon or sign a petition (i.e., PSM), he or she would be more committed to contribute to the same cause when later asked to do so (H1). On the other hand, if acceptance of the initial act is driven mainly by desire to make a good impression (i.e., IM), then signaling that one is in favor of a cause via an action that is visible but that has no real impact could be enough to fulfill this motive, and the person will be less inclined to take other actions in support of the cause (H2). Consequently, we predict that the effect predicted by commitment theory (i.e., increased behavioral engagement after an initial act compared to control conditions) is driven by PSM, while the slacktivism effect (reduced behavioral engagement after an initial act vs. control) is driven by IM.

To test this hypothesis, we experimentally manipulated IM and PSM. In a two-by-two between-subjects design, participants were randomly assigned to a PSM condition or an IM condition. They were also randomly assigned either to an “initial act of support” condition or to a control condition (i.e., with no initial act of support). We then measured behavior (Study 1 and 3) and/or behavioral intention (Studies 2 and 3). Three studies were conducted following this experimental design, all about blood donation. We then summed up the results of the studies in a small-scale meta-analysis. We report how we determined our sample size, all data exclusions, all manipulations, and all measures for all studies (Simmons et al., 2012). The research project was approved by the University ethics committee (n° PSE.20190108.MM).

Logistic regression revealed that the interaction between motivation type and initial act of support (presence/absence) on behavior was not significant [Wald = 2.20, p = 0.138, OR = 3.40, 95% CI (0.67, 17.10)]. Two additional logistic regressions were run to test the effect of the initial act of support in each condition of motivation. Data supported our first hypothesis (H1): in the PSM condition, we found that the initial act of support had a significant positive effect on volunteering, Wald = 6.23, p = 0.013, OR = 4.07, 95% CI [1.35, 12.27]. In the IM condition however, the initial act of support did not have any effect on volunteering, Wald = 0.09, p = 0.764, OR = 1.20, 95% CI [0.37, 3.90]. Thus, the hypothesis of a decrease in subsequent behavioral engagement after an initial act of support in the IM condition (H2) was not supported. Percentages of acceptance to volunteer according to experimental conditions are presented in Figure 2.

Analyses of the follow-up data revealed that 68% of participants who accepted the badge did not actually wear it. Ten percent mentioned not having received one (despite being in the badge condition). To those who said they had worn the badge, we asked for how long they did so. The lowest duration was 5 h/the whole day (four persons), and nearly all the others reported still wearing it 1 week later.

In Study 1, we found that, when PSM was induced, an initial act of support (accepting a blood donation badge) produced an increase in acceptance of the subsequent behavior (spending time at a blood donation promotion event) in comparison with a control condition. This result is consistent with the idea that the foot-in-the-door effect occurs particularly when there is a strong PSM. However, as the effect of the initial act of support was non-significant in the IM condition, this precludes giving support to the slacktivism effect. The low level of volunteering in both conditions of IM might have induced a floor effect, preventing the appearance of a slacktivism effect. Most participants did not wear the badge, according to their answers in the follow-up survey 1 week later. This result is not an issue in itself, as the dependent variable (DV) was measured just after badge acceptance, but it suggests that badge distribution is not the optimal low-cost act of support.

Manipulation checks were analyzed with 2 × 2 ANOVAs estimating main effects of motivation and act of support, as well as their interaction. This showed that participants in the PSM condition had higher scores on the prosocial items (M = 5.08, SD = 0.09) than participants in the IM condition (M = 4.80, SD = 0.10), F(1,139) = 4.16, p = 0.043, η²_p = 0.03. Main effect of the act of support and interaction effect were both non-significant respectively, [F(1,139) = 0.59, p = 0.808, η²_p < 0.01; F(1,139) = 0.03, p = 0.854, η²_p ≤ 0.01]. Even though means are descriptively consistent with expectations, the difference on the IM items was not significant between the conditions of motivation, F(1, 141) = 1.94, p = 0.166, η²_p = 0.01, (M_PSM = 3.43, SD = 0.11, M_IM = 3.17, SD = 0.12), neither was the main effect of the act of support F(1,141) = 1.837, p = 0.185, η²_p = 0.01, nor their interaction, F(1,141) = 1.837, p = 0.177, η²_p = 0.01.

An ANOVA showed that having donated in the past had a significant effect on intention to donate blood [F (1,140) = 60.03, p < 0.001, η²_p = 0.30, 90%³ CI (0.19, 0.38)]. However, the experimental manipulations did not interact with each other [F (1,140) = 0.72, p = 0.397, η²_p = 0.01, 90% CI (0, 0.04)]. Simple effects tested through ANOVAs revealed that neither H1 nor H2 were supported. In the PSM condition, intention did not significantly differ between the condition with the initial act of support and control condition, F(1, 80) = 0.60, p = 0.441, η²_p = 0.007, 90% CI [0, 0.07]. Similarly, in the IM condition, no significant effect emerged either, F(1, 59) = 0.16, p = 0.690, η²_p = 0.003, 90% CI [0, 0.06]. Results are presented in Figure 3.

Analysis of the number of thoughts listed related to blood donation showed that, in the IM condition, there was no effect from the initial act of support, F(1, 59) < 0.01, p = 0.997 (M_{act of support} = 2.95, SD = 0.25; M_control = 2.95, SD = 0.29). Similarly, in the PSM condition, the number of thoughts listed did not vary according to the experimental conditions, F(1, 80) = 0.05, p = 0.819, (M_{act of support} = 3.20, SD = 0.21; M_control = 3.13, SD = 0.21).

Study 2 did not replicate Study 1 in the PSM condition: the positive effect of the initial act of support was not found in Study 2. No effect of the initial act of support emerged when the IM motivation was activated, both in Study 1 and 2. Using behavioral intention as the sole DV in Study 2 might be a limitation, as DVs are mostly observed behavior in the literature on slacktivism and foot-in-the-door [e.g., Joule et al., 2004; Kristofferson et al., 2014; Studies 1 and 5; but see Fonte et al. (2017)]. It is possible that intention is an indicator of a more conscious level of processing and that our effects of interest are not captured by such DV. Another limitation is that the online setting might have rendered less credible the fact that the first and second part of the study were unrelated, and led participants to doubt our cover story. Finally, the power analysis based on Study 1 might not have been adequate, as the effect sizes issued from studies with small N can be part of a wide prediction interval and not replicate with the same magnitude in further studies (Patil et al., 2016). Thus, a more conservative effect size could have been used, and the absence of effect in Study 2 could also be linked to an insufficient statistical power.

Zero-order correlations between each intention items and behavior were calculated. Intention to donate in the next 4 months correlated at 0.471 (p < 0.001) with donation behavior, and intention to donate the following week correlated at 0.385 (p < 0.001) with donation behavior.

Interactions and simple effects were computed in ANOVAs for both measures of intention and in logistic regression for the measure of behavior.