## Motivation

We often encounter messages online that ask us to act, either towards personal wellness goals, reminders, or appeals for charitable donations. For example (Fig 1), Wikipedia often embeds a short text message in their banner to ask readers for a small amount of donation.

Making these short text messages more effective has important implications in stimulating behaviors, especially for organizations interested in alleviating public goods dilemmas.

There is a rich history of prior work in psychology that investigates how to construct effective persuasive messages. Goldstein et al. examined the idea of social proof, persuading by informing the persuadee how other people behave when facing the same choices. The study showed that compared to the standard text messages in hotels asking the customer to reuse towels, the message mentioned how other customer’s towel reuse behavior significantly increased towel re-usage (Fig. 2).

However, the idea of social proof does not always work well by itself. Schultz et al. experimented with the monthly power bill. The researcher incorporated the idea of social proof by informing homeowners how much energy their neighbors consumed (Fig 3) to motivate homeowners to reduce their power usage if they use more than their neighbors. Surprisingly, compared to the standard power bill, the social proof version does not work as expected. Homeowners did not reduce their power usage.

Why social proof doesn’t work? The results from the follow-up study are intriguing. By adding emoticons to the social proof version (Fig 3), the social proof version successfully reduced the power usage of homeowners who consume more power than their neighbors.

From previous studies, we know

• People are conditional cooperators.
• Communicating affect is important to make a message more persuasive.

Motivated by those two intriguing ideas, we looked at a highly expressive medium for persuasion, comics. Comics has everything we need for an effective persuasive message. It is simple, eye-catching, expressive, and memorable. Although the benefit of comics in delivering informative messages has been examined in areas, like education, comic’s persuasiveness has not been fully explored yet.

In this study, we focused on one specific type of comic, abstract comics (Fig 4). Abstract comic de-emphasizes character detail (face, eyes, etc.) or details about the locale. As Scott McCloud points out, using abstract representations for the comic not only increases universality but also allows the reader to project themselves onto the comic character and empathize with the character and with the situation. The abstract form has an additional benefit: since the form is visually spare, it allows for comic panel algorithmic synthesis and for the personalization of messages.

To examine the persuasive aspect of the abstract comic, we experimented with an online charitable donation task. We answer two research questions:

• RQ1: Does the use of the abstract comic form increase the level of donation over the plain text message?
• RQ2: What is the effect of introducing social proof in the abstract comic, when compared to the comic without the social proof?

## Choice of the Persuasion Task

There are many different compelling behavioral contexts on which to test the role of the comic form in persuasion: personal wellness goals (e.g., diet, exercise), mundane tasks (e.g. “pick up dry cleaning”), as well as broader public-goods issues (e.g. “take the flu shot;” “donate to cure cancer”). We consider four criteria when we chose our experiment context: the nature of the reward; single-shot tasks; an ecologically valid task; the absence of specialized knowledge to perform the task.

First, we would like the rewards to be distant, and non-exclusive, rather than proximal and exclusive so that individuals won’t perform the task in anticipation of the immediate reward. Second, while some longitudinal tasks (e.g., eating healthy) have distant rewards (better health condition), and can positively affect the public good (with more healthy people, in the long-run, insurance rates will fall), these tasks are prone to habit formation, a potential confound. Third, we would like to ensure that the experimental task is ecologically valid—a task that these individuals would be actually asked to perform in the real world, outside of the experimental context. Fourth, we would like the task not to require specialized knowledge (e.g. “asking doctors to make a decision”), so that other researchers could easily replicate and scale.

Online charitable donation tasks satisfy those criteria as 1) they are single-shot tasks, 2) they contribute to the public good with distant, non-exclusive rewards, 3) requests for charitable donations frequently occur online, and 4) the task allows for experimental replication.

Specifically, we asked donation for the Organization for Autism Research (OAR). We chose OAR for three reasons. First, Autism Spectrum Disorder (ASD), is a well-recognized developmental disorder that impairs communication and behavior which provides basic interest for the participants to support the related charitable organization. Second, the Organization for Autism Research is one of the most visible ASD related organizations that helps individuals with autism and provides assistance to their parents, families, teachers, and caregivers. The goal of OAR is clear and reputable so participants won’t question the authenticity of our message’s motive. Finally, we wished to avoid a charity associated with a life-threatening condition such as cancer as it may create an experimental confound: we don’t know if someone donates because of their intrinsic desire to help with a life-threatening situation. While ASD can have serious consequences on the well being of those who have it, the public perception is that ASD is not-life threatening.

## Experiment

We designed our experiment in the following way. The experiment contains three steps (Fig. 5). The first step is to create a context. In the real world scenario, the persuasive message often comes with a context. For example, you will see Wikipedia’s banner only if you are reading the Wikipedia article or you are a member of the organization. We created the context by asking the participants to watch a short promotional video produced by OAR and write a short feedback.

In the second step, we used showed our messages and asking participants to make donation decisions. We randomly assigned each participant to one of the three conditions (‘a plain text message (Fig 6a)’, ‘abstract comic (Fig 6b)’, ‘abstract comic with social proof (Fig 6c)’). We want to convey three key points: 1) Participants have a 10% chance of winning a $5 bonus. 2) Participants can use the bonus as they want. 3) We asked if the participants would be willing to donate to OAR. We designed the comic message (Fig 6b) as a three-panel comic strip. While aesthetic considerations govern the choice of the number of panels in the comic form, two factors influenced our choice of the number of panels: the number of points in the message, and communicating the idea of a conversation, over time, between two individuals. The social proof condition (Fig 6c) adds one additional phrase to the comic that reveals the norm. We obtained the norm from a pilot study. After reading the message, in the third step, the participants are asked to make a donation decision between$0-5 (from their perspective reward) for OAR.

## Result

A total of 277 participants from Amazon Mechanical Turk joined our study. Among all 277 participants, 223 (80.5%) participants donated non-zero amount to support autism research; 67 (71.3%) participants from the text condition, 75 (82.4%) participants from the comic condition, and 81 (88.0%) participants from the comic with the social proof condition.

To further analyze the result, We performed a careful Bayesian analysis to analyze the data. We agree with the observations by Kay et al., that beyond the impact on experimental replicability, shifting the question from “did it have an effect?” to “how strong was the effect?” is important to the HCI community. Two additional ideas—transparency, impact small-n studies—also motivate our use of Bayesian analysis.

We show that using the comic has a significant increase (in a Bayesian sense) in donations over the plain text conditions, with a medium to a large effect size of 0.59. Thus we can answer RQ1 affirmatively. To answer RQ2, we show that while the comic with social proof increases the donation level over the comic without the norm, the effect size is very small (0.11) and the increase is not significant. For more details about our Bayesian formulation, please refer to the paper.

To summarize, the comic form significantly increases donations over the plain text, but the presence of social proof is not effective. We caution that the result holds for single-shot, public goods tasks. To fully understand the value of incorporating the social proof in the comic, for specific longitudinal tasks such as dieting, with distant, exclusive rewards and where habits can confound, we need future research.

## Framework for Algorithmically Synthesized Abstract-Comic Persuasive Messages

Our study showed the persuasive power of abstract comics in encouraging people to make prosocial decisions. We propose a framework that allows full/semi- automatic generation of abstract persuasive messages and identifies crucial features that need to be addressed in future work. We believe such a tool will lower the barrier for the persuader to take advantage of the abstract comic (as demonstrated in our study) in encouraging individuals to act in public good dilemmas.

In our study, we created a comic generator with existing packages including “cmx.js” and “rough.js” to generate three-panel abstract comic strips. With several pre-defined character gestures (see Fig. 8), the generator only requests the text input from the persuader to create the comic message. We believe the comic generator built in this study can be further developed as a framework for algorithmic synthesis.

## Design Implications

The primary design implication of our findings: helping organizations with their online messaging strategies as they work to alleviate public goods dilemmas. In particular, public-goods dilemmas that require single-shot decisions (online charitable donations) are opportune candidates for intervention. We believe that these agencies can easily include the use of the comic form as part of their overall messaging strategy because the simplicity of the abstract comic form allows it to be easily synthesized and to additionally incorporate social proofs.

## tl;dr

We examine the use of the abstract comic form for persuading online charitable donations. Persuading individuals to contribute to charitable causes online is hard and responses to the appeals are typically low; charitable donations share the structure of public goods dilemmas where the rewards are distant and non-exclusive. In this paper, we examine if comics in abstract form are more persuasive than in the plain text form. Drawing on a rich literature on comics, we synthesized a three-panel abstract comic to create our appeal. We conducted a between-subject study with 307 participants from Amazon Mechanical Turk on the use of abstract comic form to appeal for charitable donations. As part of our experimental procedure, we sought to persuade individuals to contribute to a real charity focused on Autism research with monetary costs. We compared the average amount of donation to the charity under three conditions: the plain text message, an abstract comic that includes the plain text, and an abstract comic that additionally includes the social proof. We use Bayesian modeling to analyze the results, motivated by model transparency and its use in small-sized studies. Our experiments reveal that the message in abstract comic form elicited significantly more donations than text form (medium to large effect size=0.59). Incorporating social proof in the abstract comic message did not show a significant effect. Our studies have design implications: non-profits and governmental agencies interested in alleviating public goods dilemmas that share a similar structure to our experiment (single-shot task, distant, non-exclusive reward) ought to consider including messages in the abstract comic form as part of their online fund-raising campaign.

You can learn more about our work by reading our paper. The code is publicly available on Github. Here is the BibTex:

@article{xiao2019should,
title={Should We Use an Abstract Comic Form to Persuade? Experiments with Online Charitable Donation},
author={Xiao, Ziang and Ho, Po-Shiun and Wang, Xinran and Karahalios, Karrie and Sundaram, Hari},
journal={Proceedings of the ACM on Human-Computer Interaction},
volume={3},
number={CSCW},
pages={1--28},
year={2019},
publisher={ACM New York, NY, USA}
}