Skip to Content Skip to Footer
Playing the Broken Phone Game: How Successive News Retelling Distorts Information and Emphasizes Negativity

Playing the Broken Phone Game: How Successive News Retelling Distorts Information and Emphasizes Negativity

Narek Grigorian and Cheryl-lyn Ngoh

telephone game sized

The importance of word-of-mouth communication in marketing is well-established. Consumers tend to favor information disseminated by friends and acquaintances instead of accessing original sources. Our knowledge of products and services as well as news information (e.g., a new product release) is often based on retellings of original sources summarized by others. This brings up the question: “Can we rely on information that is consecutively retold?” Regardless of our benevolent intentions to help others through summarizing information, we end up telling less than the complete story. The process of information diffusion as it is retold across consumers comes with a price, that is, information distortion.

In a recent Journal of Marketing Research article, Shiri Melumad, Robert Meyer, and Yoon Duk Kim explore how content and tone of original information is modified over the course of sequential retelling. As news stories are sequentially retold across consumers, factual details decline while information becomes more opinionated and appears increasingly inclined toward negativity. This stylistic evolution is conceptualized as a shift toward “disagreeable personalization.” This phenomenon resembles the popular broken telephone game where the further the information gets from the starting player, the original source, the more it twists and evolves across following players.


Melumad et al. test their theory by conducting a series of experiments and text analysis techniques to study how contents of original stories evolve over different waves of retelling. They argue that audience characteristics play a crucial role. Disagreeable personalization unfolds when retellers feel they are more knowledgeable than their audience and, thus, are motivated to provide guidance and add their personal opinions in their content summaries. In providing guidance, retellers need to act persuasively and demonstrate that their opinions are worth reading. To achieve this, retellers resort to negativity. Negativity bias draws attention to retellers’ arguments as negative information tends to be more prominent and noticeable. Noteworthy, this is far from malevolent motives as retellers are unaware of the potential downside.

Can we tame disagreeable personalization and stop information distortion? Melumad et al. show that by providing consumers readable access to original sources of information may diminish the tendency for disagreeable personalization but not prevent it. Interestingly, disagreeable personalization arises even when the original story includes no negative content.

Melumad et al. extend our understanding of how a story evolves as a strategic process that is dependent on audience characteristics relative to what summarizers can retain from it. This research carries meaningful managerial implications that add new perspective to issues like polarization in the public sphere and helps managers curb information distortion.

We corresponded with the authors to dig deeper into the phenomenon of disagreeable personalization. Read on to discover research implications for practitioners and academics.

Q & A with authors Shiri Melumad, Robert Meyer, and Yoon Duk Kim

Q: What would your advice be to a firm that is going through a crisis (i.e., a product harm crisis) to curb informational distortion and prevent the potential spread of negative word of mouth among consumers? How could your work potentially inform crisis management? Do you think that the sequential retelling of news may result in deepening a crisis into a major event that would normally not be expected to happen?  

A: Our findings underscore the urgent need for firms to not just to get out in front of messaging to consumers in times of crisis, but also to stay in control of messaging over time. What we find is that when news spreads by word of mouth from one consumer to the next, it invariably becomes more negative through retelling. This implies that, even if a firm issues a press release indicating that a crisis has been resolved (e.g., via a successful recall), the positive aspects will tend to vanish over successive retellings, and what will remain are the more negative elements from the message as well as injections of skepticism. Hence, repeated “reseeding” of the message is essential, along with constant monitoring of social media to assess the extent to which the retellings are indeed drifting negative. 

“Our findings underscore the urgent need for firms to not just to get out in front of messaging to consumers in times of crisis, but also to stay in control of messaging over time. … Repeated ‘reseeding’ of the message is essential, along with constant monitoring of social media to assess the extent to which the retellings are indeed drifting negative.” 

Melumad, Meyer, and Kim

Q: How could your research explain the current situation with the pandemic, in which one can observe polarization in society and the divide between social groups who “fight” over measures of public health and the lack of trust in science? Could this toxicity in the public sphere be attributed to the tendency toward negativity observed in your study?

A: We think that our work indeed can help explain this. For example, imagine that two groups of consumers are both exposed to the same message about vaccines in a Centers for Disease Control report—one group being anti-vaccine, the other pro-vaccine. As the message is repeatedly discussed and retold across their respective networks, it becomes distorted, but in very different ways: for the pro-vaccine group, the retellings may take the form of dire warnings about a disease threat, while for the anti-vaccine group, the retellings may take form of dismissals of the threat. Note that in both groups, the tone of the retellings would become increasingly negative over time but would result in opposite, polarized beliefs. We believe that similar processes have unfolded during the pandemic.

Q: The consumers that you studied have benevolent intentions. However, unintended consequences arise that could harm consumer welfare. Do you think that reminding retellers about the risk on their own reputation or status in (irresponsibly) relaying news could help curb disagreeable personalization? Will the “desire to guide” mechanism still hold in the face of the potential to harm one’s reputation?

A: One of the paradoxes of our findings is that when consumers emphasize their opinions and negativity in their retellings, it is often because they are trying to being helpful. They believe that this convinces their audience to attend to the important details and guidance they are trying to convey. Given that disagreeable personalization bias often arises from a benevolent place, if it indeed causes reputational harm, then not only are retellers likely unaware of it but, if anything, they probably think they are improving their reputation.

Q: In Study 2, you found that participants who reported greater intention to persuade were associated with greater intentions to highlight negative aspects of the story. The study also revealed that those participants claimed they intended to highlight positive aspects of the story. Do you think this finding is something worth investigating further? For example, why do their intentions differ from their actions? Is there a potential moderator to change the magnitude of effects?

A: One of our key hypotheses is that when consumers retell a story, they try to emphasize the aspects that they think will most grab their audience’s attention; often, those stories are retold in a negative tone. There are of course contexts where the more positive (vs. negative) elements of a story would be more attention-grabbing. In some of our experiments, we explored the conditions under which people might emphasize positive versus negative information. We found that a key determinant was whether the positive or negative details in a story were viewed as the more surprising or unexpected. When the positive details about a product were more unexpected than the negative details, retellers tended to selectively emphasize positivity in their summaries—creating an agreeable personalization bias. One area of future research we would like to pursue is to identify interventions to help mitigate biases in either direction (negative or positive).

Q: What advice would you provide businesses in the news industry who have “experts” who retell stories through podcasts? To what extent do you think that retelling stories via podcasts could damage the brand’s reputation or be potentially beneficial to the brand?

A: One of our major findings is that the “disagreeable personalization” bias is most acute when retellers believe that their audience is much less knowledgeable than they are. For instance, the more retellers believe they know more relative to their audience, the more compelled they are to emphasize negativity to get their audience to pay attention and inject their opinions to help their audience interpret the information. Thus, our advice would be to encourage experts to try tempering this tendency by imagining their audience as somewhat more knowledgeable on the topic than they might be. Indeed, one can imagine that in an effort get an audience’s attention, a podcaster might discuss a brand in more subjective or sensationalist terms, and thereby cause reputational harm to the brand.

Q: Did you face any challenges when you constructed the measures of content with the automated text analysis tool and human judgment in Study 1 (e.g., when content was neutral)?

A: The central challenge is that both are noisy measures—Natural Language Processing  tools cannot perfectly capture sentiment (or other traits of text), and human judges, for their part, are subjective and, thus, can vary widely in how they interpret text. Therefore, we gathered both types of measures in the hope that they would provide convergent validity for the results—which, fortunately, they did.

Read the full article here:

Melumad, Shiri, Robert Meyer, and Yoon Duk Kim (2021), “The Dynamics of Distortion: How Successive Summarization Alters the Retelling of News,” Journal of Marketing Research, 58(6), 1058­–78.

Narek Grigorian is a doctoral candidate at Bayes Business School (formerly Cass), City | University of London.

Cheryl-lyn Ngoh is a doctoral candidate in Marketing at Ambassador Crawford College of Business and Entrepreneurship, Kent State University