Skip to Content Skip to Footer

When Online Discussions Are Social, Do They Become Less Informative?

Rebecca W. Hamilton, Ann Schlosser and Yu-Jen Chen

Marketing managers often listen to online conversations to glean customer insights. However, such conversations may be biased. Look at responses to this TripAdvisor forum question: 

My family and I will be in Seattle for a few days in June, the 1st-3rd. We are there for a few days then going to San Francisco for a few days as a part of our sons graduation trip. The only requests he has for Seattle is to see the Space Needle and Pike Place. I have narrowed it down to a few hotel choices and am hoping you experts can help me narrow it down to one. I really don’t want to spend a fortune, which seems hard in Seattle, but I want something clean.

Within minutes, there were four replies:


Notably, the first respondent provides helpful information about the location of one of the hotels listed in the original query, but she does not address cleanliness or explain why she didn’t recommend the less expensive hotels near the Space Needle. Like the first respondent, the next three respondents also mention location but ignore cleanliness, and none listed the less expensive hotels. Strikingly, the responses seem to be guided by the first respondent’s reply as much or more than the original query.

The authors of a recent article published in the Journal of Marketing Research show that patterns of responses like this are common in online discussion forums, and they find that each respondent’s desire to affiliate with others seems to be the cause. Because online discussion forums provide an opportunity for social interaction, the quality of information provided can suffer. Although most participants report that they join online discussion forums primarily to seek information, recent research has shown that people are more likely to continue participating when they identify and affiliate with other forum members (Pendry and Salvatore 2015). This change in motivation is important because the desire to affiliate with others can systematically affect the information contributed by participants in online discussions, sometimes in undesirable ways.

How Do Participants Decide What to Contribute to an Online Discussion?

Hamilton, Schlosser, and Chen’s (2017) research examines the textual content of online discussion threads to see how early responses to a query affect subsequent responses. Research on social influence in face-to-face groups has shown that group members tend to repeat information that has already been shared by others rather than contribute new information to a discussion (Larson, Foster-Fishman, and Keys 1994; Mojzisch and Schulz-Hardt 2010). Repeating information can serve the goal of affiliating with other participants in the forum, but it competes with the goal of providing comprehensive, relevant information to help the advice seeker.

Although it is more challenging to analyze text than to analyze quantitative metrics such as ratings or the volume of content generated, text analysis can provide important insights into how previously contributed content influences subsequently contributed content. Specifically, the authors code the characteristics or attributes (e.g., location, price, cleanliness) that are mentioned in the query and in each subsequent response. Next, the authors coded whether each respondent mentioned the same attributes as previous respondents or different attributes. Using this procedure, they determine whether the content of later responses is systematically biased to repeat what others have mentioned previously.

Analyzing Data from Discussion Forums

The authors collected and analyzed data from three online discussion forums. These forums were hosted by two different organizations (TripAdvisor and and spanned two different services (hotels and restaurants) in three mid-sized cities that are popular tourist destinations (Seattle; Washington, DC; and Orlando). They collected posts from discussion threads with at least two respondents that began with a query seeking advice about a specific decision (e.g., “Should I stay at hotel A or B?” or “Should I eat at restaurant X or Y?”).

Not surprisingly, the data showed that the first respondent tended to mention the same attributes that were mentioned by the advice seeker. This suggests that the first respondent tried to answer the advice seeker’s question. More interestingly, the authors also found that the second respondent was more likely to mention attributes discussed in the first respondent’s post, controlling for whether the attributes were mentioned by the advice seeker. Similarly, the third respondent was more likely to mention attributes that had been mentioned by the second respondent and the first respondent, controlling for whether they were mentioned by the advice seeker. These results show that participants in online discussion forums are more likely to mention attributes that have already been discussed by previous respondents, reducing the informativeness of their responses.

In a series of four experiments, the authors isolated the reason for this repetition: a desire to affiliate with other participants in the discussion. Across all four experiments, the attributes mentioned by previous respondents significantly influenced which attributes subsequent respondents mentioned in their posts. Participants even repeated attributes that were clearly not important to the advice seeker. The effect was stronger when participants in the experiments were given a goal to affiliate rather than a goal to provide accurate information, and when participants had a strong desire to affiliate with others.

Designing Platforms for Better Online Information Exchange

It’s worth considering how to structure discussion forums such as those on TripAdvisor and Disney forums. One option would be to show the query initiating each discussion thread but hide previous responses so that they don’t bias subsequent responses. However, if reading previous responses enhances feelings of group membership that motivate consumers to contribute their own responses, hiding previous responses could reduce participation. Managers designing discussion forums face the challenge of meeting both respondents’ affiliation needs and advice seekers’ needs for accurate information. Hamilton, Schlosser, and Chen (2017) also find that encouraging respondents to provide accurate information can reduce repetition across answers. Managers might explicitly ask respondents to provide accurate information or encourage advice seekers to acknowledge respondents who provide them with accurate information.

The results are also important for managers who use online word of mouth to gain consumer insights by tracking which product attributes are most discussed. Because the content of online discussion forums can be biased by the repetition of previously shared information, regardless of its importance, it may be misleading to infer the importance of an attribute to consumers based on its frequency of mention in online discussion forums. Managers need to take this bias into account when interpreting such information.

Article Citations

Rebecca W. Hamilton, Ann Schlosser, and Yu-Jen Chen (2017), “Who’s Driving This Conversation? Systematic Biases in the Content of Online Consumer Discussions,” Journal of Marketing Research, 54 (4), 540–55.

James R. Larson, Pennie G. Foster-Fishman, and Christopher B. Keys (1994), “Discussion of Shared and Unshared Information in Decision-Making Groups,” Journal of Personality and Social Psychology, 67 (3), 446–61.

Andreas Mojzisch and Stefan Schulz-Hardt (2010), “Knowing Others’ Preferences Degrades the Quality of Group Decisions,” Journal of Personality and Social Psychology, 98 (5), 794–808.

Louise F. Pendry and Jessica Salvatore (2015), “Individual and Social Benefits of Online Discussion Forums,” Computers in Human Behavior, 50 (September), 211–20.

Rebecca W. Hamilton

Rebecca W. Hamilton is Michael G. and Robin Psaros Chair in Business Administration and Professor of Marketing, McDonough School of Business, Georgetown University.

Ann Schlosser

Ann Schlosser is Professor of Marketing and Evert McCabe Fellow, Foster School of Business, University of Washington.

Yu-Jen Chen