AffectCheck: How Real-time Feedback on Affective Tone Influences Twitter Communication

Will Riley* Will Riley*, School of Information, University of Michigan, Ann Arbor, United States
Margaret Morris, Intel Corporation, Portland, United States
Sean Munson, School of Information, University of Michigan, Ann Arbor, United States
Paul Resnick, School of Information, University of Michigan, Ann Arbor, United States

Track: Research
Presentation Topic: e-Coaching
Presentation Type: Poster presentation
Submission Type: Single Presentation

Building: LKSC Conference Center Stanford
Room: Lower Lobby
Date: 2011-09-18 12:00 PM – 01:00 PM
Last modified: 2011-08-12

If you are the presenter of this abstract (or if you cite this abstract in a talk or on a poster), please show the QR code in your slide or poster (QR code contains this URL).


Tools that invite self-reflection of affective tone may help people communicate more effectively in online social networks, allowing them to develop and maintain the social connectedness that contributes to health and well-being. In a recent study, people described reluctance to post health experiences because they were worried they would be perceived as either complaining or boasting, and an aversion to those that they perceived as chronically complaining. In this study, we explore ways to enhance self-awareness about projected affective tone on Twitter. To help Twitter users (“Twitterers”) self-monitor affective tone, we created a tool that reflects the positivity or negativity of “tweets” as they are written. Inspired by real-time spell checking, the tool automatically color-codes words (red for negative words and green for positive words), which allows writers to edit their emotional content before publication. Like many spell checkers, the tool also allows writers to correct and personalize the automatic classification of words. By clicking on a word, writers can redefine it as negative or positive. This prototype allows us to investigate the potential for real-time feedback about affect as a tool for impression management in social media.
In this study, we examine how real-time affect-checking influences microblogging communication on Twitter. Specifically, we test whether feedback leads to more editing and a change in the balance of negative and positive words. We also measure whether affect feedback and associated editing influences the reactions of readers, that is whether tweets are more likely to be re-tweeted, whether they get more direct responses, and whether responses are more positive.
We will recruit active Twitterers who have expressed an interest in self-improvement (by following a self-improvement thought leader on Twitter). Participants in our study will install a Firefox add-on that collects keystroke-level data on how they write tweets using the website. During the first phase of the study, the add-on will gather data about how participants write tweets without any affect-checking feedback. During the second phase of the study, the add-on color codes tweets as they are composed, prior to posting. The initial dictionary of positive and negative words comprises the LIWC and ANEW collections, but users can correct and personalize their dictionaries. The affect checker stems words and reverses their valences if preceded by negation words, such as “not” or “can’t.” Effects will be evaluated using a within-subjects design, comparing baseline activity to activity when the affect-checking display is turned on. Dependent measures include the total number of tweets posted, the percentage of positive vs. negative valence words, the amount of editing during message composition, the probability of being re-tweeted, the probability receiving a reply, the affect of replies, and the number of followers.
Research in progress.
Research in progress.

Medicine 2.0® is happy to support and promote other conferences and workshops in this area. Contact us to produce, disseminate and promote your conference or workshop under this label and in this event series. In addition, we are always looking for hosts of future World Congresses. Medicine 2.0® is a registered trademark of JMIR Publications Inc., the leading academic ehealth publisher.
Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.