Do online word-of-mouth effects differ
across platforms?
A comparison between Reddit and Twitter
Huub Kuiper S2707101 University of Groningen
Topic introduction
› Customer Satisfaction § ACSI § Word-of-mouth (Anderson 1998) § Electronic word-of-mouth - Twitter - RedditPrior research on eWOM effects
§ Positive relationship with respect to stock returns.
- Hourly level (Deng et al. 2018) - Daily level (Smailovic 2013)
§ Kind of sentiment
- Magnitude is higher for negative sentiment (Lui 2017)
Hypotheses
› H1: eWOM found on Reddit performs better in predicting yearly stock returns than Twitter.
› H2: A model including both Reddit and Twitter eWOM as predictors
Research Method
› Regressions between the predictors themselves.
› Regression between the predictors and the DV Stock Return.
§ Year Dummies § Lagged DV
› Weighted AIC (Wagenmakers & Farrell 2004)
Data Collection
› Reddit Scraping
› Sentiment Analysis using qdap › Dataset size
§ 14,432 Reddit posts § 5+ million tweets
Conclusions
› Negative Sentiment found on Reddit performs best in predicting other predictors.
› Change in negative Reddit sentiment performs second best in predicting stock returns.
Discussion
› Lack of data in the used timespan
› Future research gives opportunity due to growth of platform › Improved scraping tools
› Time interval