Suggestions for Improving the Evidence Level in Analysis using Twitter Data

Abstract

　A proposal was made to address research issues regarding “A Sentiment Analysis of Tweets about Fertility Treatment: Investigating the Impact of Changing Subsides on Social Media,” which was published in Volume 41, Issue 4 of this journal. Although significant results were obtained in this research, the validity and reproducibility seemed to be diminished due to the inclusion of bot tweet, the lack of consideration of retweets, and the unclear source of tweets. In order to overcome these problems, I propose to use exclusion criteria using the “source” data of tweets, sensitivity analysis weighted by the number of retweets, and specifying the source of tweets.