This study empirically examines how a mobile app’s daily star ratings and installation/usage metrics are statistically associated with user review sentiment using daily data. Specifically, we analyze user responses through regression analysis and se...
This study empirically examines how a mobile app’s daily star ratings and installation/usage metrics are statistically associated with user review sentiment using daily data. Specifically, we analyze user responses through regression analysis and sentiment analysis, focusing on differences by app installation and activation levels. The empirical dataset covers the public transportation app KorailTalk from February 2017 to September 2025, including daily star ratings and user reviews from the Google Play Store, as well as Korail’s internal records on installed devices, active devices, and deleted (uninstalled) devices. The results show that a higher average daily star rating is significantly and positively associated with the number of positive reviews on the same day, and significantly and negatively associated with the number of negative reviews. Although app installations may be influenced by existing user reviews, KorailTalk users tend to install the app first and then write reviews. Therefore, rather than emphasizing causality, this study focuses on the statistical relationships among daily star ratings, installation/activation indicators, and review sentiment. In other words, while user reviews are shaped by various factors, users are more likely to write positive reviews when the average star rating on that day is higher. In addition, the analysis using daily installation, active-device, and deleted-device measures indicates that more daily installs are associated with increases in both positive and negative reviews, with negative reviews increasing more. For users already using the app (active devices), positive review posting decreases, while negative reviews are not significantly affected, suggesting that positive reviews are more likely to be written around the time of installation. Deleted (uninstalled) devices do not show a meaningful impact on either positive or negative reviews. Finally, sentiment analysis of review texts reveals positive sentiment toward ease of use, booking speed, and stable connectivity, whereas negative sentiment is mainly related to technical instability, payment issues, and a decline in UI/UX usability.