In this disseration, I study several topics in the field of behavioral finance. Chapter 1 empirically explores the divergent sentiment shift of partisan investors after shift in political power, by using the 2020 presidential election as an event study. With data from social media platform StockTwits, the empirical findings indicate that after the election, Republican (Democrat) investors become more pessimistic (optimistic) toward future stock returns.These partisan divergent belief shifts are more pronounced when the election results become more solidified and are more concentrated in major stock indices (SPY, QQQ, DIA) and stocks with higher market beta. Additionally, consistent with the theoretical framework of Kruger (2020), further analysis indicates that during the post-election period, partisan disagreement is associated with increased stock liquidity and intraday volatility. These results indicate that partisan investors are willing to trade against those with opposite beliefs on the financial market during the post-election period. In Chapter 2, I empirically examine the impact of local Covid spread on the net flows of locally headquartered mutual funds. The empirical findings indicate that state-level Covid spread reduces the net flows of locally headquartered mutual funds, which are more pronounced in the Covid crash period. Further analysis indicates that the reducing effect of Covid on fund net flows is more pronounced for retail fund share class, suggesting a heterogeneous response to the pandemic across investor types. Additional analysis shows that Covid-induced fund outflows are more pronounced for funds associated with higher levels of risk, implying that heightened risk aversion during the pandemic is a major driving factor behind the baseline results. Controlling local economic conditions does not significantly alter the main findings, indicating that visceral response is a more plausible explanation than economic shock. Alternative measures using state-level Google search volumes corroborate the main findings.In Chapter 3, I empirically examine the impact of social media sentiment and attention on IPO pricing. By using social media sentiment from Stocktwits as a proxy for retail valuation, I empirically examine the theoretical predictions in prior studies (Ljungqvist et al. (2006), Cornelli, Goldreich, and Ljungqvist (2006), and Derrien (2005)) that over-optimism of sentiment investors leads to initial overpricing of IPO followed by long-term reversals. Using posts on Stocktwits during the pre-IPO period, I construct investor attention and sentiment measurements. The empirical results are generally consistent with the theoretical predictions that retail investor over-optimism leads to higher IPO first-day price run-up and worse long-term performance. Additionally, using machine learning techniques to classify untagged posts, I find similar results where sentiment measures are constructed with classified untagged posts. Results with sentiment measures constructed by classified untagged posts imply that more optimistic sentiment leads to a higher turnover rate shortly after IPO, implying that informed investors are selling overpriced IPO shares to sentiment retail investors.