I show that the natural experiment approach utilized in the extant studies to control for public information may not be appropriate to the extent that information arrival itself is a function of trading. I provide a direct empirical test of the competing hypotheses by analyzing the volatility of close-to-open and open-to-close returns for NASDAQ securities with and without active extended-hours trading, while jointly and explicitly controlling for the firm-specific contemporaneous public information flow. My methodology disentangles the effects of noise, public information, and private information on stock return volatility. I also contribute to the burgeoning literature analyzing trading activity and return characteristics in the quickly growing extended-hours market.