Concerns over political sensitivity bias in survey research in stable autocracies have been raised by previous studies. However, as democracy recedes worldwide, it remains unclear how pervasive such bias is in autocratizing contexts. This research note aims to demonstrate the extent of sensitivity bias in an autocratizing context using the case of Hong Kong, where political freedoms are rapidly shrinking. In doing so, we aim to detect sensitivity bias by analyzing panel survey data without relying on list experiments. Our analyses of online panel survey data collected before and after the enactment of the Hong Kong National Security Law and its related political crackdown found that pro-democracy respondents subject to political repression (1) are more likely to drop out of the panel of a political poll, (2) even when they do not drop out, they falsify some potentially sensitive past behavior, and (3) pro-democracy moderates are more likely to engage in preference falsification. Theoretical implications for understanding public opinion in autocratizing contexts are discussed.