In this study we explore the mechanisms and factors explaining why citizens continue to trust and forgive politicians after corruption scandals. While it is well-established that voters do not consistently sanction corrupt politicians at the ballot box, the underlying mechanisms remain unclear. Using a survey experiment conducted in the United Kingdom, Switzerland, and Germany (n=6000), we employ an innovative digital visualization tool that allows participants to create their ideal politician by selecting descriptive characteristics (age, gender, ethnic background, and social class) and preferred political behaviors. Respondents are then randomly assigned to scenarios where either their ideal politician or a standard, predefined politician is involved in a corruption scandal. This design enables us to examine whether politicians’ similarity to their constituents (homophily/descriptive representation) or their history of advocating for specific group interests (substantive representation) affects citizens’ reactions to political misconduct. We show that politicians who share demographic characteristics with their constituents or have a history of advocating their specific interests are perceived as more trustworthy, are more likely to be forgiven, and hence have a higher likelihood of continued electoral support despite corruption. By integrating insights from representation and accountability research, this study deepens our understanding of voter behavior in the face of political corruption and highlights the trade-offs between personalized representation and democratic accountability.