Privacy is a ubiquitous issue in our increasingly digitalized society. This study investigates the impact
of users' privacy decisions on what data social network platforms collect about them. We build on a
data donation approach to observe people's privacy behavior on Instagram directly. Our findings
contextualize and advance previous studies that almost exclusively relied on users' self-reported
behavior. Data is collected from a German probability-based online panel in two stages: Participants
first complete a survey about their privacy behavior and well-studied predictors of that behavior, such
as privacy concerns and privacy literacy. One week after the survey, participants are asked to
download and donate their Instagram usage data. A one-week gap between the survey and donation
request minimizes the mere-measurement effect, ensuring participants are less likely to hesitate in
donating personal data due to prior privacy-related questions. The paper investigates 1) how
participants' privacy behavior observed in the donated data compares to their self-reported privacy
behavior and 2) how using observed privacy behavior enhances current models studying the
relationship between privacy concerns and privacy behavior. Ultimately, our research shows that data
donation is a valuable data collection method to understand online privacy and how that is a vital next
step to explaining people's online privacy behavior.
