Overall, our results differentiate norm inference from norm adoption and highlight behavioral prevalence as a stable normative driver. They also raise concerns about how visible peer behaviors may facilitate the diffusion of risky disclosure practices.

Link: osf.io/preprints/ps...

(9/end)

Reinforcement signals (likes and comments) showed no observable effect on norm formation or disclosure intentions. This challenges common assumptions about the centrality of engagement metrics in shaping behavior.

(8/9)

Similarity moderated these processes. Participants relied less on prevalence cues when evaluating similar others, yet similarity amplified the impact of norms once they were formed.

(7/9)

Written disclosure had a more limited role. It shifted norm perceptions modestly but did not meaningfully influence disclosure intentions, suggesting a modality-specific sensitivity to normative information.

(6/9)

The findings were quite consistent: prevalence was the dominant mechanism. Higher levels of visual disclosure among peers led to stronger descriptive and injunctive norms, which in turn increased participants’ own disclosure intentions.

(5/9)

We therefore developed a new platform called Travelgram, closely resembling Instagram. It simulated the full social media experience. Participants scrolled, posted, liked, commented and we manipulated what they saw.

(4/9)

Across two preregistered experiments (n=590; n=1337) using a ecologically valid, but simulated platform, we independently manipulated two elements: the prevalence of others’ disclosures and reinforcement via likes/comments. We further measured perceived similarity of other users.

(3/9)

Building on my previous work (doi.org/10.1371/jour...), we argue that platforms are saturated with signals about what others do, and these signals structure users' own decisions to share.

(2/9)

Oh, this is old. I think I should revisit this. :D

Thanks, Cameron!

Thanks so much, Ye!

Using specification curve analyses, we show that this relationship is sensitive to analytical decisions, highlighting the importance of transparency and replication in survey-based privacy research.

(4/4)

Only 32.5% of the original effects replicated exactly, though 67.5% were significant and in the expected direction. Interestingly, the widely reported negative link between privacy concerns and self-disclosure did not replicate—in our data, it turned positive.

(3/4)

Together with Giulia Ranzini, we closely replicated three foundational studies in privacy research:
🔹 Krasnova et al. (2010) on the privacy calculus
🔹 Vitak (2012) on context collapse
🔹 Dienlin & Trepte (2015) on the privacy paradox

(2/4)

(4) analyzing how these units interact to shape privacy expectations, policies, and behaviors.

The paper resulted from discussion within the comparative privacy research network. For more info, check out: comparativeprivacy.org