Cameron Martel
@cameronmartel.bsky.social
Assistant Professor at Johns Hopkins Carey Business School.
Studies misinformation & inauthentic behavior online.
Studies misinformation & inauthentic behavior online.
A second survey exp found that minimal social connections foster a general norm of responding, such that ppl feel more obligated to respond - and think others expect them to respond more - to ppl who follow them, even outside the context of misinfo correction
April 14, 2025 at 2:28 PM
A second survey exp found that minimal social connections foster a general norm of responding, such that ppl feel more obligated to respond - and think others expect them to respond more - to ppl who follow them, even outside the context of misinfo correction
Exploratory analyses also show that in both survey & field exps, extreme partisanship moderates the effects of social connection on engagement - social connection increases engagement for co-partisans, but decreases engagement for politically extreme counter-partisans
April 14, 2025 at 2:28 PM
Exploratory analyses also show that in both survey & field exps, extreme partisanship moderates the effects of social connection on engagement - social connection increases engagement for co-partisans, but decreases engagement for politically extreme counter-partisans
We next conducted a follow-up survey on MTurk to replicate effects in a more controlled setting (eg eliminate blocking of counter-partisan bots) & obtained similar results
April 14, 2025 at 2:28 PM
We next conducted a follow-up survey on MTurk to replicate effects in a more controlled setting (eg eliminate blocking of counter-partisan bots) & obtained similar results
To account for this we (i) compare unaffected conditions (all but social counter-partisan) & (ii) perform principal stratification (weighting obs in unaffected conditions by p(success treat delivery) had they been in social counter-partisan condit)
April 14, 2025 at 2:28 PM
To account for this we (i) compare unaffected conditions (all but social counter-partisan) & (ii) perform principal stratification (weighting obs in unaffected conditions by p(success treat delivery) had they been in social counter-partisan condit)
We sent corrections to 1,586 users & measured p(engage w correction):
(i) Among users in the co-partisan condition, social connection had a sig positive effect on engagement
(ii) Among users in the baseline (non-social) condition, no evidence of effect of shared partisanship on engagement
(i) Among users in the co-partisan condition, social connection had a sig positive effect on engagement
(ii) Among users in the baseline (non-social) condition, no evidence of effect of shared partisanship on engagement
April 14, 2025 at 2:28 PM
We sent corrections to 1,586 users & measured p(engage w correction):
(i) Among users in the co-partisan condition, social connection had a sig positive effect on engagement
(ii) Among users in the baseline (non-social) condition, no evidence of effect of shared partisanship on engagement
(i) Among users in the co-partisan condition, social connection had a sig positive effect on engagement
(ii) Among users in the baseline (non-social) condition, no evidence of effect of shared partisanship on engagement
Each user was then socially corrected by their randomly assigned bot. Social corrections were done via public reply to the tweet containing the debunked URL and included a link to the fact-check on @snopes.com
April 14, 2025 at 2:28 PM
Each user was then socially corrected by their randomly assigned bot. Social corrections were done via public reply to the tweet containing the debunked URL and included a link to the fact-check on @snopes.com
We created human-looking bots & corrected users who shared debunked URLs
We randomized whether our bots
(i) were co-partisan or counter-partisan for the to-be-corrected user
(ii) followed the user & liked some of their tweets before correcting them (creating a minimal social connection)
We randomized whether our bots
(i) were co-partisan or counter-partisan for the to-be-corrected user
(ii) followed the user & liked some of their tweets before correcting them (creating a minimal social connection)
April 14, 2025 at 2:28 PM
We created human-looking bots & corrected users who shared debunked URLs
We randomized whether our bots
(i) were co-partisan or counter-partisan for the to-be-corrected user
(ii) followed the user & liked some of their tweets before correcting them (creating a minimal social connection)
We randomized whether our bots
(i) were co-partisan or counter-partisan for the to-be-corrected user
(ii) followed the user & liked some of their tweets before correcting them (creating a minimal social connection)
🚨New in @plosone.org🚨
Corrections of misinfo are often ignored. What can drive engagement?
Twitter field exp & survey followups find
-Social ties matter: users more likely to engage w corrections from accounts who followed user
-Shared partisanship had smaller effects on engagement
shorturl.at/0Ycdp
Corrections of misinfo are often ignored. What can drive engagement?
Twitter field exp & survey followups find
-Social ties matter: users more likely to engage w corrections from accounts who followed user
-Shared partisanship had smaller effects on engagement
shorturl.at/0Ycdp
April 14, 2025 at 2:28 PM
🚨New in @plosone.org🚨
Corrections of misinfo are often ignored. What can drive engagement?
Twitter field exp & survey followups find
-Social ties matter: users more likely to engage w corrections from accounts who followed user
-Shared partisanship had smaller effects on engagement
shorturl.at/0Ycdp
Corrections of misinfo are often ignored. What can drive engagement?
Twitter field exp & survey followups find
-Social ties matter: users more likely to engage w corrections from accounts who followed user
-Shared partisanship had smaller effects on engagement
shorturl.at/0Ycdp
We also found:
-Information & making friends were most mentioned as follow-back reasons
-Curiosity also oft mentioned, esp for counter-partisan follow-back
-Not wanting info (esp from counter-partisans) & id’ing account as a stranger were most mentioned rzns for ignoring accounts
-Information & making friends were most mentioned as follow-back reasons
-Curiosity also oft mentioned, esp for counter-partisan follow-back
-Not wanting info (esp from counter-partisans) & id’ing account as a stranger were most mentioned rzns for ignoring accounts
October 16, 2024 at 4:21 PM
We also found:
-Information & making friends were most mentioned as follow-back reasons
-Curiosity also oft mentioned, esp for counter-partisan follow-back
-Not wanting info (esp from counter-partisans) & id’ing account as a stranger were most mentioned rzns for ignoring accounts
-Information & making friends were most mentioned as follow-back reasons
-Curiosity also oft mentioned, esp for counter-partisan follow-back
-Not wanting info (esp from counter-partisans) & id’ing account as a stranger were most mentioned rzns for ignoring accounts
We found:
-50% of ppl in co-partisan condition who followed-back account mentioned same partisanship as motivation
-58% of ppl in counter-partisan condition who ignored account mentioned diff partisanship as motivation
-50% of ppl in co-partisan condition who followed-back account mentioned same partisanship as motivation
-58% of ppl in counter-partisan condition who ignored account mentioned diff partisanship as motivation
October 16, 2024 at 4:21 PM
We found:
-50% of ppl in co-partisan condition who followed-back account mentioned same partisanship as motivation
-58% of ppl in counter-partisan condition who ignored account mentioned diff partisanship as motivation
-50% of ppl in co-partisan condition who followed-back account mentioned same partisanship as motivation
-58% of ppl in counter-partisan condition who ignored account mentioned diff partisanship as motivation
Ppl w ⬆️ issue polarization (& to lesser extent ⬆️out-party dislike) were less likely to follow-back neutral & counter-partisans, relative to co-partisans
In contrast, ppl w ⬆️ in-party affinity were more likely to follow-back neutral & co-partisans, relative to counter-partisans
In contrast, ppl w ⬆️ in-party affinity were more likely to follow-back neutral & co-partisans, relative to counter-partisans
October 16, 2024 at 4:20 PM
Ppl w ⬆️ issue polarization (& to lesser extent ⬆️out-party dislike) were less likely to follow-back neutral & counter-partisans, relative to co-partisans
In contrast, ppl w ⬆️ in-party affinity were more likely to follow-back neutral & co-partisans, relative to counter-partisans
In contrast, ppl w ⬆️ in-party affinity were more likely to follow-back neutral & co-partisans, relative to counter-partisans
We find:
-Pref follow-back of co-partisan vs counter-partisan explicit bot accounts (ev of simple content prefs)
-Even *greater* pref follow-back in human-looking vs bot accounts (ev of additional social motivation)
-Pref follow-back of co-partisan vs counter-partisan explicit bot accounts (ev of simple content prefs)
-Even *greater* pref follow-back in human-looking vs bot accounts (ev of additional social motivation)
October 16, 2024 at 4:19 PM
We find:
-Pref follow-back of co-partisan vs counter-partisan explicit bot accounts (ev of simple content prefs)
-Even *greater* pref follow-back in human-looking vs bot accounts (ev of additional social motivation)
-Pref follow-back of co-partisan vs counter-partisan explicit bot accounts (ev of simple content prefs)
-Even *greater* pref follow-back in human-looking vs bot accounts (ev of additional social motivation)
We examine this in a Twitter field exp. We created 3 *explicit bot* accounts and 3 *human-looking* accounts, varying only in their expressed party ID (⅓ Dem, Rep, Politically Neutral)
October 16, 2024 at 4:18 PM
We examine this in a Twitter field exp. We created 3 *explicit bot* accounts and 3 *human-looking* accounts, varying only in their expressed party ID (⅓ Dem, Rep, Politically Neutral)
🚨New in JEP:G🚨
Why do ppl preferentially reciprocate follows by co-partisans online? In a Twitter field exp & online survey exp we find:
-Both content *and* social prefs drive co-party tie-making
-Distinct roles for in-party pref & out-party dispref
dx.doi.org/10.1037/xge0...
Why do ppl preferentially reciprocate follows by co-partisans online? In a Twitter field exp & online survey exp we find:
-Both content *and* social prefs drive co-party tie-making
-Distinct roles for in-party pref & out-party dispref
dx.doi.org/10.1037/xge0...
October 16, 2024 at 4:16 PM
🚨New in JEP:G🚨
Why do ppl preferentially reciprocate follows by co-partisans online? In a Twitter field exp & online survey exp we find:
-Both content *and* social prefs drive co-party tie-making
-Distinct roles for in-party pref & out-party dispref
dx.doi.org/10.1037/xge0...
Why do ppl preferentially reciprocate follows by co-partisans online? In a Twitter field exp & online survey exp we find:
-Both content *and* social prefs drive co-party tie-making
-Distinct roles for in-party pref & out-party dispref
dx.doi.org/10.1037/xge0...
Similarly to our accuracy findings, warning labels still sig effective at reducing sharing intentions of false posts even for those distrusting of FCs - crucially again w *no* evidence of a backfire
September 5, 2024 at 5:12 PM
Similarly to our accuracy findings, warning labels still sig effective at reducing sharing intentions of false posts even for those distrusting of FCs - crucially again w *no* evidence of a backfire
In our sharing exps, warning labels:
-Decrease sharing of false headlines on average (24.7%⬇️)
-Are similarly effective for those w greater trust in FCs
-Still work on those in lowest quartile trust in FCs
-Still work on those maximally distrusting of FCs (16.7%⬇️)
-Decrease sharing of false headlines on average (24.7%⬇️)
-Are similarly effective for those w greater trust in FCs
-Still work on those in lowest quartile trust in FCs
-Still work on those maximally distrusting of FCs (16.7%⬇️)
September 5, 2024 at 5:11 PM
In our sharing exps, warning labels:
-Decrease sharing of false headlines on average (24.7%⬇️)
-Are similarly effective for those w greater trust in FCs
-Still work on those in lowest quartile trust in FCs
-Still work on those maximally distrusting of FCs (16.7%⬇️)
-Decrease sharing of false headlines on average (24.7%⬇️)
-Are similarly effective for those w greater trust in FCs
-Still work on those in lowest quartile trust in FCs
-Still work on those maximally distrusting of FCs (16.7%⬇️)
In our accuracy exps, warning labels:
-Decrease belief in false headlines on average (27.6%⬇️)
-Are more effective for those w greater trust in FCs
-Still work on those in lowest quartile trust in FCs
-Still work on those maximally distrusting of FCs (12.9%⬇️)
-Decrease belief in false headlines on average (27.6%⬇️)
-Are more effective for those w greater trust in FCs
-Still work on those in lowest quartile trust in FCs
-Still work on those maximally distrusting of FCs (12.9%⬇️)
September 5, 2024 at 5:09 PM
In our accuracy exps, warning labels:
-Decrease belief in false headlines on average (27.6%⬇️)
-Are more effective for those w greater trust in FCs
-Still work on those in lowest quartile trust in FCs
-Still work on those maximally distrusting of FCs (12.9%⬇️)
-Decrease belief in false headlines on average (27.6%⬇️)
-Are more effective for those w greater trust in FCs
-Still work on those in lowest quartile trust in FCs
-Still work on those maximally distrusting of FCs (12.9%⬇️)
Ppl were randomized to a warning label treatment or control. In treatment, a high proportion (eg 2/3, 3/4, or all) of false headlines were labeled. In control, no labels were provided. 10 exps evaluated accuracy ratings of headlines, 11 evaluated headline sharing intentions
September 5, 2024 at 5:07 PM
Ppl were randomized to a warning label treatment or control. In treatment, a high proportion (eg 2/3, 3/4, or all) of false headlines were labeled. In control, no labels were provided. 10 exps evaluated accuracy ratings of headlines, 11 evaluated headline sharing intentions
As expected, we found more Republican-leaning participants were less trusting of FCs; this was increasingly the case for Reps w higher procedural news knowledge, analytic thinking, & web-use skill
September 5, 2024 at 5:05 PM
As expected, we found more Republican-leaning participants were less trusting of FCs; this was increasingly the case for Reps w higher procedural news knowledge, analytic thinking, & web-use skill
One of the most widely used interventions against misinfo (eg on FB, Instagram) is attaching warning labels from professional fact-checkers (FCs) to content the FCs have classified as false
September 5, 2024 at 5:02 PM
One of the most widely used interventions against misinfo (eg on FB, Instagram) is attaching warning labels from professional fact-checkers (FCs) to content the FCs have classified as false
🚨New in Nature Human Behaviour🚨
Will misinfo warning labels backfire for ppl who distrust fact-checkers? No!
Labels reduce belief in & sharing of false news even for those highly distrusting of fact-checkers - warning labels are a key tool for platforms!
rdcu.be/dSHtF
Will misinfo warning labels backfire for ppl who distrust fact-checkers? No!
Labels reduce belief in & sharing of false news even for those highly distrusting of fact-checkers - warning labels are a key tool for platforms!
rdcu.be/dSHtF
September 5, 2024 at 5:01 PM
🚨New in Nature Human Behaviour🚨
Will misinfo warning labels backfire for ppl who distrust fact-checkers? No!
Labels reduce belief in & sharing of false news even for those highly distrusting of fact-checkers - warning labels are a key tool for platforms!
rdcu.be/dSHtF
Will misinfo warning labels backfire for ppl who distrust fact-checkers? No!
Labels reduce belief in & sharing of false news even for those highly distrusting of fact-checkers - warning labels are a key tool for platforms!
rdcu.be/dSHtF
Finally in our survey studies we asked *why* people chose to block accounts. The most common reason for blocking was not wanting to see the content of the blocked account’s tweets & retweets; rather than blocking so blocked users cannot see one’s own account
May 28, 2024 at 8:32 PM
Finally in our survey studies we asked *why* people chose to block accounts. The most common reason for blocking was not wanting to see the content of the blocked account’s tweets & retweets; rather than blocking so blocked users cannot see one’s own account
In a separate study on Lucid (N=3k) we found partisan diffs in reported blocking based on account *content* - eg Dems more likely to block users sharing false content or mean/nasty posts
May 28, 2024 at 8:31 PM
In a separate study on Lucid (N=3k) we found partisan diffs in reported blocking based on account *content* - eg Dems more likely to block users sharing false content or mean/nasty posts
To gain more insight on mechanism, we aimed to replicate our field results in a survey experiment. We found that participants were 3x more likely to block a counter-partisan vs politically neutral profile; but we did not observe a partisan asymmetry in blocking
May 28, 2024 at 8:31 PM
To gain more insight on mechanism, we aimed to replicate our field results in a survey experiment. We found that participants were 3x more likely to block a counter-partisan vs politically neutral profile; but we did not observe a partisan asymmetry in blocking
Looking at *blocking* behavior, we found evidence of selective blocking. Users were 12x more likely to block counter-partisan than co-partisan accounts; & 4x more likely to block counter-partisan than neutral accounts
May 28, 2024 at 8:30 PM
Looking at *blocking* behavior, we found evidence of selective blocking. Users were 12x more likely to block counter-partisan than co-partisan accounts; & 4x more likely to block counter-partisan than neutral accounts