It would be useful to have a gender breakdown for this trial. Having just read a preview copy of James Ball’s book ‘The Other Pandemic’, it seems a certain subset of men are the big drivers of online misinformation and online conspiracies, which I found interesting as it counters our societal conditioning that men are the ‘rational’ sex.
That's a good question, Heather - there's not a lot, except that they note that they tried to recruit according to benchmarks for race, gender, age and location, "but a non-random subset of these respondents opted in to installing the browser extension. Compared to national estimates, our 2018 sample had fewer people aged 18-24, a greater proportion of strong partisans [which they actually wanted], and morte participants who identified as white, male and college-educated. Among our 2020 sample, the differences in age and partisan identity groups were closer to national estimates than in 2018, and a greater proportion of participants identified as female." (I'm now trying to find the tables where they give the overall gender breakdown..) However I don't think they looked at gender breakdown for who the misled partisans were.
Keep in mind "misinformation" isn't a monolith. For example, I'd conjecture anti-vax before Covid skewed heavily female, since that's how childcaring responsibility skews. Election-denial is likely predominantly male. Also, there's social networks which aren't that visible from the perspective of the Very Online. The phrase "church lady" is a cliche for reason.
I think the system/people interaction is complicated, so taking it to either extreme isn't true (i.e. people aren't completely slaves to propaganda, but virtue-preaching yields no insight). One idea I wish I could explore more is: An aspect of this complication is that "algorithms" and such, is a mechanism which allows the chattering class to seriously discuss media issues in a way which would otherwise be considered outside their boundaries. It's like the old SF/Fantasy subgenre of using aliens or robots to talk about real-world racism. For example, the "filter bubble" is one way the chatterers want to consider the problem of why not everyone is a liberal intellectual (term used descriptively). One obvious possible answer is of course that they just don't know any better. And if they were just brought The Word, by missionaries (err, the proper pundits), they'd convert. This fails rather badly, because in the main it's just not true (in fact, the reverse is more accurate).
On the other hand, there really can be very bad echo chambers, as we've seen with e.g. vaccines. These can't ever be eliminated entirely (human nature) - but I do think there's some structures which make them better or worse Unfortunately this gets into issues of money and civic infrastructure and similar, which is a self-reinforcing problem.
Equating "partisan" with "unreliability" is a bit problematic. Relying on Newsguard is even more so. Infact using Newsguard is probably a fatal flaw in the logic chain of the analysis. Truth isn't always found in the middle of the normal distribution of one poorly measured variable.
Bravo Charles. Great work as ever.
It would be useful to have a gender breakdown for this trial. Having just read a preview copy of James Ball’s book ‘The Other Pandemic’, it seems a certain subset of men are the big drivers of online misinformation and online conspiracies, which I found interesting as it counters our societal conditioning that men are the ‘rational’ sex.
That's a good question, Heather - there's not a lot, except that they note that they tried to recruit according to benchmarks for race, gender, age and location, "but a non-random subset of these respondents opted in to installing the browser extension. Compared to national estimates, our 2018 sample had fewer people aged 18-24, a greater proportion of strong partisans [which they actually wanted], and morte participants who identified as white, male and college-educated. Among our 2020 sample, the differences in age and partisan identity groups were closer to national estimates than in 2018, and a greater proportion of participants identified as female." (I'm now trying to find the tables where they give the overall gender breakdown..) However I don't think they looked at gender breakdown for who the misled partisans were.
Keep in mind "misinformation" isn't a monolith. For example, I'd conjecture anti-vax before Covid skewed heavily female, since that's how childcaring responsibility skews. Election-denial is likely predominantly male. Also, there's social networks which aren't that visible from the perspective of the Very Online. The phrase "church lady" is a cliche for reason.
I think the system/people interaction is complicated, so taking it to either extreme isn't true (i.e. people aren't completely slaves to propaganda, but virtue-preaching yields no insight). One idea I wish I could explore more is: An aspect of this complication is that "algorithms" and such, is a mechanism which allows the chattering class to seriously discuss media issues in a way which would otherwise be considered outside their boundaries. It's like the old SF/Fantasy subgenre of using aliens or robots to talk about real-world racism. For example, the "filter bubble" is one way the chatterers want to consider the problem of why not everyone is a liberal intellectual (term used descriptively). One obvious possible answer is of course that they just don't know any better. And if they were just brought The Word, by missionaries (err, the proper pundits), they'd convert. This fails rather badly, because in the main it's just not true (in fact, the reverse is more accurate).
On the other hand, there really can be very bad echo chambers, as we've seen with e.g. vaccines. These can't ever be eliminated entirely (human nature) - but I do think there's some structures which make them better or worse Unfortunately this gets into issues of money and civic infrastructure and similar, which is a self-reinforcing problem.
Equating "partisan" with "unreliability" is a bit problematic. Relying on Newsguard is even more so. Infact using Newsguard is probably a fatal flaw in the logic chain of the analysis. Truth isn't always found in the middle of the normal distribution of one poorly measured variable.