Or more accurately...
For "Netizens are more likely to be duped by misinformation presented in text form compared to video clips created with the help of algorithms"
read "A small self selected sample of Americans appear to be more likely to be duped by misinformation about polarised politics when presented in text form than when presented in video clips created with the help of algorithms".
Given the extraordinary (or quite possibly, increasingly ordinary) propensity of some people to believe the highly improbable in relation to politics regardless of how it's presented, this would not seem to be the best subject matter to elicit objective generalisable results in respect of differential capacity of alternative media types to convince.
However I am particularly concerned that only 15% off of the self selected sample was vetted for "political preferences, trust in media and politics [... or] an individual’s tendency towards reflecting on questions before answering".