Project info
Project name
6.4 Reconciling epistemic and demographic diversity
Work package
- Theory
Sustainability threat
- Feedback Cycles
Challenge
- Dealing with diversity
Study info
Description of Study
What are effective ways to identify false information on online social networks? Professional fact-checking has shown to be a powerful tool, but it is slow and costly. Algorithmic misinformation detection, as an alternative, is untransparent and still an early field. In this paper we ask whether letting users judge content as true or false might instead be an efficient alternative. We hypothesize that when information is first judged by subgroups of susceptible, like-minded people – as they would be found in so called echo chambers – biased judgements can grant false information with the necessary early support to later convince other, initially skeptical members of a group. We test this hypothesis by letting members of experimental bipartisan communities sequentially assess the veracity of true and false informational messages. We expect our findings to show that in most cases, judgements of susceptible and skeptical community members provide checks-and-balances that increase individual ability to correctly identify true and false messages. However, when susceptible individuals judge ideologically divisive, false information first, crowd intelligence backfires and correct identification deteriorates among susceptible and skeptical community members alike. In light of the omnipresence of segregated online communities and recent considerations for crowd-based solutions against misinformation, our study may highlight a potentially detrimental effect once they appear together.
Study research question
Under what conditions may user-based content veracity ratings hinder or amplify individuals’ ability to correctly identify true over false information?
Collection provenance
- Collected during project
Collection methods
- Experiment
Personal data
No
External Source
Source description
File formats
- csv
Data types
- Structured
Languages
- English
Coverage start
Coverage end
01/07/2021
31/12/2021
Spatial coverage
United States
Collection period start
01/07/2021
Collection period end
31/12/2021
Variables
Unit
Unit description
Sample size
Sampling method
Individuals
self-identified US conservatives and liberals
2000
Recruited on Amazon Mechanical Turk
Individuals
self-identified US conservatives and liberal
2000
Recruited on Prolific
Hypothesis
Theory
If the probability of individual, independent judgements being correct exceeds 0.5, the overall fraction of correct judgements increases in the alternating-order scenario in comparison to the independence scenario.
Wisdom of crowds under social influence
If the probability of correct, independent judgements from susceptible individuals exceeds 0.5, the overall fraction of correct judgements in the susceptible-first scenario increases in comparison to the independence scenario.
Wisdom of crowds under social influence
If the probability of correct, independent judgements from susceptible individuals is lower than 0.5, the overall fraction of correct judgements in the susceptible-first scenario decreases in comparison to the independence scenario.
Wisdom of crowds under social influence
In in the susceptible-first scenario, the individual propensity of a correct judgement decreases in i’s position in the sequence for susceptible individuals (H4a) and increases in i’s position for skeptical individuals (H4b). This is only the case if the probability of a correct, independent judgement from a susceptible individual is lower than 0.5.
Wisdom of crowds under social influence
Variable type
Variable name
Variable description
Dependent variable
fraction_correct
fractions of correct judgements in a sequence
Independent variable
social_influence
whether sequence allows for social influence (yes / no)
Independent variable
susceptible_first
whether susceptible individuals give ratings first (yes / no)
Discipline-specific operationalizations
Conflict of interest
Data packages
Publications
Documents
Filename
Description
Date
Ethics
Ethical assessment
Yes
Ethical committee
Ethical committee of the European University Institute, Florence