What is the Credibull score?
The credibull score is an algorithm based on several factors, weighted according to importance. The way this is done is an art form at the moment, and a work in progress. The algorithm for creating it will evolve over time.
I welcome input on the method, and feedback via comments on posts about changes which can be found using this tag: Algorithm Changes
For Journal articles
The factors for scientific papers are
- Size (n)
- who was studied, ie. profile of the subjects
- funding source
- number of citations
- scientific impartiality of researchers
- type [5 pillars from FMD book]
- Initial study
- Survey based/ based on qualatiative information
- Survey based/ quantative measures
- RCT – Randomised Control Trial or equivelant
- Repeating a study
- Meta Analysis
At the time of writing, September 2023, these factors are just accessed by OpenAi API. A specialist machine learning tool will be trained, and every article redone.
For credibility of Journalists, podcasters or other media creators
The factors used to understand those reporting on this, the users of studies, we use:
- correctly citing / referencing of mentioned studies
- context of citations and article comparison – relevance
- credibull score of studies cited
- has the study been refuted/ disproved after the article was published?
- ie article / podcast published on Jan 2023, cited study from August 2015, but study disproved by another team in 2017
- sponsors of the media
- who is the site owner? Biases?
- is the site biassed? Left or right leaning or more impartial/ partisan
- expert / trying to be expert
- commentary / opinion
- qualifications of author
- experience in field of author
- number of articles written
- number of months writing
- length of articles
- depth of articles
- [Rob Moore – investors, years of experience, doing Tesla daily for 5 years, interviewed hundreds of people]