Skip to Main Content

Credible Resource Criteria: Evaluating Credible Resources Criteria

Evaluating Credible Resource Criteria

Identifying credible resources can be challenging because it is not something clearly stated. 

  • All sources should be evaluated for credibility.
  • Bias can exist on a spectrum of political ideology, religious views, financial influence, misinformation, and more. 
  • When you are considering which resources to use there are several criteria that you should check to verify that the resources or information are credible
  • The main points to take into consideration are is it from a reputable source, can you verify the information, is it relevant, is it up to date, and is it at an appropriate level for your needs?

The CRAAP Test

The CRAAP Test was developed as an easy to remember acronym that provides a set of criteria that can help you decide if a resource is credible. (1),(2)

  • C - Currency, • R - Relevance, • A - Authority, • A - Accuracy, • P - Purpose

-  Currency: 

  • When was the information or resource published or posted
  • Has the information been revised or updated
  • Is the information current or out of date?

- Relevance: 

  • How is the information related to your needs?
  • Who is the intended audience?
    • Is the information at an appropriate level for your needs?

       

- Authority: 

  • Who has produced the resource?
    • Is it a reputable organisation or expert in the field?
    • What are the author's qualifications, credentials or affiliations?

       

- Accuracy: 

  • Where does the information come from?
  • Is the information supported by evidence?
  • Has the information been reviewed or refereed?
  • Can you verify any of the information in another source or from personal knowledge?
  • Does the language or tone seem biased and free of emotion

  - Purpose: 

  • What is the purpose of the information? to inform? teach? sell? entertain? persuade?
    • Do the authors/sponsors make their intentions or purpose clear?
    • Are there disclosures of sponsorship or advertising
  • Is the information fact? opinion? propaganda?
    • Are opinions expressed balanced with facts?

The LOCAD Test

The LOCAD Test has been developed as a set of criteria that can help you decide if an AI resource is credible.

  • L - Limitations, • O - Ownership, • C - Confidentiality, • A - Accuracy, • D - Disclosure.

-  Limitations:  

  • AI can produce different outputs even when given the same input multiple times, leading to unpredictability in its results.
  • Is there bias in the information?
  • There is a dependency on AI data quality,
    • if the training dataset is limited in scope, so too will the generated answers.
  • Are there any training data date limits of the AI tool?

 - Ownership

  • What is the copyright of the information input or output?
  • Who owns the content that generative AI has created for you?
  • Is there intellectual property infringement?
  • Is there transparency?

- Confidentiality: 

  • Are you disclosing confidential information?
    • Re-identification Risk: A 2019 study showed that AI could re-identify 99.98% of individuals in anonymized datasets using only 15 demographic attributes. This puts patient privacy at risk, even when data is anonymized. (3)
  • AI typically uses previous user input to learn from and generate future content.

- Accuracy

  • Where does the information come from?
  • Is the information supported by evidence?
  • Can you verify any of the information in another source or from personal knowledge?
  • Do not use as a primary source, especially when there is no references.

  - Disclaimer:

  • If you use AI information, you should disclose that you have used it.
  • AI output is not always perfect. If you can not guarantee the accuracy of the AI-generated content you are responsible for the information you are disseminating.
  • Does your institution allow usage?

References

References:
(1) Sarah Blakeslee, (2004) “The CRAAP Test,” LOEX Quarterly 31, no. 3
(2) Applying the CRAAP Test. California State University-Chico. accessed 28th June 2021

(3) Rocher, L., Hendrickx, J.M. & de Montjoye, YA. Estimating the success of re-identifications in incomplete datasets using generative models. Nat Commun 10, 3069 (2019). https://doi.org/10.1038/s41467-019-10933-3