Our work methods

Cotejo.info uses fact checking as its preferred technique. Data journalism is integrated to this method of handling journalistic information. These modes of information management are largely based on Meyer’s precision and investigative journalism.

We have developed our own methodology, adapted to Venezuela’s context, where the lack of knowledge of official statistics is a constant, opacity and denial of reality in government public statements are the norm. Therefore, Cotejo.info has adapted the methodology used in South America by the leading fact-checking website of Argentina, Chequeado. Likewise, other fact-checking methodologies have been studied to conduct analysis of the public conversation and instances of mis/disinformation on social media.

Methodology for analyzing public conversation statements:

Procedures at Cotejo.info are implemented face-to-face and online. We hold regular meetings the last week of each month, and team members interact constantly in an online newsroom. Likewise, our web Coordinating Team is in daily communication to follow up on the pieces assigned, as well as to perform editing and content publication processes.

At the monthly meeting, team members tasked with preparing texts analyzing the public narrative submit topics they wish to address the upcoming month, in a deliberative and purposeful manner, with an approach that prioritizes depth over immediacy.

Journalists must follow the formal protocol below for their topic proposals:

1. Public statement or item to be fact-checked;
2. Original (video, audio, or text) source to be verified;
3. Preliminary research to ascertain the existence of sufficient data available for a successful verification process.

The Coordinating Team evaluates proposals, offers general guidelines, and formulates suggestions and approaches for those topics. Once topics are approved, our fact checkers follow the work methodology below to produce their texts, item visualizations, and multimedia content:

1. Select the statement or item from public conversation to be checked or verified.
2. Fact-check the original source that the statement or item (on video, audio, or published text) is based on. The original source is verified and, in accordance with our principles of transparency, offered to the reader;
3. Contrast the information from the original source with claims from other official sources relevant to the topic being investigated (for example, if the president gives information on oil production in Venezuela, we also investigate what state-owned oil corporation PDVSA, the Ministry of Petroleum, Venezuela’s representative in OPEC, the Vice President’s Office for Communications, and the Vice President’s Office for Economic Affairs of the Executive Cabinet have stated in this regard). In Venezuela, as is well known, official sources are currently very restricted by opacity, lack of transparency, and concealment of data, a prevalent feature of government policy;
4. Compare and verify cross-checked public-discourse items and issues with alternative, independent sources, whether domestic or foreign, of proven track record in mastering the subject under investigation, as well as with databases produced or reviewed and with previously published studies or research. This includes data and studies produced by multilaterals, NGOs, universities, among others. We find and analyze scholarly documents, primarily undergraduate and graduate theses / dissertations, scholars’ tenure and promotion papers, research conducted by specialized think-tanks, released at conferences, and/or published in refereed papers produced in academic institutions within or about the country. We build databases that are also made available to readers.
5. Once this research is completed and the relevant and timely data for the topic is collected, a news item is drafted, and context provided for the underlying story. To this end, journalists must first analyze the data collected, prioritize what is relevant so that they can organize the endless flow of data available in the 21^st century and make sense of the story by using background information and facts enabling a better understanding of the issues raised. We must process and hierarchize by refining the search with a focused process of inquiry and, finally, we must make sense of the stories we wish to report and analyze. Each piece is prepared using links connecting relevant portions to video, documents, databases, audio, photos, graphs, and/or infographics that support each of the components of the piece. In addition, by means of the links provided, we also allow our readers to directly access the sources from which the key elements for our analysis and verification for the piece originate.
6. Confirm, expose accuracies and inaccuracies of, or debunk the statements checked.

Once writers submit their notes, the Coordinating Team reviews the pieces by double-editing them. This ensures that the content meets the quality standards and format established for factual verification of information. They also review the multimedia products accompanying the piece and visualization aids prepared by the graphic designer.

Once these news pieces are approved, they are published in Cotejo.info‘s website and its social media accounts.

The verification of items and facts and the analysis performed by the journalist and the Editorial Team have allowed to prove that the statement or item assessed shows tampering with facts or data presented or an omission of relevant information that does not allow ruling what has been reviewed as a TRUTH or a LIE. These cases include, for example, those deemed improbable, inaccurate, incomplete, exaggerated, unsustainable, misleading, debatable, or premature (the statement could be true; but resulting from a prediction and not of a verifiable fact at the time the news item was released). Cotejo.info defines this rating as follows: “Even if the content components of a public statement cannot be reliably and completely proven upon verification, it will be rated as half-truth when the substance of the statement shows some consistency with factual information available and verifiable published evidence thematically connected to the fact-checked item is not sufficient to make an unequivocal ruling”.

The final purpose of a verifications or analysis on public statements and items is to evaluate and rate the public statement assessed as a truth, a lie, or a half-truth, which are the three categories we use in our fact-checking model, as explained below:

TRUTH

The verification of items and facts and the analysis performed by the journalist and the Editorial Team have allowed to prove that the statement or item assessed is veritable, authentic, accurate. Cotejo.info defines this rating as follows: “A public statement will be rated or labeled as a truth when, upon verification of its content components, it factually corresponds to verifiable evidence of public information available from entities, agencies, and across the media, or from other spokespersons with purview on the subject.”

LIE

The verification of items and facts and the analysis performed by the journalist and the Editorial Team have allowed to prove that the statement or item assessed is untrue, spurious, incorrect. Cotejo.info defines this rating as follows: “A public statement will not be admissible as truthful and should be assumed as a lie when, upon content verification, one or more of its components are refuted or disproved by instances of duly verified opposing facts found when performing a fact-check.”

HALF TRUTH

Another noteworthy aspect in the work procedure is Cotejo.info’s concern with building, preserving, and making the data used transparent. The sources on which the information is based are accessible to readers and we try to keep an archive in our cloud-hosted repositories with the databases found or built in the course of our efforts, trying to keep them updated and accessible to the whole team.

Methodology for social media fact-checks:

Since mid-March 2022, the Cotejo.info team began to verify instances of mis/disinformation circulating across social media, from its incorporation as the first local partner of Meta’s Third-Party Fact-Checker Initiative (TPFC) in Venezuela, an independent verification program in place since 2016 and currently with over 80 news partners in different locations of the world.

The methodology implemented for this fact-checking endeavor is similar to that conducted for analyzing public statements, preserving the quality and thoroughness standards typical of us.

In our online newsroom, collaborators send proposals of mis/disinformation items detected on Facebook, Instagram, or Threads, along with the components found during the preliminary research.

Our website coordinators review them and evaluate their relevance as a topic and their impact on public debate in Venezuela, as well as the scope of the publication. Once a proposal is approved, the assigned writer must prepare a rebuttal piece based on data and facts relating to the components of the mis/disinformation item.

The publication being analyzed must clearly state the hoax and must be contrasted with valid and reliable sources, in addition to being placed in context. In the piece being drafted, the steps taken towards verification and the grounds for ruling a content as fake, partly fake, altered, or missing context should be stated clearly.

For this method, links connecting relevant portions to video, documents, databases, audio, photos, graphs, and/or infographics that support each of the elements presented are also provided.

The rating categories in this regard have been adapted from those used within the TPFC program, described as follows:

FAKE

The item is not based on facts. This includes:

• Statements that are impossible or cannot be considered an interpretation of something that has happened or been claimed, for example: A statement about a natural disaster allegedly taking place when no such event has occurred; a claim that a person has created or patented something that they have not; conspiracy theories suggesting that certain events are the secret doing of individuals or groups (these theories may cite true or unverifiable information, but propose an improbable conclusion); a claim that a company is secretly engaged in drug trafficking, made on the basis that it charges high prices for its products or services, something having no connection with what is being stated; an unsubstantiated claim that some government officials are directly responsible for a terrorist attack solely for the purpose of starting a war; fabricated content from websites impersonating authentic news outlets; genuine images, audio, or video, but offered as evidence of an unrelated event (i.e., false context); an authentic photo showing that no damage has occurred in a city after an incident, being an image captured earlier; an authentic video showing a person confessing to a crime, but it is someone else; the presentation of an authentic act of government previously issued as if it were current, when a more recent act contradicts or repeals it.
• False quotes / misquoted portions;

ALTERED

Images, audio, or video content that has been edited or abridged beyond the extent necessary to provide clarity or quality, in ways that could mislead people on unfounded issues. This includes multimedia content created or edited digitally using artificial intelligence (AI). It also includes multimedia content that is linked without some portions (see the “missing context” information rating), or multimedia content for which the method of creation or editing, albeit not misleading, includes a false claim (see the “fake” information rating). It does not include multimedia content bearing one of Meta’s AI transparency tags or watermarks, or non-Meta transparency methods that clearly reflect the use of AI. Notwithstanding this, a publication may qualify if it includes an independent assertion of the use of digitally created or edited multimedia content. This category includes:

• Altered or transformed audio, video, or photos, for example: video speed modified to alter speech patterns in a deceptive manner; adding an image to an authentic photo to simulate something that has never happened;
• Synthetic images, audio, or video, including multimedia content using AI. For example: a photorealistic image showing an event that has not occurred; a video showing a person saying or doing something they did not say or do.
• Multimedia content edited with the objective of omitting or changing the order of words spoken by a person and intending the resulting statement to convey the opposite message. For example: Removing the word “not” in this sentence: “I will not do such a thing” said by someone;

PARTLY FAKE

The item contains some factual inaccuracies. This includes:

• A combination of true and false key statements in which those false do not predominate, for example: a list of several statements, some true and some false; a video containing many key statements being some true and some false;
• Content presented as opinion, but based on underlying inaccurate information. For example: Promoting a policy change supported by several key statements, one of which is demonstrably false.
• Inexact, miscalculated, wrongly transcribed figures, dates, and times, but which could be considered an interpretation of something that has happened or been stated, for example: Disclosing an incorrect number of people who registered for or attended an event; incorrectly estimating the cost of a government program;

MISSING CONTEXT

An item implying a false assertion not straightforwardly stated. This includes:

• Clips or snippets of authentic multimedia content that is neither altered (see definition for “altered” rating) nor presented in a false context (see definition for “false” rating), but distort the meaning of the original content to imply a false claim. It does not include clips of authentic political statements (e.g., text quotes or video clips; see definition of “political”), for example: an unaltered video segment of a group of people chanting and protesting peacefully, but the full-length video shows the same group instigating violence; splicing together authentic text fragments of a someone’s statements in a manner that changes, but does not reverse, the meaning of what they said;
• Reporting a false third-party claim without questioning its veracity. Learn more about our guidelines for reporting false third-party claims, for example: Broadcasting an interview with a guest making a claim that can be proved wrong, and a host not confirming or questioning its truthfulness;
• Using data or statistics leading to a wrong conclusion, for example: Hand-picking data from a study to imply a finding not reached by it; highlighting correlated data to imply causation that can be proved wrong.