Methodology
How do we verify public figures’ statements?
From the entire pool of politicians’ statements, we choose those that are substantive and pertain to objectively verifiable realities. The selection of the research sample is strictly qualitative and based on the criterion of material accessibility.
- We scrutinize statements that are subject to verification.
- We do not employ probabilistic selection, i.e. random selection. The frequency with which a politician provides objectively verifiable information determines how often they will feature on our portal.
- Our sources of statements include recordings of key programs featuring politicians on radio and television, transcripts of parliamentary sessions, as well as direct statements made by politicians on their social media profiles.
- We also verify statements of politicians upon request, submitted through the contact form (report a statement for verification), or via our social media channels.
This sampling method does not allow for drawing general conclusions about the entire class. Therefore, we cannot assert that statistically, politicians from party “A” are more likely to provide false data (but we can state that in the examined sample of politicians, representatives of party “A” are more likely to provide false data).
The value of our research also has a qualitative aspect: we address questions about how politicians construct their statements, what manipulative techniques they employ to mislead their audiences, and on which subjects they provide truthful information.
In our analyses, we refrain from determining whether a politician lied in their statement. This is because falsehoods are not equivalent to lies. We leave it to the readers to judge whether the author intentionally deviated from the truth.
01 What kind of statements do we verify?
Fact-checking applies exclusively to content that is factual and relevant for the public debate.
- Factual content must be based on references to facts (numbers, past events, etc.).
- Factual content must extend beyond the scope of so-called common knowledge.
- Declarations regarding the future and opinions are not considered factual content.
02 What sources of information do we use?
We utilize publicly available sources, including official channels for accessing public information. Whenever possible, we opt for primary, unprocessed sources where data are presented in their original context. Our preference is for the most up-to-date sources available at the time of the politician’s statement.
The sources frequently employed by us include national, international, and foreign legal documents, reports from international organizations (such as the World Bank, OECD, WHO, ECDC), data from statistical offices (Central Statistical Office, Eurostat), reports from state audit institutions (Supreme Chamber of Control), reports from public opinion research institutes (Public Opinion Research Center), peer-reviewed scientific studies published in reputable journals (e.g., Science, The Lancet).
We reference opinions from leading experts and scientists in the respective field, considering their specialization and past scientific contributions.
In cases of uncertainty, we prioritize publications from governmental and research institutions over individual expert opinions.
For verifying central claims, we conduct analyses based on at least two independent and credible sources, unless there is only a single relevant source.
03 Ratings of the statements
True
We consider a statement to be true when:
- there are two credible and independent sources (or one if it is the only one relevant to the context of the statement) confirming the claim,
- it includes the most up-to-date data available at the time of the statement,
- the data is used in accordance with its original context.
In cases where terms such as “about,” “almost,” or “more” are used, rounding should align with the norms of everyday language, considering the context of the statement and the significance of the issue. A true statement may contain minor inaccuracies that do not impact the overall context of the statement.
Partly true
We rate a statement as partially true when:
- it combines true and false information. However, the presence of false information does not distort or misrepresent the claim presented in the statement,
- the actual data further supports the author’s claim.
False
A statement is rated false when:
- it does not align with any publicly available information based on a representative and credible source,
- the author presents outdated information contradicted by more recent data,
- it contains vestigially correct data but omits key information, thus presenting a false factual state.
A statement rated as false is not the same as a lie.
Manipulation
We rate a statement as a manipulation if it contains misleading or distorted facts, particularly by:
- omitting important context,
- using accurate data to draw false conclusions,
- selectively using data that supports the thesis (cherry-picking),
- employing disparate data to create the illusion of similarity or contrast,
- exaggerating one’s achievements or downplaying the role of the opponent,
- using non-substantive modes of argumentation.
Unverifiable
A statement is considered unverifiable if it:
- is impossible to verify in any available source,
- refers to outdated sources that do not allow judgments about the present situation,
- involves rough estimates with a high degree of uncertainty,
- contains imprecise or overly general statements,
- is unverifiable for other objective reasons.
This rating also applies to statements in the process of obtaining necessary information through the public information access procedure.
How do we verify the promises?
- In the case of election promises, the primary sources for their selection are the program documents of parties and election candidates, as well as their official speeches at pre-election congresses, rallies, or press conferences.
- For promises made by the Prime Minister, their exposé is also considered a source. Since politicians also make promises while in office, their source may be public statements made, for example, during media interviews.
- The fact-checking of promises, similar to statements, is exclusively applied to substantive promises that are specific and verifiable, and whose fulfillment status is relevant from the perspective of public interest.
- Our task is not to assess the validity of promises (whether political, economic, social, etc.), but solely to verify the degree of their fulfillment in comparison to the content of the declared commitment.
- We continuously verify promises, updating their fulfillment status at least once a quarter.
01 Ratings of the promises
Made
This is the initial assessment of any promise that meets our verifiability criteria. Such an assessment of a promise indicates that we will monitor its implementation throughout the term of the relevant body.
Done
We rate a promise as done if it has been completely fulfilled within the designated period, for example, during the term of the ruling party.
Done with a delay
We rate a promise as done with a delay if its complete implementation exceeded the set deadline, for instance, a law that was supposed to be passed within the first hundred days of a new government but was enacted later.
Partially done
We rate a promise as partially done when the declared commitments have only been implemented partially, meaning that certain steps have been taken toward their fulfillment. Based on the analysis of sources, it can be determined that the final actions align with the initially declared promise but do not entirely fulfill it.
In progress
The “in progress” rating is applied to a promise whose implementation is ongoing but has not yet been conclusively closed at the time of the analysis. For example, actions have been initiated to pass a specific law (the law is in the legislative process), or changes made to an already enacted law have not yet come into effect.
Frozen
A promise is rated as frozen if external circumstances prevent its fulfillment, for example: the lack of the required majority in the Sejm to pass a particular law, objections from EU bodies hindering the implementation of certain regulations, or a decision from the Constitutional Tribunal impeding a change in the law. This rating is also applicable to promises made by opposition parties that can only take action after assuming power.
No action taken
This rating is applied to a promise for which no actions have been taken to achieve its realization, and the specified deadline for implementation has not yet passed (e.g., the end of the term of office).
Broken
We rate a promise as broken if no actions have been taken to implement it by the end of the specified time period. This rating is used to evaluate promises both during the term of office (if the promise had a clearly defined deadline for implementation, such as passing a law within the first 100 days of government) and after the end of the term of office (for promises without a clearly announced deadline).
How do we verify fake news?
We are an organization that, for the sake of the quality of public debate, combats fake news and disinformation. In our analyses, we concentrate on the most harmful examples of disinformation, including:
- Health disinformation (false information about vaccines, the COVID-19 pandemic, alternative treatments for diseases).
- War disinformation (false information about the war in Ukraine, Russian propaganda).
- Climate disinformation (false information about the causes, course, and effects of climate change).
- Disinformation targeting minorities and protected groups (refugees, national, ethnic, religious and sexual minorities).
- Technological disinformation (false information about 5G technology).
- Conspiracy theories.
- Scams.
01 How do we verify fake news?
- We use specialized Internet monitoring tools (CrowdTangle, SentiOne, Brand24) to search for false information. We collaborate with the Institute of Media Monitoring (Instytut Monitorowania Mediów). Additionally, we consider submissions from our readers received through the contact form (Report fake news) or via our social media channels.
- Our topic selection is completely independent. The sole responsibility for their selection lies with the Demagog editorial team.
- For fact-checking fake news, we apply the same substantive criteria as when verifying statements. All our analyses are grounded in the most up-to-date, primary, and credible sources
- We quote the opinions of leading experts and scholars in the respective fields.
- As a partner of Meta in the Third-Party Fact-Checking Program Program, we have the ability to verify content on Facebook and Instagram, including ads, articles, photos, videos, reels, and text posts.
02 Ratings of fake news on Facebook and Instagram
In accordance with the program guidelines, we use the following categories when evaluating content posted on Facebook and Instagram:
False
Content that has no basis in fact. This includes:
- Fake quotes.
- Claims that are impossible, or that could not be considered an interpretation of something that actually happened or was said.
- Conspiracy theories that explain events as the secret work of individuals or groups, which may cite true or unverifiable information but present an implausible conclusion.
- Fabricated content on websites falsely presented as genuine news services.
- Content in an image, sound file, or video that is authentic but presented as evidence of an unrelated event (e.g., in a false context).
Altered
Image, audio, or video content that has been edited or synthesized beyond adjustments for clarity or quality, in ways that could mislead people. This definition includes splicing, but not media excerpts or taking media out of context. This includes:
- Faked, manipulated or transformed audio, video, or photos.
- Media edited to omit or reorder the words someone said to reverse the meaning of the statement
Partly false
Content has some factual inaccuracies. This includes:
- Inaccuracies or miscalculations regarding numbers, dates, times, but that could be considered an interpretation of something that actually happened or was said.
- A mix of true and false key claims, where the false claims do not predominate.
- Content presented as an opinion but based on underlying false information.
Missing context
Content that implies a false claim without directly stating it. This includes:
- Clips or excerpts from authentic media that hasn’t been altered or presented in a false context, but distorts the meaning of the original content to imply a false claim.
- Reporting on a false claim made by a third-party without questioning the veracity of the claim.
- Use of data or statistics that implies a false conclusion.
Satire
Content that uses irony, exaggeration, or absurdity for criticism or awareness, particularly in the context of political, religious, or social issues, but that a reasonable user would not immediately understand to be satirical. This may be from sites not clearly labeled as or widely-known as satire, or presented without clear labeling. Content rated as “Satire” will include fact-checkers’ articles for more context.
True
Content that contains no inaccurate or misleading information.
03 Rules of cooperation with Meta
- Content rated by us on Facebook is labeled with a special labeling.
- We do not delete any posts from Facebook – it is not possible within the program we are a part of.
- We cannot evaluate content posted by political parties or official profiles of politicians.
- For content rated as false, partially false, or altered, Meta imposes distribution restrictions among users, thereby reducing the reach of false information.
A detailed description of the program can be found at this link.
Corrections
The credibility of our articles is ensured through our policy of open and honest corrections.
- Every article undergoes a second round of review before publication. This process is conducted by a different analyst who is not the original author of the analysis and is overseen by the Editor-in-Chief. This cross-check evaluates both the selection of sources and the final conclusions.
- We prioritize transparency and integrity in our operations, so if a reader identifies an error or a lack of crucial information, they can report it by selecting the relevant text on the page and pressing the CTRL + Enter key combination.
- Within 24 hours of receiving the error report, we conduct a preliminary review of its substantive validity.
- If the submitted comments lead to a change in the rating, we make the necessary corrections, changing the rating of the statement. In the article, we include a new, corrected version of the analysis, while also clearly indicating the erroneous version, allowing for a precise comparison of the content of both versions.
- We prepare the corrected version of the article within 5 working days of receiving the report. However, if its preparation might take more than 5 working days (e.g., due to the need to obtain information through the access to public information procedure), we provide information about it by posting a relevant update in the article.
- Individuals or institutions described in our articles are offered the opportunity of right to reply regarding our findings, particularly in situations where a statement or information is unclear and requires clarification for an accurate verification of the factual state.
- You can find all articles in which we have made corrections or updates at this link.
If you believe that we have violated the principles to which we have committed, you can report it to the International Fact-Checking Network (link) or the European Fact-Checking Standards Network (link).
