To verify a shared electricity statistic, record the claim exactly, follow its citations to the earliest available dataset, read the metadata and definition, and check how the data was created, selected, excluded, or transformed before repeating the number.
Start With the Shared Statistic and Follow Its Source Trail
Begin by preserving the statistic exactly as presented. Record the number, unit, date or period, geographic area, accompanying wording, and the account that shared it. This prevents a later source from being mistaken for the source of a slightly different claim.
Next, inspect the post for a link, named publisher, report title, chart credit, or other attribution. Follow that trail through any linked article, report, or graph, checking each item for its own citation. The goal is to reach the earliest available dataset, data release, or documentation—not merely another page repeating the figure.
Keep a simple source trail as you proceed: social post → cited article or chart → report or data release → original dataset and metadata. Compare the wording and units at every stage. If a post says “electricity supplied,” for example, do not assume that a linked source measures the same concept unless its documentation establishes that definition.
Finding a related graph does not necessarily mean that the underlying records have been found. A published number or visualization may omit the raw data, recording process, inclusion and exclusion rules, or transformations applied before presentation. When the trail ends at such a page, treat it as the earliest source you could locate, but distinguish it from an inspectable original dataset.
Use Metadata to Identify What the Number Measures
Once you locate a dataset or data release, look for its metadata. Metadata describes data and may appear as variable descriptions, questionnaires, logs, readme files, or similar documentation. These materials can clarify what a field measures and help determine whether the shared wording accurately reflects the source.
Check the statistic against a short definition checklist:
- What is the variable called, and how is it defined?
- What unit is used?
- Which population, locations, organizations, or records are covered?
- What time period does the figure represent?
- How was the information collected or measured?
- Which observations were included or excluded?
- Was the data cleaned, combined, averaged, adjusted, or otherwise transformed?
Do not assume that every publisher supplies all of this information. Instead, separate what the documentation states from what remains unknown. A number can look precise while its practical meaning remains unclear if the population, time period, unit, or calculation is missing.
The creation method also matters. Dataset quality can depend on the accuracy of measuring devices or, when human behavior is measured, on the methodology used. Note any disclosed instruments, questionnaires, recording procedures, or calculation steps. If the social post uses a broader label than the documented variable supports, repeat the source’s narrower definition rather than the post’s interpretation.
Check Whether the Dataset Supports the Shared Claim
After confirming the definition, examine whether the available data can support the way the claim is framed. Ask whether the dataset is sufficiently large for the stated conclusion, whether its observations represent the relevant population or real-world setting, and whether duplicate records could distort the result. Also check whether exclusions or transformations could explain a difference between the original figures and the number circulating online.
These are questions to investigate, not automatic reasons to reject a dataset. The supplied guidance does not set an electricity-specific threshold for adequate size or representativeness. Your conclusion should therefore reflect the documentation actually available rather than an invented quality standard.
Google’s machine-learning guidance offers a useful illustration of dataset-composition concerns: it says validation or test data should be large enough to produce statistically significant results, representative of the wider dataset and relevant real-world data, and free of examples duplicated from training data. It also recommends separating training, validation, and test sets and warns that repeatedly using validation and test sets can weaken confidence that results will generalize to new data. This guidance concerns machine learning, not electricity-sector reporting, so it should be used only to illustrate why dataset size, representativeness, separation, and duplication can affect a reported result.
Finally, compare the documented dataset with the exact social-media claim. Confirm that the figure, unit, period, coverage, and definition align. If raw records or essential documentation are unavailable, state that limitation plainly and avoid presenting the claim as fully verified.
Conclusion
A defensible verification trail moves from the social post to its cited publication or chart, then to the earliest available dataset, metadata, and methodology. At each stage, compare the number’s unit, definition, period, coverage, exclusions, and disclosed transformations. Then consider whether the data-creation method, dataset size, representativeness, or duplicate observations affect how confidently the claim can be repeated.
This process does not validate any particular Nigerian electricity statistic: the available sources provide general dataset-assessment guidance, while the second source discusses machine-learning data. Before sharing an electricity statistic, save the original dataset link and verify its variable definition, collection method, coverage, exclusions, and disclosed transformations.
Disclosures and limitations
- This article was prepared with AI assistance from the supplied research package and content plan; its factual guidance is attributed through the listed source IDs.
- The sources provide general dataset-assessment guidance and a machine-learning illustration, not verification of a specific Nigerian electricity claim. This article contains no product recommendation or disclosed affiliate promotion.
Related reading
- Digital Safety Guides for Risky Links and Online Claims
- If a Supposed DisCo Representative Requests Remote Access: What Consumers Should Consider
- How to Verify an Outage or Reconnection Message Before You Act
Sources
- LibGuides: Evaluating information & data: Assess original dataset — libguides.eur.nl
- Datasets: Dividing the original dataset | Machine Learning | Google for Developers — Google for Developers