Secondary Data in the ESS IA: Sources, Methods and Evaluation

12 Aug Secondary Data in the ESS IA: Sources, Methods and Evaluation

A suitable secondary dataset can support an IB ESS individual investigation. The challenge is to ask a focused question, explain the source and make your own justified analysis. Secondary data are not automatically inferior to primary measurements, and adding a few local readings does not automatically make an investigation stronger.

Use the current IA requirements

For the course first assessed in 2026, the report has a maximum of 3,000 words and six criteria: A, research question and inquiry; B, strategy; C, method; D, treatment of data; E, analysis and conclusion; F, evaluation. The older 1,500–2,250-word advice and alternative criterion lists do not describe this assessment.

Discuss the proposed question, available data and school approval before committing. A secondary-data method does not require proving that primary collection is impossible. Explain why the evidence is appropriate for the question and what limitations follow from its original purpose.

Match the question to available evidence

Published monitoring, satellite or survey records may support time trends, spatial comparisons and environmental–social relationships. A local question can also use secondary data if the location, timing and measurement are suitable. Conversely, a global dataset may be too coarse to answer a question about one street or habitat.

Before finalising a question, check that the variables, years, geographical coverage and observation frequency actually exist. Do not assume a familiar organisation’s name guarantees a complete dataset. Avoid choosing a question whose answer is simply copying the source’s existing graph or conclusion.

Evaluate the source

• Provenance: Who collected the observations, who processed them and who publishes the file?
• Definitions: What do the variables, units and flags mean?
• Coverage: Which locations, people or periods are represented or missing?
• Methods: Did instruments, classifications, sampling or reporting change?
• Uncertainty: What limitations or revisions does the source document?
• Access: Can you cite the exact dataset version and use it under its terms?

Save the original file and relevant documentation. Keep the essential selection and processing explanation in the report; an appendix alone cannot replace the method.

Useful starting points

Potential sources include Our World in Data, World Bank Open Data, and relevant national monitoring agencies. Follow curated datasets back to the original producer and inspect the documentation for the particular variable.

For US air quality, the EPA’s AQS access guidance explains available routes. The AQS data documentation describes summary records and completeness information. A reported annual summary is not automatically evidence that the full year has sufficient observations.

Make processing reproducible

1. Keep an unchanged source copy and identify the version or access date.
2. Define inclusion criteria, locations, period and variables before selecting favourable results.
3. Inspect missing values, duplicates, quality flags and changes in measurement method.
4. Document conversions and calculations, preserving the meaning of each measure.
5. Choose summaries and graphs that answer the question, with uncertainty and sampling structure considered.
6. Interpret the result at the scale supported by the data, including alternative explanations.

Missing values are not zero. Interpolation or replacement by a mean requires a specific justification and may distort trends or uncertainty; it is not a routine cleaning step. Do not silently discard incomplete periods.

CO₂ and CO₂-equivalent represent different quantities and cannot be made comparable through a simple mass-unit conversion. Likewise, a temperature anomaly requires its stated baseline. Check definitions before combining sources.

Aggregation needs care. A mean of monthly means may not equal an observation-weighted or time-weighted annual mean. Use the documented method for the quantity you intend to estimate. Averaging every station in a city without considering coverage, repeated records and weighting can create a misleading city-level indicator.

Connect the work to the six criteria

A fixed number of limitations or a sentence claiming secondary data are “better” does not automatically satisfy a criterion. Apply the current descriptors to your investigation.

Cite data and respect reuse terms

Record the creator or organisation, dataset title, date or version, publisher, DOI or URL and access date where appropriate. Cite both the data and relevant methods. Check each file’s terms, including third-party content; a public website is not a blanket licence to republish everything it hosts.

Present your own processing clearly and credit reused figures. A properly attributed background figure may be useful, but must not be represented as your own analysis. Do not include identifiable participant information in the report or upload it to unapproved tools.

Evaluate comparability and limitations

Explain how coverage, definitions or missing records affect the conclusion. Distinguish a documented bias from a possible one; do not invent its size or direction. Historical data can be appropriate for a historical question, even when they are not the latest release.

Cross-checking another source can help when the quantities, places, periods and methods are comparable. Three short local sensor readings cannot validate a monitoring station’s annual mean. Differences might reflect sampling duration, instrument behaviour or location rather than errors in the published dataset.

Illustrative planning example

Question idea: “How did a defined PM₂.₅ measure change at selected monitoring sites in [region] during [period]?” This is a starting point, not a claim that a particular set of complete observations has been located.

• Check the available sites, parameter definitions, observation duration and completeness flags.
• Choose a justified period and selection rule, and document changes in instruments or site coverage.
• Use the source’s appropriate annual measure or justify a transparent calculation.
• Discuss a relevant air-quality management strategy and competing perspectives without assuming a time trend proves the policy caused it.
• Evaluate how coverage and other factors limit the conclusion, and describe a meaningful unresolved question.

Understand the evidence before interpreting it

A large file does not guarantee independent observations or an appropriate design. Your contribution is the focused question, justified treatment and interpretation, not simply the volume downloaded. Do not present invented numerical examples or an expected decline as results.

Marija Maletz is an ESS teacher, examiner and tutor, with 13 years of teaching experience and more than 10 years as an IB ESS examiner. Request a FREE 20-minute trial lesson to discuss your needs and a starting point. Your analysis and assessed writing remain your own.

References

• IB ESS guide, first assessment 2026
• EPA: about AQS data
• EPA: obtaining AQS data