Internal Assessment · ESS · 11 min read
The ESS IA: the six criteria, 15 research questions and how to score well
What the new individual investigation asks for, criterion by criterion, with 15 ideas that link data to a real environmental issue and a strategy.
On this page
The ESS IA, officially the individual investigation, is a report of up to 3,000 words in which you answer your own research question about an environmental issue, using data you collect or select, and then connect the issue to a real strategy and the tensions it creates. It is marked out of 30 on six criteria and counts for 25% of your final grade at SL and 20% at HL. Below: what each criterion rewards, what makes an ESS IA different from a Biology one, 15 research questions that work, and the statistics and timeline to go with them.
What is the ESS IA on the new course?
ESS was rebuilt for first exams in May 2026 and now runs at SL and HL. The IA is the same task at both levels:
- One open-ended investigation, about 10 hours of class time, answering a research question you wrote yourself.
- Any kind of data: primary or secondary, quantitative or qualitative. A literature review on its own does not count, because it is not a method someone else could repeat.
- 3,000 words maximum. Graphs, data tables, equations and calculations, citations and the bibliography are not counted, as in the science IAs.
- 30 marks across six criteria (A to F, below).
- Group work is optional. Up to three students can develop the question and method together, but each needs a unique research question, method and data, and the report is always individual.
- One round of teacher feedback. Your teacher advises but cannot choose your topic or method, and comments on one draft.
The IA is also the only place ESS assesses its fourth objective, investigating sustainability issues. That explains why the criteria care so much about the issue and the strategy, not just the data.
What do the six criteria reward?
| Criterion | Marks | What it rewards, in short |
|---|---|---|
| A. Research question and inquiry | 4 | A named local or global environmental issue, enough background to understand it, and a focused question tied to that issue |
| B. Strategy | 4 | An existing or emerging strategy that addresses an issue linked to your question, and a tension between perspectives that the strategy creates |
| C. Method | 4 | A repeatable method that gathers enough data: set-up, sampling or survey technique, how data were recorded |
| D. Treatment of data | 6 | Raw and processed data shown clearly; correct processing that actually answers the question |
| E. Analysis and conclusion | 6 | Every relevant pattern explained, with bias, reliability, validity and uncertainty; a conclusion that answers the question and follows from the analysis |
| F. Evaluation | 6 | Specific weaknesses and their effect on the conclusion, realistic improvements, and new questions the work raises |
The marks are front-light and back-heavy: A to C give 12 marks, D to F give 18. Students who spend half their words on background are spending them where the marks are thinnest.
What lifts each criterion
- A. Name the issue at a real scale ("nitrate run-off from arable fields into the River X catchment", not "water pollution"). Then state a question with a measurable independent and dependent variable, a place and a range. Background should explain the mechanism you expect, so your conclusion has something to test. How to sharpen a research question works the same way in ESS.
- B. Pick a strategy that really exists or is being proposed (a low-emission zone, a fertiliser buffer strip rule, a deposit-return scheme), describe what it does, and then explain who gains and who loses. "Farmers lose a strip of cropland; anglers and water companies gain cleaner water" is a tension. "Some people disagree" is not.
- C. Write it so a classmate could repeat it without asking you anything: quadrat size, number of samples, how points were chosen (random coordinates or systematic along a transect), times of day, exact database filters, survey wording. Justify the sample size.
- D. Show raw data (or a representative extract with the full set in an appendix if your teacher allows), then processed data with units and consistent decimal places. Choose processing that answers the question: a diversity index, a correlation, a comparison of means, a percentage change.
- E. Describe the trend with numbers, explain it with ESS reasoning, and say how far you can trust it: sampling bias, survey response bias, whether your measurement measured what you claimed (validity), how consistent your repeats were (reliability), and the uncertainty in your instruments.
- F. Name weaknesses specific to your investigation, say which way each pushed the result and how much it matters, and give an improvement that fixes that weakness. Finish with an unresolved question your data raised.
How is an ESS IA different from a Biology IA?
A Biology IA can be a clean lab experiment that never leaves the bench. An ESS IA is judged on whether it helps you understand an environmental issue. Three habits make that visible.
Think in systems
Sketch your issue as a system before you choose variables: storages, flows, inputs, outputs and feedback. For footpath trampling, that might be: trampling compacts soil, compaction reduces infiltration, more water runs off, run-off erodes the path edge, a wider bare strip pushes walkers further onto vegetation. That last link is a positive feedback loop, and it tells you what to measure (compaction, infiltration, vegetation cover at distances from the path) and what to discuss.
Use perspectives and value systems, not just data
Criterion B needs a tension between perspectives (economic, social, cultural, political or environmental). Environmental value systems give you the vocabulary: a technocentric view might favour engineering a solution, an ecocentric view might favour restricting the activity. Use them to explain why stakeholders disagree about your strategy, not as a definition section.
Connect to a real issue and strategy
The strategy can address a different issue from the one you measured, as long as the link to your research question is credible. A duckweed-and-nitrate lab experiment links credibly to rules on fertiliser use near rivers. It does not link credibly to plastic bag charges.
15 ESS IA ideas with research questions
Every question below names its variables and range. Replace the bracketed place with your own, and check the method with your teacher before you collect anything. If you are choosing between several, how to choose an IA topic has a quick filter.
| Type | Research question | Strategy for criterion B |
|---|---|---|
| Field | How does distance from the main footpath (0 to 10 m, every 1 m, on five transects) affect plant species richness per 0.25 m² quadrat and soil compaction (penetrometer reading) on [heath]? | Boardwalks or path rerouting: access vs habitat protection |
| Field | How does a macroinvertebrate biotic index from standard one-minute kick samples change from upstream of [a field drain] to 50, 100, 200 and 400 m downstream? | Riparian buffer strips: cropland lost vs water quality |
| Field + secondary | How does lichen diversity (Simpson's index, 20 trunks of one tree species per site) relate to the annual mean NO₂ concentration at the nearest official monitor, across six sites in [city]? | Low-emission zone: drivers and traders vs public health |
| Field | How does air temperature at 1.5 m, at 14:00 on three clear days, vary with percentage impervious surface within 20 m at ten sites in [town]? | Urban tree planting or green roofs: cost and space vs cooling |
| Field | How does infiltration rate (time for 500 cm³ of water to soak in through a 10 cm diameter cylinder) differ between lawn, compacted path and permeable paving on [school site]? | Sustainable urban drainage rules for new car parks |
| Field | How does litter density (items per 100 m²) change with distance (0 to 500 m) from the main access point of [beach]? | Deposit-return scheme or bin provision: cost vs clean-up |
| Lab | How does nitrate concentration (0, 5, 10, 20 and 40 mg dm⁻³) affect the growth of duckweed (Lemna minor), measured as frond number after 14 days? | Limits on fertiliser near watercourses: yields vs eutrophication |
| Lab | How does vegetation cover (0, 25, 50, 75 and 100% turf on a tray at 20°) affect the mass of soil washed off by 1 dm³ of simulated rainfall? | Cover crops on bare winter fields |
| Lab | How does salinity (0, 25, 50, 100 and 150 mmol dm⁻³ NaCl) affect the percentage germination of radish seeds after 7 days? | Drip irrigation to reduce salinisation: equipment cost vs yield |
| Lab | How does bag material (conventional plastic, "compostable" plastic, paper) affect percentage mass loss after eight weeks buried in garden compost? | Plastic bag charges or bans: retailers and shoppers vs waste reduction |
| Secondary | How did annual mean NO₂ at roadside monitoring stations in [city] change in the three years before and after the low-emission zone began, compared with background stations? | The low-emission zone itself |
| Secondary | What is the relationship between GDP per capita and municipal waste generated per capita across the 27 EU member states in [year], using Eurostat data? | Landfill taxes or producer-responsibility schemes |
| Secondary | How has the spawning stock biomass of [a named fish stock] changed relative to the agreed catch limits from 1990 to 2020, using published stock-assessment data? | Catch quotas or marine protected areas: livelihoods vs stock recovery |
| Survey | How does parents' support for a car-free zone outside [school] (Likert score) relate to their position on a ten-statement ecocentric–technocentric questionnaire? | School streets: convenience vs air quality and safety |
| Observation | How does food waste per student (g) differ between pre-plated and self-service lunch days over four weeks in [school canteen]? | Canteen waste-reduction policies: choice vs waste |
Surveys need informed consent, anonymous responses, a right to withdraw and no sensitive personal questions; your school may need parental consent for under-18 participants, so ask first. In the field, get landowner permission, avoid damaging the habitat you are studying, and do not remove or harm animals. Your teacher will know the IB's rules on experiments involving animals.
Should you use primary data, secondary data or both?
Either is allowed, and combining them is often the strongest design:
- Primary data gives you control over the method and plenty to evaluate. The risk is too few samples, collected on one day, at one site.
- Secondary data gives you years of records and many sites that you could never collect yourself. The risk is a weak method section. Criterion C still applies: state the source, the exact filters (stations, years, variables), how you handled gaps, and why you chose that subset. Evaluate the source too: who collected it, with what instruments, and whether methods changed over the period (see evaluating sources for the IA).
- Mixed designs use secondary data to supply a variable you cannot measure well. Lichen counts you made yourself, paired with official NO₂ readings, are a good example. Say clearly which data are yours.
For field sampling decisions (transects vs random quadrats, controlling confounders, a missing-data rule), follow the ecology field IA walkthrough; it applies to ESS unchanged.
Which statistics suit an ESS IA?
| Your question asks… | Use | Watch out for |
|---|---|---|
| How diverse is each site? | Simpson's reciprocal index, D = N(N − 1) ÷ Σn(n − 1); compare sites | Only compare sites sampled with the same effort |
| How many animals are there? | Lincoln index (mark–release–recapture): N = (M × C) ÷ R | Assumes no births, deaths or migration between samples, and that marks don't affect survival |
| Does Y change along a gradient? | Spearman's rank correlation | Aim for at least 10 pairs; correlation is not cause |
| Do two groups differ? | t-test (roughly normal data) or Mann–Whitney U | Report means or medians with spread, not just the p-value |
| Are two categories associated (e.g. survey answer vs age group)? | Chi-squared test of association | Expected counts should be at least 5 in each cell |
| Did something change over time? | Percentage change, a time-series graph, means before vs after | Natural year-to-year variation; pick periods of equal length |
Likert scores are ordinal: report medians and use Mann–Whitney or Spearman rather than averaging "agree" and "strongly agree". More detail on choosing and reporting tests is in statistical tests for the IA, and on presentation in graphs for the IA.
What does a realistic timeline look like?
Your school sets the actual deadlines. This order works whatever the dates are:
- Weeks 1–2: pick the issue, read enough to explain the mechanism, draft two or three candidate questions. Check that data can really be collected or downloaded.
- Week 3: write the method, risk assessment and ethics plan. Run a pilot: one transect, five survey responses, a test download. Fix what breaks.
- Weeks 4–6: collect data, with backup field dates. Enter everything into one spreadsheet the same day.
- Week 7: process and graph the data, run the tests, write the analysis.
- Week 8: write the strategy section and evaluation, then hand in the full draft. Your teacher comments on one draft only, so make it complete.
- Weeks 9–10: act on the feedback, cut to 3,000 words, check every graph and citation.
If you use AI tools at any stage, read the rules on AI in the IA first; the report must be your own work.
What sinks ESS IAs?
- A topic, not a question. "Plastic pollution in oceans" has no variables. Narrow it to something you can measure.
- An issue that disappears after page one. If the conclusion never returns to the issue from criterion A, the investigation reads as a lab exercise.
- A strategy section that only describes. Two paragraphs on how a carbon tax works, with no stakeholders and no tension, leaves most of criterion B unearned.
- Too little data. Three quadrats per site cannot support a correlation. Work out the number of samples you need before collection day.
- Generic evaluation. "Human error" and "more repeats" score little. "The penetrometer was read by eye to ±0.25 kg cm⁻², which is larger than the difference between 6 m and 8 m" is specific.
- Silently changing the question. If the data forced a change of plan, say so and explain it. Never adjust or invent data to fit your hypothesis.
Once the IA is in, the same skills turn up in the exams: Paper 1 is built around unseen case-study data. See the ESS exam format guide and the most common ESS mistakes.
Questions students ask
Can my ESS IA use only secondary data?
Yes. Primary or secondary data are both allowed, and so are qualitative or quantitative data. What you cannot do is submit a literature review: you still need a method someone else could repeat, which for secondary data means stating exactly which database, which years or sites, and how you selected and processed the records.
How long is the ESS IA?
At most 3,000 words. As in the science IAs, graphs, data tables, equations and calculations, citations and the bibliography do not count towards the limit. Your teacher can confirm how your school wants the opening details (title, candidate code, word count) set out.
Is the ESS IA different at HL?
No. The task, the six criteria and the 30 marks are the same at both levels. Only the weighting differs: 25% of the final grade at SL and 20% at HL.
Can I do my IA with friends?
Up to three students may develop a research question and method together, but each of you needs your own research question, method and data (for example a different independent variable, a different dependent variable or a different subset of a shared data set), and every report is written individually.
Does the ESS IA have to be fieldwork?
No. Lab experiments, secondary data from databases, surveys and modelling all work, as long as the investigation answers a focused question tied to a real environmental issue.
Try one: ESS
A student counts ground beetles in pitfall traps at a woodland site and records three species: species X 10 individuals, species Y 5, species Z 5. Using Simpson's reciprocal index, D = N(N − 1) ÷ Σn(n − 1), what is the value of D for this site?
- 1.24
- 2.92
- 0.34
- 3.00
Show the answer and why
Answer: B
N = 20, so N(N − 1) = 20 × 19 = 380. Σn(n − 1) = (10 × 9) + (5 × 4) + (5 × 4) = 90 + 20 + 20 = 130. D = 380 ÷ 130 = 2.92. Option C (0.34) is the same calculation upside down: the original probability form of Simpson's index, not the reciprocal form ESS uses. Option D is just the species richness. Option A is what a second site with the same three species but very uneven counts (18, 1, 1) would give: 380 ÷ 306 = 1.24. Same richness, much lower diversity, because evenness has dropped.