5 ESS SL IA sampling errors that quietly cap a 7 at a 5
Practical IB ESS SL Internal Assessment planning: how to design fieldwork, sample cleanly, and write the report so it scores the top band on moderation.
The IB Environmental Systems and Societies (ESS) Internal Assessment is the single piece of work where a Standard Level candidate can pull a 5 up to a 7, or lose marks they cannot recover in Paper 1 or Paper 2. It is a 2,250-word investigation worth 25% of the final IB Diploma grade, and it is the only ESS component that is marked against a moderation grid rather than a mark scheme. Most students read the word count and the topic list, then write a report that reads like a textbook chapter. A band-7 report reads like a scientific argument with visible method choices and visible limits.
What the ESS SL IA is really testing
The IA is a scientific investigation, not a research essay. Candidates pick a research question, collect primary data from an environmental system, analyse the data, and evaluate the methods. The IB markbands for the IA at SL score four criteria: research question and design, methodology and data collection, data analysis, and discussion and evaluation. Each band is described in terms that require visible methodology: hypotheses stated in operational form, controlled variables named, sampling strategy defended, uncertainty estimated, anomalies explained.
For most candidates reading this, the practical difficulty is not the topic; it is justifying the design. A topic such as "the effect of urbanisation on river invertebrate diversity" is workable only if the student can argue why Kick Sampling was chosen, why three transects were used, why abiotic variables were measured at the same time, and why abiotic data matters to diversity at all. A topic such as "the impact of plastic pollution on the ocean" is unworkable because the candidate cannot collect primary data within the available fieldwork window. The IB guide explicitly says personal fieldwork is required; secondary data is permitted only as a complement. The report must therefore sit on top of a piece of work a candidate actually did in a riverbed, a forest, a car park or a greenhouse.
Designing the research question so the report can score 7
The research question is the load-bearing wall of the whole report. A weak research question is unanswerable, too broad, or descriptive. A workable one is comparative, quantified, and tied to a specific place and time window. The strongest ESS SL IA questions I have seen look like three sentence-shaped moves at once: a dependent variable, an independent variable, and an environmental system bounded by a site.
- Operationalise the variable: "average number of macroinvertebrate taxa per kick sample at three sites along [named river]" beats "the biodiversity of the river".
- Make the independent variable measurable: "distance downstream of [named outflow]" beats "urbanisation".
- Lock the system in place and time: "sampled on three dates in May 2024 between 09:00 and 11:00" beats "in spring".
A common mistake is writing a research question that is interesting but cannot be answered with the equipment available. If the question is about primary productivity, the candidate should have a dissolved oxygen probe in the kit, not a Secchi disc alone. If the question is about soil organic matter, the candidate should have a plan for loss-on-ignition, not a promise of "further investigation". The markband for the first criterion explicitly rewards a question that can actually be tested in the field window.
Choosing a system you can reach on foot
Fieldwork access is the limiting reagent of the ESS SL IA. Candidates who plan a woodland study but attend a school two hours from any woodland are in trouble. The 12-week revision plan that sits around the IA is meaningless if the candidate does not have a site they can revisit. The system has to be physically reachable, legally accessible, and seasonally workable.
| System | Typical equipment | Fieldwork window | Common pitfall |
|---|---|---|---|
| Local river or stream | Kick net, identification key, conductivity meter, thermometer | Year-round, peak invertebrate signal in late spring | Storm events in the days before sampling wipe out the data; candidates forget to record weather |
| School grounds (soil, vegetation, light) | Quadrat, light meter, soil pH strips, soil moisture probe | Term-time only; repeatable across consecutive weeks | Variables are weakly differentiated; one site looks like the other and the graph shows nothing |
| Coastal rock pool or shoreline | Quadrat, identification key, salinity refractometer, tape measure | Tide-dependent; plan two visits at least one tidal cycle apart | Treating the rock pool as a closed system and ignoring freshwater input |
| Urban air or noise | Sound meter, particulate monitor, GPS | Daytime, with paired traffic counts | Conflating one site visit with a trend; the data is a snapshot, not a pattern |
For most candidates, a river or stream is the strongest option because the system varies along a visible gradient, the protocol is well documented, and the IB guide for Topic 2 provides the conceptual scaffolding the report can lean on. A school-grounds study is workable but the candidate has to engineer the gradient, which is harder than finding one.
Sampling strategy: the part moderators actually read first
In practice, moderators open the report and look for two things in the first two pages: a clear independent variable and a defended sample size. A sample of three sites, three visits, three replicates per visit is the standard for a workable SL study, which is 27 data points on the dependent variable plus the abiotic data. That is enough for a t-test, a Spearman rank, or a basic regression line. It is not enough for a chi-square, and the report should not pretend otherwise.
Common pitfalls and how to avoid them
- Mixing qualitative and quantitative observations. If a candidate writes "the water looked polluted" in the data table, the moderator will mark down the methodology band. Observations are photos, not data. The fix is to record the observation as a measured variable: turbidity, dissolved oxygen, ammonia, or a single 0–10 scale with the criteria defined in the appendix.
- Skipping the abiotic variables. ESS is a systems course, and the rubric expects the candidate to connect the biotic data to a measured abiotic driver. A report on invertebrates without temperature, pH, or dissolved oxygen loses the band-7 discussion even if the biotic data is strong.
- Reporting averages without a measure of spread. A column of means and nothing else suggests the candidate did not look at the raw data. Standard deviation or range should be in the table even if the candidate does not know what to do with it statistically; the discussion can use the spread to talk about uncertainty.
- Writing the conclusion before the analysis. The IB markbands reward discussion of evidence that actually challenges the hypothesis. A candidate who concludes that "pollution reduced biodiversity" without a single sentence of "the data does not show this because…" cannot score a 7 in the discussion band.
Writing the report so the markbands are obvious
The best ESS SL IA reports I have read look like they were written for a peer, not for a teacher. The structure is conventional, the figures are captioned, the method reads as a recipe that a stranger could follow, and the discussion is argument-shaped: a claim, a piece of evidence, a limit, a back-link to the system model from the syllabus. A report that reads as a textbook chapter usually signals to a moderator that the candidate did not analyse the data; the candidate transcribed it.
For the data analysis band, the report needs at least one figure where the x-axis and y-axis are visible, the units are stated, the trend is named in words below the figure, and the line of best fit is not drawn through the noise by hand. For the discussion band, the report needs a minimum of three explicit back-links: to the model (Topic 2 ecosystem flows, Topic 4 water, Topic 6 pollution), to the data, and to a published source or dataset that contextualises the magnitude of the result. For most candidates, the back-link to the published source is the weakest part, and it is the part that separates a 5 from a 7.
Time, word count, and the moderation trap
The 2,250-word limit is on the report body, not the appendices. Appendices can hold the raw data, the protocol, the photographs, the calibration notes and the full statistical test output. A candidate who puts the analysis into the appendix and the discussion into the body is misreading the markbands. The discussion is where the marks live, because four of the four IA criteria reward visible evaluation. Candidates who write a 1,200-word method and a 200-word discussion are spending their time on the lowest-weighted band of the rubric.
For most candidates I work with, the IA is more stressful than Paper 1 or Paper 2 because it has no revision and no second sitting. The way to make it less stressful is to plan the report in skeleton before the first fieldwork day, so the candidate knows what the discussion will look like before they take the data. The IB guide explicitly rewards planning, and the moderation grid for the first criterion treats a written plan as a partial credit signal even before the data is collected.
Conclusion and next steps
The ESS SL Internal Assessment rewards a small, well-bounded piece of fieldwork, written as an argument rather than a report. A candidate who treats the IA as a study in moderation, rather than a study in environment, will outscore a candidate who knows more ecology but writes in generalities. The strongest next move is to draft a research question, defend the sample size in writing, and lock the fieldwork dates into the school calendar before the data window closes. IB Courses' one-to-one ESS SL IA programme stress-tests each candidate's research question against the moderation grid and converts a 5-target into a concrete fieldwork plan.