IB Chemistry IA Research Question and Pilot: Make the Study Testable

One of the most expensive mistakes in an IB Chemistry IA—the Scientific Investigation in the current course—is collecting full data too early. A student finds an interesting topic, writes a broad question and only then discovers that the change is too small, the reaction too fast, the instrument insufficiently sensitive or a supposed control variable moves with another factor.

Use this order: interest → filtered research question → pilot → decision → full data collection. A pilot is not a miniature final experiment. It is inexpensive information about which parts of the design work.

Eight filters for a viable research question

Before writing a method, ask these eight questions.

1. Is it chemically meaningful? Does answering it require a chemical model rather than a purely descriptive comparison?
2. Is the system defined? Are the substances, reaction, medium and relevant conditions named?
3. Is the independent variable measurable? Does it have a numerical range, unit and feasible interval?
4. Can the dependent variable be measured directly or derived defensibly? Is instrument resolution sufficient for the expected change?
5. Can the major confounding variables be controlled? Listing them is insufficient; the method of control must be known.
6. Can enough quantitative data be collected? Is there room for repeats and several levels of the independent variable?
7. Is there an analysis route? Can you anticipate which graph, model or derived quantity will answer the question?
8. Is it safe, ethical and realistic in resources? Does the plan fit school equipment, time and rules?

An “I don’t know” is not failure. It identifies exactly what background research or pilot work must resolve.

Use the eight filters three times: with the initial idea, after drafting the method and at the end of the pilot. A dependent variable may look measurable on paper, yet the pilot can show that the expected difference is indistinguishable at the instrument’s resolution. The chemical interest is not necessarily poor; the operational form of the question needs revision.

Attach evidence to each filter. Instead of “safe”, record concentration, hazard and risk reduction. Instead of “enough repeats”, state the number and the spread those repeats are intended to estimate. This makes the decision auditable by the supervising teacher.

From interest to testable wording

Too broad: “The effect of temperature on reaction rate.”

The system, measurement method, range and rate quantity are all undefined.

Better, but incomplete: “How does temperature affect the rate of reaction between sodium thiosulfate and hydrochloric acid?”

There is now a system, but “rate” remains unoperationalised and the range is open.

Testable starting point: “How does temperature from 20.0 to 50.0 °C affect the reciprocal of the time required to reach a defined optical threshold in the reaction of sodium thiosulfate and hydrochloric acid of specified concentrations, while total volume and initial reactant concentrations are held constant?”

This is not automatically final. The pilot must test whether the optical endpoint is reproducible, the 20–50 °C range is usable and the reciprocal-time analysis is defensible.

Good wording is not good because it is long. It is good because it exposes the major method and analysis decisions.

What a pilot must find out

A useful pilot is not a full data set. A few extreme and middle points with limited repetition can test five decisions.

• Range: is there a measurable change between the lowest and highest values?
• Resolution: is the signal between neighbouring levels larger than measurement scatter?
• Timing: is there enough time for reliable readings, or is the investigation too slow for the available session?
• Control: can temperature, concentration, mixing, pH or another condition actually remain stable?
• Data path: will the derived quantity and graph from the raw measurements answer the question?

Suppose that in a 20–50 °C pilot the difference between 20 and 25 °C is barely larger than scatter, while at 50 °C the reaction has already advanced substantially before timing can begin reliably. Do not hide these as “bad data”. They reveal that the method provides unequal information quality at the two ends of its range.

A defensible response might be to use 25–45 °C, reduce intervals where the response changes quickly, replace a subjective visual endpoint with an automated light sensor, or add repeats where signal and scatter are similar. The right decision depends on available equipment and the chemical question.

Do not change everything at once. If the pilot leads to a new range, concentration, endpoint and instrument, run a short second pilot; otherwise the new combination remains untested. Record the date and reason for each change, then apply the approved method consistently during full data collection.

Preserve pilot data even if they do not enter the final analysis. The notes document how range and repeat count were chosen and later support a specific evaluation. Do not automatically combine pilot and final data when the method changed materially between them.

Five red flags

1. The question contains no numerical range. A major method decision is probably still unresolved.
2. The dependent variable is “reaction”, “colour” or “effect”. It is unclear which measurement will become data.
3. Control variables appear only as a list. There is no plan for how, or to what precision, they remain constant.
4. The pilot is expected to look “good”. Its purpose is not to confirm a hypothesis; it is to find weaknesses in the method.
5. Analysis will be chosen later. If there is no visible route from data to answer, extensive measurement may produce little evidence.

Also stop if the plan assumes a hazardous condition, equipment that cannot be authorised or precision beyond the instrument available in school.

One-page decision sheet before full data collection

Complete this on one page, then discuss it with the supervising teacher.

Finish with one status: approved, approved with changes, or redesign required. Attach pilot evidence to every change.

Before full collection, write one sentence describing what data pattern would support the expectation and what pattern would weaken it. This does not invent a result; it records the interpretation rule in advance and reduces the risk of noticing only a convenient trend later. If the real data show another pattern, that pattern is what must be analysed.

For the complete path—from raw data through uncertainty, conclusion, evaluation and integrity review—use the IB Chemistry IA / Scientific Investigation guide. Support can help a student ask and test better questions, but the research decisions and submitted writing must remain the student’s own.