IB Chemistry Scientific Investigation: the 20% shaped before the report is written

Most students begin by looking for an impressive topic. That is usually the wrong first move.

The Scientific Investigation is worth 20% of the final IB Chemistry grade, but many of the decisions that shape its quality happen before the final report is written. Can the school laboratory measure the expected change? Is the selected range wide enough to reveal a relationship? Can the procedure be repeated consistently? Does the resulting evidence answer a genuinely chemical question?

A reaction that finishes in three seconds, a colour change smaller than the instrument can distinguish, or a method that alters several variables at once can weaken an investigation before the student creates the first graph.

The encouraging part is that an excellent investigation does not require an exotic topic. A modest experiment with a thoughtful pilot, trustworthy data and careful chemical reasoning is usually stronger than an ambitious project that cannot produce interpretable evidence.

The answer in 60 seconds

The current Scientific Investigation is an individual research task assessed at both SL and HL. The student formulates a research question, gathers or selects quantitative data, analyses the evidence and communicates a conclusion and evaluation in an individual written report.1 2

Four facts are worth knowing immediately:

• it contributes 20% of the final Chemistry grade;
• the recommended time allocation is 10 hours;
• the report has a maximum overall length of 3,000 words; and
• it is assessed through four criteria worth six marks each: Research design, Data analysis, Conclusion and Evaluation.1 2

The strongest investigations are not necessarily the most complicated. They are the ones in which the research question, method, measurements, analysis and conclusion fit together.

If you are a student, your first priority is not writing the introduction. It is testing whether your proposed investigation can generate useful evidence.

If you are a parent, your most valuable role is to protect planning time, ask about feasibility and encourage early teacher feedback—without taking ownership of the assessed work.

Four problems to find before they damage the investigation

Pilot work is valuable because it reveals problems while they are still inexpensive to solve.

The response is too fast to measure

Suppose a reaction is effectively complete before the reactants can be mixed, the apparatus sealed and the first measurement recorded. The topic may be chemically interesting, but the method cannot capture the part of the process needed for a reliable rate calculation.

The solution might be to reduce concentrations, change the temperature range, use automated data logging or choose a different way to measure the response. The correct choice depends on the chemistry and the equipment available.

The signal is smaller than the experimental variation

Imagine expecting a change of 0.01 absorbance units when repeated measurements under the same conditions vary by approximately ±0.02. More decimal places in the spreadsheet will not solve the problem. The method cannot distinguish the expected effect from the variation in the measurement.

The student may need a wider range, a more sensitive technique, better control of conditions or a different dependent variable.

The range is too narrow

Five values do not automatically make a useful investigation. Temperatures of 20, 21, 22, 23 and 24°C provide five levels, but the expected change may be small compared with uncertainty and random variation.

An extremely wide range can be equally problematic. Assumptions may stop being valid, evaporation may become important, the reaction mechanism may change, or the fastest conditions may become impossible to measure. A useful range must be wide enough to reveal a relationship while remaining chemically sensible and experimentally controllable.

Several important variables change together

If temperature, concentration and surface area all change between conditions, it becomes difficult to decide which factor produced the observed result. A good investigation isolates a meaningful relationship and controls the factors that could otherwise distort it.

Finding any of these problems during a pilot is not failure. It is evidence that the pilot has done its job.

What the current Scientific Investigation requires

The current task can be summarized as follows.1 2

Feature	Current Chemistry requirement
Weighting	20% of the final grade
SL and HL	Common task and assessment criteria
Recommended time allocation	10 hours
Final product	Individual written report
Maximum overall word count	3,000 words
Evidence	Quantitative data
Total marks	24
Criteria	Research design, Data analysis, Conclusion, Evaluation

Students and schools often still call this task the IA. That remains useful shorthand, but the current Chemistry course describes it as the Scientific Investigation.

This distinction matters because many online templates still describe the previous assessment model. A resource that organizes the report around Personal engagement, Exploration, Analysis, Evaluation and Communication is referring to the old five-criterion model. Before following any online checklist, verify that it explicitly addresses the current course with first assessment in 2025.

Where the 24 marks actually come from

The criteria should not be treated as four isolated writing sections. Together, they describe one connected scientific argument.

Research design — 6 marks

Can the reader identify a focused research question, understand the chemical context and see a method capable of answering the question?

Good research design is not measured by the length of the method. It is shown through purposeful choices: the variable range, the measurements, the controls, the quantity of data, the treatment of safety and the practical detail needed to reproduce the investigation in principle.

Data analysis — 6 marks

Have the measurements been transformed into evidence carefully, correctly and transparently?

This includes clear recording and processing, appropriate units and significant figures, meaningful treatment of uncertainty and graphs or models that are justified by the data and the chemistry.

Conclusion — 6 marks

Does the final claim answer the research question, use the processed data and apply correct chemical reasoning?

A conclusion is not a summary of the method. It is a judgement about what the evidence shows, how the result relates to chemistry and how confidently the relationship can be stated.

Evaluation — 6 marks

Has the student identified the limitations that materially affected this investigation and proposed improvements that would genuinely address them?

Evaluation requires more than a memorized list of “human errors.” It asks the student to connect a specific weakness in the procedure to an effect on the measurement, an effect on the result and a realistic improvement.

Conclusion and Evaluation together account for 12 of the 24 marks. Collecting data is therefore only the middle of the investigation—not its final intellectual task.

A successful investigation is a connected chain

It is tempting to think of the Scientific Investigation as an experiment followed by a report. That description misses what is really being assessed.

A successful investigation is a chain of decisions:

chemical idea → focused question → defensible method → suitable measurements → transparent processing → chemical interpretation → conclusion → evaluation

Each stage places a limit on the next one. A vague question usually produces an unfocused method. A poor method produces weak data. Weak data cannot be rescued by a sophisticated graph, and a confident conclusion cannot be justified by evidence that the method was unable to measure reliably.

This is why the best time to strengthen an investigation is usually before final data collection begins.

Start with a question that can survive contact with the laboratory

Before perfecting the wording, identify the relationship the experiment is meant to investigate.

Ask four questions:

1. What chemical system is being studied?
2. What factor will be deliberately changed?
3. What response can be measured quantitatively?
4. What chemical principle or model will help interpret the result?

Too broad

How does temperature affect reaction rate?

There is no single experiment that can answer that question. It does not identify the chemical system, the temperature range, the measured response or the measurement approach.

A stronger structure

How does temperature (20.0, 30.0, 40.0, 50.0 and 60.0°C) affect the initial rate of oxygen formation during the iodide-catalysed decomposition of hydrogen peroxide, determined from the gradient of oxygen volume against time during the first 20 seconds?

This example identifies:

• the independent variable and proposed range;
• a quantitative dependent variable;
• the chemical system; and
• the way the dependent variable will be determined.

It is still not a finished investigation. Exact concentrations, volumes, controls, equilibration time, mixing method, safety considerations and the usable temperature range would need to be established through research and pilot work.

The example above is provided to demonstrate question structure. It should not be treated as a ready-made topic or procedure.

Original does not mean “never done before”

The Scientific Investigation is school-level research. Students are not expected to discover a new chemical law or find a chemical system that has never been studied.

Originality is better understood as ownership of the scientific decisions. A familiar system can support a strong investigation when the student develops a focused question, adapts the approach appropriately, collects their own evidence and explains the chemistry independently.

What is not enough is a simple repetition of a standard classroom practical in which every important decision has already been made and the expected result is merely reproduced.

Trying too hard to appear unusual can also create avoidable problems:

• the reagent is not available;
• the school does not have the necessary instrument;
• the expected change is below the instrument's useful range;
• the reaction is too fast;
• the procedure introduces several uncontrolled variables;
• the chemistry becomes more advanced than the student can explain accurately; or
• the investigation creates unnecessary safety or environmental concerns.

A project is not strong because its title sounds complicated. It is strong because the evidence is good enough to support a careful chemical conclusion.

Keep the chemistry at the centre

Before writing a long background section, ask:

What chemical relationship am I actually investigating?

The investigation might concern kinetics, energetics, equilibrium, acid–base chemistry, redox, electrochemistry, spectroscopy, solubility or a structure–property relationship. Whatever the area, there should be a clear connection between what is measured and the chemical model used to interpret it.

If the chemistry appears only in the introduction and the rest of the report becomes generic statistics, the investigation has lost its centre.

Background chemistry should therefore be selective. Include the concepts, equations, assumptions and accepted information needed to justify the design and interpret the result. Do not turn the opening of the report into a textbook chapter that has little influence on the investigation itself.

Research design: make every choice defensible

Good design is not about producing the longest possible list of apparatus or controls. It is about making choices that lead to useful data.

Independent variable

What will be deliberately changed? Why is the proposed range chemically sensible? Can every value be prepared or maintained accurately?

Dependent variable

What will be measured in response? Does the instrument measure that quantity directly, or must it be calculated from another measurement? Is the response large enough to distinguish from uncertainty and random variation?

Controlled variables

Which other factors could change the dependent variable enough to affect the conclusion? How will each important factor be controlled, and how effective is that control likely to be?

Students often create long tables of controlled variables without explaining why the variables matter. A more useful test is:

If this factor changed during the experiment, could it alter the measured response enough to affect the conclusion?

If the answer is yes, the variable deserves attention.

Range and repetition

A common mistake is to think, “I have five values and three repeats, so I have enough data.” Those numbers may be reasonable in some investigations, but they are not universal guarantees of quality.

The range must be wide enough to reveal the expected relationship without creating new chemical or practical problems. Repetition should be sufficient to judge consistency and variation under the conditions of the particular method. Both decisions should be supported by the pilot rather than copied mechanically from a template.

The pilot: the most valuable small experiment

A pilot is not simply a smaller version of final data collection. It is a decision-making experiment.

Its purpose is to answer questions such as:

1. Can the dependent variable be measured reliably?
2. Is the response large enough compared with the instrument resolution and trial-to-trial variation?
3. Is the independent-variable range wide enough to reveal the relationship?
4. Can the fastest or most extreme condition still be measured properly?
5. Does the apparatus capture the most important part of the process?
6. Is each run practical within the available time?
7. What must change before final data collection?

The student should record the conditions tested, approximate results, unexpected observations, instrument limits, timing problems and the changes made in response. Those notes provide evidence of scientific decision-making and protect the student from repeating unsuccessful conditions later.

Starting early does not mainly mean writing early. It means discovering problems early enough to solve them.

Collect data that another scientist could audit

A good raw-data record makes the investigation transparent. Another reader should be able to see what was measured, in what units, with what resolution or uncertainty and in which trial.

Separate what the instrument actually measured from quantities calculated later. If a gas syringe records oxygen volume against time, those volumes and times are raw data. The initial rate determined from the gradient is processed data.

Raw-data tables should make clear:

• the independent-variable condition for each run;
• individual trial values;
• units;
• instrument resolution or measurement uncertainty where appropriate;
• consistent decimal places that reflect the measurement; and
• relevant observations that help interpret unexpected behaviour.

Qualitative observations can add useful context—for example, foaming, precipitation, incomplete dissolution or an unexpected colour change—but the current Scientific Investigation requires quantitative evidence.2

Avoid false precision. If an instrument records two decimal places, later spreadsheet calculations should not imply that six decimal places are experimentally meaningful.

Keep the original data. Do not overwrite raw measurements with corrected values or means. If a value is identified as anomalous, retain it, mark it transparently and explain how it was treated.

Repeats have a purpose

Repeats help the student examine consistency, random variation and unusual results. They also help judge whether a mean is representative.

They cannot repair a biased method, an unsuitable independent-variable range or a systematic loss of material. Repeating the same flawed measurement five times produces a more precise description of the flawed measurement—not necessarily a more accurate answer to the research question.

Turn measurements into evidence—not decoration

Students sometimes over-process data because complexity appears scientific. But calculations do not earn value simply by being difficult.

For every processing step, ask:

What does this calculation allow me to conclude that I could not conclude from the raw measurements alone?

Depending on the investigation, useful processing might include:

• calculating a mean and a measure of variation;
• converting an instrument response into concentration through a calibration relationship;
• determining an initial rate or gradient;
• calculating an equilibrium or thermodynamic quantity;
• propagating uncertainty through a calculated value;
• transforming variables to test a model; or
• comparing an experimental result with a literature or theoretical value.

Each calculation should be traceable. Show enough working for the reader to understand the method, then apply it consistently. The goal is not to fill the report with repeated arithmetic but to make the route from measurement to conclusion auditable.

Uncertainty is part of the argument

Uncertainty is not a decorative column added after the experiment. It helps determine how much confidence can be placed in the result.

Suppose two mean values differ only slightly. If the difference is small compared with instrument uncertainty or the variation between repeated trials, the evidence may not support a confident distinction between the conditions. If the observed trend is large and consistent relative to those sources of uncertainty, a stronger conclusion may be justified.

The relevant treatment depends on the measurement and processing. It may involve instrument resolution, absolute or percentage uncertainty, propagation through calculations, variation between trials, error bars or comparison with an accepted value.

The important question is not “Have I included uncertainty?” but:

How does uncertainty change what I am justified in claiming?

If the uncertainty analysis has no effect on the interpretation, it is probably being treated mechanically.

Graphs should reveal the chemistry

A graph is useful when it makes a relationship easier to interpret. It is not automatically better than a table, and a more complicated graph is not automatically more scientific.

Check:

• whether both axes are labelled with quantity and unit;
• whether the scale shows the data honestly;
• whether error bars are meaningful and clearly defined;
• whether the selected trendline or model is justified;
• whether a transformed plot has a chemical purpose; and
• whether the graph reveals the relationship relevant to the research question.

Do not force a straight line because linear graphs look tidy. If chemical theory predicts a curve, a curved relationship may be exactly what the investigation should reveal. Conversely, do not fit a high-order polynomial simply because the software produces a larger coefficient of determination. A mathematical fit must have a defensible scientific role.

Write a conclusion that answers the question

A conclusion should begin by answering the research question—not by retelling the method.

A strong conclusion usually performs four jobs:

1. Claim: State the relationship found.
2. Evidence: Support it with processed numerical results.
3. Chemistry: Explain the relationship using an appropriate chemical model.
4. Confidence: Judge how strongly the evidence supports the interpretation.

Using the illustrative data in Figure 7, a conclusion might begin:

Across 20–60°C, the mean initial rate increased from 1.20 to 5.00 × 10⁻⁴ mol dm⁻³ s⁻¹, a 4.17-fold increase. The change was large relative to the variation between repeated trials and was not linear across the investigated range. This is consistent with a larger fraction of collisions having energy equal to or greater than the activation energy at higher temperatures. However, these data alone do not demonstrate that the reaction follows the Arrhenius model; an appropriate transformed analysis would be needed to test that relationship.

This conclusion does not claim that the data are perfect. It distinguishes the observed trend from the model that might explain it and states what additional analysis would be required to make a stronger claim.

Where relevant, compare an experimental value with a literature value or theoretical expectation. The comparison should contribute to the scientific argument rather than become a disconnected paragraph.

Evaluation: explain what actually limited the evidence

“Human error occurred” is not an evaluation. Neither is “use more accurate equipment next time.” Both statements are too general to show how the limitation affected this investigation.

A useful evaluation follows a causal chain:

specific limitation → effect on the measurement → effect on the result or conclusion → realistic improvement

Consider a rate investigation in which the apparatus is sealed only after the reactants are mixed.

Weak evaluation

Human error occurred when the reaction was started.

Stronger evaluation

There was a variable delay between mixing the reactants and recording the first oxygen-volume measurement. Oxygen produced during this interval was not captured, causing the calculated initial rate to be underestimated. The effect was likely greater at higher temperatures because the early reaction was faster. A sealed vessel connected to a pressure sensor could be used, with data logging started before the catalyst is injected through a septum. This would capture the earliest part of each run more consistently.

The improvement matches the limitation. It explains what would change in the measurement and why that change would make the resulting evidence more reliable.

Not every imperfection deserves equal attention. Prioritize the limitations that had the greatest plausible effect on the result or on the confidence of the conclusion.

What about the 3,000-word limit?

The current Chemistry task has a maximum overall report length of 3,000 words.1

Do not treat 3,000 as a target. A strong scientific report is concise because its reasoning is clear, not because important reasoning has been hidden inside oversized tables or unexplained figures.

Schools should provide current instructions on what is and is not included in the word count. Follow those instructions rather than relying on an old IA template found online.

Edit for purpose. Each paragraph should help the reader understand the question, the design, the evidence, the chemical interpretation or the evaluation. Remove background information that never affects the investigation and repeated explanations that do not advance the argument.

Does the investigation have to be a laboratory experiment?

No. The current sciences framework supports different forms of scientific inquiry where they are appropriate, including hands-on experimentation, databases, simulations, modelling or combinations of approaches.2 3

A database or simulation investigation is not automatically easier. It still needs:

• a focused research question;
• justified selection of data or model conditions;
• appropriate quantitative processing;
• chemical interpretation;
• consideration of uncertainty and assumptions; and
• critical evaluation of limitations.

The standard of reasoning remains the same. Students should confirm that the proposed approach meets their school's current instructions before investing substantial time in it.

Can students collaborate?

The final report and the student's scientific argument remain individual. The current sciences framework may allow limited collaboration during parts of the investigative process where the school considers it appropriate, but each student's investigation must remain distinguishable and the submitted report must be their own work.2

Students should not infer permitted collaboration from what another school allowed. Follow the instructions of the student's own teacher and school.

A realistic ten-week workflow

Schools organize the Scientific Investigation differently, so the schedule below is a planning model rather than an official IB timetable.

Week 1 — Understand the task

Read the school's instructions, deadlines and assessment criteria. Confirm what equipment, software, reagents and data sources are genuinely available.

Week 2 — Explore possible questions

Generate several ideas. Reject those that are too broad, unsafe, unmeasurable or impractical. Discuss the shortlist with the teacher.

Week 3 — Design the investigation

Define the variables, proposed range, measurements, controls, quantity of data and intended processing. Complete the required safety, ethical and environmental considerations.

Week 4 — Pilot

Test whether the method produces a measurable and useful response. Refine the range, timing, concentrations or measurement technique before final data collection.

Weeks 5–6 — Collect the final data

Record raw data carefully, retain individual trials and note unexpected behaviour. Back up the original dataset in more than one secure location.

Week 7 — Process the evidence

Complete calculations, graphs and appropriate uncertainty treatment. Check units, formulae, spreadsheet references and significant figures.

Week 8 — Interpret the chemistry

Decide what the processed evidence shows, what chemical model explains it and what the evidence does not establish.

Week 9 — Evaluate

Identify the limitations that most affected the measurements or conclusion. Trace their effects and propose realistic, matched improvements.

Week 10 — Edit and check integrity

Check the logic, concision, figures, references, word count and consistency of the report. Confirm that the submitted work authentically reflects the student's own decisions and understanding.

Five mistakes that repeatedly weaken investigations

Choosing a topic before checking whether it can be measured

A good-sounding idea is not useful if the available equipment cannot produce data that distinguish the proposed conditions.

Writing pages of theory before testing the method

Background chemistry matters, but a long theoretical introduction cannot compensate for a method that fails to generate usable evidence.

Changing several important variables at once

When multiple factors change together, the result becomes difficult to interpret chemically.

Treating uncertainty as a box to tick

Uncertainty should influence the confidence and wording of the conclusion. If it changes nothing in the interpretation, it has probably been included mechanically.

Writing generic evaluation points

The evaluation must show understanding of this experiment, these measurements and this conclusion—not a memorized list of laboratory problems.

What parents can help with—and what they should not do

Parents do not need to understand every calculation to provide valuable support.

They can help by:

• protecting regular time before deadlines;
• asking whether the pilot produced a measurable change;
• encouraging the student to seek teacher feedback early;
• helping keep milestones and appointments visible;
• supporting sleep, workload management and realistic planning; and
• listening while the student explains their reasoning in their own words.

Useful questions include:

• “What did your pilot show?”
• “Can your equipment distinguish the conditions you want to compare?”
• “What is the next decision you need to make?”
• “When will you discuss this with your teacher?”
• “What evidence would make you change your conclusion?”

Parents should not:

• choose the research question;
• rewrite sections of the report;
• turn the student's writing into an adult voice;
• create, remove or modify data;
• perform calculations or analysis for the student; or
• outsource any part of the assessed work.

The principle is simple: support the process without taking ownership of the product.

What appropriate tutoring support looks like

Appropriate support teaches the student how to make scientific decisions independently.

A tutor can help a student learn how to:

• narrow a research question;
• reason about variables, range and controls;
• judge whether a proposed measurement is feasible;
• understand the underlying chemistry;
• process data correctly;
• interpret uncertainty;
• understand teacher feedback; and
• evaluate a limitation scientifically.

A tutor should not produce text, calculations, graphs or conclusions for submission, nor fabricate or repair data. The student should be able to explain and defend every important decision in the finished investigation.

IB academic-integrity guidance emphasizes that submitted work should authentically represent the student's own ability and that contributions from other sources must be acknowledged appropriately.4

What about AI tools?

AI tools may help a student understand a chemical concept, generate questions to consider or identify an error for the student to investigate. They can also produce confident but chemically incorrect explanations, inappropriate statistical advice, invented references or prose that does not represent the student's own work.

Students must follow their school's current AI and academic-integrity policies. Work produced by an AI tool should not be presented as if it were the student's own, and any permitted use must be acknowledged in the way required by the school and the IB.4 5

The safest principle is the same as for tutoring: use support to improve the student's thinking, not to replace it.

Final pre-submission check

Research question and design

• Is the research question focused and clearly chemical?
• Can the chosen method genuinely answer it?
• Is the range wide enough and still chemically sensible?
• Are the important variables controlled or addressed?
• Is the procedure reproducible in principle?
• Have relevant safety, ethical and environmental considerations been addressed?

Data and analysis

• Are raw and processed data clearly distinguished?
• Are individual trials retained?
• Are units, decimal places and significant figures appropriate?
• Can the reader follow the processing and sample calculations?
• Are error bars, transformations or fitted models justified?
• Does uncertainty affect the interpretation?

Conclusion

• Does the conclusion answer the research question directly?
• Does it use processed numerical evidence?
• Is the chemical explanation correct and relevant?
• Is the confidence of the claim justified?
• Does the conclusion avoid claiming more than the data demonstrate?

Evaluation

• Are the most important limitations specific to this investigation?
• Is the effect of each limitation traced through the measurement to the result?
• Does each improvement address the stated limitation?
• Are the improvements realistic in a school or comparable laboratory?

Communication and integrity

• Is the report concise and within the permitted length?
• Are tables and figures readable and numbered consistently?
• Are all sources and permitted external contributions acknowledged?
• Does the report authentically represent the student's own understanding and work?

Frequently asked questions

Is the Scientific Investigation the same as the Chemistry IA?

“IA” remains common shorthand because the task is the internally assessed component of Chemistry. The current course describes the task itself as the Scientific Investigation. When searching for guidance, check that the resource refers to the current assessment model rather than the previous five-criterion IA.

Do SL and HL students complete the same task?

Yes. The Scientific Investigation uses common requirements and assessment criteria at SL and HL. The chemical context and depth should still be appropriate to the student's course and investigation.2

How many independent-variable values and repeats are required?

There is no universal number that guarantees a strong investigation. The quantity of data should be appropriate to the research question, expected relationship, method, variability and available time. A pilot should inform the range, spacing and repetition plan.

Does the investigation need a completely new topic?

No. Students are not expected to discover a new chemical law. However, the investigation should not be an unthinking repetition of a standard practical in which every meaningful decision has already been made. The student must take ownership of the question, design, evidence and reasoning.

Does it have to be performed in a laboratory?

Not necessarily. Database, simulation, modelling and combined approaches may be appropriate. They still require a focused question, justified data selection, quantitative analysis, chemical interpretation and critical evaluation. The student should obtain school approval before committing to an approach.2 3

Should the report be exactly 3,000 words?

No. The 3,000-word figure is a maximum, not a target. The report should be as concise as possible while containing the reasoning needed to make the investigation clear and defensible. Follow the school's current guidance on what is included in the count.1

Can students collaborate?

The school may permit limited collaboration during aspects of the investigative process, but the student's investigation must remain distinguishable and the final report must be individual. Follow the school's instructions rather than assumptions based on another student's experience.2

Can a parent or tutor review the work?

They may support planning and help the student understand chemistry or feedback, within the school's rules. They should not rewrite, calculate, analyse, fabricate data or create material for submission. The student must remain the author and scientific decision-maker.

Can AI be used?

Any use must follow the school's current policy and IB academic-integrity expectations. AI-generated material must not be presented as the student's own work. Even when a use is permitted and acknowledged, the student remains responsible for checking the chemistry, evidence and references.4 5

Need structured support without outsourcing assessed work?

The ChemistryTutor Scientific Investigation programme is designed to teach the process while keeping every scientific decision and every line of the report under the student's ownership.

Students learn how to develop a feasible question, pilot a method, analyse evidence, reason about uncertainty and evaluate limitations—without outsourcing assessed work.