1.5 Identifying the Research Question & Selecting the Best Procedure
Key Takeaways
- The research question of an investigation is reconstructed by identifying what was deliberately varied (independent variable) and what was measured as an outcome (dependent variable).
- A research question is distinct from the background motivation, the hypothesis, the procedure, and the conclusion; HiSET inquiry items routinely offer all five as answer choices.
- A question about a factor the passage says was held constant can never be the research question, because the design cannot answer it.
- The best procedure is the simplest design that varies the right factor, measures the right outcome directly, includes a comparison group, holds other variables constant, and is adequately replicated.
- Only a controlled experiment supports a causal claim; observational and comparative studies establish association when manipulation is impossible or unethical.
Why the Research Question Comes First
Every stimulus on the HiSET Science subtest describes an investigation, and a large share of the questions attached to those stimuli never ask you to recall a fact at all. Instead they ask you to work out what the investigation was actually trying to find out, or to pick which of four procedures would answer that question best. These are distinct, learnable skills, and they are worth practicing separately from content review because they transfer across Life Science, Physical Science, and Earth Science stimuli without any change in method.
A research question is the single testable question an investigation is designed to answer. It is not the same thing as the hypothesis, the conclusion, the procedure, or the reason anyone cared in the first place. Confusing these five is the most common way test takers lose points on inquiry items.
| Element | What it is | Typical signal language |
|---|---|---|
| Background/motivation | Why the topic matters | "Crop yields have fallen across the region..." |
| Research question | The testable question being answered | "How does X affect Y?" |
| Hypothesis | A predicted answer, stated before testing | "The researchers predicted that..." |
| Procedure | How data were gathered | "Twelve plots were treated with..." |
| Conclusion | The answer the data support | "Therefore, the treatment..." |
Recovering the Research Question from a Study Description
You almost never get the research question handed to you. You reconstruct it from the design using a two-step diagnostic:
- Find what was deliberately varied. That is the independent variable.
- Find what was measured as an outcome. That is the dependent variable.
The research question is then simply: How does [independent variable] affect [dependent variable] in [the system studied]?
Worked example. A researcher fills 20 identical tanks with pond water, adds 0, 5, 10, or 20 milligrams per liter of dissolved phosphate to five tanks each, holds temperature at 22 degrees Celsius and light at 14 hours per day for all tanks, and records algal cell density after 14 days.
- Varied: phosphate concentration.
- Measured: algal cell density.
- Held constant: temperature, light, water source, duration.
Research question: How does dissolved phosphate concentration affect algal cell density in pond water? Notice what the question is not. It is not "Is phosphate pollution bad for lakes?" (too broad and not directly testable by this design), and it is not "How does temperature affect algal growth?" (temperature was deliberately held constant, so this design cannot answer it).
Four Recurring Traps in Research-Question Items
- Too broad to test. "Does fertilizer harm the environment?" names no measurable outcome.
- Names a controlled variable. If the passage says a factor was held constant, no question about that factor can be the research question.
- Names an outcome that was never measured. If the study recorded cell density, a question about oxygen production is out of scope.
- Restates the conclusion. A conclusion is an answer; a research question is open.
Selecting the Best Procedure
The second half of this skill runs in the other direction: you are given a research question and asked which experimental procedure answers it best. Score each candidate against a fixed checklist rather than judging by how impressive it sounds.
| Criterion | Question to ask |
|---|---|
| Right variable manipulated | Does the procedure actually vary the factor named in the question? |
| Right outcome measured | Does it measure the named outcome directly, not a loose proxy? |
| Comparison group present | Is there a control or baseline to compare treated groups against? |
| Other variables held constant | Is everything except the manipulated factor kept the same? |
| Adequate replication | Are there enough subjects or trials that one odd result cannot drive the finding? |
| Feasible and ethical | Can it actually be carried out safely? |
Worked comparison. Research question: Does adding powdered limestone to acidified soil increase tomato seedling growth?
| Candidate procedure | Verdict |
|---|---|
| Grow 30 seedlings in limestone-treated acidified soil and 30 in untreated acidified soil, same seed lot, same light and water, measure height at 4 weeks | Best. Varies limestone, measures growth directly, has a control, holds other factors constant, replicated |
| Grow 30 seedlings in limestone-treated soil only and compare their height to the seed packet's stated average | Weak. No concurrent control grown under the same conditions |
| Grow 2 seedlings in treated soil and 2 in untreated soil | Weak. Replication far too small to separate signal from individual variation |
| Survey 50 gardeners about whether they use limestone and how their tomatoes did | Weak for this question. Observational, uncontrolled, and self-reported |
Matching the Design to the Question
Not every question calls for a controlled experiment.
- Controlled experiment — required whenever the question asks whether one factor causes a change in another. It is the only design that supports a causal claim.
- Observational or correlational study — appropriate when manipulation is impossible or unethical, such as studying the effect of a volcanic eruption or of a human disease. It can establish that variables move together but not that one causes the other.
- Comparative field study — uses naturally occurring variation across sites; higher external validity, lower internal validity than a lab experiment.
The single most common wrong answer in this item type is the most elaborate procedure. The best procedure is the simplest one that isolates the variable named in the research question and includes a fair comparison group. Extra instruments, extra treatments, and larger budgets do not make a design better if they do not address the question asked.
Exam Strategy
- Read the stimulus once for the design, not the science. Underline what was varied and what was measured before you look at the answer choices.
- If an answer choice mentions a variable the passage says was held constant, eliminate it.
- If a procedure lacks a comparison group, eliminate it, no matter how many subjects it uses.
- Match the design type to the verb in the question: cause demands an experiment; associated with only demands an observational study.
A soil scientist fills 24 identical planters with the same batch of sterilized loam, mixes earthworm castings into eight planters at 5% by volume, into eight more at 15% by volume, and leaves eight untreated. All planters receive identical radish seed, 200 milliliters of water daily, and 12 hours of light at 21 degrees Celsius. After 28 days the scientist harvests the radishes and records dry root mass. Which of the following best states the research question this investigation was designed to answer?
A public health researcher wants to determine whether a new mouth rinse reduces the number of oral bacteria colonies in adults. Which of the following procedures would best answer this specific research question?
Researchers want to know whether long-term exposure to volcanic ash fall reduces lung capacity in nearby residents. They cannot ethically assign people to breathe volcanic ash. Which study design is most appropriate, and what limitation must accompany any conclusion drawn from it?