On this page · 5 sections
Experimental-design questions ask you to plan, read, or judge a scientific investigation. They show up in two places on the AP Environmental Science exam. In the multiple-choice section (80 questions in 90 minutes), a stem gives you a hypothesis and asks which design would best test it, or which result would support or weaken it. In the free-response section, FRQ 1 is Design an Investigation, worth 10 points. You get 70 minutes for all three FRQs, and the section is 40% of your score.
FRQ 1 is scored in small pieces. Each part asks for one specific thing: a testable hypothesis, the independent variable, the dependent variable, a control, a constant, or a way to collect data. Most parts earn a single point, and readers look for the exact element the verb asks for. An answer that names the right idea but never ties it to the scenario usually earns nothing. The same question also mixes in content parts, such as explaining why nitrate runoff lowers dissolved oxygen, so you need the science as well as the method.
The core skill is the same in both sections. You must find the one variable being tested, keep everything else the same, and measure a result that actually answers the question.
The steps
- 1
Turn the scenario into one cause and one effect
Before you write anything, find the claim hiding in the stem. Rewrite it as "Does X change Y?" For example, "Does road salt concentration change radish seed germination?" X is the factor the researcher changes on purpose, which is the independent variable. Y is what gets measured, which is the dependent variable. If the stem gives a graph or a table, the x-axis is usually X and the y-axis is usually Y. Spend 30 seconds here, because every later part depends on getting this pair right.
- 2
Write a hypothesis you could prove wrong
A testable hypothesis names both variables and the direction of the effect. "Higher salt concentration will lower the percent of seeds that germinate" works. "Salt is bad for plants" does not, because it names no measurement. If the prompt asks for a null hypothesis, state that the independent variable has no effect: "Salt concentration will not change germination rate." Do not write a question when the prompt asks for a hypothesis. Do not write a prediction about something the experiment never measures.
- 3
Name the control and the constants separately
The control group gets no treatment or the normal condition, such as seeds watered with distilled water and no salt. It gives you a baseline to compare against. Constants are the conditions you keep the same in every group, such as seed type, light, temperature, soil, and water volume. Students lose points by calling a constant "the control." They are different things. The control is a group. A constant is a condition shared by all groups. Name at least one specific constant from the scenario, not a vague phrase like "same conditions."
- 4
Describe a method someone could repeat
When a part says "describe," give steps with numbers. Say how many groups, what levels of the independent variable, how many replicates, how long, and what you measure. For example, use five dishes at each of 0, 1, 2 and 4 grams of salt per liter, with 20 seeds per dish, and count germinated seeds after 7 days. Replication matters because one dish could fail by chance. More trials make the average more reliable and let you spot outliers.
- 5
Check that the design isolates one variable
On multiple-choice questions, test each option against the hypothesis. The right design changes only the factor named in the hypothesis and compares it to a group without that factor. Wrong options usually fail in one of three ways. They measure a variable the hypothesis never mentions. They compare places with different histories, so many things differ at once. Or they only observe natural variation, such as warm versus cold summers, when the hypothesis is about something else entirely.
- 6
Predict the result, then judge support versus weakening
Decide what the data would look like if the hypothesis were true. Then ask which result would break that pattern. Evidence weakens a claim when the effect appears even without the proposed cause. If hare populations cycle on islands with no lynx, predation cannot be the whole story. A result that matches what the hypothesis already expects does not weaken it, even if it sounds critical. Correlation alone also does not prove cause, because a third factor may drive both variables.
- 7
Answer the content parts with a mechanism
FRQ 1 often ends with parts like "explain why" or "propose a solution." Here the readers want a cause-and-effect chain from the course. For example: fertilizer runoff adds nitrogen and phosphorus, algae bloom and die, bacteria decompose the algae and use up oxygen, and fish suffocate. Give each link. One correct chain of three or four steps earns more than a list of five loose facts.
Worked through
Scenario (invented for practice): A city spreads sodium chloride on roads every winter. Meltwater drains into a field where farmers plant radishes in spring. A student wants to know if the salt is hurting the crop.
(a) Identify a testable hypothesis. Answer: If the concentration of salt in irrigation water increases, then the percent of radish seeds that germinate will decrease. This names the cause, the measured effect, and the direction, so it can be shown false.
(b) Identify the independent variable and the dependent variable. Answer: The independent variable is salt concentration in the water, measured in grams per liter. The dependent variable is the percent of seeds that germinate after 7 days. Notice that the independent variable is the thing the student sets, and the dependent variable is the thing the student counts.
(c) Identify the control group and one variable held constant. Answer: The control group is seeds watered with distilled water that has no added salt. One constant is the volume of water each dish gets, 10 mL per day. Other good constants are seed variety, light hours, and temperature. The control is a group, and the constant is a shared condition, so they go in separate sentences.
(d) Describe a method for collecting data. Answer: Set up 20 petri dishes, each with paper towel and 20 radish seeds. Give five dishes each of 0, 1, 2 and 4 g/L salt water. Keep all dishes in the same growth chamber at 22 °C with 12 hours of light. On day 7, count the seeds with a visible root and calculate the percent germinated for each dish. Average the five dishes at each level. The five dishes are replicates, which reduce the effect of one bad dish.
(e) Describe the expected results if the hypothesis is supported. Answer: Mean germination will be highest in the 0 g/L control and drop as salt rises, for example 90%, 75%, 50% and 20%. A downward trend across all four levels, not just a difference between two, is the strongest support.
(f) Explain how salt reduces germination. Answer: Salty water outside the seed has a lower water potential than the seed. So water moves out of the seed by osmosis, or at least does not move in. Without enough water, the seed cannot activate the enzymes that start growth, so fewer seeds sprout. This links the result to a mechanism instead of just restating it.
(g) Identify one limitation of the study. Answer: A petri dish is not a field. Real soil can bind or leach salt, and rain dilutes it, so field germination may differ. A follow-up could test soil plots near and far from the road while holding soil type constant.
Common mistakes
Mixing up the variables: calling germination the independent variable when it is the thing being measured. If you can set it with a dial, it is independent; if you count or measure it at the end, it is dependent.
Calling a constant the control. "Same amount of sunlight" is a constant shared by all groups. The control is a separate group that gets no treatment, such as seeds watered with plain distilled water.
Writing a hypothesis that cannot be tested, like "pollution harms ecosystems," or writing a question when the prompt asks for a hypothesis. Name both variables and the direction of the effect.
Picking a design that measures something the hypothesis never mentions. If the claim is about soil nitrogen from alder, measuring rainfall does not test it, because rainfall does not change with alder.
Choosing "evidence that weakens" a claim when that evidence is actually what the claim predicts. Prey rising when predators are scarce fits a predation hypothesis; prey cycling with no predators at all weakens it.
Describing a method with no numbers or no replication. "Water some plants with salt and see what happens" earns nothing. Give the levels, the number of trials, the time span and the measurement.
Checklist
- Can you rewrite any scenario as "Does X change Y?" in under 30 seconds?
- Can you write a hypothesis that names both variables and a direction, and a null hypothesis that says there is no effect?
- Can you name the independent variable as the factor you set and the dependent variable as the result you measure?
- Can you state a control group and a constant as two different things, each tied to the scenario?
- Can you describe a method with specific levels, number of replicates, time span and units?
- Can you explain why replication makes results more reliable?
- Can you spot a multiple-choice design that changes more than one factor, or that measures an unrelated variable?
- Can you tell evidence that weakens a hypothesis from evidence that only sounds critical but fits the prediction?