How to use the simulator
The matcher shows one study at a time, labelled Scenario 1 of 7 and so on. The seven scenarios are shuffled each time the page loads.
Under the question Which research design is this? there are five answer buttons:
- Experiment
- Correlational
- Case study
- Naturalistic observation
- Meta-analysis
You get one attempt per scenario. Once you click, all five buttons lock. The correct design turns green, and if you picked something else your choice turns red. A feedback panel then opens with a verdict line (Correct or Not quite, naming the right design) and four short notes:
- Why this design: the feature of the scenario that identifies the method.
- Possible confound: a variable that could distort the result.
- Controls: what the researchers did, or could not do, to limit that problem.
- Ethics: the consent, privacy or harm issue the study raises.
The Score line beside the button keeps a running total, such as Score: 5 / 6 correct. Click Next scenario to move on. After the seventh scenario the list starts again, and the score keeps counting rather than resetting, so a second lap shows whether you fixed earlier mistakes.
The key ideas
These are the five designs in the matcher:
- Experiment. The researcher manipulates an independent variable and randomly assigns participants to conditions, then measures a dependent variable. It is the only design here that supports a cause-and-effect conclusion.
- Correlational study. Two or more variables are measured as they naturally occur. The result describes how strongly they are related, not why.
- Case study. One person, or a very small group, is examined in depth, often because the condition is rare. They give rich detail but generalize poorly.
- Naturalistic observation. Behavior is recorded in its everyday setting without any interference. The main threats are observer bias and participants changing their behavior because they know they are being watched.
- Meta-analysis. Statistical results from many earlier studies on the same question are combined into one overall effect size. No new participants are tested.
In an experiment, a confounding variable is anything besides the independent variable that differs between groups. Random assignment, a control group, a placebo and single- or double-blind procedures are the usual defences. Non-experimental designs cannot rule out third variables or reverse causation.
Every design must meet ethical guidelines: informed consent (parental consent plus the child's assent for minors), protection from harm, confidentiality, the right to withdraw, and a debriefing, especially after deception. An institutional review board checks these in advance.
Worked example
Scenario: A psychologist recruits 90 college volunteers. Using a random number generator, she puts 30 in each of three conditions: reading a passage while listening to songs with lyrics, reading while listening to instrumental music, or reading in silence. Everyone then takes the same 20-question comprehension quiz.
Step 1: Apply the two diagnostic questions. Was a variable manipulated? Yes, the type of background sound. Was assignment random? Yes, the random number generator decided each person's condition. Two yeses mean this is an experiment.
Step 2: Name the variables. The independent variable is background sound, with three levels. The dependent variable is reading comprehension, operationally defined as the number correct out of 20. The silence group is the control group.
Step 3: Look for a confound. If the lyrics group were tested in a noisier room, the noise rather than the music could explain their scores. Using one room for every session removes that; random assignment handles individual differences such as reading skill.
Step 4: Check the ethics. The risk is minimal, but volunteers still need informed consent, the right to withdraw and a debriefing.
Step 5: Change one detail. Suppose instead she surveyed 90 students about whether they usually study with music and compared their GPAs. Nothing is manipulated and nobody is assigned, so it becomes a correlational study. Any link could then reflect a third variable, such as study habits.
Common mistakes on the AP exam
- Calling any comparison of groups an experiment. Comparing groups that already differ, such as smokers and non-smokers, is not an experiment, because the researcher did not assign them.
- Confusing random assignment with random sampling. Random sampling (choosing who is in the study) helps you generalize to a population. Random assignment (deciding which condition each participant gets) is what allows a causal conclusion.
- Treating a survey as its own design. A survey is a way of collecting data. Most survey studies are correlational unless a variable is manipulated.
- Generalizing from a case study. Saying that a finding from one patient applies to everyone ignores the main limitation of the method.
- Naming an ethical issue without applying it. Writing "informed consent" is not enough. Explain what these participants would need to be told, or how this study could harm them.
When the AP exam uses this
Identifying and evaluating research designs is Science Practice 2 in the AP Psychology framework, and it runs through all five units rather than belonging to one topic. Multiple-choice questions describe a study and ask for its design or the conclusion it supports. The Article Analysis Question in the free-response section gives you a summary of a real study and typically asks you to identify the research method, a variable's operational definition, an ethical guideline and whether the findings generalize.