Research Methods: Experimental Design

How do psychologists move from an interesting idea to a definitive conclusion about human behavior? The bridge between hypothesis and knowledge is built through rigorous experimental design—the deliberate planning and structure of a study to test causal relationships. For IB Psychology, mastering experimental design is not just about passing Paper 3; it’s about developing the critical thinking skills to evaluate any claim about why people think, feel, or act as they do.

Core Components: Variables and Control

Every experiment is an investigation into a presumed cause-and-effect relationship. To conduct this investigation cleanly, you must first identify and control your variables.

The independent variable (IV) is the factor the researcher systematically manipulates. It is the proposed "cause." The dependent variable (DV) is the outcome that is measured. It is the proposed "effect." For a study to be a true experiment, the IV must be actively manipulated (e.g., one group receives a treatment, another does not) and participants must be randomly assigned to conditions.

The major threat to a clear result is the confounding variable. This is an extraneous variable (an unintended factor) that systematically changes along with the IV, making it impossible to know which variable caused the change in the DV. If you test the effect of a new study technique (IV) on test scores (DV), but the group using the technique also has more sleep, sleep becomes a confounding variable.

Researchers control extraneous variables—all variables other than the IV that could affect the DV—through standardization and randomization. Standardization involves keeping all conditions (instructions, environment, measurement tools) identical for all participants except for the manipulation of the IV. Random allocation of participants to conditions helps ensure that individual differences (like intelligence or motivation) are evenly distributed, preventing them from becoming confounding variables. This control is what allows researchers to establish cause-and-effect relationships; by isolating the IV and observing its impact on the DV, we can infer that changes in the IV caused changes in the DV.

Independent Groups Design

In an independent groups design (also called between-subjects design), different participants are used in each condition of the IV. For example, one group of participants learns a list of words in a silent room (Condition A), while a separate group learns the same list in a noisy room (Condition B). Their recall scores are then compared.

The primary strength of this design is that it avoids order effects, such as practice or fatigue, because participants only experience one condition. There is also no risk of participants figuring out the aim of the study (demand characteristics) from experiencing multiple manipulations.

However, its major limitation is participant variability. Differences in DV scores between groups could be due to the IV, or simply because one group happened to have people with better memories. This threat is mitigated by using a large sample and, crucially, random allocation. A second limitation is that it requires more participants than other designs to achieve the same statistical power, as data is collected from separate individuals for each condition.

Repeated Measures Design

A repeated measures design (within-subjects design) uses the same participants in all conditions of the experiment. Using the previous example, the same group of participants would learn word lists first in silence and then in noise (or vice versa), and their recall scores in each condition are compared.

The greatest strength of this design is the control it provides over participant variables. Since the same people are in all conditions, differences in memory ability are automatically controlled. This often leads to more sensitive detection of the effect of the IV and requires fewer participants.

The critical weakness is the high risk of order effects. Performance in the second condition might be better due to practice or worse due to fatigue. To counter this, researchers use counterbalancing. Half the participants experience Condition A then B; the other half experience Condition B then A. This balances out order effects across the experiment. Other limitations include the increased risk of demand characteristics and the fact that it cannot be used for experiments where participation in one condition permanently alters the participant (e.g., studying the effect of a life-altering therapy).

Matched Pairs Design

The matched pairs design attempts to combine the strengths of both previous designs. Different participants are used in each condition, but they are paired together on key variables relevant to the DV before the experiment begins. For instance, in a study on reaction time, participants might be given a pre-test. The two highest scorers are paired, and one is randomly assigned to Condition A, the other to Condition B. This continues for all participants, creating two groups that are very similar on the matching variable.

This design successfully controls for participant variables (better than independent groups) and avoids order effects (unlike repeated measures). It is useful when repeated measures is inappropriate.

Its limitations are practical. Matching can be time-consuming, expensive, and imperfect—you can only match on variables you know to measure and that you believe are relevant. It also requires access to a large initial pool of participants to form the pairs, and if one participant drops out, the entire matched pair’s data is often lost.

Internal Validity: The Goal of Good Design

The overarching aim of controlling variables and choosing an appropriate design is to achieve high internal validity. This refers to the degree to which we can be confident that the IV, and nothing else, caused the observed change in the DV. A study with high internal validity has successfully minimized the impact of confounding and extraneous variables.

Each design choice directly impacts internal validity. Independent groups designs threaten it through participant variability but protect it from order effects. Repeated measures designs threaten it through order effects but protect it from participant variability. Matched pairs seeks a balance. The researcher’s task is to select the design that best controls the most significant threats for their specific research question, always using randomization and standardization as foundational tools.

Common Pitfalls

1. Confusing Repeated Measures with Matched Pairs: A frequent error is to label a study as "repeated measures" simply because participants are tested twice. The key is whether the same IV condition is applied twice or different conditions are applied. In matched pairs, participants are tested once each, but in carefully constructed pairs.
2. Misidentifying the IV and DV: Students often mistake a measured variable for a manipulated one. Remember: if the researcher actively changes it to create groups, it’s the IV. If it’s simply recorded as an outcome, it’s the DV. For example, in a study correlating age and memory, age is not an IV because it is not manipulated; it is a co-variable in a correlational analysis.
3. Stating Controls Vaguely: It is insufficient to say "extraneous variables were controlled." You must state how. Specify the method: "Participant variables were controlled by using random allocation to conditions," or "Order effects were controlled by using counterbalancing in a repeated measures design."
4. Overlooking the Operationalization Gap: When designing a study, a major pitfall is failing to operationalize variables in a measurable way. "Happiness" cannot be an IV or DV until you define it as, for example, "a score above 24 on the Subjective Happiness Scale."

Summary

• Experimental design is the structured plan for testing causal hypotheses by manipulating an independent variable (IV) and measuring its effect on a dependent variable (DV), while controlling for confounding variables.
• The three core experimental designs each manage the trade-off between participant variables and order effects: Independent groups uses different people per condition (risk: participant variability; strength: no order effects), repeated measures uses the same people (risk: order effects; strength: controls participant variables), and matched pairs uses pre-tested pairs (balances both, but is complex).
• Control is achieved through standardization of procedures and random allocation of participants, which are essential for establishing cause-and-effect relationships.
• The ultimate objective of a sound experimental design is to maximize internal validity—the certainty that changes in the DV are caused by the manipulation of the IV and not by other factors.
• For the IB exam, you must be able to identify, evaluate, and design experiments using these concepts, always providing specific methodological details rather than general statements.