Research and scientific reasoning
Research design and validity
Variables, experimental and quasi-experimental designs, between-, within- and mixed-group designs, single-case and correlational methods, and threats to validity.
What this module draws on:
- Textbook concept Scientific method, experimental designs, validity threats, case studies, correlational, epidemiological, cross-sectional and longitudinal designs (Ch. 4).
- Original elaboration Design decision tree, research-reasoning exercises and fictional studies.
- Contemporary update Validity terminology aligned with Shadish, Cook & Campbell (2002).
Learning objectives
After this module you should be able to:
- Identify independent and dependent variables in a research question.
- Explain why random assignment allows causal inference and why it is often impractical in clinical research.
- Distinguish true experimental, quasi-experimental, between-, within- and mixed-group, analogue, single-case, correlational and epidemiological designs.
- Recognise common threats to internal and external validity and suggest controls.
- Choose an appropriate design for a given clinical question.
Introduction
Research is the foundation of clinical psychology: it tells clinicians which assessments are trustworthy and which treatments help. Good research design is about ruling out alternative explanations. Every design choice is a trade-off between control, realism, ethics and cost.
Variables and the logic of experiments Textbook concept
An independent variable (IV) is what the researcher manipulates or compares (for example, type of treatment). A dependent variable (DV) is the outcome measured (for example, symptom severity). Experiments try to show that changes in the IV cause changes in the DV, while minimising experimenter bias and error.
Random assignment is the hallmark of a true experiment: every participant has an equal chance of being in each condition, so pre-existing differences are spread evenly. Clinical research often cannot randomise for ethical, legal or practical reasons — you cannot randomly assign people to experience trauma, for instance.
Families of designs Textbook concept
| Design | Key feature | Main strength | Main limitation |
|---|---|---|---|
| True experiment | IV manipulated; random assignment | Strongest causal inference | Often impractical or unethical in clinical settings |
| Quasi-experiment | IV manipulated or compared without random assignment | Feasible in real settings | Groups may differ before the study |
| Between-group | Different participants in each condition | No carry-over effects | Needs larger samples; group differences |
| Within-group | Same participants measured across conditions or time | Each person is their own control | Order and carry-over effects |
| Mixed-group | Combines between and within factors | Tests group × time interactions | More complex analysis |
| Analogue | Simulates clinical conditions (e.g., students with mild fears) | High control | May not generalise to clinical populations |
| Case study | Intensive description of one person | Rich detail; useful for rare problems and hypothesis generation | Cannot establish cause; generalisation limited |
| Single-case (e.g., ABAB, multiple baseline) | Repeated measurement of one person across phases | Can show functional control experimentally | Generalisation needs replication |
| Correlational | Measures association between variables | Studies variables that cannot be manipulated | Correlation is not causation |
| Epidemiological | Studies incidence, prevalence and risk factors in populations | Population-level picture | Causal inference limited |
Time: cross-sectional and longitudinal designs Textbook concept
Cross-sectional designs compare different groups at one point in time (for example, people of different ages) — quick but vulnerable to cohort effects. Longitudinal designs follow the same people over time — they show development and change but are costly and suffer attrition.
Threats to validity Textbook concept
Internal validity asks whether the IV really caused the change in the DV. External validity asks whether results generalise to other people, settings and times. The textbook lists the threats shown below.
- History — outside events during the study (a festival, an exam period, a news event).
- Maturation — natural change over time (recovery, growing older, fatigue).
- Testing — taking a test changes later scores.
- Instrumentation — the measure or the raters change.
- Statistical regression — extreme scorers tend to score closer to the mean when retested.
- Selection bias — groups differ at the outset.
- Attrition (experimental mortality) — participants drop out unevenly.
Culture and generalisability Original elaboration
Much of the research the textbook reviews was conducted with Western, often university-based samples. Before assuming a finding applies in Baramulla, Bengaluru or Bogotá, ask: Were similar people studied? Were measures translated and validated? Do the outcomes matter in this cultural context? These are external-validity questions.
Worked example: A sleep study with a hidden problem Original, fictional
A fictional research team offers a sleep-hygiene workshop to night-shift nurses in one hospital and compares them with day-shift nurses in another hospital who receive nothing. After six weeks, the workshop group reports better sleep. Before concluding the workshop worked, consider: the groups differ in shift pattern and hospital (selection bias); a new rota was introduced in one hospital during the study (history); and only nurses who slept well kept completing diaries (attrition). Randomising nurses within the same hospital, and analysing everyone who started, would strengthen the conclusion.
Interactive activity
Research-design reasoning
Work through each fictional scenario. You will get feedback as soon as you choose.
Knowledge check
5 questions. Every option has an explanation. Your answers are not sent anywhere when JavaScript is available; without it, they are checked by the server and immediately discarded.
Summary
- The IV is manipulated or compared; the DV is measured.
- Random assignment allows causal inference but is often impossible in clinical research.
- Each design trades control against feasibility and realism.
- Internal validity concerns causation; external validity concerns generalisation.
- Cultural and contextual fit is part of external validity.
Questions for reflection
- Think of a question you care about in clinical psychology. What is the strongest ethical design you could use to answer it?
Key terms
Further reading
- Textbook Plante, T. G. (2005). Contemporary clinical psychology (2nd ed.). John Wiley & Sons.
- Contemporary Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental designs for generalized causal inference. Houghton Mifflin.
- Further reading Kazdin, A. E. Research design in clinical psychology (latest edition). Pearson.