Experimental Design
A true experiment is the gold standard for establishing causal relationships. It requires three elements: (1) random assignment of participants to conditions, (2) manipulation of an independent variable (the treatment), and (3) a control or comparison group. When these conditions hold, differences on the dependent variable can be attributed to the treatment rather than to pre-existing group differences.
Core Features
| Feature | Requirement |
|---|---|
| Random assignment | Participants allocated to groups by chance |
| Treatment manipulation | Researcher controls the independent variable |
| Control group | At least one group receives no treatment or a placebo |
| Pre/post measurement | Outcome measured before and after intervention |
Random assignment is the critical distinction from Quasi-Experimental Design. It distributes confounding variables (motivation, aptitude, proficiency) roughly equally across groups, strengthening Internal Validity.
In SLA Research
True experiments are rare in classroom SLA research for practical and ethical reasons:
- Intact classes — schools rarely allow random assignment of individual students to groups mid-term
- Ethical constraints — withholding a potentially beneficial treatment from a control group raises concerns, particularly with minors
- Ecological validity — lab-based random assignment may not generalise to real classroom conditions (see Ecological Validity)
As a result, the vast majority of classroom intervention studies use quasi-experimental designs with intact groups, and true RCTs remain largely confined to laboratory settings or large-scale educational trials.
Variations
- Between-groups design — different participants in each condition (Between-Group Study)
- Within-subjects (repeated measures) — same participants experience all conditions, controlling for individual differences but introducing order effects
- Factorial design — two or more independent variables manipulated simultaneously (e.g., feedback type × task type), allowing interaction effects to be examined
Strengths and Limitations
Strengths: highest Internal Validity; permits causal inference; results are interpretable via standard statistical tests; Effect Size comparisons are straightforward.
Limitations: low Ecological Validity when conducted in labs; difficult to implement in educational contexts; Hawthorne Effect may confound results; randomisation does not guarantee equivalence in small samples (common in SLA).
Key References
- Campbell & Stanley (1963) — foundational taxonomy of experimental and quasi-experimental designs
- Shadish, Cook & Campbell (2002) — updated framework for causal inference in field settings
- Loewen & Plonsky (2016) — guidance on experimental design in SLA research