Quasi-Experimental Design — Previous Year Questions

📊 Topic Overview

Metric	Value
Papers Appeared	16 of 28 (57.1%)
Priority	🔴 HIGH — heavily December
Section Distribution	Sec A: 8
Most Repeated	Types of quasi-experimental design (12 times)
Probability	Extremely High in December

🎯 Must-Know Concepts

1. Definition and meaning of quasi-experimental design
2. Types: non-equivalent control group, interrupted time-series, regression-discontinuity
3. Difference from true experimental design (no random assignment)
4. Advantages and limitations
5. Threats to internal validity specific to quasi-experiments

🔥 Most Repeated Question: Types of Quasi-Experimental Design / vs True Experimental (12 appearances)

Answer (Section A — ~450 words):

A quasi-experimental design resembles a true experimental design but lacks full experimental control — most critically, random assignment of participants to conditions. The term "quasi" (Latin: as if) signals that these designs approximate experiments without meeting all their criteria. They are widely used when random assignment is unethical, impractical, or impossible.

Distinction from True Experimental Design:

Feature	True Experimental	Quasi-Experimental
Random assignment	Yes	No
Manipulation of IV	Yes	Yes
Control group	Yes	Sometimes
Internal validity	High	Moderate (more threats)
Ethical feasibility	Sometimes limited	Often higher
Real-world applicability	Lower	Higher

Types of Quasi-Experimental Design:

1. Non-Equivalent Control Group Design Two pre-existing groups are compared — one receives the treatment, the other serves as control — but participants were not randomly assigned to groups. Pre-tests are administered to assess baseline equivalence.

Design: O₁ X O₂ (experimental group) O₃ O₄ (control group, no X)

Threat: Selection bias — groups may differ at baseline on important characteristics. Pre-test comparison helps identify but cannot eliminate this threat.

2. Interrupted Time-Series Design Multiple observations are made over time before and after an intervention. The researcher looks for a discontinuity — a change in level or slope — at the point of intervention.

Design: O₁ O₂ O₃ O₄ X O₅ O₆ O₇ O₈

Advantages: Controls for maturation and regression threats (because the trend before intervention can be projected). Useful for evaluating policy interventions (e.g., the effect of a new law on traffic fatalities). Threat: History — any event occurring at the time of intervention could explain the discontinuity.

3. Regression-Discontinuity Design Participants are assigned to treatment based on a cutoff score on a continuous assignment variable. Those above/below the cutoff receive the treatment; the discontinuity at the cutoff is the treatment effect.

Example: Students scoring below 50 on an entrance test receive remedial education; the researcher looks for a discontinuity in outcomes at the cutoff.

Advantage: Assignment rule is known and modelled, making causal inference relatively strong for a non-experimental design.

4. Separate Sample Pre-test Post-test Design Different (non-equivalent) samples are measured at pre-test and post-test. Less powerful than repeated measurement but avoids testing effects.

Advantages of Quasi-Experimental Designs: Greater ecological validity; feasible when randomisation is impossible; often the only option in applied policy and programme evaluation research.

Limitations: Multiple threats to internal validity; cannot definitively establish causality; requires more statistical sophistication to handle pre-existing group differences.

---

📝 Other Questions

Advantages and disadvantages of quasi-experimental design (Section B — ~250 words)

Advantages: Feasibility: When random assignment is impractical (e.g., studying effects of natural disasters, government policies, or clinical interventions), quasi-experiments are the best available option. Ethical appropriateness: Cannot withhold beneficial treatments from a control group — quasi-experimental designs allow study without this ethical violation. External validity: Studies are conducted in real-world settings, increasing generalisability compared to laboratory experiments. Longitudinal feasibility: Time-series designs track natural change over time.

Disadvantages: Weak internal validity: Selection bias is the primary threat — non-equivalent groups may differ on unmeasured variables. Confounding: Without random assignment, it is difficult to rule out alternative explanations for observed effects. Interpretation difficulty: Multiple plausible explanations for findings; statistical corrections (matching, ANCOVA) are imperfect. Regression to mean: Particularly problematic when groups are selected on extreme scores.

---

🧠 Memory Aids

Types (NIRRS): Non-equivalent control group, Interrupted time-series, Regression-discontinuity, Separate sample

Key difference from true experiment: No random assignment → selection bias is the primary internal validity threat

---

📚 Related Content

• Quasi-Experimental Design — Meaning and Types
• Quasi-Experimental Design — Specific Types
• Quasi-Experimental Advantages and Limitations

✅ Self-Assessment

1. What is the single most important way quasi-experimental designs differ from true experiments?
2. Draw the notation for a non-equivalent control group design.
3. What threat does the interrupted time-series design control for that other quasi-experiments do not?
4. How does the regression-discontinuity design achieve causal inference without random assignment?
5. When is a quasi-experimental design preferred over a true experiment?

💡 Exam Tips

• Section A: Define → Compare with true experiment (table) → Four types (each with design notation, example, key threat) → Advantages/Limitations.
• The design notation (O₁ X O₂ etc.) earns marks — write it for each type.
• December papers almost always include quasi-experimental design — this is the highest-priority December design topic.