Kinds of Quantitative Research & Experimental/Non-Experimental Designs

Definition & Purpose of Research Design

• Overall strategy that integrates all components of a study so they fit together coherently & logically
• Always chosen in line with:
  • The researcher’s aim (describe, predict, determine causality, compare, etc.)
  • How extensively the findings will be used (policy decisions, clinical protocols, theory-building, classroom practice, etc.)

Main Branches of Quantitative Research Design

• Experimental
  • Researcher manipulates an independent variable, controls extraneous factors, observes causal effect on dependent variable
• Non-Experimental
  • No manipulation or treatment; phenomena observed in their natural state; focus on description, relationships, trends, comparisons

Experimental Research Design – Core Concepts

• Strict adherence to the scientific method; highest level for testing cause-and-effect
• Key symbols (always written in LaTeX):
  • $R$ → Random assignment/randomization
  • $X$ → Treatment / Intervention
  • $O$ → Observation / Test (pre-test or post-test)
• Groups
  • Experimental Group – receives $X$
  • Control Group – receives no treatment, standard treatment or placebo
• Variables
  • Independent (IV) – presumed cause/manipulated factor
  • Dependent (DV) – measured effect/outcome

Examples of Experimental Questions

• “Is there an impact of social-media usage on high-school students’ academic performance?”
• “What is the influence of music therapy on anxiety reduction among cancer patients?”

Hierarchy of Experimental Designs

1. Pre-Experimental Designs (weakest internal validity)

• One-Shot Case Study
  • Structure: $X \; O$
  • Single group → treatment → single post-test only
  • No baseline, no control, minimal control of threats
  • Ex: Teacher uses Numbered Heads Together (NHT) in one class, then measures average speaking score.
• One-Group Pre-Test–Post-Test
  • Structure: $O<em>1 \; X \; O</em>2$
  • Same group measured before & after treatment; improvement interpreted as effect
  • Ex: 20-item attitude scale before 10-week counselling → counselling → same scale after.
• Static-Group Comparison
  • Structure: $X \; O<em>1 \,\, /\,\, \; \; O</em>2$
  • Two intact groups (non-random): experimental gets $X$, control does not, both post-tested
  • Ex: Class A taught with NHT vs. Class B without; compare post-test means.

2. Quasi-Experimental Designs (moderate control)

• Non-Equivalent Control Group
  • With pre- & post-test: $O<em>1 \; X \; O</em>2 \quad / \quad O<em>1 \; - \; O</em>2$
  • Without pre-test: $X \; O \quad / \quad - \; O$
  • Groups NOT randomly assigned; often intact classes, wards, communities.
  • Post-tests compared; pre-test helps check baseline equivalence.
• Time-Series Design
  • Repeated measures over time; pattern inspected for shifts after treatment.
  • Experimental group: $O<em>1 O</em>2 O<em>3 X O</em>4 O<em>5 O</em>6$
  • Optional control group with identical observations but no $X$.

3. True Experimental Designs (highest internal validity)

• Common features: randomization ($R$), control group, manipulation
• Pre-Test & Post-Test Control-Group Design
  • Structure: $R \; O<em>1 \; X \; O</em>2 \quad / \quad R \; O<em>1 \; - \; O</em>2$
  • Evaluates change while controlling for initial differences.
• Post-Test-Only Control-Group Design
  • Structure: $R \; X \; O \quad / \quad R \; - \; O$
  • Removes pre-test threat of sensitization; still random.
  • Example diagram: 100 teachers → $R$ → 50 $X$ (sensitivity training) vs 50 no $X$; measure morale.
• Solomon Four-Group Design
  • Combines both previous designs to detect pre-test effects
  • Groups:
  1. $R \; O<em>1 \; X \; O</em>2$
  2. $R \; O<em>1 \; - \; O</em>2$
  3. $R \; - \; X \; O_2$
  4. $R \; - \; - \; O_2$
  • Provides richest validity checks but needs large sample

Non-Experimental Research Designs

• No treatment/intervention; researcher observes variables as they occur.
• Descriptive
  • Describe nature, characteristics, components of a population/phenomenon.
• Survey Research
  • Large-scale data via questionnaires/interviews; can be:
  • Cross-Sectional – data at one time; compare groups (e.g., student nurses’ knowledge of neonatal resuscitation).
  • Longitudinal – repeat measures over extended period; track trends/changes in same population.
• Ex-Post Facto (Causal-Comparative)
  • “After the fact”; look backwards to possible causes of existing outcome (e.g., pill use & ovarian cancer).
• Correlational
  • Measure two (or more) variables; compute strength/direction of relationship (e.g., support system use & labour outcome).
• Comparative
  • Compare two or more pre-existing groups on one or more variables (e.g., rural vs urban older people’s health problems).

Quick Comparison of Major Non-Experimental Types

• Purpose: Describe (Descriptive) vs Relate (Correlational) vs Explain past causal conditions (Ex-Post Facto) vs Compare groups (Comparative) vs Track over time (Longitudinal).
• Time orientation: Cross-sectional (single snapshot) vs Longitudinal (repeated).

Practical & Ethical Notes

• True experiments often impossible/unethical in education & medicine → quasi or non-experimental used instead.
• Randomization increases internal validity but may reduce external validity if sample becomes unrepresentative.
• Placebo use must obey ethical standards—participants informed in consent if deception is involved.

Symbols & Notations Recap

• $R$ – Randomization
• $X$ – Intervention/Treatment
• $O$ – Observation/Test
• Subscripts (e.g., $O<em>1, O</em>2$) identify measurement sequence

Key Takeaways / Study Tips

• Always match design choice with research purpose & ethical feasibility.
• Verify if a study possesses three hallmarks of true experiment: manipulation, control group, random assignment.
• When critiquing studies:
  1. Identify notation pattern (e.g., $R\;O<em>1\;X\;O</em>2$) to categorize design.
  2. Check threats to validity (selection bias, maturation, testing, instrumentation, regression, mortality).
• Memorize hierarchy: Pre-Experimental < Quasi-Experimental < True Experimental for causal strength.
• Know common examples & be able to map research questions onto appropriate design.