Experimental Research and

Observation

• Overview of Experimental Design, Validity, and

Quasi-experimental Design
• Understanding Experimental Research
Decisions
Experimental Research and
Observation

Dinh Thi Le Trinh

 What is Experimental Research?
• A research approach to establish cause-and-
effect relationships
• Involves manipulation of variables and
control extraneous variables
• Often used in psychology, education,
medicine, and social sciences.
Purpose of Experimental Research
• Test hypotheses about cause-and-effect
relationships.
• Examine the effectiveness of interventions or
treatments.
• Identify causal mechanisms influencing
outcomes.
Key Features of Experimental Research
• Control group vs. experimental group
• Random assignment
• Manipulation of the independent variable
 Types of Variables
• Independent Variable (IV)
• Dependent Variable (DV)
• Control Variables
 Types of Experimental Research
Designs
• True Experimental Design: Random
assignment used.
• Quasi-experimental Design: No random
assignment.
• Pre-experimental Design: Lacks randomization
and control groups.
True Experimental Design Features
• Random assignment to groups ensures
comparability.
• Control group receives no treatment or
standard treatment.
• Experimental group receives the
treatment/intervention.
Types of True Experimental Designs
• Pretest-Posttest Control Group Design
• Posttest-Only Control Group Design
• Solomon Four-Group Design
 Example: Pretest-Posttest Control
Group Design
• Study testing effectiveness of a new teaching
method.
• Group A (experimental) and Group B (control)
are both pretested.
• Group A receives new method, Group B
follows standard.
• Posttest administered to both groups.
Example: Posttest-Only Control Group
Design
• Evaluate training impact on customer service
skills.
• Participants randomly assigned to two groups.
• Only post-training performance is measured.
Example: Solomon Four-Group Design
• Used in evaluating a drug’s effectiveness.
• Two groups receive pretest, two do not.
• Allows analysis of testing effects along with
treatment effects.
 Internal Validity
• The degree to which changes in the
dependent variable are due to the
independent variable.
• Threats include: history, maturation, testing,
instrumentation, and regression.
Controlling Internal Validity Threats
• Use of control groups and random
assignment.
• Standardization of procedures.
• Blinding and repeated measures.
 External Validity
• Extent to which results can be generalized to
other populations, settings, and times.
• Threats: interaction effects, selection bias, and
setting-specific effects.
 Quasi-Experimental Design
• Used when random assignment is impractical
or unethical.
• Attempts to establish cause-effect
relationships with reduced control.
 Types of Quasi-Experimental Designs
• Nonequivalent Control Group Design
• Time Series Design
• Interrupted Time Series with Switching
Replications
 Nonequivalent Control Group Design
• Participants not randomly assigned to groups.
• Pretest and posttest measures are used.
• Groups may differ on key characteristics.
 Example: Nonequivalent Control
Group Design
• Study comparing two schools with different
math curricula.
• No random assignment possible.
• Pretest and posttest scores analyzed to
measure differences.
 Time Series Design
• Multiple pretests and posttests are used.
• Trends before and after intervention are
analyzed.
• Helpful in real-world policy evaluation.
 Example: Time Series Design
• Assessing policy impact on hospital wait
times.
• Monthly data collected 12 months before and
after policy implementation.
• Changes in trend analyzed post-intervention.
Example: Interrupted Time Series with
Switching Replications
• Smoking ban effect studied in two cities.
• City A gets ban first, City B later.
• Measure long-term health effects with
staggered implementation.
 Advantages of Quasi-Experiments
• Practical for real-world settings.
• Feasible when random assignment is not
possible.
• Can still provide evidence of causal
relationships.
 Limitations of Quasi-Experiments
• Lack of randomization reduces internal
validity.
• Groups may differ in important ways.
• Difficult to rule out confounding variables.
 Choosing Between True and Quasi-
experiments
• Availability of random assignment
• Ethical and practical considerations
• Control over confounding variables
Experimental Research Decision Steps
• 1. Define research problem and hypotheses.
• 2. Select participants and sampling method.
• 3. Choose appropriate experimental design.
• 4. Implement treatment or intervention.
• 5. Collect and analyze data.
• 6. Interpret results and draw conclusions.
 Role of Observation in Research
• Observational methods involve watching
behavior in natural or controlled settings.
• Used when manipulation is not possible or
ethical.
• Includes naturalistic, participant, and
structured observation.
 Example: Naturalistic Observation
• Observing children’s play behavior in a
preschool setting.
• No manipulation, behavior recorded in real
environment.
 Example: Participant Observation
• Researcher joins a community meeting to
study group dynamics.
• Both interaction and participation are
observed and recorded.
Advantages of Experimental Research
• High internal validity and control over
variables.
• Ability to infer causality.
• Precision and replicability.
 Limitations of Experimental Research
• Ethical constraints on manipulation.
• Artificial settings may reduce external validity.
• Requires careful planning and resources.
 Applications in Real Research
• Education: Comparing teaching methods.
• Health Sciences: Evaluating a new treatment.
• Business: Testing effects of training on
performance.
 Example: Single-Subject Design
• Study of a student with ADHD receiving
behavior therapy.
• Repeated measures before, during, and after
intervention.
 Example: Field Experiment
• Test of flyer design on donation rates in a
public space.
• Different flyer versions randomly distributed
across days.
• Donations recorded anonymously to measure
effectiveness.
 Summary and Conclusion
• Experimental research is essential for causal
inference.
• Design choice depends on research goals,
feasibility, and ethics.
• Validity considerations are key to rigorous
research.