Experimental Research Overview

Experimental Research Overview

• Experimental research is a carefully controlled procedure, manipulating variables to determine cause and effect.
• Key features: independent variable (manipulated) and dependent variable (measured); experimental and control groups.
• Correlation is not causation, only experiments can determine cause and effect

What is an Experiment?

• Carefully controlled scientific procedure that manipulates variables to determine cause and effect.
• Key features include independent (manipulated) and dependent variables (measured).
• Experimental group receives treatment, control group receives no treatment.

Manipulation of the IV

• To manipulate an independent variable (IV) means systematically changing its level to expose different participant groups to different levels of the variable.
• Examples include whether a child watches violent or non-violent television.
• Designs include single factor two-level and multi-level designs.

Two Features of an Experiment

• Manipulation of the independent variable: this variable is the "cause" of the effect (dependent variable)
• Control of extraneous variables: keeping other variables constant (e.g. random assignment of participants).

Treatment and Control Conditions

• Treatment: intervention intended to change behavior for the better (e.g., psychotherapy).
• Treatment group receives treatment.
• Control group does not receive treatment.
• Randomized controlled trials experiment the effectiveness of psychotherapies and medical treatments.

Types of Control Conditions

• No-treatment control condition (receives nothing)
• Placebo control group (receives a placebo, checks for placebo effect).
• Wait-list control condition (treatment delayed, checks for "expectation of improving").

Types of Designs (Between-Subjects)

• Each participant tested only once.
• Relies on random assignment
• Block randomization, or matching participants on the dependent variable or extraneous variable

Types of Designs (Within-Subjects)

• Each participant tested in all conditions.
• Order Effects: carryover effects, practice effects, fatigue effects, context effects.
• Counterbalancing: complete, or random counterbalancing

Order Effects

• Carryover effect: effect of testing condition on participant's behavior in later conditions.
• Practice effect: improved performance in later conditions due to practice.
• Fatigue effect: worsened performance in later conditions due to tiredness or boredom.
• Context effect: testing condition changes how participant perceives stimuli or interprets task.

Counterbalancing

• Counterbalance testing order for different participants, to control for order effects.
• Methods include complete and random counterbalancing

Internal Validity

• A valid experiment supports the conclusion that the IV caused the observed DV differences.
• Strong support through IV manipulation and control of extraneous variables (e.g., random assignment).

External Validity

• External validity supports generalizability to people and situations beyond the study.
• Two types: mundane and psychological realism.

Construct Validity

• Construct validity is when the research question is clearly operationalized.
• Operationalization is the specification of exactly how the research question will be studied.

Statistical Validity

• Statistical validity concerns the proper statistical treatment of data and the researchers' statistical conclusions.

Prioritizing Validities

• High internal and construct validity but may compromise external validity in some studies is common.

Last Class – Face Validity

• Face validity is an informal assessment of how accurate a measure looks on the surface.
• Can be assessed quantitatively, but generally evaluated informally.

Last Class – Criterion Validity

• Criterion validity measures correlation between a measure and other variables (criteria) it should correlate with.
• Types include concurrent (same time) and predictive (future) validity, as well as convergent (correlated with existing, established measures of similar concepts) validity.

Recruiting Participants

• Subject pools are established groups of people agreeing to participate in research.
• Volunteers show differences in demographics/characteristics.

Standardizing the Procedure

• Experimenter expectancy effect: researchers’ expectations affect participants’ behavior.
• Double-blind studies prevent both researcher and participant from knowing condition assignments.
• Procedures should be standardized to replicate.

Manipulation Checks & Pilot Testing

• Manipulation check: separate measure confirming IV success.
• Pilot test: small-scale study of procedure and/or stimuli/materials.