EXPERIMENTAL RESEARCH DESIGNS EXPERIMENTAL DESIGN Advantages Best establishes
EXPERIMENTAL RESEARCH DESIGNS
EXPERIMENTAL DESIGN Advantages • Best establishes cause-and-effect relationships Disadvantages • Artificiality of experiments • Feasibility • Ethical considerations
CAUSALITY q Temporal precedence q Co-variation between IV and DV q Eliminate alternative explanations
TYPES OF EXPERIMENTAL DESIGNS v True Experimental Research Designs v Quasi-Experimental Research Designs
TRUE EXPERIMENTAL RESEARCH DESIGN
TYPES OF TRUE EXPERIMENTAL DESIGNS q Simple True Experimental Research Design q Complex True Experimental Research Design (also referred to as factorial designs)
CHARACTERISTICS OF TRUE DESIGNS Ø Manipulation (treatment) Ø Randomization Ø Control Characteristics of simple true designs Ø One IV with 2 or more levels Ø One DV
EXAMPLES In a study of the effectiveness of a new drug in treating depression, some participant received the drug whereas others received placebo. A third group is not treated at all. After the program is completed the participants level of depression is assessed using beck depression inventory.
In a study of cell phone use and driving some participants try to perform as accurately as they can in a driving simulator (i. e keep the car on a narrow road) while talking on a hand held cell phone, others while talking on a hands free, yet others without talking on a phone at all. Half the subjects have 2 years of driving experience and the remaining subjects have four years of driving experience.
WHERE DO WE CONDUCT EXPERIMENTS v Laboratory/ Controlled Experiments (Bandura’s “Bobo doll Experiment”, Loftus experiments on “Leading questions and eye-witness report” v Field Experiments (Experiment by Darley and Batson on the “ Good Samartian Parable”)
VARIATIONS OF TRUE EXPERIMENTS q Independent Measure Design (between groups design) q Repeated Measure Design (within groups design)
INDEPENDENT GROUPS DESIGN • Also called between subject design • Each participant is in one and only one of the treatment conditions. • Different group of participants are in each treatment condition. • typically used to study differences
CHARACTERISTICS • Each group represents a different condition • Balancing technique is used to control individual differences and to enhance internal validity. • Researcher makes sure that one group is not smarter, more motivated or intelligent than other • By balancing subject characteristics across group, alternative explanations are ruled out.
TYPES OF INDEPENDENT GROUP DESIGN § Random Group Design § Matched Group Designs
RANDOM GROUP DESIGN Random assignment of subjects to different treatment conditions. (random assignment vs random selection) Block Randomization: Subjects are assigned to conditions one block at a time. CAEBD ECDAB DBEAC BACED ACEBD BCADE
MATCHED GROUP DESIGNS Used to create comparable groups when • There are too few subjects available • Matching task is available • Heterogeneous population Ø Participants are matched on some trait and then distributed randomly to different groups. Ø Researcher forms comparable groups by matching subjects on a pre-test task.
REPEATED MEASURES DESIGN • Also called within subject designs • Each participant is in all of the treatment conditions so there is no confound due to individual differences. Typically used when • few participants are available • to study changes in individual subjects. • To increase experiments sensitivity • Study not frequently occurring phenomenon
CHARACTERISTICS Ø Each participant experiences all the conditions in experiment. Ø Subjects serve as their own control because they participate in both experimental and control conditions.
Advantages: • Fewer subjects needed (less costly) • Sensitive to finding statistical differences Disadvantages: • Order effect
ORDER EFFECT Once a participant has complete the first part of the study the experience or altered circumstances could improve performance in later parts of the study. There are 3 forms of order effect: Ø Practice effect Ø Fatigue Ø Carry-over effect
TYPES OF RMD q Complete Design q Incomplete Design
COMPLETE DESIGN In complete design practice effects are balanced for each participant, which is accomplished by administering the conditions to each participant several times using different order each time.
BALANCING PRACTICE EFFECT IN COMPLETE DESIGN Block Randomization: the number of blocks is equal to the number of times each condition is administered and the size of each block is equal to the number of conditions in the experiment. Reverse counterbalancing: also called ABBA design presenting the conditions in one sequence than in reverse for example ABCD DCBA Ineffective in case of anticipation effect (participants anticipate which condition should occur in the next sequence)
INCOMPLETE DESIGN Ø Each condition is administered to the participant only once. The order of administering the condition is varied across participants rather than for each participant. Ø Rule for balancing practice effect is that each condition of the experiment must be presented in each ordinal position equally often.
BALANCING PRACTICE EFFECT IN INCOMPLETE DESIGN Ø All possible orders (complete counter balancing) Ø Partial Counterbalancing
§ Each participant experiences all treatment conditions but in different order. All possible orders are used equal number of times. § The trick in counterbalancing is to make sure that each condition appears in each position an equal number of times. § One way of counterbalancing the conditions would be to use the following three orders of presentation: Order 1 A B C Order 2 B C A Order 3 C A B Notice that all three conditions are performed once first, once second, and once third.
Partial Counterbalancing: using subset of all possible sequence. Latin Square: Each condition appears at each ordinal position only once. And each condition precedes and follows the other only once. For example: Exp having four conditions (ABCD)
LATIN SQUARES Row 1 Row 2 Row 3 Row 4 1 2 3 4 A (60) B (0) C (180) D (120) A (60) B (0) C (180) D (120) A (60) B (0)
TYPES OF TRUE EXPERIMENTAL RESEARCH DESIGNS
q Randomized posttest-only control group design. q Randomized pretest-posttest control group design q Randomized Solomon four group design q Randomized posttest-only control group design using matched subjects. q Randomized pretest-posttest control group design using matched subjects.
RANDOMIZED POSTTEST-ONLY CONTROL GROUP DESIGN. Treatment Group Control Group R X O (Random assignment of 50 employees to experimental group) Treatment (motivational workshops) Posttest (employee morale questionnaire) R C O (Random assignment of 50 employees to control goup) No Treatment Posttest (employee morale questionnaire)
RANDOMIZED PRETEST-POSTTEST CONTROL GROUP DESIGN R Treatment Group Control Group O (Random Pretest assignment of (employee 50 employees morale to questionnaire) experimental group) X O Treatment (motivational workshops) Posttest (employee morale questionnaire) R O C O (Random assignment of 50 employees to control goup) Pretest (employee morale questionnaire) No treatment. Posttest (employee morale questionnaire)
RANDOMIZED SOLOMON FOUR GROUP DESIGN
Treatment Group Comparison Group R O X O (Random assignment of 25 employees to experimental group) Group I Pretest (employee morale questionnaire) Treatment (motivational workshops) Posttest (employee morale questionnaire) R O C O (Random Pretest Workshop that do Posttest assignment of 25 (employee morale include (employee morale employees to questionnaire) motivational questionnaire) control group) training. Group II R X O (Random assignment of 25 employees to experimental group) Group III Treatment (motivational workshops) Posttest (employee morale questionnaire) R C O (Random assignment of 25 employees to control group) Workshop that do Posttest include (employee morale motivational questionnaire) training.
RANDOMIZED POSTTEST-ONLY CONTROL GROUP DESIGN USING MATCHED SUBJECTS. Treatment Group M X O Control / Comparison Group M C O
RANDOMIZED PRETEST-POSTTEST CONTROL GROUP DESIGN USING MATCHED SUBJECTS. Treatment Group M O X O Control / Comparison Group M O C O
BIASES IN EXPERIMENTAL SITUATION Experimenter Bias Experimenter Effect: Ø Biases in the experiment due to expectation of the experimenter. Ø Also referred to as Rosenthal effect or Pygmalion effect.
HOW TO CONTROL FOR EXPERIMENTER BIASES Double-Blind Procedure: both the experimenter and subjects are blind to which treatment is being administered.
BIASES IN EXPERIMENTAL SITUATION Participant Bias Ø Hawthorne Effect Ø Demand Characteristics Ø Evaluation Apprehension
HOW TO CONTROL FOR PARTICIPANT BIASES Ø Placebo control group Ø Expectancy control groups: half of the participants are informed about the expected outcome half are not informed.
- Slides: 41