Ch. 10 Research Designs for Special Circumstances PDF

Ch.10 Research designs for special circumstances: What is a true experiment? A basic experimental design that has high internal validity and the IV is the only cause of the DV. Casual relationship: 1. Temporal precedence 2. Covariation of cause and effect 3. Eliminate other explanations...

Ch.10 Research designs for special circumstances: What is a true experiment? A basic experimental design that has high internal validity and the IV is the only cause of the DV. Casual relationship: 1. Temporal precedence 2. Covariation of cause and effect 3. Eliminate other explanations What is the purpose f a basic experimental design for special circumstances? Not all research topics can follow the basic design. Therefore solutins must be adapted to the design. Some examples are: A. Single case design : one participant appropriate for design B. Quasi-experimental desing: random assignment not possible as what is being measured is not interchangeable (such as race or age) ** as designs move away from qualities of a basic experimental design causality cannot be claimed Program Evaluation: What is the purpose? True experiments face many practical difficulties that prevent researchers from using them. What is program evaluation? Research on programs that target to achieve some positive effect on a group of people. Why use this? Although true experiment may not be possible an experiment needs to be done.  Context applied to: Schools Work settings Communities Ex: program to target reducing obesity in humans at school. The program educates children on healthy eating to avoid obesity. (Positive effect on a group of people) The research on how to reduce obesity is Program Evaluation. Research on program evaluation: Five broad questions to guide. 1. Needs assessment 2. Program theory assessment 3. Process evaluation 4. Outcome evaluation 5. Efficiency assessment Single case experimental designs: What is it? A design with only one participant. Ex: therapy effectiveness Features: No random assignment How? Baseline phase = no change to environment (the problem) Treatment phase = apply a change to see its impact on DV ** if redo baseline phase = reversal design 1. Participant’s behavior is first measured during a baseline control period (no manipulation or intervention occurs during this time). 2. Manipulation is applied during the treatment period. 3. Participant’s behavior is obeserved continuously. How to indicate manipulation effectiveness? Change in participant’s behavior from baseline to treatment periods offers evidence for effectiveness of manipulation. Solutions for being more confident for causality: Reversal designs: Redo baseline phase to see if it has reliability. Multiple baseline designs: Replication in single case designs: Developmental research designs: What is it? Experimenting age with other variables. Age effects = main variable (how this impact other variables) Ex. How habits change as we get older. Confounding effects: interfere with age effect a. Cohort effects: 1. Unique experiences cause differences 2. Circumstances to a particular generation b. Time of measurement effects 1. the time of data collection due to sociocultural, environmental, historical or other events ** effects are often confounded Why use this? Unable to do random assignment as cannot change age. What are the features? No random assignment. There are three main designs. 1. Cross-sectional method Using different participants at various ages to measure an effect. Problem with design:  Cohort effects benefit with design: Less expensive Yield result immediately 2. Longitudinal method Use same participants in their aging to measure an effect. Problem with design: Loss of participants (death/drop out) Expensive Confounds: time measurement, results may not generalize, measures may change Benefit with design: Evidence of developmental change greater Differences between Cross and Longitudinal? 3. Sequential method design What? Purpose? Can control cohorts and time measurement confounds. How? Use multiple cohorts. (Different groups of participants) Use different times of measurement. Quasi-experimental design: What is it? Type of design used to study IVs that cannot use true experimental design. (Cannot manipulate the IV) Ex: impacts of drugs on babies, concussions impact on cognition Disadvantages- Features of this design: Unable to make casual inferences (only correlational) Used when cannot make random assignments (ethically or practically unfeasible) Unable to control conditions Advantages: Quasi experimental designs fall short of many qualities of a true experimental design yet they help use identify flaws in a study that claims to be a true experimental design. How quasi-experimental designs help? Allow to distinguish if: 1. Quasi is used intentionally and out of necessity 2. A true experimental design has flaw that make it a Quasi in nature Designs of a quasi-experimental design: (6) There are many designs within a Quasi but will cover 6 major ones. 1. One-group post test only (poor design) What? Expose group to one IV condition and measure the DV. Problem: Effect of condition unknown on DV. No comparison. 2. One-group pretest-posttest (slight improvement) What? First receive a piece of comparison, then expose group to one IV condition and measure the DV.  Compare DV from initial condition to new condition. Problem: Threat to internal validity. Other things may have caused a change in IV. Threats to Internal validity: How to overcome threats to internal validity? 1. Maturation: change in pretest and posttest is due to something other than the treatment. (Ex; changed on their own)  2. History: concurrent events that influenced the IV. (Other events) 3. Testing: process of testing and order effects change DV. (Ex. Taking a pretest) 4. Instrument decay: over repeated measures or treatments may change effect on DV. 5. Regression toward the mean: tendency to move towards the middle. (High to mean or low to mean) Continued Quasi-experimental designs: 3. Non-equivalent control group posttest only design: Use of two different groups. Apply treatment to one group. Compare their posttest results of DV. 4. Non-equivalent control group pretest-posttest Use of two different groups. Record DV before treatment. Apply treatment. Record DV after treatment. Compare DVs. Multiple repeated measures: **no random assignment so unable to ensure IV causing DV. Interrupted time series: What? A treatment within a time frame. It interrupts the DV. Purpose? When no random assignment available. Apply a treatment and analyze the effect through a time frame. Control series: What? A time frame of controlling a variable. Purpose? When no random assignment available. Control the variable to bring experiment closer to a true experiment. How? Add a control group. Interrupted time series design: Control series design: Practice: True experiment: need random assignment (could be repeated measures) Assign to complex and simple controls in the airplane. In a flight simulator. Quasi-experimental: (comparison but recruit both conditions) Recruit Complex airplanes – measure # of crashes Recruit Simple airplanes- measure # of crashes Control series: (comparison but recruit one condition) Time series: change condition of one airplane. Collect rates of crash. Compare with rate of crash of the airplane before treatment. Which one is best? Out of all