Quasi-Experimental Designs and Longitudinal & Cross-Sectional Designs
Exam Review: Quasi-Experimental Designs and Longitudinal & Cross-Sectional Designs
Nonexperimental vs Experimental Designs
Nonexperimental Designs:
Characteristics:
No random assignment
No manipulation of variables
Relatively lower levels of control
Examples include:
Quasi-experimental designs
Correlational studies
Survey designs
Experimental Designs:
Characteristics:
Random assignment of participants
Manipulation of variables
High control
Example: Simple experiments
Quasi-Experiments
Definition:
A study where the researcher selects participants for different conditions from preexisting or self-selected groups, lacking random assignment but possibly involving manipulation of an independent variable (IV).
Example: Comparing test scores from two different high schools (IV = school).
True vs Quasi Experiments
Strengths and Weaknesses of True Experiments:
Strengths:
High internal validity
Complete control by experimenter
Weaknesses:
Relatively lower external validity
Strengths and Weaknesses of Quasi Experiments:
Strengths:
Greater external validity
More ethical when random assignment is not feasible
Weaknesses:
Lack of equivalence between groups
Confounding of manipulated IV with group membership
Non-Equivalent Control Group Design
Description:
Quasi-experimental designs without random assignment employing nonequivalent control groups.
Comparisons must address the initial equivalence of groups.
Interpretability:
Dependence on whether results can be explained by group differences or other factors.
Potential Threats:
Ceiling Effects: Performance already near maximum limits further improvement.
Floor Effects: Very low performance levels make differences hard to detect.
Enhancing Interpretability
Procedures:
Matching: Pair participants in experimental and control groups based on characteristics relevant to the outcome.
Identifying Moderator Variables: Include extraneous variables that may influence outcomes as moderator variables in the study.
Example: Motivations or prior training.
Pretesting: Measure participants on outcome variables prior to treatment to document nonequivalence.
Statistical Control: Use statistical methods (e.g., ANCOVA) to control for preexisting differences based on pretest data.
Delayed Control Group Design
Definition:
A design where comparison groups are tested sequentially with time intervals; the experimental group and control group are not tested at the same time.
Example: Data collected on moral beliefs six weeks before and one month after the Al-Qaeda bombing with a total gap of ten weeks between measurements.
Mixed Factorial Designs
Definition:
Designs that include one within-subject variable (manipulated or measured over time) and one between-subject variable (preexisting factors like gender).
Example: Examining a trait (between-subject) with state anxiety (within-subject) and its impact on test performance.
Time Series Designs
Interrupted Time-Series Design:
Comparison of the same group over time by analyzing data trends before and after a treatment, involving multiple data points collected.
Multiple Time-Series Design:
Inclusion of both control and experimental groups to mitigate the history effect as a rival hypothesis; both groups measured repeatedly over time.
Example: Assessing life satisfaction for married individuals multiple times before and after marriage, also distinguishing stays married vs divorced groups as control.
Repeated Treatment Design
Definition:
Design without control groups where the same subjects are measured before and after receiving repeated treatments.
Example: Testing the effect of calming music on time taken to fall asleep, comparing how long it takes to fall asleep with and without music.
Longitudinal vs Cross-Sectional Designs
Longitudinal Design:
A study where the same cohort is selected and measured repeatedly over time; it allows for the establishment of temporal sequences between IV and DV.
Cross-Sectional Design:
Testing different cohorts at a single point in time, eliminating issues like practice effects, but introducing the cohort effect.
Within vs Between Subjects Designs
Within Subject Designs:
The same participants are exposed to all levels of the IV, reducing variability and requiring fewer participants.
Allows for documented participant equivalence, but risks testing/practice effects.
Between Subject Designs:
Different participants for each condition with each participant experiencing only one level of the IV.
Requires more participants and ensures no testing effects or order effects.
Selective Survival and Dropout
Selective Survival:
Occurs when some individuals in a population are no longer part of the sample due to various reasons, threatening internal validity especially in older adult studies.
Selective Dropout:
When participants leave the study for reasons such as relocation, loss of interest, or death, affecting the representativeness of the remaining sample.
Practice/Retesting Effects
Definition:
Occurs when participants improve on repeated measures due to familiarity with the test instead of actual improvement in the measured behavior.
Example: The Berkeley Growth Study, focusing on intelligence over a long period; performance inflation on IQ tests due to repeated exposure.
Other Effects Influencing Study Outcomes
Cohort Effects:
Differences arising from historical or social contexts impacting one generation, affecting the generalizability of findings across different cohorts.
Cohorts:
Groups identified by shared characteristics, often age-related.
Generation Variability:
Shaped by unique cultural and social trends, which confound the interaction between age and ability or behavior, necessitating consideration in research analysis.