In-Depth Notes on Experimental Design and Observational Studies
Experimental Design
Observational Studies
Researchers do not assign treatments or choices; instead, they observe them.
Retrospective Observational Study
Subjects selected based on past conditions or behaviors.
Focus on estimating differences between groups or associations between variables.
Drawbacks include lack of random sampling.
Prospective Observational Study
Subjects are followed to observe future outcomes without deliberate treatment.
Focus on estimating differences as groups are followed throughout the study.
Experiments
Purpose: Study the relationship between two or more variables by imposing treatments on subjects.
Key Elements in Experimental Design
Factors
Explanatory variables involved in the experiment.
Levels
Various treatments tested within each factor.
Treatment
Each combination of factors and levels applied to experimental units.
The Best Experiments
Characteristics include:
Randomized
Comparative
Double-blind
Placebo-controlled
Four Principles of Experimental Design
Control
Helps reduce extraneous variation by making conditions similar across treatment groups.
Randomization
Balances uncontrolled variability and reduces bias by equalizing effects spread.
Replication
Involves repeating treatment applications to multiple subjects for reliability.
Blocking
Groups subjects by a similar characteristic to reduce variability across treatment effects.
Statistical Significance and Confounding
Statistically Significant Results
Occur when outcome differences cannot be attributed to chance, thus linked to treatments.
Confounding Variables
When effects of two variables on a response cannot be distinguished.
Matched Pairs Design and Blinding
Matched Pairs Design
Compares subjects that are similar on untested attributes, ensuring each subject experiences all treatments sequentially.
Placebo
A control treatment that mimics the experimental treatment, enhancing the validity of the results through blinding.
Experimental Write-up Components
Plan: Identify factors and levels.
Response: Define the response variable being measured.
Treatments: Outline the treatments used.
Experimental Design: Explain selection processes for treatments (Control, Randomization, Blocking).
Analysis: Measure and compare data collected to draw conclusions.
Repeat and Replicate: Ensure validity through repeated trials.
Applying Experimental Design Concepts
Examples:
Tsunami Construction Materials
Considerations for building materials that withstand natural disasters while controlling for variables.
OptiGro Plant Fertilizer
Experiment to test a claim using different treatment groups with controlled application.
Responding to Questions:
Identify if a scenario is an observation or an experiment, determine subjects and response variables, check for hidden variables, and suggest improvements in methodologies.
Case Studies
Example of Exercise Testing Effectiveness
Test on insomnia treatment comparing diet restrictions and exercise among volunteers to assess effectiveness.
Reflecting on Results
Assessing how external variables may influence outcome measures in various studies (e.g., blood pressure, omega-3 fatty acids, etc.).