Zoo212 - The Scientific Method L10 - Planning Experiments
The Scientific Method Overview
Date of Lecture: April 21, 2021
Course: Zoo212
Instructor: Professor Niall Vine
Planning Experiments
Objective: Understand the planning required to conduct experiments in scientific research.
Key Topics Covered:
Types of Experiments
Experimental Approaches
Principles of Designing an Experiment
Identifying Study Types
Data Sheet Design
Pre-Experiment Preparations
Sampling Considerations
What is a Scientific Experiment?
Definition: An organized and detailed series of steps to validate or reject a hypothesis.
Types of Experiments
Experimental: Direct manipulation and control of variables.
Quasi-experimental: Examines effects without random assignment.
Observational: Studies behaviors or conditions without interference.
Experimental Approaches
Laboratory Experiments:
Easily controlled settings.
Limited inference about real-world applications.
Natural Experiments:
Occur naturally without manipulation.
Weak conclusions due to lack of replication.
Field Experiments:
Manipulative treatments applied in natural environments.
Field Studies:
No random assignments; important but have limitations.
Deciding on Study Type
Experiment Study:
Mostly laboratory-based.
Correlation Study:
Often observation-based; field studies are a common example.
Principles of Designing an Experiment Study
Manipulation of Variables: Investigator artificially manipulates a variable within different groups to test effects.
Goal: Seek explanation rather than mere description.
Steps in Designing an Experiment
Identify the Groups:
Start with a research hypothesis.
Based on the null hypothesis, determine required groups (Control vs Experimental).
Control Group: Not exposed to experimental treatment.
Establish the Variables:
Clearly define independent and dependent variables.
Independent Variable: What you change (the cause).
Dependent Variable: What changes as a result (the effect).
Controlled Variables: Kept constant across all groups.
Determine the Number of Replicates:
Minimum of three replicates for each group recommended.
More replicates enhance reliability; seek stable means and confidence intervals.
Typically aim for n ≈ 30 to avoid pseudoreplication.
Randomise Subjects Across All Groups:
Ensure random assignment of subjects, not random selection.
Tools: Random number generators, coins, dice, Excel.
Principles of Designing a Correlation Study
Focus: A correlation study seeks the relationship between two or more variables.
Goal: Description rather than causation/itemization.
Key Steps:
Establish variables for comparison.
Determine the required number of replicates.
Randomly allocate measuring equipment.
Next Steps After Designing the Experiment
Consider how to properly record data to ensure accurate results.