3.1
Lesson Overview
Welcome to lesson 3.1 on variables and research design.
Key Topics Covered:
Basics of variables and their classifications
Distinction between independent, dependent, and extraneous variables
Understanding confounds and their effect on internal validity
Strategies for controlling confounds
Overview of different research designs: true experimental, quasi-experimental, and non-experimental
Discussion of internal and external validity in research designs
Understanding Variables
Definition of a Variable:
Anything that can vary, e.g., age, wellbeing, experimental group membership.
Importance of Variables in Research:
Variables describe sample characteristics and facilitate understanding of measured changes.
Symbols Used in Research:
Independent Variable (IV): Usually represented as X, the variable that predicts change.
Dependent Variable (DV): Usually represented as Y, the outcome being measured.
Types of Variables:
Independent Variables (X): Predictors or factors causing changes.
Multiple independent variables can be analyzed.
Dependent Variables (Y): Outcomes believed to be influenced by IVs.
Analyses in this course focus on one dependent variable, though multiple DVs are possible.
Graphical Representation
X-axis: Represents the Independent Variable (horizontal axis).
Y-axis: Represents the Dependent Variable (vertical axis).
Example:
IV: Age, affecting scores on a questionnaire.
DV: General health questionnaire results.
Extraneous Variables
Definition: Variables that are not the focus of the research but can influence results.
Importance of Control:
Research design can help control these variables, either through design or statistical inclusion.
Failure to account for significant extraneous variables may undermine conclusions.
Types of Extraneous Variables:
Covariate: A variable that only affects the dependent variable and can be controlled statistically.
Confound: A variable that varies with the independent variable and affects the dependent variable, potentially providing alternative explanations for results.
Study Design Considerations
Common Issues with Variables:
Misinterpretation of causal relationships, reverse associations between IV and DV.
Importance of ensuring covariates are not associated with predictors to maintain clarity in results.
Research Designs
Types of Research Designs
True Experimental Design:
Most rigorous type of design, allowing causal inferences.
Random assignment to conditions controls extraneous variables.
Homogeneous sample of participants.
Quasi-Experimental Design:
Similar to experimental but lacks random assignment.
Capable of drawing causal conclusions only if designed thoroughly.
Often used when randomization is unethical or impractical (e.g., cannot assign smokers versus non-smokers).
Non-Experimental Design:
Observational, measuring variables without manipulation.
Lower ability to assert causation; can only suggest relationships.
Internal vs. External Validity
Internal Validity: Confidence in concluding that changes in the DV are due to manipulation of the IV.
External Validity: Generalisability of findings to other settings or populations.
Forms of validity to consider:
Measurement Validity (Construct Validity)
Internal Validity
External Validity
Threats to Internal Validity
Examples of threats include:
Selection Bias
History Effects
Maturation Effects
Testing Effects
Instrumentation Effects
Mortality (attrition)
Environmental and Contextual Factors
Generalisability can be impacted by the overall context of the study and participant demographics.
Practical and ethical considerations in research affect which designs are chosen.
Experimental Designs
Between Subjects Design: Different participants in each group.
Within Subjects Design: Participants undergo all conditions.
Mixed Design: Combines aspects of both designs.
Statistical Tests
To analyze data from the study:
T-tests for comparing two groups.
ANOVAs for three or more groups.
ANCOVAs when including covariates in the analysis.
Need to understand the level of measurement and structure of data for analysis.
Example Context - Shared Reading and Language Development
Types of Designs to Investigate Impact:
Non-Experimental: Survey on reading practices and language abilities over time.
Quasi-Experimental: Using existing conditions (e.g., childcare centers with varying reading practices).
True Experimental: Randomly assigning groups to different exposure levels of reading time.
Mixed Design: Investigating the pre-test and post-test impact of shared reading.
General Recommendations
Designs should balance between obtaining robust internal validity while allowing for generalization of relevant external validity.
Researchers must manage trade-offs effectively to ensure quality study designs that can yield actionable insights without violating ethical guidelines or practical limitations.