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:

    1. Covariate: A variable that only affects the dependent variable and can be controlled statistically.

    2. 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.