Comprehensive Guide to Simple Experimental Design and Psychological Methodology

Theoretical Foundations of Research and Proverbs

• Proverbs are defined as traditional sayings that express a perceived truth based on common sense, experience, or observation. They are rarely intended to be literal; instead, they capture human behavior in meaningful ways and apply to a variety of circumstances.
• In the context of psychological research, proverbs are treated as theories. Experiments are designed to test their accuracy rather than assuming they are inherently true.
• Through research questions and hypotheses, the accuracy of the "perceived truth" found in a proverb is rigorously tested.

Limitations of Correlational Designs

• Correlational designs involve observing and measuring the relationship between a predictor variable and a criterion variable without manipulating any factors.
• A primary example involves testing the proverb "An apple a day keeps the doctor away."
  • Variables: Number of apples consumed and number of doctor visits.
  • Observed Relationship: A negative correlation. As the number of apples increases ($A \uparrow$), the number of doctor visits decreases ($D \downarrow$). This indicates a negative directionality.
• Two major limitations prevent correlational studies from establishing causation:
  • Directionality Problem: It is impossible to determine if variable $A$ affects variable $B$, or if variable $B$ affects variable $A$. For instance, observing a duck walking through a door simultaneously with the door opening does not reveal if the duck pushed the door or if the door opening prompted the duck to enter.
  • Third Variable Problem: An unmeasured third factor could be causing the relationship between the two variables. For example, a gust of wind could have blown the door open and pushed the duck inside simultaneously.

Simple Experimental Design and Causal Claims

• To draw causal conclusions, a true experiment is required. In a simple experimental design, one variable is manipulated while its effects on another variable are measured.
• Independent Variable (IV): The variable manipulated by the researcher. It is the hypothesized cause.
• Dependent Variable (DV): The variable being measured. It is called "dependent" because its value is expected to depend on the manipulation of the IV.
• Control: To make causal claims, the researcher must hold all other variables constant. If the only difference between groups is the manipulation of the IV, and a difference is observed in the DV, one can conclude the IV caused the change in the DV.
• Definition of "Simple": The word "simple" in this context refers to a design involving a single IV and a single DV.

Construct Selection and Operationalization

• Constructs: Abstract, psychological concepts (usually one to three words). These must be validated using the APA Dictionary of Psychology. Examples include "Nutrition," "Health," "Aggression," or "Personality."
• Operational Definitions: The specific, concrete method used to measure or manipulate a construct within a study. For example, the construct "Nutrition" can be operationally defined as the "number of apples eaten per day over one year."
• Variable Types:
  • Quantitative Variable: Involves numbers or amounts that can be counted (e.g., number of apples, height, weight).
  • Qualitative (Categorical) Variable: Involves types or categories (e.g., types of fruit, organic vs. inorganic apples).

Scales of Measurement

• Ratio Scale: A measurement scale that possesses a true, meaningful zero, representing the complete absence of the property being measured. Ratios between numbers are accurate.
  • Examples: Weight ($20\,kg$ is exactly twice as heavy as $10\,kg$), height, heart rate, and number of doctor visits.
• Interval Scale: A scale where the intervals between numbers are meaningful and equal, but there is no true zero point. Ratios are not meaningful.
  • Examples: Temperature in Fahrenheit or Centigrade (zero does not mean "no warmth"), and IQ scores (there is no "$0$ IQ," and a score of $160$ is not precisely "twice as smart" as an $80$).
  • Note on Temperature: Kelvin is a ratio scale because its zero point represents the absolute absence of molecular movement causing heat.

Hypothesis Articulation

• A well-articulated hypothesis must contain three specific components:
  1. Significance: The use of the word "significantly" to indicate that the expected differences are statistically meaningful (e.g., "significantly fewer monthly doctor visits").
  2. Directionality: A clear statement of which group is expected to have higher or lower scores on the DV (e.g., "Group A will have fewer visits than Group B").
  3. Rationale/Theory: A brief explanation connecting the prediction to the theoretical basis (the proverb).
• Example Hypothesis (If-Then Statement): "If apple consumption promotes health, then patients who eat two apples a day for one year will have significantly fewer monthly doctor visits than patients who eat one apple a day for one year."

Between-Subjects vs. Within-Subjects Designs

• Between-Subjects Design: Each participant is assigned to only one level of the independent variable.
  • Pros: No risk of carryover effects, order effects, or practice effects. Requires fewer materials per person.
  • Cons: Potential for individual differences between groups (e.g., one group might be naturally healthier or smarter). Lower statistical power, requiring more total participants.
• Within-Subjects Design: All participants experience every level of the independent variable.
  • Pros: Controls for individual differences because each person acts as their own control. Higher statistical power ($higher\,power \rightarrow fewer\,participants\,needed$). Cost-effective.
  • Cons: Risk of fatigue, practice effects, and carryover effects (e.g., health benefits from year one's apples carrying into the "no apple" year two). Requires counterbalancing and unique materials.
• Counterbalancing: Reversing the order of conditions for half of the participants to control for sequence or order effects (e.g., half get one apple then zero apples; the other half get zero apples then one apple).

Internal Validity and Control of Extraneous Variables

• Extraneous Variables: Every variable in a study that is not the IV or the DV (e.g., age, genetics, socioeconomic status, exercise habits).
• Confounding Variables: Extraneous variables that were not properly controlled and vary systematically with the IV, providing an alternative explanation for the results and ruining internal validity.
• Three Methods of Control:
  1. Keeping it Constant: Ensuring the variable is the same across all groups (e.g., only using Fuji apples or only recruiting college students). This can hurt external validity (generalizability).
  2. Varying Randomly: Using random assignment so that extraneous factors are distributed equally across groups on average.
  3. Counterbalancing: Specifically used in within-subjects designs to control for order effects.

Visualizing Experimental Results

• Line Graphs: Used when the independent variable is quantitative and continuous.
  • Example: Measuring the effect of $1\,apple$, $2\,apples$, and $3\,apples$ on health.
• Bar Graphs: Used when the independent variable is qualitative or categorical.
  • Example: Comparing "Organic Apples" vs. "Non-Organic Apples" vs. "GMO Apples."
• Graph Requirements: The IV goes on the x-axis, the DV goes on the y-axis. Graphs must include a descriptive title and correct axis labels. If using an interval scale with no true zero, the y-axis minimum should be adjusted accordingly (e.g., scale of $1$ to $10$).

Functional Relationships and the Third Level of the IV

• A simple experiment with only two levels cannot establish a functional relationship or trend (e.g., linear vs. curvilinear).
• Need for Three Levels: At least three levels of a quantitative IV are required to see if a relationship plateaus or reverses.
  • Example: While $2\,apples$ might be better than $1\,apple$, adding a third level ($4\,apples$) might show that the benefits eventually level off.
• Range Sensitivity: The range of the IV must be realistic. A range that is too narrow (e.g., $1\,minute$ vs. $2\,minutes$ of TV watching) might miss an effect, while a range that is too wide (e.g., $0\,hours$ vs. $24\,hours$ of TV) might produce unrealistic, exaggerated variance in the DV.

Scenario: Two Heads Are Better Than One

• Context: The Director of UCLA Office of Waste Management wants to test if groups brainstorm more creative ideas than individuals.
• Variables:
  • Independent Variable: Number of people brainstorming together (levels: one person/individual vs. group/two people).
  • Dependent Variable: Number of creative ideas generated.
• Constructs:
  • IV Construct: Collaboration (or Social Interaction).
  • DV Construct: Creativity (or Productivity).