Exhaustive Study Notes on Correlational and Descriptive Research Methods

Foundations of Research Methods: Experimental vs. Correlational Designs

• Experimental Research Foundations:

  • In experimental research, researchers always manipulate the independent variable.
  • The independent variable is presumed to be the direct cause of changes in the dependent variable.
  • Researchers observe and measure changes in the dependent variable following the manipulation of the independent variable.

• Real-World Limitations of Experimental Manipulation:

  • In real-world settings, many variables of interest cannot be ethically, practically, or physically manipulated.
  • Because direct manipulation is not always possible, alternative non-experimental research methods are utilized in psychological research.

• Correlational Research Methods:

  • Correlational studies serve as a primary alternative to experimental designs.
  • Correlational studies address two distinct types of scientific questions:
    • Examining the degree of relationship or association between two measured variables.
    • Comparing preexisting groups formed on the basis of existing traits.
  • Strict Rule of Correlational Studies: In no correlational study is there ever a manipulation of a variable. Variable manipulation occurs exclusively in experimental designs.
  • Most correlational studies collect data by utilizing surveys.

• Two Approaches to Correlational Design:

  1. Measuring Two Continuous Variables:
     • Researchers measure two distinct characteristics in participants to evaluate whether a relationship exists between them.
     • Example: Examining whether a person's age is related to their preference for chocolate ice cream.
     • Measurement Parameters: Preference for chocolate ice cream might be rated on a continuous scale from $1$ to $100$. Participant age might span continuous values from $1\text{ day}$ to $99\text{ years}$ (or $100\text{ years}$).
     • Because neither variable is forced into artificial categories, both variables are measured quantitatively across a continuous continuum to calculate a correlation.
  2. Comparing Preexisting Groups:
     • This approach examines individuals possessing different preexisting traits or self-selected characteristics to determine if they differ on a secondary outcome variable.
     • Preexisting traits represent characteristics participants carry with them that cannot be randomly assigned by a researcher.
     • Example: Participants categorize themselves via a categorical survey question (e.g., answering "Yes" or "No" to the prompt: "Do you like vanilla ice cream?").
     • Researchers then compare the group of individuals who like vanilla ice cream against the group who do not, observing whether the two groups differ systematically on another measured trait.

Correlational Research Mechanics and Pearson's r

• Statistical Differentiation Across Correlational Designs:

  • When assessing the statistical relationship between two continuous measured variables, the primary statistical test employed is Pearson's $r$ correlation coefficient.
  • When comparing two preexisting groups (e.g., categorical group classifications), Pearson's $r$ correlation coefficient is not used.
  • Instead, group comparison involves evaluating the score distribution of the first group against the score distribution of the second group to ascertain if one group scores significantly higher.
  • Statistical tests used for comparing groups are mathematically distinct and more complex than Pearson's $r$.

• Mathematical Properties of Pearson's $r$:

  • Pearson's $r$ is a single numerical statistic created by mathematicians to summarize the linear relationship between two variables.
  • Bounded Range: Pearson's $r$ is strictly bounded mathematically between $-1$ and $+1$:
    • The minimum possible value is $-1$ ($-1.00$).
    • The maximum possible value is $+1$ ($+1.00$).
    • A value of $0$ ($0.00$) indicates the complete absence of a linear relationship between the two measured variables.
    • Every calculated Pearson's $r$ value must fall within the inclusive numerical interval $[-1.00, +1.00]$.
  • Example Value: A textbook reporting a Pearson's $r$ of $-0.15$.

• Deconstructing Pearson's $r$:

  • Fully understanding a Pearson's $r$ value requires evaluating two distinct, independent components:
    1. The sign (positive or negative dash in front of the number), which indicates direction.
    2. The absolute numerical value (magnitude), which indicates strength.

Direction of Correlation: Positive vs. Negative Relationships

• Determining Direction via the Sign:

  • To determine the direction of a relationship, pay attention exclusively to the sign in front of the Pearson's $r$ statistic ($+$, $-$) and ignore the numerical value entirely.

• Positive Correlations:

  • Indication: Represented by a plus sign ($+$) or the absence of a sign preceding the numerical statistic.
  • Definition: A positive relationship signifies that higher amounts of the first variable are associated with higher amounts of the second variable.
  • Symmetrical Definition: Equivalently, a positive correlation also signifies that lower amounts of the first variable are associated with lower amounts of the second variable.
  • Core Rule: In a positive correlation, both quantitative descriptors move in the same direction ("more with more", "greater with greater", "less with less", or "lower with lower").
  • Concrete Example: Examining the relationship between human height and human weight.
    • Study Design: A sample of $1000\text{ individuals}$ is recruited, ranging from infants $1\text{ day}$ old to elderly adults $99\text{ years}$ old. Height and weight are measured for every individual to calculate Pearson's $r$.
    • Interpretation: A positive Pearson's $r$ indicates that greater height in individuals is systematically associated with greater weight in those individuals. Rephrased, lower height is systematically associated with lower weight.
  • Population-Level Limitation Caveat: Correlations describe traits across an entire studied population or group, not a single specific individual. A correlation does not allow one to assert that a specific individual is tall because they are heavy, or short because they are light.

• Negative Correlations:

  • Indication: Represented explicitly by a negative sign ($-$) preceding the numerical statistic.
  • Definition: A negative relationship signifies that the quantitative descriptors of the two variables move in opposite directions (an inverse relationship).
  • Core Rule: Higher amounts of one variable are associated with lower amounts of the second variable ("more with less", "higher with lower", or "lower with higher").
  • Concrete Example: Examining the relationship between age and overall physical health.
    • Study Design: Participants complete standardized questionnaires, and the resulting data yields a Pearson's $r$ statistic with a negative sign.
    • Interpretation: A negative correlation indicates that higher age is associated with lower health (or greater health problems), and lower age is associated with higher health (or fewer health problems).
    • The two descriptive quantitative terms mismatch intentionally to reflect inverse movement.

Magnitude and Strength of Correlation

• Determining Strength via the Numerical Value:

  • To evaluate the strength (magnitude) of a relationship, pay attention exclusively to the absolute numerical value, ignoring the positive or negative sign entirely.
  • Example: For a Pearson's $r = -0.15$, evaluate only the value $0.15$ to assess strength.

• Quantitative Thresholds for Correlation Strength:

  • While real-world scientific thresholds can be flexible, strict definitive cutoffs are established for evaluation as follows:
    • Weak (Small) Relationship: Any Pearson's $r$ absolute value strictly smaller than $0.25$ ($|r| < 0.25$).
      • Example: A calculated relationship between depression and anxiety of $r = 0.05$ indicates an extremely weak, negligible relationship.
    • Moderate (Intermediate) Relationship: Values falling between $0.25$ and $0.75$ ($0.25 ≤ |r| < 0.75$).
      • Example: An absolute numerical value around $0.45$ represents a moderate relationship.
    • Strong (Large) Relationship: Any Pearson's $r$ absolute value equal to or greater than $0.75$ ($|r| ≧ 0.75$).
      • Example: A calculated Pearson's $r = 0.99$ indicates an extraordinarily strong, vital relationship.

The Standard Sentence Template for Correlational Statements

• Standardized Verbal Construction:

  • To accurately report correlational findings without implying causation, use the standardized sentence structure:
    • [Level Descriptor 1] [Variable 1 Name] is [Strength Adverb] associated with [Level Descriptor 2] [Variable 2 Name]
  • Components of the Formula:
    • [Level Descriptor 1] and [Level Descriptor 2]: Selected as directional adjectives ("higher", "lower", "more", "less").
      • If Pearson's $r$ is positive, use matching descriptors (e.g., "lower" with "lower" or "higher" with "higher").
      • If Pearson's $r$ is negative, use opposing descriptors (e.g., "lower" with "higher" or "higher" with "lower").
    • [Variable Names]: Pulled directly from the operational definitions in the research problem.
    • [Strength Adverb]: Insert "weakly" if $|r| < 0.25$ or "strongly" if $|r| ≧ 0.75$ based on the magnitude.
    • is associated with: Mandatory non-causal linking phrase (synonyms such as "is related to" are permitted).

• Applied Template Examples:

  • Example 1 (Height and Weight): "Lower height is strongly associated with lower weight."
  • Example 2 (Age and Health): "Lower age is strongly associated with higher health."

Non-Causality and the Third Variable Problem

• The Inability to Infer Causation:

  • Correlational studies cannot determine causality. They state only that variables co-occur, not that one variable causes the other in the real world.
  • Examples: Age does not directly cause health deterioration; height does not cause weight; weight does not cause height.
  • Prohibited Causal Terminology: Language such as "causes", "leads to", or "having variable X means you will have variable Y" must never be used when describing correlational findings because these terms explicitly imply causation.

• The Third Variable Problem Defined:

  • Causality cannot be established in correlational studies because of the potential presence of a third variable.
  • A third variable is an unmeasured, external variable not accounted for in the primary research question that is correlated with both measured variables in the study and accounts for their observed statistical association.

• Real-World Case Study 1: Ice Cream Sales and Homicide Rates:

  • Observed Correlation: Higher levels of ice cream sales are strongly associated with higher numbers of homicides/murders.
  • Erroneous Causal Inference: Concluding that eating ice cream causes violence and attempting to eliminate ice cream stands to stop homicides.
  • Identifying the Third Variable: Warm weather (outdoor temperature/seasonality).
    • When weather is warm, ice cream sales increase significantly.
    • Warm weather also causes people to leave their homes, congregate in large outdoor crowds, and interact socially, increasing potential friction, antagonism, and homicides.
    • During cold, wet winter seasons, individuals stay indoors, sharply reducing opportunities for interpersonal violence regardless of ice cream consumption.

• Real-World Case Study 2: Social Media Usage and Child Physical Health:

  • Context: A television news broadcast in Portland reported on a study regarding youth internet usage. The news anchor concluded she would remove her children's Facebook access to prevent them from getting sick.
  • Erroneous Causal Inference: Assuming Facebook access directly causes physical illness or elevated cholesterol.
  • Identifying the Third Variable: Physical activity level.
    • Sedentary habits, such as spending excessive time indoors or in a basement using social media, reduce time available for physical exercise.
    • A low level of physical activity directly impacts body health and cholesterol levels, explaining the observed relationship between social media time and poor health.

Descriptive Research Methods: Case Studies, Naturalistic Observation, and Single-Variable Surveys

• Overview of Descriptive Methods:

  • Descriptive research methods aim to systematically observe and describe behavior without manipulating independent variables or necessarily evaluating two-variable correlations.

• Method 1: Case Studies:

  • Definition: An in-depth research design where a clinician or researcher spends extensive time examining a single individual.
  • Primary Application in Psychology: Used predominantly to investigate psychological phenomena or disorders that are extremely rare, novel, or previously unseen.
  • Procedure: The researcher conducts comprehensive interviews, asks detailed qualitative questions, gathers exhaustive data, and writes a thorough descriptive narrative report summarizing the single case.

• Method 2: Naturalistic Observation:

  • Definition: Observing and systematically recording subjects' spontaneous behaviors in their natural environment without researcher interference or manipulation.
  • Exemplar Case Study: Jane Goodall's field research studying chimpanzees in Zimbabwe.
  • Details: Jane Goodall relocated to Zimbabwe approximately $16\text{ years}$ prior to this analysis (building on observational programs extending across $6\text{ decades}$ or $60\text{ years}$ of continuous scientific data).
  • Methodology: Goodall lived alongside the primates, tracking social interactions, relationships, and behavioral patterns completely non-intrusively without disrupting the subjects.

• Method 3: Single-Variable Surveys:

  • Definition: Deploying survey instruments to measure and describe a single standalone variable within a sample, rather than finding correlations between two variables or conducting an experiment.
  • Example: Administering a post-lecture questionnaire asking students: "How annoying was class today?", scored on a single numeric rating scale from $1$ to $5$.
  • Surveys are versatile tools that can be adapted for experimental designs, correlational designs, or simple single-variable descriptive assessments.

Statistical Significance, Decision Thresholds, and Researcher Bias

• Statistical Significance Criteria:

  • For a scientific finding to be considered meaningful or important within psychology, it must pass formal inferential statistical testing.
  • The Probability Criterion ($p < 0.05$):
    • Scientific disciplines overwhelmingly establish statistical significance using a strict maximum acceptable error threshold.
    • A statistical result is deemed significant only if the probability of making a false inference (a false positive or bad choice) is strictly less than $0.05$ ($p < 0.05$).
    • The proportion $0.05$ corresponds precisely to a $5\text{\%}$ error rate.
    • To assert that variable $X$ and variable $Y$ are meaningfully related, researchers must be at least $95\text{\%}$ confident that the observed outcome did not occur due to random chance.

• Clinical Significance vs. Statistical Significance:

  • In applied and clinical sciences, researchers must differentiate statistical significance from practical or clinical significance.
  • An outcome must not merely pass the mathematical threshold of $p < 0.05$; it must also demonstrate practical, real-world utility and clinical value.

• Human Bias in Scientific Inquiry:

  • Researchers conducting psychological and clinical scientific studies are human beings.
  • Because scientists are human, they are inherently subject to cognitive biases, expectations, and personal perspectives.
  • Critical thinking requires acknowledging that all scientific literature across psychology and related fields is produced by biased human individuals, requiring careful evaluation of methodologies and conclusions.

Course Resources and Exam Preparation Guidance

• Study Guide Usage:

  • The provided study guide represents the primary starting reference point for exam preparation, as it explicitly defines what material will and will not be evaluated on examinations.

• Navigating Course Materials:

  • If a concept is listed on the study guide but was omitted from lecture discussions, students must consult the textbook to review that information.
  • If a topic is covered thoroughly within classroom PowerPoint presentations, consulting the textbook for that specific topic may be optional.
  • The textbook serves as an essential secondary resource to find additional examples, seek clarification, and obtain precise, formal definitions for core vocabulary terms.