2.1 Research and Descriptive Methods

Operational Definitions

  • An operational definition specifies how a researcher will measure and manipulate variables in the study.
  • Why this matters: avoids vague terms like simply saying 'music' or 'affection' and ensures all participants perform the same procedures.
  • Music example:
    • Instead of saying music, define it operationally as: listening to an hour of a specific artist/album under controlled conditions, or exposure to that music in a defined setting.
    • This could be quantified as a fixed listening duration, a specific playlist, or a particular volume level, etc.
  • Abstract constructs require concrete measurement:
    • Affection as a construct can be quantified with proxies such as:
    • Number of hugs per day, or
    • Heart rate changes, cortisol levels, or
    • A composite score from a validated affection scale.
    • The measurement method must be defined (e.g., count hugs, use a wearable to measure heart rate, or use a standardized questionnaire).
  • Measurement of studying and prior knowledge:
    • Amount of time studying could be quantified (e.g., minutes/hours).
    • Prior knowledge could be measured with a pretest score before any manipulation (e.g., pretest out of 100).
  • Environment and noise:
    • Could quantify study environment with a rating scale (e.g., 1–5) obtained by a trained evaluator who visits the home to assess noise and other factors.
    • Example: one-to-five rating based on predefined criteria (noise level, interruptions, lighting, etc.).
  • Example using IQ/intelligence:
    • Intelligence is a complex construct; for quantification, researchers often use an IQ test score (e.g., score out of 100).
    • Acknowledgement: this is a simplification of a more complex construct, but it provides a measurable criterion for analysis.
  • The general point: operational definitions make abstract constructs measurable and observable by specifying exact procedures and measurements.
  • Typical structure of an operational definition:
    • The variable name, the exact measurement approach, the unit of measurement, and the procedure to standardize across participants.
  • Real-world implication:
    • Without precise operational definitions, experiments risk inconsistent implementation, biased results, and poor reproducibility.

APA: Role and Purpose

  • APA stands for the American Psychological Association.
  • Function: a professional organization that approves and standardizes research practices in psychology; publishes guidelines and research findings.
  • Analogy: similar to the Associated Press in journalism as a standard-bearer for quality and consistency (bias-free reporting, standardized formats).
  • Relevance: following APA guidelines helps ensure ethical conduct, measurement consistency, and credible reporting of results.

Correlation vs Causation

  • Correlation:
    • Definition: two factors vary together; as one increases, the other increases or decreases, or both move in tandem.
    • Not necessarily indicating one causes the other.
    • Example concept: scarves and hot chocolate both increase in cold weather; correlation does not imply causation.
  • Causation:
    • Definition: one factor causes a change in another.
    • Not guaranteed by correlation; there can be reverse causation or a third variable influencing both.
  • Third-variable problem:
    • A third factor (C) may influence both X and Y, producing a spurious correlation between X and Y.
    • Example from lecture: cold weather drives both scarf-wearing and hot chocolate consumption, creating a correlation between the two without direct causation.
  • Correlation coefficient (r):
    • Numerical measure of the strength and direction of a linear relationship between two variables.
    • Range: 1r1-1 \le r \le 1
    • Interpretation:
    • |r| close to 1 indicates a strong linear relationship; |r| close to 0 indicates a weak or no linear relationship.
    • r = 0 means no linear correlation.
    • Example interpretation: a correlation of r=0.75r = -0.75 is relatively strong (negative) correlation.
    • Important caveat: correlation does not imply causation; a strong correlation could be due to a third variable or be coincidental.
  • Formal note (optional formula):
    • Pearson correlation coefficient:
      r=<em>i(X</em>iXˉ)(Y<em>iYˉ)</em>i(X<em>iXˉ)2  </em>i(YiYˉ)2r = \frac{\sum<em>i (X</em>i - \bar{X})(Y<em>i - \bar{Y})}{\sqrt{\sum</em>i (X<em>i - \bar{X})^2}\; \sqrt{\sum</em>i (Y_i - \bar{Y})^2}}
    • This formula quantifies the degree to which two variables linearly co-vary.

Correlational Research (non-experimental)

  • Definition:
    • A scientific method used to examine the relationship between two variables without manipulating them.
    • Researchers observe and measure naturally occurring associations.
  • Key features:
    • No experimental manipulation of variables.
    • Identifies patterns and makes predictions but does not establish causation.
  • Practical example:
    • Measure average exam scores and hours of sleep per night across students and look for a pattern.
    • A scatter plot would visually display the relationship between sleep and exam scores.
  • Limitations:
    • Cannot infer causation due to potential third-variable problems and bidirectionality.
    • Observational data may be influenced by uncontrolled confounds.
  • Third-variable problem (revisited):
    • A correlation between X and Y may be driven by a third variable Z (e.g., stress, environment) that affects both.
  • Common spurious correlations (illustrative, not causal):
    • Example pairs that show strong correlations with no causal link (e.g., Nicolas Cage film appearances vs drowning in pools; per capita cheese consumption vs bed sheet tangling) to illustrate that correlation can be coincidental or driven by a common external factor.

Descriptive, Correlational, and Experimental: Three Broad Study Types

  • Descriptive studies:
    • Describe phenomena and collect detailed information about a specific situation, population, or event.
    • Include case studies, naturalistic observations, and surveys without testing causal hypotheses.
  • Correlational studies:
    • Examine relationships between two variables without manipulation.
    • Identify patterns, associations, and potential predictors, but cannot establish cause-and-effect.
  • Experimental studies:
    • Investigate causal relationships by manipulating an Independent Variable (IV) and observing effects on a Dependent Variable (DV).
    • Key components: random assignment, control of variables, and manipulation of the IV.
  • Data types involved:
    • Quantitative data: numerical measurements (scores, counts, scales).
    • Qualitative data: descriptions, experiences, and observations (often from interviews or open-ended observations).

Four Main Research Methods (data collection tools)

  • Survey:
    • Collects data from participants via questionnaires or interviews.
    • Frequently uses Likert scales (e.g., 1–5 or 1–7) or true/false/multiple choice items.
    • Strengths: efficient, inexpensive, scalable, low risk to participants.
    • Limitations: potential misinterpretation of questions, social desirability bias, reliance on self-report, cannot establish causation, limited to asked questions.
  • Naturalistic observation:
    • Observe behavior in real-world settings without interference.
    • Strengths: high ecological validity; captures authentic behavior.
    • Limitations: less control over variables; observer bias; cannot easily infer causation; risk of confirmation bias.
  • Case studies:
    • In-depth examination of a single person or small group.
    • Strengths: rich, holistic data; useful for rare or sensitive topics; generates hypotheses.
    • Limitations: limited generalizability; time-consuming; potential bias; cannot establish causation.
  • Experiments:
    • Manipulate one or more IV(s) and measure effects on DV(s).
    • Strengths: can establish causal relationships; high control over variables.
    • Limitations: ethical and practical constraints; may have limited external validity; sometimes higher risk to participants.
  • Data types in these methods:
    • Quantitative data: numerical results from surveys, tests, or coded observations.
    • Qualitative data: textual or narrative information from interviews, open-ended responses, or observations.

Surveys: Design, Wording, and Biases

  • Likert scales:
    • Common format: strongly agree to strongly disagree (or similar ordered categories).
  • Structured interviews:
    • Pre-determined questions asked in the same order to all participants.
  • Wording effects:
    • Subtle phrasing changes can influence responses.
    • Examples:
    • "Do you feel stressed about your upcoming exams?" vs. "Are you worried about failing your upcoming exams?"—the latter may elicit stronger negative emotions.
  • Social desirability bias:
    • Participants respond in ways they think will be viewed favorably by others.
    • Example: answering that you always hold doors for others or are always polite when asked in a survey, even if not always true.
  • Observational setting biases:
    • Participant bias: people alter behavior when they know they are being watched.
  • Practical advantages and limits:
    • Pros: large samples, cost-efficient, broad reach.
    • Cons: self-report limitations, fixed question sets limit data scope, causality cannot be inferred.

Other Data Collection Considerations

  • Two-way mirrors in child development research:
    • Used to observe interactions (e.g., sharing or bullying) in a controlled setting without participants knowing they are being observed.
    • Ethical and practical implications: allows naturalistic observation with some level of observer presence.
  • Confirmation bias risk in observational research:
    • Researchers may notice data that confirm prior expectations and overlook contradictory data.
  • Participant bias:
    • Knowing they are being observed can change how participants behave.
  • Case studies and ethics:
    • Often used for topics where experimentation would be unethical or impractical (e.g., trauma).
  • Generalizability concerns:
    • Case studies and individualized observations may not generalize to broader populations.
  • Establishing causation:
    • Correlational and descriptive methods cannot establish causation; experiments are designed to test causal relationships.

Experimental Design: IVs, DVs, and Causality

  • Key terms:
    • Independent Variable (IV): the variable the researcher deliberately changes/manipulates.
    • Dependent Variable (DV): the variable observed/measured to assess the effect of the IV.
  • Example framework (music and studying):
    • IV: exposure to music during study (music vs. no music).
    • DV: test scores after studying.
    • Hypothesis: listening to music during study will affect test scores compared to not listening to music.
  • Strengths of experiments:
    • Tight control over environment allows causal inference.
  • Limitations:
    • Potential ethical concerns and participant risk.
    • Generalizability to real-world settings may be limited; artificial laboratory conditions may affect results.

Cross-Sectional vs Longitudinal Data Collection

  • Cross-sectional studies:
    • Study many people at one point in time.
    • Pros: efficient, cheaper, broad population snapshot.
    • Cons: comparisons across different individuals; cannot track changes within individuals; potential confounding variables across groups.
    • Example: survey stress levels across freshmen, sophomores, juniors, and seniors at a single point in time.
  • Longitudinal studies:
    • Study the same people over an extended period.
    • Pros: measures changes within individuals; can reveal development and causality trends over time.
    • Cons: time-consuming; attrition risk; external events or policy changes can confound results.
    • Example: measure stress from freshman year through senior year, repeatedly.
  • Trade-offs:
    • Cross-sectional is quicker and cheaper but less able to track individual change.
    • Longitudinal provides rich developmental data but requires more time and resources and may face participant dropout.

Quick Practice Note

  • You will do a 2.1 practice with your group on Google Classroom.
  • Focus on applying these concepts to practice items about operational definitions, correlation vs causation, and research design.

Summary of Key Concepts

  • Operational definitions convert abstract constructs into measurable procedures.
  • APA provides ethical and reporting standards for psychological research.
  • Correlation measures association; causation requires experimental manipulation and control.
  • Correlational research identifies patterns but cannot prove causation due to third variables and bidirectionality.
  • Descriptive, correlational, and experimental designs each serve different research purposes.
  • Four main data collection methods: surveys, naturalistic observation, case studies, experiments.
  • Data can be quantitative or qualitative.
  • Surveys have practical advantages but several biases and limitations (misinterpretation, social desirability, cannot prove causation).
  • Cross-sectional vs longitudinal designs offer different strengths for studying changes and patterns over time.
  • Understanding IVs and DVs is essential for interpreting experimental results and causal claims.