PSYCH: Scientific Reasoning, Bias, & Measurement

Read ahead when possible; overlaps between slides and text aid recognition.
Spaced repetition helps long-term memory; better to revisit after a forgetting interval.
One module per class planned; four modules in Chapter 2; the last module (statistical primer) will be cherry-picked.
Participation questions to ensure accessibility for all; raise hands to check access issues with eText (Nexus/Pearson).

Question: does full moon increase crime, hospital visits, or erratic behavior?
Why people believe it (despite debunking):
- Confirmation bias: memorable, salient events align with belief.
- Cultural stories/media keep myth alive.
- Illusion of correlation: coincidence mistaken for causation.
- Preference for a simple, cosmetic explanation over randomness.

Numerology basics: life path numbers (main vs secondary), enemy years; examples discussed (e.g., life path 33).
People report practical benefits regardless of peer review status; anecdotes vs evidence.
Various perspectives: entertainment vs meaningful guidance for some individuals.
Important takeaway: consider biases and lack of robust scientific validation.

Understand key research terms: variable, hypothesis, etc.
Identify five qualities of good quality research; apply reliability and validity concepts.
Distinguish anecdotes, authority, and common sense; evaluate claims critically.
Recognize the role of bias, sampling, replication, and generalizability.

Objectivity: measurements and conclusions not dependent on personal opinion.
Validity: measures what it is intended to measure.
Reliability: consistency of results across time and observers.
Generalizability: applicability of findings to broader populations.
Reducing bias and enabling replication/peer scrutiny: transparency, openness, and conditions for replication.
Application: findings should be usable in real-world settings with acknowledged limitations.

Objective: fixed definitions and criteria; e.g., a ruler always reads the same length.
Subjective: interpretation-dependent (e.g., projective tests like inkblots).
Some measures mix both; objective measures are generally preferred for scientific conclusions.

Variable: what you measure or manipulate (e.g., memory, stress, personality).
Independent vs. Dependent: independent is manipulated; dependent is the outcome.
Operational definitions: specify exact measurement rules (e.g., depression = a score ext{BDI} \ge ext{20}).
Clarity in definitions prevents “telephone game” drift across studies.

Validity vs Reliability:
- Validity = does the measure assess the intended construct?
- Reliability = are measurements consistent across time/raters/forms?
Examples:
- A tape measure: reliable and valid for length.
- Head circumference: reliable but not necessarily valid as a proxy for intelligence.
- Lie detectors: show reliability in some contexts but limited validity for deception.
Types of reliability:
- Test-retest reliability: scores stay similar over time.
- Alternate forms reliability: different versions yield similar results.
- Inter-rater reliability: different observers rate the same event similarly.

Bias sources: researcher bias, participant bias, conflicts of interest, sampling bias.
Social desirability: participants tailor answers to appear favorable.
Placebo effect: improvement due to expectations rather than the intervention.
Context matters: culture, baseline differences, and environment affect results.

Generalizability depends on sample diversity and size.
- Example: n=400{,}000 vs n=40; larger, diverse samples improve generalizability.
Replication is essential: other researchers should obtain similar results using the same methods.
Studies have limitations; explicit limitations guide future research and improvements.

Objective measures (e.g., physiological data) vs subjective assessments (e.g., projective tests).
PCR test discussion (contextual): validity and reliability debates; inventor cautions about diagnostic use.
Takeaway: always consider what a measurement truly captures and its limitations.

Use clear operational definitions for all constructs.
Aim for objectivity and minimize subjective interpretation.
Prefer reliable and valid measures; assess both where possible.
Ensure sample size and diversity support generalizability.
Be vigilant about biases; design studies to mitigate them and enable replication.