Notes on Observational Methods, Psychological Testing, Surveys, and Correlation vs Causation

Operationalizing aggression and the importance of definition

• Aggression must be defined operationally before observation: you must specify what counts as aggression in the study.
• Why define? Different researchers might label the same behavior differently (e.g., rumors as aggression to some, not to others).
• Train the research team to ensure everyone is measuring the same variable in the same way.
• Then proceed to observe, using a clearly defined target behavior.

Covert observation and how to observe without bias

• Observers should be covert: avoid influencing behavior by being conspicuous (no legal pads, obvious note-taking, etc.).
• Rationale: if participants know they are being watched, their behavior changes (Hawthorne effect).
• Bar experiment example: to study whether people drink more alone or in a group, you could not simply sit with a clipboard in a bar; you should blend in and collect data unobtrusively.
  • Strategy: go with a research partner, order non-alcoholic drinks (or sodas) and pretend to be social; use cell phones or discreet methods to record observations.
  • Spatial strategy: have team members observe different parts of the bar to capture different social contexts.
  • Ethical caveat: do not let covert methods compromise consent or safety; the data quality depends on non-disruptive presence.
• Practical warning: some bars or settings may require more subtle approaches to avoid discovery and data contamination.
• Query: does the setting (type of bar, crowd size) affect results? Yes—different environments (day bars, nightclubs, piano bars) may yield different drinking patterns. A representative sample of settings improves generalizability.

Naturalistic field observation vs. laboratory observation

• Naturalistic (field) observation: observe behavior as it occurs in the real world with minimal interference.
  • Pros: high ecological validity, captures natural behavior.
  • Cons: little control over environment, measurement imprecision (e.g., exact amount of alcohol consumed is hard to quantify).
  • Example: observing 32 bars in a region to compare solitary vs group drinking requires substantial time, resources, and a representative sample.
• Laboratory observation (in a controlled setting): increase control, precise measurement, but may reduce natural behavior.
  • Example of controlled lab scenario: a classroom or simulated bar where the researcher controls variables such as serving method, age verification, and tracking drinks via bartender-recorded data.
  • Trade-off: greater precision and control vs loss of naturalistic behavior.
• Hybrid approach: if resources allowed, combine both methods to balance naturalism with precision, but it is resource-intensive.
• Key takeaway: observational studies are systematic and require clear operational definitions, covert observation when appropriate, and an explicit acknowledgment of limitations.

Sampling and the representativeness problem

• Important concept: representative sample is needed to generalize beyond the observed group.
• Example issue: generalizing findings from white participants to non-white populations is problematic without representative sampling.
• The bar study illustrates why sampling all types of venues (day bars, nightclubs, etc.) is necessary for generalizability.
• Resource considerations: comprehensive observational studies (e.g., 32 bars) demand substantial time and personnel.

Observational study details and measurement challenges

• Observational data can be naturalistic but may lack precise measures (e.g., exact alcohol intake, whether a drink was finished).
• Observer must decide how to record ambiguous situations (e.g., a drink that is half-full, finished in the bathroom, or spilled).
• When you bring observers into a highly controlled setting (e.g., a classroom with named cups and bartenders recording pours), you gain precision but lose naturalism.
• Balance: acknowledge trade-offs and openly state limitations when publishing results.

Types of psychological tests: objective vs. projective

• Objective tests (paper-and-pencil): standardized questions with numerical scoring; examples include IQ tests, personality inventories (e.g., MBTI in this context as a standard self-report measure).
• Projective tests: ambiguous stimuli designed to elicit projection of unconscious thoughts; examples include inkblot-like tasks where participants describe what they see.
• Projective tests rely on interpretation of responses by the researcher, introducing potential bias.
• Pattern approach: for projective tests, look for recurring themes across multiple stimuli rather than focusing on a single response.
• Practical note: projective tests are less standardized and historically viewed as less scientifically robust than objective tests; they are often used in clinical settings.

The inkblot projection example and interpretation bias

• Demonstration: an ambiguous image (inkblot-like) can be interpreted as various things (animals, objects, etc.).
• The idea: different participants project their own unconscious content onto the ambiguous stimuli.
• The researcher looks for patterns across many stimuli (10–12 in a set) to identify latent themes, rather than relying on a single interpretation.
• Caution: interpretation is subjective and can introduce bias; not as objective or standardized as some tests.
• In class examples: students discuss which image they saw (e.g., two dogs, animals, insects, etc.), illustrating projection and subjectivity.

Reliability, standardization, and validity of psychological tests

• Reliability: consistency of a test over time or across different measures.
  • Example: test-retest reliability or parallel-form reliability; a highly reliable test yields consistent results across administrations.
• Standardization: uniform procedures for administering and scoring the test.
  • Ensures that all test-takers are treated the same way and scores are comparable.
• Validity: whether the test measures what it is intended to measure.
  • A test can be reliable but not valid (e.g., a bathroom scale that consistently weighs you 20 pounds off).
  • Validity has multiple facets (content, construct, criterion-related, etc.), but the key idea is alignment with the intended construct.
• Practical example: a depression scale should reflect depressive symptoms (changes in mood, activities, concentration, sleep, appetite, etc.).
  • Question design matters: asking about crying a lot might not accurately indicate depression (some depressed individuals cry less); better items target core symptoms and functioning.
• Takeaway: a useful psychological test must be reliable, standardized, and valid; lacking any one of these undermines usefulness.

Distinguishing objective and projective tests, with examples

• Objective test example: a self-report inventory asking about frequency of certain behaviors on a Likert scale.
• Projective test example: ambiguous stimuli requiring interpretation (e.g., inkblot-type images).
• Discussion point: even widely used objective measures (like MBTI) have validity debates; reliability and cross-sample stability are essential for usefulness.

Survery methods: advantages, challenges, and ethics

• Surveys offer a quick way to collect data from large samples with limited resources.
• They involve sampling from a population and generalizing to that population; potential problem if the sample isn’t representative.
• Common settings: mall or public spaces to collect responses, targeting high-traffic areas for data collection.
• Demographic and sensitive topics: surveys can gather age, ethnicity, sexual orientation, upbringing, relationships, etc., while attempting to preserve anonymity by avoiding identifying information.
• Sensitive topics (e.g., human sexuality, sexual activity, number of partners, masturbation frequency) require careful, non-threatening wording and a clinical, non-judgmental approach to minimize social desirability bias.
• Techniques to reduce bias:
  • Use non-threatening, non-identifying questions early in the survey to build comfort.
  • Provide anonymity and reassure confidentiality.
  • Train interviewers to be non-judgmental and neutral; avoid signaling judgment or surprise.
  • Use standardization in question phrasing and order when possible.
• Masters and Johnson and Alfred Kinsey's contributions:
  • Kinsey trained researchers to maintain clinical, non-judgmental stance with non-threatening initial questions and then progressively more sensitive questions.
  • Avoid showing judgment; do not react to surprising answers to prevent skewing responses.
• Biases and data quality issues:
  • Social desirability: respondents may underreport socially undesirable behaviors.
  • Volunteer bias: those who volunteer for surveys may differ from the general population.
  • Question wording and survey design can strongly influence responses.
• Survey design takeaway: carefully craft questions, ensure anonymity, minimize bias, and acknowledge limits when generalizing results.

Correlation versus causation and common examples

• Core concept: correlation does not equal causation.
• Example critique: more churches in an area correlating with higher crime rate does not imply churches cause crime; the relationship may reflect population density and other socio-economic factors.
  • In dense areas: more churches, more crime, more services (food pantries, thrift shops, counseling) provided by churches.
  • Population density can explain both higher church presence and higher crime rates due to more people and more poverty.
• Another example: higher ice cream sales correlating with higher crime rates; again likely a density-related or confounding factor rather than a causal link.
• Cautionary note for interpretation:
  • Always consider alternative explanations and potential confounders.
  • Avoid drawing causal conclusions from simple correlations.

Linking to broader themes and exam-ready takeaways

• Always define variables clearly (operational definitions) before observation.
• Choose observational approach based on research goals and resources; recognize trade-offs between naturalism and control.
• Ensure samples are representative to generalize findings beyond the observed group.
• Distinguish between types of data collection: observational, test-based (objective vs projective), and survey data; each has strengths and limitations.
• For tests:
  • Reliability: consistency across time or forms.
  • Standardization: uniform administration and scoring.
  • Validity: measuring what the test intends to measure.
• In surveys:
  • Be mindful of volunteer bias and social desirability.
  • Use non-identifying questions, ensure privacy, and train interviewers to minimize bias.
• In data interpretation:
  • Distinguish correlation from causation and consider potential confounding factors.
  • Use multiple data sources or designs (e.g., combination of field and lab studies) when possible to triangulate findings.
• Exam patterns you should anticipate:
  • Awareness of how practice effects can influence repeated testing (e.g., SAT-like contexts) and how reliability, standardization, and validity are interrelated.
  • Understanding that highly reliable tests can still be invalid if they do not measure the intended construct.
  • The need to discuss limitations openly when presenting research findings.

${32}$ bars, ${5-7}$ observer teams or sample sizes, ${12}$ questions in a Kinsey-style sequence, ${1300}$ or ${1200}$ SAT-like scores as examples of practice effects, and other numeric references used to illustrate points throughout the lecture.