PSY CHAPTER 9

🧠 Chapter Overview

This chapter explains survey and correlational research methods, focusing on how psychologists gather descriptive data and analyze relationships between variables without manipulating them experimentally. It contrasts experimental designs (cause-and-effect) with non-experimental, descriptive, or correlational designs that study existing differences and relationships.

🧾 1. Nature and Purpose of Survey Research

  • Survey research seeks to describe attitudes, beliefs, values, or behaviors by directly asking people through structured questionnaires or interviews.

  • It is descriptive, but the data can later support correlational or predictive analyses (e.g., regression).

⚙ Key Goals

  • Accurately describe mental processes and behaviors.

  • Explore how individuals relate to their environments.

  • Generate ideas for future experimental studies.

🧩 Relation to Other Methods

  • Experimental method: Manipulates variables to determine causality.

  • Correlational method: Measures naturally occurring variables and examines relationships (no manipulation).

Robert Woodworth emphasized that both methods are equally valuable for psychological research.

📜 2. Historical Origins

🧑‍🔬 Sir Francis Galton (1822–1911)

  • Credited with originating the survey method.

  • Distributed questionnaires about traits such as scientific ability.

  • Believed intelligence and interests were innate — emphasizing the nature vs. nurture debate.

🌍 Charles Darwin (1809–1882)

  • Sent questionnaires on emotional expressions to colleagues globally — possibly the first cross-cultural study.

  • Used leading questions like “Is astonishment expressed by open eyes and mouth?” → biased results but pioneering.

🎓 G. Stanley Hall

  • Introduced surveys in the child study movement (“Questionary”).

  • Gathered data from children and teachers to understand development.

  • His surveys revealed differences in knowledge between urban and rural children.

  • Although groundbreaking, early survey research faced criticism from figures like William James and E. B. Titchener, who considered it unscientific.

🎯 3. Sampling in Survey Research

Sampling is crucial in survey work to ensure results generalize to a population.

🧮 Types of Sampling

  • Probability sampling: Every member of the population has a known chance of selection (e.g., random, stratified, cluster sampling).

  • Non-probability sampling: Participants are not chosen randomly (used more often in experiments).

⚠ Biased Sampling Example — Literary Digest (1936)

  • Predicted Alf Landon would beat F.D. Roosevelt using mailed surveys from car registries and phone books → biased toward wealthy Republicans.

  • Also suffered self-selection bias — only those who chose to reply were counted.

💡 Lesson:

Large numbers don’t ensure representativeness. Random, representative samples are essential to valid survey results.

🧩 4. Surveys vs. Psychological Tests

Surveys

Psychological Tests

Measure attitudes, opinions, beliefs

Assess psychological constructs (personality, intelligence, etc.)

Descriptive focus

Diagnostic focus

No formal validation required

Require proven reliability and validity

Example: campus climate survey

Example: Beck Depression Inventory-II

✍ 5. Creating an Effective Survey

📌 Steps:

  1. Start with a clear empirical question.

  2. Define variables operationally.

  3. Choose item type and wording carefully.

🎓 Examples of Empirical Questions:

  • “Do students feel safe crossing campus at night?”

  • “Are students satisfied with their education?”

🧮 6. Types of Survey Questions

🔹 Open-ended Questions

  • Allow free responses.

  • Example: “How should we finance public education?”

  • Pros: Rich data, unexpected insights.

  • Cons: Hard to score, time-consuming.

🔹 Closed Questions

  • Provide limited response options.

  • Example: “Should lottery money fund education? (Yes/No).”

  • Efficient and quantifiable.

🔹 Partially Open Questions

  • Checklist + “Other (please specify).”

🔹 Likert Scales

  • Measure attitudes on a continuum (typically 5–7 points).

  • Example:
    “The instructor was well-prepared for class.”
    (1 = Strongly Disagree … 5 = Strongly Agree)

  • Use both favorable and unfavorable statements to avoid response acquiescence (tendency to agree with everything).

  • Keep scale type consistent throughout the survey.

🔄 7. Ordering and Sensitivity

  • Start with easy, non-personal questions.

  • Place sensitive questions (e.g., sexual behavior, drug use) at the end.

  • Group similar topics together.

  • Pilot test with different audiences to check clarity and interpretation.

🧠 8. Assessing Memory & Knowledge

  • Don’t overburden memory — use short time frames (e.g., “past week” instead of “past month”).

  • Use DK (“Don’t Know”) options sparingly:

    • Pros: Avoids guessing.

    • Cons: Overuse makes data less meaningful.

    • Can reword as “Using your best guess…” to minimize DK responses.

👤 9. Demographic Information

  • Examples: age, gender, income, education, marital status.

  • Include only relevant demographics.

  • Place at the end of the survey.

  • Follow U.S. Census Bureau standards (e.g., “never married,” not just “single”).

⚠ 10. Common Wording Problems

❌ Ambiguity

“Visiting relatives can be fun.” → Who’s visiting whom?

❌ Double-barreled Questions

“It’s wrong for women in bars to curse and buy drinks for strangers.”
→ Two issues in one.

❌ Leading Questions

“Given the importance of clean air, should the Clean Air Act be strengthened?”
→ Biased phrasing that pushes respondents toward agreement.

🧰 Tips for Clear Wording

  • Use simple, complete sentences.

  • Avoid negatives, jargon, slang, and abbreviations.

  • Use balanced items (“support or oppose”).

  • Avoid carry-over effects — earlier questions influencing later answers.

📦 11. Collecting Survey Data

🗣 a. In-Person Interviews

  • Example: Kinsey Reports on sexual behavior.

  • Pros: Detailed, clarifiable responses.

  • Cons: Expensive, interviewer bias, limited reach.

📬 b. Mailed Surveys

  • Require high return rates (≥70% is good).

  • Problems: Nonresponse bias (those who reply differ from those who don’t).

  • Improve returns: brief survey, advance notice, prepaid postage, personalized cover letter, small incentives.

💭 Social Desirability Bias

Respondents may present themselves favorably (e.g., overreporting handwashing or voting).
→ Reduce with anonymity.

☎ c. Phone Surveys

  • Used random-digit dialing; peaked in the 1980s.

  • Declined due to telemarketing and cell phone complications.

  • “Sugging” = Selling Under the Guise of a Survey.

  • Mixed-mode (mail + phone) approaches help improve responses.

💻 d. Online Surveys

  • Platforms: Qualtrics, SurveyMonkey, Amazon MTurk.

  • Pros: Fast, low-cost, wide reach.

  • Cons: Self-selection bias, unverified respondents, spam filters, potential multiple entries.

⚖ 12. Ethical Considerations

  • Informed Consent: Not always required for anonymous, minimal-risk questionnaires (APA 8.05).

  • Still include cover letters or opening screens explaining purpose and confidentiality.

  • Misuse of surveys (e.g., biased corporate polls) can cause harm.

📦 Ethics Box — Christine Craft Case

  • TV anchor fired based on a biased survey about appearance.

  • Survey wording, sampling, and analysis were manipulated.

  • Demonstrates how surveys can be abused to justify discrimination.

📚 13. Research Example: Study Strategies Survey (Hartwig & Dunlosky, 2012)

  • Assessed college students’ study strategies via written survey (324 students).

  • Replicated Kornell & Bjork (2007) findings.

  • Most common strategies:

    • Highlighting (72%)

    • Self-testing (71%)

    • Rereading (66%)

    • Cramming (66%)

    • Flashcards (62%)

🧾 Key Results

  • Practice testing correlated with higher GPA.

  • Flashcards and cramming did not.

  • Planning study schedules predicted higher GPA.

  • Spaced vs. crammed study: No significant difference here, but other research shows spacing > cramming.

📈 14. Correlation and Regression

🔹 Correlation

  • Describes how two variables relate.

  • Positive correlation: both increase (e.g., study time ↑, GPA ↑).

  • Negative correlation: one increases while the other decreases (e.g., TV time ↑, GPA ↓).

📊 Scatterplots

  • Visual representation of relationships.

  • Tight clustering = strong correlation.

🔹 Pearson’s r

  • Ranges from −1.00 to +1.00.

  • Measures strength and direction of relationship.

  • Cohen’s effect sizes:

    • .10 = small

    • .30 = medium

    • .50 = large

🔹 Coefficient of Determination (r²)

  • Represents shared variance between variables.

  • Example: r = .50 → r² = .25 → 25% of variance explained.

⚠ Outliers

  • Extreme scores that distort correlation.

  • Check scatterplots to identify and remove before analysis.

🔮 15. Regression Analysis (Prediction)

  • Regression uses correlations to make predictions.

  • Regression line (line of best fit) shows predicted relationship.

  • Y = criterion variable (e.g., GPA).

  • X = predictor variable (e.g., study hours).

  • Stronger correlation = higher prediction accuracy (narrower confidence intervals).

🧩 Types

  • Simple linear regression: One predictor (e.g., study time → GPA).

  • Multiple regression: Two or more predictors (e.g., study time + motivation → GPA).

🧠 16. Causality Problems in Correlation

🚫 Directionality Problem

  • Unsure which variable causes the other.
    Example: Does happiness cause success, or does success cause happiness?

Solution: Cross-lagged panel design — measure both variables at multiple time points to infer direction.

⚙ Third-Variable Problem

  • An external factor affects both variables.
    Example: Stress might affect both sleep and grades.

Solution: Partial correlation — statistically control for the third variable.

🔗 Mediators and Moderators

  • Mediator: Explains how/why two variables are related (e.g., stress → anxiety → poor sleep).

  • Moderator: Affects the strength or direction of a relationship (e.g., coping skills reduce stress’s impact).

🧩 17. Key Takeaways

  • Survey research offers descriptive insights but requires careful design to ensure reliability and validity.

  • Good surveys: unbiased sampling, clear wording, thoughtful order, balanced scales.

  • Correlation ≠ causation, but regression allows for useful prediction.

  • Always evaluate directionality, third variables, and ethical use of survey data.