सSaral Shiksha Yojna
Courses/Behavioral Research: Statistical Methods

Behavioral Research: Statistical Methods

CG3.402
Vinoo Alluri•Monsoon 2025-26•4 credits
Sample Papers/200-mark mock paper (Set 2) · Paper TWO

200-mark mock paper (Set 2) · Paper TWO

Duration: 180 min • Max marks: 200

Section A — 0.5 mark MCQs (20 × 0.5 = 10 marks)

10 marks
  1. 1.Distance walked in kilometers is which scale? (a) Nominal (b) Ordinal (c) Interval (d) Ratio0.5 m
  2. 2.In a normal distribution, the **mean, median, and mode** are: (a) Different (b) All equal (c) Cannot be determined (d) Equal only when SD = 10.5 m
  3. 3.A study with **internal validity but low external validity** is: (a) Causally interpretable but limited in generalization (b) Both strong (c) Both weak (d) Easily replicable0.5 m
  4. 4.A test with α = 0.10 (rather than 0.05) is: (a) More conservative (b) More liberal — easier to reject H₀ (c) Always preferred (d) Equivalent0.5 m
  5. 5.Kurtosis of a Normal distribution is: (a) 0 (b) 1 (c) 3 (excess = 0) (d) Always positive0.5 m
  6. 6.A scatter plot with data falling perfectly on a downward line has Pearson r = (a) +1 (b) −1 (c) 0 (d) Cannot say0.5 m
  7. 7.Adira draws 30 sample means from a uniform population. The sampling distribution of those means is: (a) Uniform (b) Approximately normal by CLT (c) Bimodal (d) Skewed0.5 m
  8. 8.A research design where participants know they're being studied is: (a) Naturalistic observation (b) Experimental (c) Reactive (d) Always invalid0.5 m
  9. 9."Generalizability across measures" most relates to: (a) Test-retest reliability (b) Construct validity (c) Inter-rater reliability (d) Internal consistency0.5 m
  10. 10.Aagam's data show **kurtosis = 5**. Distribution is: (a) Platykurtic (b) Mesokurtic (c) Leptokurtic / heavy-tailed (d) Skewed0.5 m
  11. 11.Regression with **R² = 0.95**: (a) Strong fit (b) Overfitting (c) Always biased (d) Insufficient data0.5 m
  12. 12.**Fisher's exact test** appropriate when: (a) Chi-square assumptions fail (expected cells < 5) (b) Sample size huge (c) Always for proportions (d) Only for paired data0.5 m
  13. 13.A within-subjects design tests whether **scores change** as a function of: (a) Different participants (b) Conditions within the same participants (c) Random sampling (d) Sample size0.5 m
  14. 14.McNemar test is used for: (a) Two independent proportions (b) Two **related** proportions (paired binary) (c) ANOVA (d) Continuous DVs0.5 m
  15. 15.Cohen's d between two means: (a) (M₁ − M₂) / SD_pooled (b) (M₁ − M₂) / SE (c) M₁ − M₂ (d) (M₁ − M₂) × √n0.5 m
  16. 16.**Eta-squared (η²)** is: (a) Proportion of total variance explained by effect in ANOVA (b) Effect size for chi-square (c) Correlation coefficient (d) Always > 10.5 m
  17. 17.99% CI is **wider** than 95% CI on same data because: (a) Higher confidence needs wider interval (b) Different data (c) Larger sample (d) Random0.5 m
  18. 18.**Mode** for a continuous distribution: (a) Highest density point (b) Mean (c) Median (d) Always equal to mean0.5 m
  19. 19.A **leverage** point in regression has: (a) Extreme predictor values (b) High residual (c) Always influential (d) Outlier in DV only0.5 m
  20. 20.A **conjugate prior** in Bayesian inference yields: (a) Non-informative posterior (b) Posterior in same family as prior (c) Frequentist result (d) Always normal0.5 m

Section B — 1 mark MCQs (20 × 1 = 20 marks)

20 marks
  1. 1.Mehek compares **stress** in 4 hospital shifts (Morning, Afternoon, Evening, Night) using **different nurses** in each (between-subjects). Continuous, normal. Best test: (a) RM ANOVA (b) One-way ANOVA (c) Mixed ANOVA (d) Friedman1 m
  2. 2.A 4 × 5 chi-square test of independence, n = 200. df and expected-cell concern: (a) df = 20; n > 200 needed (b) df = 12; expected counts must be ≥ 5 (c) df = 9 (d) df = 12; sample fine1 m
  3. 3.Manvi has a 3-level **ordinal DV** and a continuous IV. Best analysis: (a) Linear regression (b) Ordinal logistic regression (c) Chi-square (d) Pearson r1 m
  4. 4.Regression's F-test F(3, 96) = 5.2, p = .002 indicates: (a) Overall model significant; at least one predictor matters (b) Each predictor significant (c) Model fits perfectly (d) Multicollinearity1 m
  5. 5.Disha sees Cook's distance > 1 for one observation: (a) Normality (b) Multicollinearity (c) Highly influential observation (d) Insufficient data1 m
  6. 6.`Outcome = β₀ + β₁X + β₂X² + ε` tests: (a) Linear only (b) **Non-linear (quadratic) relationship** (c) Multicollinearity (d) Heteroscedasticity1 m
  7. 7.A **2 × 2 × 3 factorial ANOVA**. Possible main + interaction effects: (a) 5 (b) 6 (c) 7 (d) 81 m
  8. 8.Friedman's test statistic is approximately χ² with df = (a) n − 1 (b) k − 1 (k = conditions) (c) (r−1)(c−1) (d) n − k1 m
  9. 9.Anaya models `time_to_event` (days until customer churn). Best model: (a) Linear regression (b) Logistic (c) Survival analysis / Cox proportional hazards (d) ANOVA1 m
  10. 10.Logistic regression's pseudo-R² (e.g., McFadden's): (a) True variance explained (b) Interpreted similarly to R² but different scale; 0.2–0.4 = good fit (c) Always 1 (d) Cannot compute1 m
  11. 11.Yashvi: 5 items, Cronbach's α = 0.95: (a) Excellent reliability (b) Possibly redundant items (c) Both — excellent + may indicate redundancy (d) Poor reliability1 m
  12. 12.Two-way ANOVA: Factor A (2 levels) × Factor B (3 levels), n = 15 per cell. Total N and df_total: (a) N = 90, df = 89 (b) N = 15, df = 14 (c) N = 30, df = 29 (d) N = 6, df = 51 m
  13. 13.Mann-Whitney U test tests: (a) Means (b) Stochastic dominance — whether one distribution gives larger values (c) Variances (d) Independence1 m
  14. 14.Confounders are best controlled by: (a) Randomization in experiments (b) Statistical adjustment in observational studies (c) Both, with randomization being more effective (d) Increasing n1 m
  15. 15.AIC = 250 vs AIC = 245. Lower preferred if: (a) Δ AIC > 10 (strong) (b) Δ = 5 (substantial) (c) Δ = 5 (some preference) (d) Equally good1 m
  16. 16.Hriday reports BF₁₀ = 0.4: (a) Anecdotal evidence for H₀ (b) Moderate for H₁ (c) Anecdotal for H₁ (d) Strong for H₀1 m
  17. 17.Mediator vs moderator: (a) Mediator explains mechanism; moderator changes effect's strength/direction (b) Same thing (c) Always categorical (d) Only used in ANOVA1 m
  18. 18.Multilevel modeling (LMM) preferred when: (a) Observations independent (b) Data nested (students in classrooms, repeated measures) (c) DV categorical (d) Small sample1 m
  19. 19.Kavin's RM-ANOVA shows sphericity violation. Greenhouse-Geisser with ε = 0.65, original df = (3, 60): (a) (3, 60) (b) (1.95, 39) (c) (2, 60) (d) (3, 39)1 m
  20. 20.Eshan: p = 0.049, d = 1.5, n = 8. Best interpretation: (a) Robust and confirmed (b) Significant, but with n = 8, d = 1.5 likely inflated (winner's curse); replicate (c) Type I error (d) No effect1 m

Section C — 2 mark short answers (15 × 2 = 30 marks)

30 marks
  1. 1.State **Bayes' theorem** with each term named.2 m
  2. 2.Differentiate **between-subjects** and **within-subjects** designs, with one advantage of each.2 m
  3. 3.Define **mediation**; contrast with **moderation** (with one-line examples).2 m
  4. 4.Define **convergent** and **discriminant** validity with example pair.2 m
  5. 5.Why is **stratified random sampling** often preferred?2 m
  6. 6.Reyaan studies whether an app affects daily steps. Identify IV, DV, scales, and design maximizing internal validity.2 m
  7. 7.Why does **double-blind** matter in behavioral research?2 m
  8. 8.Define **regression to the mean** with example.2 m
  9. 9.State **homoscedasticity** assumption and how to check.2 m
  10. 10.Define **interaction effect** in ANOVA with one-line example.2 m
  11. 11.Define **p-hacking** with two specific practices.2 m
  12. 12.Define **publication bias** and one consequence for meta-analysis.2 m
  13. 13.Two purposes of a **manipulation check** in an experiment.2 m
  14. 14.Difference between **FWER** and **FDR**.2 m
  15. 15.Two differences between **paired t-test** and **Wilcoxon signed-rank**.2 m

Section D — 5 mark questions (12 × 5 = 60 marks)

60 marks

    Section E — 10 mark long descriptive (8 × 10 = 80 marks)

    80 marks

      Track your attempt locally — score and time are recorded in your browser. (Coming soon: timed-attempt mode.)