Saral Shiksha Yojna
Courses/Behavioral Research: Statistical Methods

Behavioral Research: Statistical Methods

CG3.402
Vinoo Alluri•Monsoon 2025-26•4 credits
Sample Papers/Mock Paper 2 — All Units Mix

Mock Paper 2 — All Units Mix

Duration: 120 min • Max marks: 100

Section A — MCQ (20 × 1 = 20)

20 marks
  1. 1.Likert responses (Strongly Disagree → Strongly Agree) are best described as: (a) Nominal (b) Ordinal (c) Interval (d) Ratio1 m
  2. 2.SEM is computed as: (a) σ × √n (b) σ / √n (c) σ² / n (d) σ × n1 m
  3. 3.Which is a confound, not a treatment effect? (a) Random-assignment differences (b) Pre-treatment group differences due to a third variable (c) Sampling variability (d) Measurement noise1 m
  4. 4.A 95% CI for the difference of means is [−0.5, 3.2]. The null value 0 is: (a) Inside → fail to reject (b) Outside → reject (c) Interval too wide (d) Cannot tell1 m
  5. 5.Which test assumes sphericity? (a) Independent t (b) Pearson r (c) Repeated-measures ANOVA (d) χ²1 m
  6. 6.About F-distribution: (a) Symmetric around 0 (b) ≥ 0, right-skewed (c) Identical to t (d) For categorical data1 m
  7. 7.r = −0.85 indicates: (a) Weak negative (b) Strong positive (c) Strong negative linear (d) No relationship1 m
  8. 8.Cohen's d = 0.5 is: (a) Small (b) Medium (c) Large (d) No effect1 m
  9. 9.True of an unbiased estimator: (a) Always equals the parameter (b) Expected value equals the parameter (c) Has smallest variance (d) Always larger1 m
  10. 10.Boxplot point at 1.7 × IQR above Q3 — flagged outlier? (a) Yes — beyond 1.5×IQR (b) No — must exceed 3×IQR (c) Depends on skew (d) No — must exceed 2 SD1 m
  11. 11.Best describes a heat map: (a) 3D bar chart (b) Matrix where colour encodes cell value (c) A pie chart variant (d) Density scatter1 m
  12. 12.Anscombe's quartet teaches: (a) Mean + SD are sufficient summaries (b) Different datasets can share summaries but differ visually — always plot data (c) Outliers should be removed (d) Pearson r is the only valid correlation1 m
  13. 13.Factor loading 0.85 on a single factor: (a) Weak (b) Strong association of item with factor (c) Item is irrelevant (d) Negative correlation1 m
  14. 14.Parallel analysis is used to: (a) Test linearity (b) Decide # factors/components to retain (c) Compute correlations (d) Adjust p-values1 m
  15. 15.α and Type I error: (a) α = P(Type II) (b) α = P(Type I) if H₀ true (c) α decreases with n (d) α is FDR1 m
  16. 16.5-IQ-point gap between two groups, p < .001, n = 50,000 per group. Likely: (a) Practically important (b) Likely real but practically small (c) p-hacking artefact (d) Sampling error1 m
  17. 17.Welch t differs from Student's t in: (a) Assumes equal variances (b) Does not assume equal variances (c) Uses ranks (d) For paired data1 m
  18. 18.BF₀₁ = 25 implies: (a) Strong for H₁ (b) Strong for H₀ (c) Inconclusive (d) Null definitely true1 m
  19. 19.Cortisol levels in same 40 participants, before and after a stress task — appropriate test: (a) Independent t (b) Paired t (c) χ² (d) ANOVA1 m
  20. 20.Intercept β₀ in Y = β₀ + β₁X + ε represents: (a) The slope (b) Predicted Y when X = 0 (c) Mean of X (d) Residual1 m

Section B — MSQ (10 × 2 = 20)

20 marks
  1. 1.True about p-values: (a) p = P(H₀ true) (b) p depends on test statistic and n (c) p is computed assuming H₀ true (d) p < .05 means H₁ true (e) Smaller p ≠ larger effect2 m
  2. 2.Properties of Normal distribution: (a) Symmetric (b) Mean = median = mode (c) Defined by μ and σ (d) Heavier tails than t-distribution (e) ~95% within ±2σ2 m
  3. 3.Apply to CFA: (a) Theory-driven (b) Number of factors pre-specified (c) Used to discover latent structure with no prior model (d) Tests model fit (e) Hypothesis testing about factor structure2 m
  4. 4.Valid uses of χ²: (a) Goodness-of-fit (b) Independence in a contingency table (c) Comparing two group means (d) Two-by-two categorical association (e) Correlation between continuous vars2 m
  5. 5.Detectable via residual plots in regression: (a) Non-linearity (b) Heteroscedasticity (c) Outliers / leverage (d) Multicollinearity (e) Normality of errors2 m
  6. 6.Why Tukey's HSD after a significant one-way ANOVA: (a) Tukey controls FWER for all pairwise comparisons (b) Tukey is the only post-hoc test available (c) Tukey adjusts for multiple comparisons (d) Tukey is appropriate after a significant omnibus F (e) Tukey replaces the F-test2 m
  7. 7.Conditions favoring parametric over nonparametric: (a) Interval / ratio DV (b) Approximately normal data (c) Large n (d) Ordinal DV (e) Roughly equal variances2 m
  8. 8.BF preferred to p-value when: (a) Evidence for null is of interest (b) Optional stopping desired (c) Want to incorporate priors (d) Want to fail-to-reject H₀ definitively (e) Effect size needs reporting2 m
  9. 9.Valid forms of sampling bias: (a) Selection (b) Self-selection (c) Survivorship (d) Random sampling (e) Non-response2 m
  10. 10.Sources that inflate Type I error: (a) Many tests without correction (b) Optional stopping (c) Post-hoc subgroup analyses (d) Larger n (e) Selecting predictors after looking at data2 m

Section C — Short descriptive (6 × 5 = 30)

30 marks
  1. 1.Why is running multiple pairwise t-tests across 3+ groups problematic? How does ANOVA fix it?5 m
  2. 2.Define and contrast Type I and Type II errors. Explain the trade-off.5 m
  3. 3.Outline the steps of a hypothesis test (specific example: testing whether average sleep < 7 hours).5 m
  4. 4.What is multicollinearity? Two ways to detect and two ways to address.5 m
  5. 5.Within-subjects vs between-subjects designs. Advantage and disadvantage of each.5 m
  6. 6.Scree plot — what does it show and how is it used? Compare with parallel analysis.5 m

Section D — Long descriptive (3 × 10 = 30)

30 marks
  1. 1.Clinical trial: three drug doses (low, medium, high) vs placebo; DV is BP reduction (mmHg); 20 patients/group. (i) Design + IV/DV. (ii) Analysis plan: omnibus, post-hoc, assumption checks, effect size. (iii) Heterogeneous variances. (iv) Bayesian alternative.10 m
  2. 2.Replication crisis in behavioural science. (i) What is it? (ii) Three statistical contributors. (iii) Five reforms. (iv) Compare NHST and Bayesian in this context.10 m
  3. 3.Reviewing a paper claiming 'social media use causes depression' from a cross-sectional survey of 5,000 adolescents, r = 0.18, p < .001. Evaluate: (i) causal inference, (ii) effect size, (iii) measurement validity, (iv) sampling, (v) recommendations.10 m

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