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 ONE

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

Duration: 180 min • Max marks: 200

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

10 marks
  1. 1.Time taken (in seconds) to solve a Rubik's cube is which scale? (a) Nominal (b) Ordinal (c) Interval (d) Ratio0.5 m
  2. 2.P(X) without conditioning denotes: (a) Conditional (b) Marginal probability of X (c) Joint (d) Posterior0.5 m
  3. 3.Variance is to standard deviation as: (a) Square root is to square (b) Square is to square root (c) Sum is to mean (d) Range is to IQR0.5 m
  4. 4.A "false alarm" in detection-theory corresponds to which statistical error? (a) Type I (b) Type II (c) Both (d) Neither0.5 m
  5. 5.A distribution with a long left tail is: (a) Right-skewed (b) Left-skewed (c) Symmetric (d) Bimodal0.5 m
  6. 6.Spearman's rho is preferred over Pearson's r when: (a) Both variables are continuous and normal (b) The relationship is monotonic but non-linear, or data are ordinal (c) Sample size is large (d) Variance is constant0.5 m
  7. 7.The CLT requires the population to have: (a) Normal distribution (b) Finite variance (c) Mean = median (d) Zero skewness0.5 m
  8. 8.Sampling entire intact groups (e.g., classrooms, hospital wards) is: (a) Stratified (b) Simple random (c) Cluster (d) Snowball0.5 m
  9. 9.A rater consistently gives higher scores to attractive participants regardless of their work quality. This is: (a) Halo effect (b) Anchoring (c) Recency (d) Confirmation bias0.5 m
  10. 10.A **stem-and-leaf plot** is best for: (a) Time-series (b) Showing actual distribution while preserving individual values (c) Categorical comparison (d) Multivariate0.5 m
  11. 11.**Predictive validity** is established when: (a) Measure predicts a future theoretically related outcome (b) Measure correlates with itself (c) Two raters agree (d) Items hang together0.5 m
  12. 12.For 10 independent tests at family-wise α = 0.05, the FWER if uncorrected is approximately: (a) 0.05 (b) 0.10 (c) 0.40 (d) 0.500.5 m
  13. 13.Mauchly's test checks: (a) Normality (b) Homogeneity of variance (c) Sphericity in repeated-measures designs (d) Outliers0.5 m
  14. 14.Compared to AIC, BIC: (a) More lenient on parameters (b) Penalizes complexity more, especially with large samples (c) Always agrees with AIC (d) Cannot be used for nested models0.5 m
  15. 15.β in hypothesis testing denotes: (a) Effect size (b) Type I error (c) Type II error rate (d) Regression coefficient0.5 m
  16. 16.A **likelihood** function evaluates: (a) Prior probability (b) Probability of data given parameter values (c) Posterior (d) Marginal evidence0.5 m
  17. 17.A Cohen's d of 0.8 is conventionally: (a) Small (b) Medium (c) Large (d) Negligible0.5 m
  18. 18.The R command for area under chi-square distribution left of value q: (a) `pchisq(q, df)` (b) `qchisq(q, df)` (c) `dchisq(q, df)` (d) `rchisq(q, df)`0.5 m
  19. 19.Bessel's correction (dividing by n−1) makes which estimator unbiased? (a) Sample mean (b) Sample variance s² (c) Sample SD s (d) Sample median0.5 m
  20. 20.In Bayes' rule odds form: Posterior Odds = Prior Odds × ___. (a) Likelihood (b) Bayes Factor (c) Evidence (d) Posterior probability0.5 m

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

20 marks
  1. 1.Aarush measures the **same** students' attention scores at 4 weekly time points. Data continuous, normal. Best test: (a) One-way ANOVA (b) Repeated-measures ANOVA (c) Mixed ANOVA (d) Chi-square1 m
  2. 2.A 4 × 4 chi-square test of independence df: (a) 16 (b) 12 (c) 9 (d) 81 m
  3. 3.Vidhi compares **median ratings** between two related observations (same raters, two products), non-normal data. Best test: (a) Mann-Whitney U (b) Wilcoxon signed-rank (c) Independent t-test (d) Friedman1 m
  4. 4.`Score = 40 + 4·Hours_Studied − 1·Hours_Phone + 6·Hours_Studied×Hours_Phone`. The interaction coefficient (6) means: (a) Phone and study are uncorrelated (b) The slope of study hours on score changes by 6 per unit of phone hours (c) Phone use causes higher scores (d) Multicollinearity1 m
  5. 5.Rule of thumb for problematic multicollinearity is VIF > (a) 1 (b) 2 (c) 5–10 (d) 1001 m
  6. 6.Arnav's data violate homogeneity of variances between two groups. Best alternative to Student's t: (a) Mann-Whitney U (b) Welch's t-test (c) Paired t-test (d) Sign test1 m
  7. 7.A 2 × 3 mixed ANOVA produces how many distinct F-tests? (a) 2 (b) 3 (c) 5 (d) 61 m
  8. 8.Kiara has **3 related ordinal measurements** per participant. Best test: (a) Mann-Whitney (b) Kruskal-Wallis (c) Friedman (d) Wilcoxon1 m
  9. 9.Counts with **overdispersion** (variance > mean) are better modeled with: (a) Poisson regression (b) Negative binomial regression (c) Linear regression (d) Logistic regression1 m
  10. 10.Drishti's logistic regression of `purchase (1/0)` on `ad_exposure_count` gives β = 0.6. OR for one additional exposure: (a) 0.6 (b) e^0.6 ≈ 1.82 (c) ln(0.6) ≈ −0.51 (d) 1/0.6 ≈ 1.671 m
  11. 11.CFA differs from EFA in that: (a) CFA tests a pre-specified factor structure; EFA discovers structure (b) EFA is confirmatory (c) CFA always uses oblique rotation (d) EFA requires SEM software1 m
  12. 12.For a one-sample t-test at α = .05, two-tailed, power = 0.80, expected d = 0.5, approximate required n is: (a) 16 (b) 32 (c) 64 (d) 1001 m
  13. 13.The **sign test** uses: (a) Magnitude and direction of differences (b) Only direction (sign) of differences (c) Raw means (d) Pearson r1 m
  14. 14.Only mothers with high incomes participated in a survey on parenting; their children had better academic outcomes. The threat is: (a) Selection bias (b) Confounding (c) Both (a) and (b) (d) Maturation1 m
  15. 15.Adjusted R² is strictly less than R² when: (a) At least one predictor (b) The model perfectly fits (c) Sample size is very large (d) Predictors are uncorrelated1 m
  16. 16.BF₀₁ = 12 means: (a) Strong evidence for H₁ (b) Strong evidence for H₀: data 12× more likely under null (c) Weak evidence (d) Inconclusive1 m
  17. 17.Holm's step-down method is: (a) Type of factor analysis (b) Correction more powerful than Bonferroni while controlling FWER (c) Non-parametric test (d) Alternative to PCA1 m
  18. 18.Games-Howell post-hoc is used when: (a) Variances are equal (b) Variances are unequal across groups (c) Data are paired (d) Within-subjects design1 m
  19. 19.Approximate Cohen's d from a two-group t-test: (a) d = t × √(1/n₁ + 1/n₂) (b) d = t / √n (c) d² = t²/(t² + df) (d) d = p × n1 m
  20. 20.A researcher selects features then fits regression on the same data. The reported p-values are likely: (a) Correct (b) Too small — selection step not accounted for → inflated Type I (c) Too large (d) Unaffected1 m

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

30 marks
  1. 1.Define **construct validity** with one example of evidence.2 m
  2. 2.State the **independence assumption** and an example of violation.2 m
  3. 3.When use **Wilcoxon signed-rank** over **Mann-Whitney U**?2 m
  4. 4.Define **operationalization** with an example for "academic burnout."2 m
  5. 5.Differentiate **Maximum Likelihood (ML)** from **Principal Axis Factoring (PAF)** in factor analysis.2 m
  6. 6.Niharika studies whether **pre-class material type** (Video / Reading / Both) affects **quiz score (0–20)** at IIIT-H. Identify IV, DV, design, test.2 m
  7. 7.Define **history effects** with an example.2 m
  8. 8.Relationship between **effect size, sample size, α, and power**.2 m
  9. 9.When is the **mode** more informative than mean/median?2 m
  10. 10.How to detect autocorrelation in time-series regression residuals?2 m
  11. 11.Why is **effect size more informative than p-value** in many contexts?2 m
  12. 12.Define **sampling distribution** in one sentence.2 m
  13. 13.Distinguish **construct validity** from **criterion validity**.2 m
  14. 14.State two differences between **frequentist** and **Bayesian** inference.2 m
  15. 15.Define **sensitivity** and **specificity**.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.)