Behavioral Research: Statistical Methods
CG3.402Vinoo Alluri•Monsoon 2025-26•4 credits
PYQ-style paper · Paper D
Duration: 120 min • Max marks: 50
Section 1 — Objective Questions (15 marks)
0 marks- 1.Which is an example of **nominal-scale** data? (a) Heights in cm (b) Temperature in °C (c) Blood types (A, B, AB, O) (d) Test scores (0–100)
- 2.Aakash administers the same depression scale on day 1 and day 14 to the same patients and computes correlation across the two scores. He is assessing: (a) Inter-rater reliability (b) Test-retest reliability (c) Internal consistency (d) Convergent validity
- 3."P(rain | dark clouds)" reads as: (a) Joint probability of rain and dark clouds (b) Conditional probability of rain given dark clouds (c) Probability of dark clouds (d) Bayes Factor
- 4.Megha records reaction times across 200 trials. To summarise her data, the most informative single number for "typical reaction time" when data are skewed is: (a) Mean (b) Median (c) Mode (d) Standard deviation
- 5.A "violin plot" shows: (a) Pie chart of categories (b) The full density of a distribution along with summary stats (c) Linear regression (d) Heat map
- 6.A **two-tailed test** is appropriate when: (a) Direction of effect is pre-specified (b) Effect could meaningfully occur in either direction (c) Sample size is small (d) Variance is unknown
- 7.In experimental design, **counterbalancing** is used to control: (a) Sample size (b) Order effects in within-subjects designs (c) Multiple comparisons (d) External validity
- 8."Convergent validity" is established when: (a) A measure correlates with conceptually similar measures (b) A measure has high reliability (c) Two raters agree (d) The test predicts future performance
- 9.AIC penalizes: (a) Model complexity / number of parameters (b) Sample size (c) Effect size (d) p-value
- 10.Which value of Pearson's r indicates the strongest association? (a) 0.45 (b) −0.78 (c) 0.20 (d) −0.30
- 11.Nikita finds that her measure of "creativity" correlates very strongly (r = 0.85) with a measure of "intelligence." If she had hoped creativity would be **distinct** from intelligence, this is evidence of: (a) Good convergent validity (b) Poor discriminant validity (c) High reliability (d) Strong predictive validity
- 12.The probability of getting **exactly 3 heads in 5 tosses** of a fair coin is computed using: (a) Normal distribution (b) Binomial distribution with n = 5, p = 0.5 (c) Poisson distribution (d) Uniform distribution
- 13.Sahil computes a Pearson r = 0.62 between hours studied and exam score. He reports this as evidence that "studying causes higher scores." His main error is: (a) Computation error (b) Correlation ≠ causation; many possible confounders or reverse paths (c) r is too small to interpret (d) Sample size is unknown
- 14.A researcher claims her effect size is "trivial but significant" with p = 0.003 and n = 12,000. This is most likely because: (a) p-hacking (b) Large samples amplify even tiny true effects to significance (c) Wrong test used (d) Type II error
- 15.In the regression equation `Score = 30 + 5×(Hours_Studied) − 2×(Hours_Slept)`, the negative coefficient on `Hours_Slept` means: (a) Sleep causes lower scores (b) Holding hours studied constant, each additional hour of sleep is associated with 2 fewer points (controlling for study time, in this sample) (c) Sleep has no effect (d) The model is misspecified
- 16.Which test is the **nonparametric** counterpart of the **paired t-test**? (a) Mann-Whitney U (b) Wilcoxon Signed-Rank (c) Kruskal-Wallis (d) Friedman
- 17.Rohan reports a Bayes Factor BF₀₁ = 25, suggesting: (a) Strong evidence for H₁ (b) Strong evidence for H₀ (data ~25× more likely under null) (c) Inconclusive (d) Power was low
- 18.A clinical trial fails to find a significant effect (p = .25) with n = 28. The researcher claims "no effect of the treatment." This conclusion is: (a) Definitively correct (b) Premature — could be underpowered; absence of evidence ≠ evidence of absence (c) Significant at α = .10 (d) Always correct in clinical trials
- 19.A logistic regression yields a coefficient of 1.4 for `daily_exercise_hours`. The **odds ratio** for a one-unit increase is: (a) 1.4 (b) e^1.4 ≈ 4.06 (c) 0.4 (d) Cannot be computed
- 20.Mauchly's test in repeated-measures ANOVA tests: (a) Normality of the DV (b) Sphericity — equality of variances of differences across conditions (c) Independence (d) Linearity
Section 2 — Short Descriptive (15 marks)
0 marksSection 3 — Long Descriptive (20 marks)
0 marksTrack your attempt locally — score and time are recorded in your browser. (Coming soon: timed-attempt mode.)