1. What is clustered data?

• ❌ Clustered data is when observations are the exact same.

• ❌ Data has low variance.

• ❌ One group is over-represented.

• ✅ Data is considered clustered when observations are correlated within groups, sometimes related to study designs.

Explanation:
Clustered (or grouped) data means subjects within the same group share similarities → intra-cluster correlation.

---

2. Which feature has the highest correlation between two observations in the same cluster?

• ❌ BPXSY1

• ✅ SDMVSTRA

• ❌ RIDAGEYR

• ❌ BMXBMI

• ❌ smq

Explanation:
SDMVSTRA is the stratum indicator used in NHANES sample design.
Within the same stratum, values are identical for all members, so correlation is effectively 1.0, the maximum possible.

---

3. What is true about multiple linear regression vs marginal linear models when dependence is present?

• ❌ Standard errors tend to be the same

• ❌ Multiple linear regression is theoretically justified with dependent data

• ✅ Marginal linear model estimates and SEs are meaningful when dependence is strictly within groups

• ❌ Standard error in multiple linear regression tends to be higher

Explanation:
MLR assumes independent observations. Marginal models (like GEE) correct SEs when there is intra-cluster dependence.

---

4. Multilevel models are expressed in terms of ________.

• ❌ Mixed effects

• ❌ Correlation coefficients

• ❌ Fixed effects

• ✅ Random effects

Explanation:
Multilevel (hierarchical) models are formally expressed through random effects, though they also include fixed effects.
The defining feature is random-effects structure.

---

5. Which is NOT true about reasons why we fit marginal models?

• ❌ Quicker computation

• ❌ Robust SEs for correlated data

• ❌ Easier accommodation of non-normal outcomes

• ✅ All the above are true

Explanation:
All statements listed are correct advantages of marginal models (GEEs), so the correct answer is: all are true.

---

🧾 Summary Table

---