**Core Concept**
Stratified randomization is a method used in epidemiological studies to reduce confounding variables by ensuring that the distribution of confounders is similar across all groups being compared. This is crucial in observational studies where randomization is not feasible.

**Why the Correct Answer is Right**
Stratified randomization involves dividing the study population into subgroups (strata) based on relevant confounding variables and then randomly assigning participants within each stratum to the study groups. This approach helps to balance the distribution of confounders across the groups, reducing the likelihood of confounding bias. For instance, in a study comparing the effects of a new treatment for hypertension, stratified randomization might involve dividing the participants into strata based on age, sex, and baseline blood pressure. By randomly assigning participants within each stratum, the study can minimize the impact of these confounders on the results.

**Why Each Wrong Option is Incorrect**
**Option A:** Randomization alone is not sufficient to remove confounding, as it may not account for all relevant confounders.
**Option B:** Restriction involves excluding participants with certain characteristics, which can lead to selection bias and is not a reliable method for removing confounding.
**Option D:** Multivariate analysis can adjust for confounders in the analysis, but it does not prevent confounding from occurring in the first place; stratified randomization is more effective in reducing confounding.

**Clinical Pearl / High-Yield Fact**
When designing observational studies, consider using stratified randomization to minimize confounding bias. This can be particularly important in studies with multiple confounders, as it can help to ensure that the results are generalizable to the target population.

**✓ Correct Answer: C. Stratified randomization**