This thesis investigates the jackknife empirical likelihood (JEL) method for estimating confidence intervals for the difference between two correlated Kendall tau coefficients. Simulation studies showed that JEL outperformed Bootstrap, influence-based empirical likelihood (IEL), and normal approximation methods (NA). It also showed that JEL gave the coverage probabilities closest to the nominal level 0.95 and the shortest interval lengths, making it the most efficient. The Iris dataset from real life con- firmed the results of the simulations, which establishes the method’s reliability in practice. Although JEL performed well compared to other methods, further studies could focus on investigating the robust- ness of JEL associated with different correlation structures, missing data situations, or rank correlation measures. Furthermore, further computational optimization of JEL could enhance its utility with large datasets. The results clearly show that JEL has great potential as an inference tool for correlated nonparametric settings.