Empirical likelihood and uniform convergence rates for dyadic kernel density estimation

This article studies the asymptotic properties of and alternative inference methods for kernel density estimation (KDE) dyadic data. We first establish uniform convergence rates KDE. Second, we propose a modified jackknife empirical likelihood procedure inference. The proposed test statistic is asymptotically pivotal regardless presence clustering. results are further extended to cover practically relevant case incomplete Simulations show that this likelihood-based delivers precise coverage probabilities even with modest sample sizes Finally, illustrate method by studying airport congestion in United States.