In studies of efficacies of treatments or intervention modalities, Quality of Life (QOL) scales, multi-dimensional constructs, are routinely used as primary endpoints. The standard data analysis strategy computes overall and domain scores, and conducts a mixed-model analysis as if the scores were continuous. Three major problems arise with this approach: non-metric responses, violation of model assumptions and failure in comparison across treatment groups. Furthermore, data arising from randomized or observational studies involve complex designs. The problem is further complicated due to missing responses on QOL items or missing visits in pre/post intervention period. These missing values occur for various reasons and may have irregular pattern. So we propose a purely nonparametric approach in the sense that meaningful nonparametric effect size measures are developed. Our methods are shown to hold, particularly effective in the presence of some form of clustering and/or missing values. Inferential procedures for the effect size measures are derived from their asymptotic theory. Pediatric Asthma QOL data from a three-arm randomized trial will be used to illustrate the applications.