Is US Federal Law applied consistently? This is difficult to determine, owing to the unique circumstances of each case, compartmented appellate system, and diversity of judges. The legal principle of stare decisis encourages judges to rule according to prior decisions made by higher courts, and such decisions are typically cited in written opinions. But the large number of prior decisions requires judges to exercise discretion when choosing the most appropriate citations. This allows for selective citing pursuant to a legal agenda. We examine US Circuit Court of Appeals case citations of prior Supreme Court precedents. These citations comprise a temporal directed bipartite network.To identify unusual citations, we sample bipartite networks with respect to a null statistical model that quantifies apriori more likely citations based on well-known patterns (e.g. temporal similarity). We build off previous work on importance sampling methods to require the sampled networks match the empirical in and out degrees and respect temporal constraints. We use this to test for the effects of several covariates, including judge political alignment, case issue area, and US Circuit.