We compare precision relation between object-sensitivity and call-site-sensitivity in generalized k-limited context abstraction. In conventional k-limited context abstraction, where method calls maintain only the last k context elements, object-sensitivity and call-site-sensitivity complement each other. There exist counterexamples that k-call-site-sensitivity is less precise than 1-object-sensitivity and vice versa (e.g., k-object-sensitivity is less precise than 1-call-site-sensitivity). The examples, however, are no longer valid in the generalized k-limited context abstraction where method calls can keep arbitrary k context elements. For example, in the generalized k-limited policy, call-site-sensitivity can analyze the counterexamples precisely. Object-sensitivity, however, is still unable to precisely analyze the examples where conventional k-object-sensitivity is hard to handle. Moreover, in the generalized setting, we observed that k-call-site-sensitivity can simulate k-object-sensitivity. The simulated one usually produces equal or more precise context abstraction than the given object-sensitivity. In the poster, we summarize the findings and details about the precision relation between the two contexts.