Our science is limited and we should apply it to the best of our ability. The dubious methodology employed by Dr. Patterson undercuts many of the scientific claims she makes.
But in the fields of primatology (and anthropology) first hand accounts ("thick description" practiced in ethnology) are considered data though not quite in the sense of quantifiable, numerical data we are used to in engineering and computer science.
The people who have direct contact with Koko and provide descriptions of their communication with Koko are documenting their empirical experience, observations that provide detail and bias.
Taking another approach, communicating with Koko is basically an organism-to-organism example of a Turing test and, from what I understand, human evaluators of Koko's communication tend to conclude Koko was communicating intelligently.
Also, labeling emotional reactions, as they pertain to communication, "garbage" is probably a mistake. Emotions are quite important in human communication, and emotions are increasingly important, for example, in the fields of robotics and artificial intelligence. Dismissing emotions wholesale as "garbage" is probably a scientific error.
EDIT: Add last paragraph. Revise first sentence of last paragraph.