This study reports on a content course for science immersion teacher candidates that emphasized authentic practice of science and thinking scientifically in the context of introductory astrophysics. We explore how 122 science teacher candidates spanning three cohorts did and did not reason scientifically and how this evolved in our program. Our primary method of exploring teacher candidate reasoning is through analysis of responses to an apparently simple multiple-choice question. The question asks for the relative motion between the Sun and Earth and then for a scientific argument supporting the response. To explore these participants' reasoning and its potential impact on classroom practice, we also describe qualitative observations of how the teacher candidates were reasoning while participating in a science course, participant evaluations, and long-term follow-up with select program graduates after they had taught science in middle or high school. Our results suggest that participant ability to reason scientifically improved significantly and that this can impact classroom practice in a positive way. (Contains 5 tables and 2 figures.)