To support student construction of knowledge during class meetings, there is a growing demand to incorporate active learning practices. However, 'active learning' can range from occasionally voting on a multiple-choice question, to spending all of class time on worksheets. Consequently, the outcomes of active learning course reforms vary widely. Here we present an exploratory study comparing student outcomes from two introductory geology courses that both use active learning practices. In one course, students regularly respond to 'Clicker' questions with no peer interaction, using a personal response system. In the other course, students engage in Clicker questions as well as collaborative predict-observe-explain (POE) activities. Our results showed no statistically significant difference in content knowledge gains between these two courses POE activities. However, we find that students who experienced both Clickers and POE activities have enhanced self-efficacy and interest in the subject. As self-efficacy and science interest are integral for retention in STEM beyond first-year courses, we argue that some active learning strategies, such as POE, may be better suited to recruiting and retaining STEM majors. Thus we highlight the need for more studies that compare active learning strategies while monitoring student feelings and perceptions in addition to content knowledge.