Literacy practices in science classrooms have been traditionally limited to the provision of macroscaffolds (writing templates like Question-Hypothesis-Methodology-Results). This paper explores the allowances and shortcomings of such practice by means of a systematic examination of a corpus of lab reports written by two small groups of college students taught to write scientifically through a macroscaffold-based approach. Despite reporting the same experience and being supported by the same macroscaffold, students' science writing differed in important ways. Group A's impersonal inferences expressed social detachment and objectivity (students positioned themselves as distant and objective knowledge producers), whereas Group B adopted a position of social closeness and subjectivity more typical of personal genres (e.g., personal diaries). Atypical of what is expected of science writers, Group B's personal inferences was taken as indicative of an alternative conception of what it meant to scientifically infer from one's empirical observations. Such a different style pointed to the possibility of some students holding alternative conceptions about what it means to scientifically infer from one's empirical observations. It is argued that, although macroscaffolding may be a helpful starting point, students need additional guidance on specific linguistic aspects of science writing, and possibly engage in genre-based literacy activities.