In this paper, we summarise several components of our recent research into students' conceptions of statistics, their learning of statistics, our teaching of statistics, and their perceptions of their future professional work. We have obtained this information on the basis of phenomenographic analyses of several series of interviews with students studying statistics, both as statistics majors and as service students. In each of these cases, the broadest views relate in some way to personal connection, growth and change – in other words, they contain a strong ontological component above and beyond the standard epistemological component of learning. We discuss the importance of personal change in becoming a statistician – or an informed user of statistics – and investigate the pedagogical conditions under which such change is likely to occur. PRELUDE: LEARNING STATISTICS – KNOWLEDGE AND SKILLS Teaching and assessing statistical thinking at tertiary level is a very broad theme, and one, we believe, that can be addressed by looking at information from studies carried out both within the field of statistics education and also from beyond this particular field. In the broad endeavour of understanding and improving statistics pedagogy, different research questions and different research approaches shine the spotlight on different facets – for instance, important technical content, effective teaching techniques, valid and reliable assessment, characteristics of statistical thinking, anxiety about statistics, or conceptions of statistics and learning statistics. We would agree with Ramsden (1992, p.5) that "the aim of teaching is simple: it is to make student learning possible " and also with Barnett (2007, p.10) that: "a part – perhaps the main part – of teaching is that of nurturing in the student a will to learn." So we find ourselves focusing less on teaching and assessment, and more on learning itself, and particularly on students' views of learning. When we have discovered what makes it possible for a student to learn statistics and how to nurture in a student the 'will to learn' statistics, we have gone a long way towards answering any questions about teaching and even assessment of statistical thinking. In the last decades of the 20 th century, a sector-wide interest in improving the quality of learning and teaching in higher education resulted in research that was aimed specifically at identifying those features of learning that could be linked with improvement in learning and teaching. From the phenomenographic tradition, the description of surface and deep approaches to learning, originally identifie d in the context of reading a text, was already developed (Marton & Säljö, 1976), and the relation between this and conceptions of learning (Marton, Dall'Alba & Beaty, 1993) formed the research basis for Ramsden's (1992) book, Learning to Teach in Higher Education, and was summarised in Marton and Booth's (1997) Learning and Awareness. Academics set about investigating the variety of ways that students (and teachers) understood various topic areas in order to design better learning experiences. Some early examples were the mole concept in chemistry (Lybeck et al., 1988) and recursion in programming (Booth, 1992) : a more recent example is web-based information seeking (Edwards, 2006). This research was supported by studies that looked at students' (and teachers') experience of entire subject areas such as science (Prosser, Walker & Millar, 1995) and mathematics (Crawford et al., 1994): a more recent example is computer programming (Bruce et al., 2004). Some of these and other studies also investigated the ways that students understood learning in their discipline. Prosser and Trigwell based their text Understanding Learning and Teaching (1999) on empirical explorations of aspects of students' learning, incorporating Biggs' presage-process-product model of learning (for details see Biggs, 1999), to develop "a constitutionalist model of student learning" (Prosser & Trigwell, 1999, p.17) that indicated that a student's learning situation simultaneously included his or her prior experience, approaches to learning, perceptions of the situation and learning outcomes.