As data are 'numbers with context' (Cobb & Moore, 1997), contextual knowledge plays a prominent role in dealing with statistics. While insights about a specific context can further the depth of interpreting and evaluating outcomes of data analysis, research shows how it can also hinder relying on data especially if results differ from expectations. In this article, the aim is to investigate how young students informally deal with empirical evidence, which differs from their initial expectations in a specific context. We present a case study with three pairs of students at the age of 9 to 10 who compare groups in survey datasets. The interpretative analysis shows how conjectures of varying degrees of confidence shape the students' statistical expectations and can play different roles in interpreting results from data analysis.