Much work is currently being undertaken to investigate the effectiveness of informal, "drop-in" mathematics support services. This can be a difficult task; students are individuals with different expectations and backgrounds, and support that is effective for one student, may be less effective in relation to another. This article investigates the effectiveness of the mathematics support service at Cardiff University by statistically analysing the relationship between student confidence and ability before and after attending a session. Some covariates, such as the effect of time and the student rating of the tutor, are also explored.