Empirical evidence has accrued suggesting that we are able to evaluate our own thoughts and actions by means of metacognitive judgements. We are interested in how these are formed and what evidence they are based on. It has often been assumed that decision time is a frugal cue for confidence judgements: the longer it takes us to form a decision, the less certain we are. It could be, however, that this association is just a by-product of the underlying mechanisms, one of which could be variability in the accumulation of evidence. In our experiment, participants had to judge whether the average colour of an array of eight coloured shapes was either red or blue. We critically manipulated the variability of information in this multi-element array. Our results suggest that for conditions with matched difficulty, variability had a significant influence on confidence with more variable arrays leading to less confident judgements.