Confidence ratings (CR) are often used to evaluate the metacognitive processes that occur during reasoning and problem solving. Typically CR are elicited with the assumption that they do not affect participants' underlying cognitive processes. However, recent evidence suggests that eliciting CR can cause changes in cognitive performance. What is not yet clear, are the metacognitive pathways by which CR affect overall performance in older individuals. In order to better understand the mechanisms driving reactivity to CR, we evaluated the impact of eliciting CR in an older sample (N = 89) on two aspects of the metacognitive framework - monitoring and control. Participants first rated their prospective confidence before performing the Latin Square Task either with or without confidence ratings. Participants subsequently self-appraised their performance. We found evidence that eliciting CR leads to "poorer" metacognitive monitoring. In addition, we found that participants with high initial prospective self-confidence who perform CR adopt a more immediate performance-orientated control strategy, which improves short-term performance but has no effect on overall performance in a timed Latin Square Task.