I already talked about this issue previously, but I recently read a post about the same topic at InfoQ that brings more insights to the question:

Forces that tend to Shorten

• No Changes: The rule of no scope changes during the current sprint. This means the organization must be able to wait on average 1 1/2 sprints before asking for a change.
• Closure: The end of a sprint creates a good feeling, it’s a chance to celebrate the team’s accomplishments before starting all over again (Ilja Preuss).
• Feedback: This is the chance to reflect on the work completed and how the team performed. More frequent feedback means smaller course corrections each time. (Ilja Preuss)
• ROI: Every sprint provides an opportunity to deploy new features. (Ilja Preuss)
• Reliability of Commitment: With shorter sprints it’s easier to tell if the commitment can be meet. With longer sprints team the team is more likely to over-commit, thinking they should be able to get that story done. (Paul Oldfield).

Forces that tend to Lengthen

• Getting to “Done”: In some environments it can be technically challenging to get a story finished in a short sprint. (Ash Tengshe).