Dave Munger at Cognitive Daily found a fun study about procrastination, Do deadlines help procrastinators?
The Social Science Statistics blog (new to me, but it's been around for a while) has a good writeup of a 2002 study by Dan Ariely and Klaus Wertenbroch which systematically examines the effectiveness of deadlines in preventing procrastination.
The study asked students to proofread three articles over three weeks and paid them for each error discovered. The participants were split into three different deadline groups: one due every 7 days; all three due at the end of 3 weeks; self-imposed deadlines within the three weeks. In addition there was no bonus for early completion and a penalty for late delivery.
Participants who were given only the final deadline performed the worst both from an accuracy perspective and a deadline-meeting perspective. Those who had the evenly-spaced deadlines performed best, and those with self-imposed deadlines came out in the middle. I note that the evenly-spaced group averaged more errors detected than were inserted into the sample text, while the self-imposed group averaged on par with the number of purposeful errors. Since earnings are tied to both accuracy and delay, they show the impact of both factors together. The final-deadline group averaged less than 25% of the earnings of the evenly-spaced group. The self-imposed group averaged 60% of the evenly-spaced group.
This is a classic test of the Student Syndrome. Participants that had only the final deadline performed the worst because they waited to the last possible minute. They didn't get everything done (scope change), and they were quite late (time change).
The Student Syndrome comes about in projects because we always over-estimate the time required for an activity, and then we waste that over-estimate by not starting the activity until our best-case estimate. Then Murphy strikes, and we have no time to recover and hit the deadline.
Another factor in projects (and procrastination) is Parkinson's Law, which says that the work expands to fill the time available. This is in evidence with the evenly-spaced group: they found more errors on average than were inserted into the texts.
If I were king, the next study should look at the impact on rewards for early delivery. Even better, create the reward mechanism to reward early completion of the entire assignment, rather than for the individual assignments.