A strong GDP is great, but it is a metric mostly transparent to the average working American. To voters, employment and wage increases are much more tangible indicators of a strong economy.
We had a rising GDP during the '92 elections, but slow job growth. There was still lingering frustration among the voters about the recession they felt still was going on. Bill Clinton "felt their pain", and the incumbent was defeated.
Bush is fortunate. Last year at this time, the job market looked pathetic. However, if job growth unfolds the way it's looking like it's going to, come November, we'll have logged over a year's worth of significant positive job growth. The voters will notice.
.
Here's a link to a 1995 research paper that models voters as caring about two presidential abilities: the ability to make war and the ability to manage the economy. Unfortunately, the paper is only available by subscription, but the basic idea is this: To get reelected an incumbent President must convince voters that his combined abilities make him better than a challenger. Apparently this model has incredible predictive power.
You can understand why the Democrats are so desperate to paint the Iraq war as a failure, and why they have invented a bizarre metric called the "Misery Index" in a desperate attempt to describe a booming economy as anything but what it is.
If the Dems critisms lose credibility on either front defined in this model, the election is essentially over.